diff options
author | Daniel Baumann <daniel.baumann@progress-linux.org> | 2024-04-07 19:33:14 +0000 |
---|---|---|
committer | Daniel Baumann <daniel.baumann@progress-linux.org> | 2024-04-07 19:33:14 +0000 |
commit | 36d22d82aa202bb199967e9512281e9a53db42c9 (patch) | |
tree | 105e8c98ddea1c1e4784a60a5a6410fa416be2de /third_party/rust/http/src/header | |
parent | Initial commit. (diff) | |
download | firefox-esr-36d22d82aa202bb199967e9512281e9a53db42c9.tar.xz firefox-esr-36d22d82aa202bb199967e9512281e9a53db42c9.zip |
Adding upstream version 115.7.0esr.upstream/115.7.0esr
Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>
Diffstat (limited to 'third_party/rust/http/src/header')
-rw-r--r-- | third_party/rust/http/src/header/map.rs | 3496 | ||||
-rw-r--r-- | third_party/rust/http/src/header/mod.rs | 174 | ||||
-rw-r--r-- | third_party/rust/http/src/header/name.rs | 1868 | ||||
-rw-r--r-- | third_party/rust/http/src/header/value.rs | 795 |
4 files changed, 6333 insertions, 0 deletions
diff --git a/third_party/rust/http/src/header/map.rs b/third_party/rust/http/src/header/map.rs new file mode 100644 index 0000000000..13cbc3a3f9 --- /dev/null +++ b/third_party/rust/http/src/header/map.rs @@ -0,0 +1,3496 @@ +use std::collections::HashMap; +use std::collections::hash_map::RandomState; +use std::convert::TryFrom; +use std::hash::{BuildHasher, Hash, Hasher}; +use std::iter::{FromIterator, FusedIterator}; +use std::marker::PhantomData; +use std::{fmt, mem, ops, ptr, vec}; + +use crate::Error; + +use super::HeaderValue; +use super::name::{HdrName, HeaderName, InvalidHeaderName}; + +pub use self::as_header_name::AsHeaderName; +pub use self::into_header_name::IntoHeaderName; + +/// A set of HTTP headers +/// +/// `HeaderMap` is an multimap of [`HeaderName`] to values. +/// +/// [`HeaderName`]: struct.HeaderName.html +/// +/// # Examples +/// +/// Basic usage +/// +/// ``` +/// # use http::HeaderMap; +/// # use http::header::{CONTENT_LENGTH, HOST, LOCATION}; +/// let mut headers = HeaderMap::new(); +/// +/// headers.insert(HOST, "example.com".parse().unwrap()); +/// headers.insert(CONTENT_LENGTH, "123".parse().unwrap()); +/// +/// assert!(headers.contains_key(HOST)); +/// assert!(!headers.contains_key(LOCATION)); +/// +/// assert_eq!(headers[HOST], "example.com"); +/// +/// headers.remove(HOST); +/// +/// assert!(!headers.contains_key(HOST)); +/// ``` +#[derive(Clone)] +pub struct HeaderMap<T = HeaderValue> { + // Used to mask values to get an index + mask: Size, + indices: Box<[Pos]>, + entries: Vec<Bucket<T>>, + extra_values: Vec<ExtraValue<T>>, + danger: Danger, +} + +// # Implementation notes +// +// Below, you will find a fairly large amount of code. Most of this is to +// provide the necessary functions to efficiently manipulate the header +// multimap. The core hashing table is based on robin hood hashing [1]. While +// this is the same hashing algorithm used as part of Rust's `HashMap` in +// stdlib, many implementation details are different. The two primary reasons +// for this divergence are that `HeaderMap` is a multimap and the structure has +// been optimized to take advantage of the characteristics of HTTP headers. +// +// ## Structure Layout +// +// Most of the data contained by `HeaderMap` is *not* stored in the hash table. +// Instead, pairs of header name and *first* associated header value are stored +// in the `entries` vector. If the header name has more than one associated +// header value, then additional values are stored in `extra_values`. The actual +// hash table (`indices`) only maps hash codes to indices in `entries`. This +// means that, when an eviction happens, the actual header name and value stay +// put and only a tiny amount of memory has to be copied. +// +// Extra values associated with a header name are tracked using a linked list. +// Links are formed with offsets into `extra_values` and not pointers. +// +// [1]: https://en.wikipedia.org/wiki/Hash_table#Robin_Hood_hashing + +/// `HeaderMap` entry iterator. +/// +/// Yields `(&HeaderName, &value)` tuples. The same header name may be yielded +/// more than once if it has more than one associated value. +#[derive(Debug)] +pub struct Iter<'a, T> { + inner: IterMut<'a, T>, +} + +/// `HeaderMap` mutable entry iterator +/// +/// Yields `(&HeaderName, &mut value)` tuples. The same header name may be +/// yielded more than once if it has more than one associated value. +#[derive(Debug)] +pub struct IterMut<'a, T> { + map: *mut HeaderMap<T>, + entry: usize, + cursor: Option<Cursor>, + lt: PhantomData<&'a mut HeaderMap<T>>, +} + +/// An owning iterator over the entries of a `HeaderMap`. +/// +/// This struct is created by the `into_iter` method on `HeaderMap`. +#[derive(Debug)] +pub struct IntoIter<T> { + // If None, pull from `entries` + next: Option<usize>, + entries: vec::IntoIter<Bucket<T>>, + extra_values: Vec<ExtraValue<T>>, +} + +/// An iterator over `HeaderMap` keys. +/// +/// Each header name is yielded only once, even if it has more than one +/// associated value. +#[derive(Debug)] +pub struct Keys<'a, T> { + inner: ::std::slice::Iter<'a, Bucket<T>>, +} + +/// `HeaderMap` value iterator. +/// +/// Each value contained in the `HeaderMap` will be yielded. +#[derive(Debug)] +pub struct Values<'a, T> { + inner: Iter<'a, T>, +} + +/// `HeaderMap` mutable value iterator +#[derive(Debug)] +pub struct ValuesMut<'a, T> { + inner: IterMut<'a, T>, +} + +/// A drain iterator for `HeaderMap`. +#[derive(Debug)] +pub struct Drain<'a, T> { + idx: usize, + len: usize, + entries: *mut [Bucket<T>], + // If None, pull from `entries` + next: Option<usize>, + extra_values: *mut Vec<ExtraValue<T>>, + lt: PhantomData<&'a mut HeaderMap<T>>, +} + +/// A view to all values stored in a single entry. +/// +/// This struct is returned by `HeaderMap::get_all`. +#[derive(Debug)] +pub struct GetAll<'a, T> { + map: &'a HeaderMap<T>, + index: Option<usize>, +} + +/// A view into a single location in a `HeaderMap`, which may be vacant or occupied. +#[derive(Debug)] +pub enum Entry<'a, T: 'a> { + /// An occupied entry + Occupied(OccupiedEntry<'a, T>), + + /// A vacant entry + Vacant(VacantEntry<'a, T>), +} + +/// A view into a single empty location in a `HeaderMap`. +/// +/// This struct is returned as part of the `Entry` enum. +#[derive(Debug)] +pub struct VacantEntry<'a, T> { + map: &'a mut HeaderMap<T>, + key: HeaderName, + hash: HashValue, + probe: usize, + danger: bool, +} + +/// A view into a single occupied location in a `HeaderMap`. +/// +/// This struct is returned as part of the `Entry` enum. +#[derive(Debug)] +pub struct OccupiedEntry<'a, T> { + map: &'a mut HeaderMap<T>, + probe: usize, + index: usize, +} + +/// An iterator of all values associated with a single header name. +#[derive(Debug)] +pub struct ValueIter<'a, T> { + map: &'a HeaderMap<T>, + index: usize, + front: Option<Cursor>, + back: Option<Cursor>, +} + +/// A mutable iterator of all values associated with a single header name. +#[derive(Debug)] +pub struct ValueIterMut<'a, T> { + map: *mut HeaderMap<T>, + index: usize, + front: Option<Cursor>, + back: Option<Cursor>, + lt: PhantomData<&'a mut HeaderMap<T>>, +} + +/// An drain iterator of all values associated with a single header name. +#[derive(Debug)] +pub struct ValueDrain<'a, T> { + first: Option<T>, + next: Option<::std::vec::IntoIter<T>>, + lt: PhantomData<&'a mut HeaderMap<T>>, +} + +/// Tracks the value iterator state +#[derive(Debug, Copy, Clone, Eq, PartialEq)] +enum Cursor { + Head, + Values(usize), +} + +/// Type used for representing the size of a HeaderMap value. +/// +/// 32,768 is more than enough entries for a single header map. Setting this +/// limit enables using `u16` to represent all offsets, which takes 2 bytes +/// instead of 8 on 64 bit processors. +/// +/// Setting this limit is especially benificial for `indices`, making it more +/// cache friendly. More hash codes can fit in a cache line. +/// +/// You may notice that `u16` may represent more than 32,768 values. This is +/// true, but 32,768 should be plenty and it allows us to reserve the top bit +/// for future usage. +type Size = u16; + +/// This limit falls out from above. +const MAX_SIZE: usize = 1 << 15; + +/// An entry in the hash table. This represents the full hash code for an entry +/// as well as the position of the entry in the `entries` vector. +#[derive(Copy, Clone)] +struct Pos { + // Index in the `entries` vec + index: Size, + // Full hash value for the entry. + hash: HashValue, +} + +/// Hash values are limited to u16 as well. While `fast_hash` and `Hasher` +/// return `usize` hash codes, limiting the effective hash code to the lower 16 +/// bits is fine since we know that the `indices` vector will never grow beyond +/// that size. +#[derive(Debug, Copy, Clone, Eq, PartialEq)] +struct HashValue(u16); + +/// Stores the data associated with a `HeaderMap` entry. Only the first value is +/// included in this struct. If a header name has more than one associated +/// value, all extra values are stored in the `extra_values` vector. A doubly +/// linked list of entries is maintained. The doubly linked list is used so that +/// removing a value is constant time. This also has the nice property of +/// enabling double ended iteration. +#[derive(Debug, Clone)] +struct Bucket<T> { + hash: HashValue, + key: HeaderName, + value: T, + links: Option<Links>, +} + +/// The head and tail of the value linked list. +#[derive(Debug, Copy, Clone)] +struct Links { + next: usize, + tail: usize, +} + +/// Access to the `links` value in a slice of buckets. +/// +/// It's important that no other field is accessed, since it may have been +/// freed in a `Drain` iterator. +#[derive(Debug)] +struct RawLinks<T>(*mut [Bucket<T>]); + +/// Node in doubly-linked list of header value entries +#[derive(Debug, Clone)] +struct ExtraValue<T> { + value: T, + prev: Link, + next: Link, +} + +/// A header value node is either linked to another node in the `extra_values` +/// list or it points to an entry in `entries`. The entry in `entries` is the +/// start of the list and holds the associated header name. +#[derive(Debug, Copy, Clone, Eq, PartialEq)] +enum Link { + Entry(usize), + Extra(usize), +} + +/// Tracks the header map danger level! This relates to the adaptive hashing +/// algorithm. A HeaderMap starts in the "green" state, when a large number of +/// collisions are detected, it transitions to the yellow state. At this point, +/// the header map will either grow and switch back to the green state OR it +/// will transition to the red state. +/// +/// When in the red state, a safe hashing algorithm is used and all values in +/// the header map have to be rehashed. +#[derive(Clone)] +enum Danger { + Green, + Yellow, + Red(RandomState), +} + +// Constants related to detecting DOS attacks. +// +// Displacement is the number of entries that get shifted when inserting a new +// value. Forward shift is how far the entry gets stored from the ideal +// position. +// +// The current constant values were picked from another implementation. It could +// be that there are different values better suited to the header map case. +const DISPLACEMENT_THRESHOLD: usize = 128; +const FORWARD_SHIFT_THRESHOLD: usize = 512; + +// The default strategy for handling the yellow danger state is to increase the +// header map capacity in order to (hopefully) reduce the number of collisions. +// If growing the hash map would cause the load factor to drop bellow this +// threshold, then instead of growing, the headermap is switched to the red +// danger state and safe hashing is used instead. +const LOAD_FACTOR_THRESHOLD: f32 = 0.2; + +// Macro used to iterate the hash table starting at a given point, looping when +// the end is hit. +macro_rules! probe_loop { + ($label:tt: $probe_var: ident < $len: expr, $body: expr) => { + debug_assert!($len > 0); + $label: + loop { + if $probe_var < $len { + $body + $probe_var += 1; + } else { + $probe_var = 0; + } + } + }; + ($probe_var: ident < $len: expr, $body: expr) => { + debug_assert!($len > 0); + loop { + if $probe_var < $len { + $body + $probe_var += 1; + } else { + $probe_var = 0; + } + } + }; +} + +// First part of the robinhood algorithm. Given a key, find the slot in which it +// will be inserted. This is done by starting at the "ideal" spot. Then scanning +// until the destination slot is found. A destination slot is either the next +// empty slot or the next slot that is occupied by an entry that has a lower +// displacement (displacement is the distance from the ideal spot). +// +// This is implemented as a macro instead of a function that takes a closure in +// order to guarantee that it is "inlined". There is no way to annotate closures +// to guarantee inlining. +macro_rules! insert_phase_one { + ($map:ident, + $key:expr, + $probe:ident, + $pos:ident, + $hash:ident, + $danger:ident, + $vacant:expr, + $occupied:expr, + $robinhood:expr) => + {{ + let $hash = hash_elem_using(&$map.danger, &$key); + let mut $probe = desired_pos($map.mask, $hash); + let mut dist = 0; + let ret; + + // Start at the ideal position, checking all slots + probe_loop!('probe: $probe < $map.indices.len(), { + if let Some(($pos, entry_hash)) = $map.indices[$probe].resolve() { + // The slot is already occupied, but check if it has a lower + // displacement. + let their_dist = probe_distance($map.mask, entry_hash, $probe); + + if their_dist < dist { + // The new key's distance is larger, so claim this spot and + // displace the current entry. + // + // Check if this insertion is above the danger threshold. + let $danger = + dist >= FORWARD_SHIFT_THRESHOLD && !$map.danger.is_red(); + + ret = $robinhood; + break 'probe; + } else if entry_hash == $hash && $map.entries[$pos].key == $key { + // There already is an entry with the same key. + ret = $occupied; + break 'probe; + } + } else { + // The entry is vacant, use it for this key. + let $danger = + dist >= FORWARD_SHIFT_THRESHOLD && !$map.danger.is_red(); + + ret = $vacant; + break 'probe; + } + + dist += 1; + }); + + ret + }} +} + +// ===== impl HeaderMap ===== + +impl HeaderMap { + /// Create an empty `HeaderMap`. + /// + /// The map will be created without any capacity. This function will not + /// allocate. + /// + /// # Examples + /// + /// ``` + /// # use http::HeaderMap; + /// let map = HeaderMap::new(); + /// + /// assert!(map.is_empty()); + /// assert_eq!(0, map.capacity()); + /// ``` + pub fn new() -> Self { + HeaderMap::with_capacity(0) + } +} + +impl<T> HeaderMap<T> { + /// Create an empty `HeaderMap` with the specified capacity. + /// + /// The returned map will allocate internal storage in order to hold about + /// `capacity` elements without reallocating. However, this is a "best + /// effort" as there are usage patterns that could cause additional + /// allocations before `capacity` headers are stored in the map. + /// + /// More capacity than requested may be allocated. + /// + /// # Examples + /// + /// ``` + /// # use http::HeaderMap; + /// let map: HeaderMap<u32> = HeaderMap::with_capacity(10); + /// + /// assert!(map.is_empty()); + /// assert_eq!(12, map.capacity()); + /// ``` + pub fn with_capacity(capacity: usize) -> HeaderMap<T> { + if capacity == 0 { + HeaderMap { + mask: 0, + indices: Box::new([]), // as a ZST, this doesn't actually allocate anything + entries: Vec::new(), + extra_values: Vec::new(), + danger: Danger::Green, + } + } else { + let raw_cap = to_raw_capacity(capacity).next_power_of_two(); + assert!(raw_cap <= MAX_SIZE, "requested capacity too large"); + debug_assert!(raw_cap > 0); + + HeaderMap { + mask: (raw_cap - 1) as Size, + indices: vec![Pos::none(); raw_cap].into_boxed_slice(), + entries: Vec::with_capacity(raw_cap), + extra_values: Vec::new(), + danger: Danger::Green, + } + } + } + + /// Returns the number of headers stored in the map. + /// + /// This number represents the total number of **values** stored in the map. + /// This number can be greater than or equal to the number of **keys** + /// stored given that a single key may have more than one associated value. + /// + /// # Examples + /// + /// ``` + /// # use http::HeaderMap; + /// # use http::header::{ACCEPT, HOST}; + /// let mut map = HeaderMap::new(); + /// + /// assert_eq!(0, map.len()); + /// + /// map.insert(ACCEPT, "text/plain".parse().unwrap()); + /// map.insert(HOST, "localhost".parse().unwrap()); + /// + /// assert_eq!(2, map.len()); + /// + /// map.append(ACCEPT, "text/html".parse().unwrap()); + /// + /// assert_eq!(3, map.len()); + /// ``` + pub fn len(&self) -> usize { + self.entries.len() + self.extra_values.len() + } + + /// Returns the number of keys stored in the map. + /// + /// This number will be less than or equal to `len()` as each key may have + /// more than one associated value. + /// + /// # Examples + /// + /// ``` + /// # use http::HeaderMap; + /// # use http::header::{ACCEPT, HOST}; + /// let mut map = HeaderMap::new(); + /// + /// assert_eq!(0, map.keys_len()); + /// + /// map.insert(ACCEPT, "text/plain".parse().unwrap()); + /// map.insert(HOST, "localhost".parse().unwrap()); + /// + /// assert_eq!(2, map.keys_len()); + /// + /// map.insert(ACCEPT, "text/html".parse().unwrap()); + /// + /// assert_eq!(2, map.keys_len()); + /// ``` + pub fn keys_len(&self) -> usize { + self.entries.len() + } + + /// Returns true if the map contains no elements. + /// + /// # Examples + /// + /// ``` + /// # use http::HeaderMap; + /// # use http::header::HOST; + /// let mut map = HeaderMap::new(); + /// + /// assert!(map.is_empty()); + /// + /// map.insert(HOST, "hello.world".parse().unwrap()); + /// + /// assert!(!map.is_empty()); + /// ``` + pub fn is_empty(&self) -> bool { + self.entries.len() == 0 + } + + /// Clears the map, removing all key-value pairs. Keeps the allocated memory + /// for reuse. + /// + /// # Examples + /// + /// ``` + /// # use http::HeaderMap; + /// # use http::header::HOST; + /// let mut map = HeaderMap::new(); + /// map.insert(HOST, "hello.world".parse().unwrap()); + /// + /// map.clear(); + /// assert!(map.is_empty()); + /// assert!(map.capacity() > 0); + /// ``` + pub fn clear(&mut self) { + self.entries.clear(); + self.extra_values.clear(); + self.danger = Danger::Green; + + for e in self.indices.iter_mut() { + *e = Pos::none(); + } + } + + /// Returns the number of headers the map can hold without reallocating. + /// + /// This number is an approximation as certain usage patterns could cause + /// additional allocations before the returned capacity is filled. + /// + /// # Examples + /// + /// ``` + /// # use http::HeaderMap; + /// # use http::header::HOST; + /// let mut map = HeaderMap::new(); + /// + /// assert_eq!(0, map.capacity()); + /// + /// map.insert(HOST, "hello.world".parse().unwrap()); + /// assert_eq!(6, map.capacity()); + /// ``` + pub fn capacity(&self) -> usize { + usable_capacity(self.indices.len()) + } + + /// Reserves capacity for at least `additional` more headers to be inserted + /// into the `HeaderMap`. + /// + /// The header map may reserve more space to avoid frequent reallocations. + /// Like with `with_capacity`, this will be a "best effort" to avoid + /// allocations until `additional` more headers are inserted. Certain usage + /// patterns could cause additional allocations before the number is + /// reached. + /// + /// # Panics + /// + /// Panics if the new allocation size overflows `usize`. + /// + /// # Examples + /// + /// ``` + /// # use http::HeaderMap; + /// # use http::header::HOST; + /// let mut map = HeaderMap::new(); + /// map.reserve(10); + /// # map.insert(HOST, "bar".parse().unwrap()); + /// ``` + pub fn reserve(&mut self, additional: usize) { + // TODO: This can't overflow if done properly... since the max # of + // elements is u16::MAX. + let cap = self + .entries + .len() + .checked_add(additional) + .expect("reserve overflow"); + + if cap > self.indices.len() { + let cap = cap.next_power_of_two(); + assert!(cap <= MAX_SIZE, "header map reserve over max capacity"); + assert!(cap != 0, "header map reserve overflowed"); + + if self.entries.len() == 0 { + self.mask = cap as Size - 1; + self.indices = vec![Pos::none(); cap].into_boxed_slice(); + self.entries = Vec::with_capacity(usable_capacity(cap)); + } else { + self.grow(cap); + } + } + } + + /// Returns a reference to the value associated with the key. + /// + /// If there are multiple values associated with the key, then the first one + /// is returned. Use `get_all` to get all values associated with a given + /// key. Returns `None` if there are no values associated with the key. + /// + /// # Examples + /// + /// ``` + /// # use http::HeaderMap; + /// # use http::header::HOST; + /// let mut map = HeaderMap::new(); + /// assert!(map.get("host").is_none()); + /// + /// map.insert(HOST, "hello".parse().unwrap()); + /// assert_eq!(map.get(HOST).unwrap(), &"hello"); + /// assert_eq!(map.get("host").unwrap(), &"hello"); + /// + /// map.append(HOST, "world".parse().unwrap()); + /// assert_eq!(map.get("host").unwrap(), &"hello"); + /// ``` + pub fn get<K>(&self, key: K) -> Option<&T> + where + K: AsHeaderName, + { + self.get2(&key) + } + + fn get2<K>(&self, key: &K) -> Option<&T> + where + K: AsHeaderName, + { + match key.find(self) { + Some((_, found)) => { + let entry = &self.entries[found]; + Some(&entry.value) + } + None => None, + } + } + + /// Returns a mutable reference to the value associated with the key. + /// + /// If there are multiple values associated with the key, then the first one + /// is returned. Use `entry` to get all values associated with a given + /// key. Returns `None` if there are no values associated with the key. + /// + /// # Examples + /// + /// ``` + /// # use http::HeaderMap; + /// # use http::header::HOST; + /// let mut map = HeaderMap::default(); + /// map.insert(HOST, "hello".to_string()); + /// map.get_mut("host").unwrap().push_str("-world"); + /// + /// assert_eq!(map.get(HOST).unwrap(), &"hello-world"); + /// ``` + pub fn get_mut<K>(&mut self, key: K) -> Option<&mut T> + where + K: AsHeaderName, + { + match key.find(self) { + Some((_, found)) => { + let entry = &mut self.entries[found]; + Some(&mut entry.value) + } + None => None, + } + } + + /// Returns a view of all values associated with a key. + /// + /// The returned view does not incur any allocations and allows iterating + /// the values associated with the key. See [`GetAll`] for more details. + /// Returns `None` if there are no values associated with the key. + /// + /// [`GetAll`]: struct.GetAll.html + /// + /// # Examples + /// + /// ``` + /// # use http::HeaderMap; + /// # use http::header::HOST; + /// let mut map = HeaderMap::new(); + /// + /// map.insert(HOST, "hello".parse().unwrap()); + /// map.append(HOST, "goodbye".parse().unwrap()); + /// + /// let view = map.get_all("host"); + /// + /// let mut iter = view.iter(); + /// assert_eq!(&"hello", iter.next().unwrap()); + /// assert_eq!(&"goodbye", iter.next().unwrap()); + /// assert!(iter.next().is_none()); + /// ``` + pub fn get_all<K>(&self, key: K) -> GetAll<'_, T> + where + K: AsHeaderName, + { + GetAll { + map: self, + index: key.find(self).map(|(_, i)| i), + } + } + + /// Returns true if the map contains a value for the specified key. + /// + /// # Examples + /// + /// ``` + /// # use http::HeaderMap; + /// # use http::header::HOST; + /// let mut map = HeaderMap::new(); + /// assert!(!map.contains_key(HOST)); + /// + /// map.insert(HOST, "world".parse().unwrap()); + /// assert!(map.contains_key("host")); + /// ``` + pub fn contains_key<K>(&self, key: K) -> bool + where + K: AsHeaderName, + { + key.find(self).is_some() + } + + /// An iterator visiting all key-value pairs. + /// + /// The iteration order is arbitrary, but consistent across platforms for + /// the same crate version. Each key will be yielded once per associated + /// value. So, if a key has 3 associated values, it will be yielded 3 times. + /// + /// # Examples + /// + /// ``` + /// # use http::HeaderMap; + /// # use http::header::{CONTENT_LENGTH, HOST}; + /// let mut map = HeaderMap::new(); + /// + /// map.insert(HOST, "hello".parse().unwrap()); + /// map.append(HOST, "goodbye".parse().unwrap()); + /// map.insert(CONTENT_LENGTH, "123".parse().unwrap()); + /// + /// for (key, value) in map.iter() { + /// println!("{:?}: {:?}", key, value); + /// } + /// ``` + pub fn iter(&self) -> Iter<'_, T> { + Iter { + inner: IterMut { + map: self as *const _ as *mut _, + entry: 0, + cursor: self.entries.first().map(|_| Cursor::Head), + lt: PhantomData, + }, + } + } + + /// An iterator visiting all key-value pairs, with mutable value references. + /// + /// The iterator order is arbitrary, but consistent across platforms for the + /// same crate version. Each key will be yielded once per associated value, + /// so if a key has 3 associated values, it will be yielded 3 times. + /// + /// # Examples + /// + /// ``` + /// # use http::HeaderMap; + /// # use http::header::{CONTENT_LENGTH, HOST}; + /// let mut map = HeaderMap::default(); + /// + /// map.insert(HOST, "hello".to_string()); + /// map.append(HOST, "goodbye".to_string()); + /// map.insert(CONTENT_LENGTH, "123".to_string()); + /// + /// for (key, value) in map.iter_mut() { + /// value.push_str("-boop"); + /// } + /// ``` + pub fn iter_mut(&mut self) -> IterMut<'_, T> { + IterMut { + map: self as *mut _, + entry: 0, + cursor: self.entries.first().map(|_| Cursor::Head), + lt: PhantomData, + } + } + + /// An iterator visiting all keys. + /// + /// The iteration order is arbitrary, but consistent across platforms for + /// the same crate version. Each key will be yielded only once even if it + /// has multiple associated values. + /// + /// # Examples + /// + /// ``` + /// # use http::HeaderMap; + /// # use http::header::{CONTENT_LENGTH, HOST}; + /// let mut map = HeaderMap::new(); + /// + /// map.insert(HOST, "hello".parse().unwrap()); + /// map.append(HOST, "goodbye".parse().unwrap()); + /// map.insert(CONTENT_LENGTH, "123".parse().unwrap()); + /// + /// for key in map.keys() { + /// println!("{:?}", key); + /// } + /// ``` + pub fn keys(&self) -> Keys<'_, T> { + Keys { + inner: self.entries.iter(), + } + } + + /// An iterator visiting all values. + /// + /// The iteration order is arbitrary, but consistent across platforms for + /// the same crate version. + /// + /// # Examples + /// + /// ``` + /// # use http::HeaderMap; + /// # use http::header::{CONTENT_LENGTH, HOST}; + /// let mut map = HeaderMap::new(); + /// + /// map.insert(HOST, "hello".parse().unwrap()); + /// map.append(HOST, "goodbye".parse().unwrap()); + /// map.insert(CONTENT_LENGTH, "123".parse().unwrap()); + /// + /// for value in map.values() { + /// println!("{:?}", value); + /// } + /// ``` + pub fn values(&self) -> Values<'_, T> { + Values { inner: self.iter() } + } + + /// An iterator visiting all values mutably. + /// + /// The iteration order is arbitrary, but consistent across platforms for + /// the same crate version. + /// + /// # Examples + /// + /// ``` + /// # use http::HeaderMap; + /// # use http::header::{CONTENT_LENGTH, HOST}; + /// let mut map = HeaderMap::default(); + /// + /// map.insert(HOST, "hello".to_string()); + /// map.append(HOST, "goodbye".to_string()); + /// map.insert(CONTENT_LENGTH, "123".to_string()); + /// + /// for value in map.values_mut() { + /// value.push_str("-boop"); + /// } + /// ``` + pub fn values_mut(&mut self) -> ValuesMut<'_, T> { + ValuesMut { + inner: self.iter_mut(), + } + } + + /// Clears the map, returning all entries as an iterator. + /// + /// The internal memory is kept for reuse. + /// + /// For each yielded item that has `None` provided for the `HeaderName`, + /// then the associated header name is the same as that of the previously + /// yielded item. The first yielded item will have `HeaderName` set. + /// + /// # Examples + /// + /// ``` + /// # use http::HeaderMap; + /// # use http::header::{CONTENT_LENGTH, HOST}; + /// let mut map = HeaderMap::new(); + /// + /// map.insert(HOST, "hello".parse().unwrap()); + /// map.append(HOST, "goodbye".parse().unwrap()); + /// map.insert(CONTENT_LENGTH, "123".parse().unwrap()); + /// + /// let mut drain = map.drain(); + /// + /// + /// assert_eq!(drain.next(), Some((Some(HOST), "hello".parse().unwrap()))); + /// assert_eq!(drain.next(), Some((None, "goodbye".parse().unwrap()))); + /// + /// assert_eq!(drain.next(), Some((Some(CONTENT_LENGTH), "123".parse().unwrap()))); + /// + /// assert_eq!(drain.next(), None); + /// ``` + pub fn drain(&mut self) -> Drain<'_, T> { + for i in self.indices.iter_mut() { + *i = Pos::none(); + } + + // Memory safety + // + // When the Drain is first created, it shortens the length of + // the source vector to make sure no uninitialized or moved-from + // elements are accessible at all if the Drain's destructor never + // gets to run. + + let entries = &mut self.entries[..] as *mut _; + let extra_values = &mut self.extra_values as *mut _; + let len = self.entries.len(); + unsafe { self.entries.set_len(0); } + + Drain { + idx: 0, + len, + entries, + extra_values, + next: None, + lt: PhantomData, + } + } + + fn value_iter(&self, idx: Option<usize>) -> ValueIter<'_, T> { + use self::Cursor::*; + + if let Some(idx) = idx { + let back = { + let entry = &self.entries[idx]; + + entry.links.map(|l| Values(l.tail)).unwrap_or(Head) + }; + + ValueIter { + map: self, + index: idx, + front: Some(Head), + back: Some(back), + } + } else { + ValueIter { + map: self, + index: ::std::usize::MAX, + front: None, + back: None, + } + } + } + + fn value_iter_mut(&mut self, idx: usize) -> ValueIterMut<'_, T> { + use self::Cursor::*; + + let back = { + let entry = &self.entries[idx]; + + entry.links.map(|l| Values(l.tail)).unwrap_or(Head) + }; + + ValueIterMut { + map: self as *mut _, + index: idx, + front: Some(Head), + back: Some(back), + lt: PhantomData, + } + } + + /// Gets the given key's corresponding entry in the map for in-place + /// manipulation. + /// + /// # Examples + /// + /// ``` + /// # use http::HeaderMap; + /// let mut map: HeaderMap<u32> = HeaderMap::default(); + /// + /// let headers = &[ + /// "content-length", + /// "x-hello", + /// "Content-Length", + /// "x-world", + /// ]; + /// + /// for &header in headers { + /// let counter = map.entry(header).or_insert(0); + /// *counter += 1; + /// } + /// + /// assert_eq!(map["content-length"], 2); + /// assert_eq!(map["x-hello"], 1); + /// ``` + pub fn entry<K>(&mut self, key: K) -> Entry<'_, T> + where + K: IntoHeaderName, + { + key.entry(self) + } + + /// Gets the given key's corresponding entry in the map for in-place + /// manipulation. + /// + /// # Errors + /// + /// This method differs from `entry` by allowing types that may not be + /// valid `HeaderName`s to passed as the key (such as `String`). If they + /// do not parse as a valid `HeaderName`, this returns an + /// `InvalidHeaderName` error. + pub fn try_entry<K>(&mut self, key: K) -> Result<Entry<'_, T>, InvalidHeaderName> + where + K: AsHeaderName, + { + key.try_entry(self) + } + + fn entry2<K>(&mut self, key: K) -> Entry<'_, T> + where + K: Hash + Into<HeaderName>, + HeaderName: PartialEq<K>, + { + // Ensure that there is space in the map + self.reserve_one(); + + insert_phase_one!( + self, + key, + probe, + pos, + hash, + danger, + Entry::Vacant(VacantEntry { + map: self, + hash: hash, + key: key.into(), + probe: probe, + danger: danger, + }), + Entry::Occupied(OccupiedEntry { + map: self, + index: pos, + probe: probe, + }), + Entry::Vacant(VacantEntry { + map: self, + hash: hash, + key: key.into(), + probe: probe, + danger: danger, + }) + ) + } + + /// Inserts a key-value pair into the map. + /// + /// If the map did not previously have this key present, then `None` is + /// returned. + /// + /// If the map did have this key present, the new value is associated with + /// the key and all previous values are removed. **Note** that only a single + /// one of the previous values is returned. If there are multiple values + /// that have been previously associated with the key, then the first one is + /// returned. See `insert_mult` on `OccupiedEntry` for an API that returns + /// all values. + /// + /// The key is not updated, though; this matters for types that can be `==` + /// without being identical. + /// + /// # Examples + /// + /// ``` + /// # use http::HeaderMap; + /// # use http::header::HOST; + /// let mut map = HeaderMap::new(); + /// assert!(map.insert(HOST, "world".parse().unwrap()).is_none()); + /// assert!(!map.is_empty()); + /// + /// let mut prev = map.insert(HOST, "earth".parse().unwrap()).unwrap(); + /// assert_eq!("world", prev); + /// ``` + pub fn insert<K>(&mut self, key: K, val: T) -> Option<T> + where + K: IntoHeaderName, + { + key.insert(self, val) + } + + #[inline] + fn insert2<K>(&mut self, key: K, value: T) -> Option<T> + where + K: Hash + Into<HeaderName>, + HeaderName: PartialEq<K>, + { + self.reserve_one(); + + insert_phase_one!( + self, + key, + probe, + pos, + hash, + danger, + // Vacant + { + drop(danger); // Make lint happy + let index = self.entries.len(); + self.insert_entry(hash, key.into(), value); + self.indices[probe] = Pos::new(index, hash); + None + }, + // Occupied + Some(self.insert_occupied(pos, value)), + // Robinhood + { + self.insert_phase_two(key.into(), value, hash, probe, danger); + None + } + ) + } + + /// Set an occupied bucket to the given value + #[inline] + fn insert_occupied(&mut self, index: usize, value: T) -> T { + if let Some(links) = self.entries[index].links { + self.remove_all_extra_values(links.next); + } + + let entry = &mut self.entries[index]; + mem::replace(&mut entry.value, value) + } + + fn insert_occupied_mult(&mut self, index: usize, value: T) -> ValueDrain<'_, T> { + let old; + let links; + + { + let entry = &mut self.entries[index]; + + old = mem::replace(&mut entry.value, value); + links = entry.links.take(); + } + + let raw_links = self.raw_links(); + let extra_values = &mut self.extra_values; + + let next = links.map(|l| { + drain_all_extra_values(raw_links, extra_values, l.next) + .into_iter() + }); + + ValueDrain { + first: Some(old), + next: next, + lt: PhantomData, + } + } + + /// Inserts a key-value pair into the map. + /// + /// If the map did not previously have this key present, then `false` is + /// returned. + /// + /// If the map did have this key present, the new value is pushed to the end + /// of the list of values currently associated with the key. The key is not + /// updated, though; this matters for types that can be `==` without being + /// identical. + /// + /// # Examples + /// + /// ``` + /// # use http::HeaderMap; + /// # use http::header::HOST; + /// let mut map = HeaderMap::new(); + /// assert!(map.insert(HOST, "world".parse().unwrap()).is_none()); + /// assert!(!map.is_empty()); + /// + /// map.append(HOST, "earth".parse().unwrap()); + /// + /// let values = map.get_all("host"); + /// let mut i = values.iter(); + /// assert_eq!("world", *i.next().unwrap()); + /// assert_eq!("earth", *i.next().unwrap()); + /// ``` + pub fn append<K>(&mut self, key: K, value: T) -> bool + where + K: IntoHeaderName, + { + key.append(self, value) + } + + #[inline] + fn append2<K>(&mut self, key: K, value: T) -> bool + where + K: Hash + Into<HeaderName>, + HeaderName: PartialEq<K>, + { + self.reserve_one(); + + insert_phase_one!( + self, + key, + probe, + pos, + hash, + danger, + // Vacant + { + drop(danger); + let index = self.entries.len(); + self.insert_entry(hash, key.into(), value); + self.indices[probe] = Pos::new(index, hash); + false + }, + // Occupied + { + append_value(pos, &mut self.entries[pos], &mut self.extra_values, value); + true + }, + // Robinhood + { + self.insert_phase_two(key.into(), value, hash, probe, danger); + + false + } + ) + } + + #[inline] + fn find<K: ?Sized>(&self, key: &K) -> Option<(usize, usize)> + where + K: Hash + Into<HeaderName>, + HeaderName: PartialEq<K>, + { + if self.entries.is_empty() { + return None; + } + + let hash = hash_elem_using(&self.danger, key); + let mask = self.mask; + let mut probe = desired_pos(mask, hash); + let mut dist = 0; + + probe_loop!(probe < self.indices.len(), { + if let Some((i, entry_hash)) = self.indices[probe].resolve() { + if dist > probe_distance(mask, entry_hash, probe) { + // give up when probe distance is too long + return None; + } else if entry_hash == hash && self.entries[i].key == *key { + return Some((probe, i)); + } + } else { + return None; + } + + dist += 1; + }); + } + + /// phase 2 is post-insert where we forward-shift `Pos` in the indices. + #[inline] + fn insert_phase_two( + &mut self, + key: HeaderName, + value: T, + hash: HashValue, + probe: usize, + danger: bool, + ) -> usize { + // Push the value and get the index + let index = self.entries.len(); + self.insert_entry(hash, key, value); + + let num_displaced = do_insert_phase_two(&mut self.indices, probe, Pos::new(index, hash)); + + if danger || num_displaced >= DISPLACEMENT_THRESHOLD { + // Increase danger level + self.danger.to_yellow(); + } + + index + } + + /// Removes a key from the map, returning the value associated with the key. + /// + /// Returns `None` if the map does not contain the key. If there are + /// multiple values associated with the key, then the first one is returned. + /// See `remove_entry_mult` on `OccupiedEntry` for an API that yields all + /// values. + /// + /// # Examples + /// + /// ``` + /// # use http::HeaderMap; + /// # use http::header::HOST; + /// let mut map = HeaderMap::new(); + /// map.insert(HOST, "hello.world".parse().unwrap()); + /// + /// let prev = map.remove(HOST).unwrap(); + /// assert_eq!("hello.world", prev); + /// + /// assert!(map.remove(HOST).is_none()); + /// ``` + pub fn remove<K>(&mut self, key: K) -> Option<T> + where + K: AsHeaderName, + { + match key.find(self) { + Some((probe, idx)) => { + if let Some(links) = self.entries[idx].links { + self.remove_all_extra_values(links.next); + } + + let entry = self.remove_found(probe, idx); + + Some(entry.value) + } + None => None, + } + } + + /// Remove an entry from the map. + /// + /// Warning: To avoid inconsistent state, extra values _must_ be removed + /// for the `found` index (via `remove_all_extra_values` or similar) + /// _before_ this method is called. + #[inline] + fn remove_found(&mut self, probe: usize, found: usize) -> Bucket<T> { + // index `probe` and entry `found` is to be removed + // use swap_remove, but then we need to update the index that points + // to the other entry that has to move + self.indices[probe] = Pos::none(); + let entry = self.entries.swap_remove(found); + + // correct index that points to the entry that had to swap places + if let Some(entry) = self.entries.get(found) { + // was not last element + // examine new element in `found` and find it in indices + let mut probe = desired_pos(self.mask, entry.hash); + + probe_loop!(probe < self.indices.len(), { + if let Some((i, _)) = self.indices[probe].resolve() { + if i >= self.entries.len() { + // found it + self.indices[probe] = Pos::new(found, entry.hash); + break; + } + } + }); + + // Update links + if let Some(links) = entry.links { + self.extra_values[links.next].prev = Link::Entry(found); + self.extra_values[links.tail].next = Link::Entry(found); + } + } + + // backward shift deletion in self.indices + // after probe, shift all non-ideally placed indices backward + if self.entries.len() > 0 { + let mut last_probe = probe; + let mut probe = probe + 1; + + probe_loop!(probe < self.indices.len(), { + if let Some((_, entry_hash)) = self.indices[probe].resolve() { + if probe_distance(self.mask, entry_hash, probe) > 0 { + self.indices[last_probe] = self.indices[probe]; + self.indices[probe] = Pos::none(); + } else { + break; + } + } else { + break; + } + + last_probe = probe; + }); + } + + entry + } + + /// Removes the `ExtraValue` at the given index. + #[inline] + fn remove_extra_value(&mut self, idx: usize) -> ExtraValue<T> { + let raw_links = self.raw_links(); + remove_extra_value(raw_links, &mut self.extra_values, idx) + } + + fn remove_all_extra_values(&mut self, mut head: usize) { + loop { + let extra = self.remove_extra_value(head); + + if let Link::Extra(idx) = extra.next { + head = idx; + } else { + break; + } + } + } + + #[inline] + fn insert_entry(&mut self, hash: HashValue, key: HeaderName, value: T) { + assert!(self.entries.len() < MAX_SIZE, "header map at capacity"); + + self.entries.push(Bucket { + hash: hash, + key: key, + value: value, + links: None, + }); + } + + fn rebuild(&mut self) { + // Loop over all entries and re-insert them into the map + 'outer: for (index, entry) in self.entries.iter_mut().enumerate() { + let hash = hash_elem_using(&self.danger, &entry.key); + let mut probe = desired_pos(self.mask, hash); + let mut dist = 0; + + // Update the entry's hash code + entry.hash = hash; + + probe_loop!(probe < self.indices.len(), { + if let Some((_, entry_hash)) = self.indices[probe].resolve() { + // if existing element probed less than us, swap + let their_dist = probe_distance(self.mask, entry_hash, probe); + + if their_dist < dist { + // Robinhood + break; + } + } else { + // Vacant slot + self.indices[probe] = Pos::new(index, hash); + continue 'outer; + } + + dist += 1; + }); + + do_insert_phase_two(&mut self.indices, probe, Pos::new(index, hash)); + } + } + + fn reinsert_entry_in_order(&mut self, pos: Pos) { + if let Some((_, entry_hash)) = pos.resolve() { + // Find first empty bucket and insert there + let mut probe = desired_pos(self.mask, entry_hash); + + probe_loop!(probe < self.indices.len(), { + if self.indices[probe].resolve().is_none() { + // empty bucket, insert here + self.indices[probe] = pos; + return; + } + }); + } + } + + fn reserve_one(&mut self) { + let len = self.entries.len(); + + if self.danger.is_yellow() { + let load_factor = self.entries.len() as f32 / self.indices.len() as f32; + + if load_factor >= LOAD_FACTOR_THRESHOLD { + // Transition back to green danger level + self.danger.to_green(); + + // Double the capacity + let new_cap = self.indices.len() * 2; + + // Grow the capacity + self.grow(new_cap); + } else { + self.danger.to_red(); + + // Rebuild hash table + for index in self.indices.iter_mut() { + *index = Pos::none(); + } + + self.rebuild(); + } + } else if len == self.capacity() { + if len == 0 { + let new_raw_cap = 8; + self.mask = 8 - 1; + self.indices = vec![Pos::none(); new_raw_cap].into_boxed_slice(); + self.entries = Vec::with_capacity(usable_capacity(new_raw_cap)); + } else { + let raw_cap = self.indices.len(); + self.grow(raw_cap << 1); + } + } + } + + #[inline] + fn grow(&mut self, new_raw_cap: usize) { + assert!(new_raw_cap <= MAX_SIZE, "requested capacity too large"); + // This path can never be reached when handling the first allocation in + // the map. + + // find first ideally placed element -- start of cluster + let mut first_ideal = 0; + + for (i, pos) in self.indices.iter().enumerate() { + if let Some((_, entry_hash)) = pos.resolve() { + if 0 == probe_distance(self.mask, entry_hash, i) { + first_ideal = i; + break; + } + } + } + + // visit the entries in an order where we can simply reinsert them + // into self.indices without any bucket stealing. + let old_indices = mem::replace( + &mut self.indices, + vec![Pos::none(); new_raw_cap].into_boxed_slice(), + ); + self.mask = new_raw_cap.wrapping_sub(1) as Size; + + for &pos in &old_indices[first_ideal..] { + self.reinsert_entry_in_order(pos); + } + + for &pos in &old_indices[..first_ideal] { + self.reinsert_entry_in_order(pos); + } + + // Reserve additional entry slots + let more = self.capacity() - self.entries.len(); + self.entries.reserve_exact(more); + } + + #[inline] + fn raw_links(&mut self) -> RawLinks<T> { + RawLinks(&mut self.entries[..] as *mut _) + } +} + +/// Removes the `ExtraValue` at the given index. +#[inline] +fn remove_extra_value<T>( + mut raw_links: RawLinks<T>, + extra_values: &mut Vec<ExtraValue<T>>, + idx: usize) + -> ExtraValue<T> +{ + let prev; + let next; + + { + debug_assert!(extra_values.len() > idx); + let extra = &extra_values[idx]; + prev = extra.prev; + next = extra.next; + } + + // First unlink the extra value + match (prev, next) { + (Link::Entry(prev), Link::Entry(next)) => { + debug_assert_eq!(prev, next); + + raw_links[prev] = None; + } + (Link::Entry(prev), Link::Extra(next)) => { + debug_assert!(raw_links[prev].is_some()); + + raw_links[prev].as_mut().unwrap() + .next = next; + + debug_assert!(extra_values.len() > next); + extra_values[next].prev = Link::Entry(prev); + } + (Link::Extra(prev), Link::Entry(next)) => { + debug_assert!(raw_links[next].is_some()); + + raw_links[next].as_mut().unwrap() + .tail = prev; + + debug_assert!(extra_values.len() > prev); + extra_values[prev].next = Link::Entry(next); + } + (Link::Extra(prev), Link::Extra(next)) => { + debug_assert!(extra_values.len() > next); + debug_assert!(extra_values.len() > prev); + + extra_values[prev].next = Link::Extra(next); + extra_values[next].prev = Link::Extra(prev); + } + } + + // Remove the extra value + let mut extra = extra_values.swap_remove(idx); + + // This is the index of the value that was moved (possibly `extra`) + let old_idx = extra_values.len(); + + // Update the links + if extra.prev == Link::Extra(old_idx) { + extra.prev = Link::Extra(idx); + } + + if extra.next == Link::Extra(old_idx) { + extra.next = Link::Extra(idx); + } + + // Check if another entry was displaced. If it was, then the links + // need to be fixed. + if idx != old_idx { + let next; + let prev; + + { + debug_assert!(extra_values.len() > idx); + let moved = &extra_values[idx]; + next = moved.next; + prev = moved.prev; + } + + // An entry was moved, we have to the links + match prev { + Link::Entry(entry_idx) => { + // It is critical that we do not attempt to read the + // header name or value as that memory may have been + // "released" already. + debug_assert!(raw_links[entry_idx].is_some()); + + let links = raw_links[entry_idx].as_mut().unwrap(); + links.next = idx; + } + Link::Extra(extra_idx) => { + debug_assert!(extra_values.len() > extra_idx); + extra_values[extra_idx].next = Link::Extra(idx); + } + } + + match next { + Link::Entry(entry_idx) => { + debug_assert!(raw_links[entry_idx].is_some()); + + let links = raw_links[entry_idx].as_mut().unwrap(); + links.tail = idx; + } + Link::Extra(extra_idx) => { + debug_assert!(extra_values.len() > extra_idx); + extra_values[extra_idx].prev = Link::Extra(idx); + } + } + } + + debug_assert!({ + for v in &*extra_values { + assert!(v.next != Link::Extra(old_idx)); + assert!(v.prev != Link::Extra(old_idx)); + } + + true + }); + + extra +} + +fn drain_all_extra_values<T>( + raw_links: RawLinks<T>, + extra_values: &mut Vec<ExtraValue<T>>, + mut head: usize) + -> Vec<T> +{ + let mut vec = Vec::new(); + loop { + let extra = remove_extra_value(raw_links, extra_values, head); + vec.push(extra.value); + + if let Link::Extra(idx) = extra.next { + head = idx; + } else { + break; + } + } + vec +} + +impl<'a, T> IntoIterator for &'a HeaderMap<T> { + type Item = (&'a HeaderName, &'a T); + type IntoIter = Iter<'a, T>; + + fn into_iter(self) -> Iter<'a, T> { + self.iter() + } +} + +impl<'a, T> IntoIterator for &'a mut HeaderMap<T> { + type Item = (&'a HeaderName, &'a mut T); + type IntoIter = IterMut<'a, T>; + + fn into_iter(self) -> IterMut<'a, T> { + self.iter_mut() + } +} + +impl<T> IntoIterator for HeaderMap<T> { + type Item = (Option<HeaderName>, T); + type IntoIter = IntoIter<T>; + + /// Creates a consuming iterator, that is, one that moves keys and values + /// out of the map in arbitrary order. The map cannot be used after calling + /// this. + /// + /// For each yielded item that has `None` provided for the `HeaderName`, + /// then the associated header name is the same as that of the previously + /// yielded item. The first yielded item will have `HeaderName` set. + /// + /// # Examples + /// + /// Basic usage. + /// + /// ``` + /// # use http::header; + /// # use http::header::*; + /// let mut map = HeaderMap::new(); + /// map.insert(header::CONTENT_LENGTH, "123".parse().unwrap()); + /// map.insert(header::CONTENT_TYPE, "json".parse().unwrap()); + /// + /// let mut iter = map.into_iter(); + /// assert_eq!(iter.next(), Some((Some(header::CONTENT_LENGTH), "123".parse().unwrap()))); + /// assert_eq!(iter.next(), Some((Some(header::CONTENT_TYPE), "json".parse().unwrap()))); + /// assert!(iter.next().is_none()); + /// ``` + /// + /// Multiple values per key. + /// + /// ``` + /// # use http::header; + /// # use http::header::*; + /// let mut map = HeaderMap::new(); + /// + /// map.append(header::CONTENT_LENGTH, "123".parse().unwrap()); + /// map.append(header::CONTENT_LENGTH, "456".parse().unwrap()); + /// + /// map.append(header::CONTENT_TYPE, "json".parse().unwrap()); + /// map.append(header::CONTENT_TYPE, "html".parse().unwrap()); + /// map.append(header::CONTENT_TYPE, "xml".parse().unwrap()); + /// + /// let mut iter = map.into_iter(); + /// + /// assert_eq!(iter.next(), Some((Some(header::CONTENT_LENGTH), "123".parse().unwrap()))); + /// assert_eq!(iter.next(), Some((None, "456".parse().unwrap()))); + /// + /// assert_eq!(iter.next(), Some((Some(header::CONTENT_TYPE), "json".parse().unwrap()))); + /// assert_eq!(iter.next(), Some((None, "html".parse().unwrap()))); + /// assert_eq!(iter.next(), Some((None, "xml".parse().unwrap()))); + /// assert!(iter.next().is_none()); + /// ``` + fn into_iter(self) -> IntoIter<T> { + IntoIter { + next: None, + entries: self.entries.into_iter(), + extra_values: self.extra_values, + } + } +} + +impl<T> FromIterator<(HeaderName, T)> for HeaderMap<T> { + fn from_iter<I>(iter: I) -> Self + where + I: IntoIterator<Item = (HeaderName, T)>, + { + let mut map = HeaderMap::default(); + map.extend(iter); + map + } +} + +/// Try to convert a `HashMap` into a `HeaderMap`. +/// +/// # Examples +/// +/// ``` +/// use std::collections::HashMap; +/// use std::convert::TryInto; +/// use http::HeaderMap; +/// +/// let mut map = HashMap::new(); +/// map.insert("X-Custom-Header".to_string(), "my value".to_string()); +/// +/// let headers: HeaderMap = (&map).try_into().expect("valid headers"); +/// assert_eq!(headers["X-Custom-Header"], "my value"); +/// ``` +impl<'a, K, V, T> TryFrom<&'a HashMap<K, V>> for HeaderMap<T> + where + K: Eq + Hash, + HeaderName: TryFrom<&'a K>, + <HeaderName as TryFrom<&'a K>>::Error: Into<crate::Error>, + T: TryFrom<&'a V>, + T::Error: Into<crate::Error>, +{ + type Error = Error; + + fn try_from(c: &'a HashMap<K, V>) -> Result<Self, Self::Error> { + c.into_iter() + .map(|(k, v)| -> crate::Result<(HeaderName, T)> { + let name = TryFrom::try_from(k).map_err(Into::into)?; + let value = TryFrom::try_from(v).map_err(Into::into)?; + Ok((name, value)) + }) + .collect() + } +} + +impl<T> Extend<(Option<HeaderName>, T)> for HeaderMap<T> { + /// Extend a `HeaderMap` with the contents of another `HeaderMap`. + /// + /// This function expects the yielded items to follow the same structure as + /// `IntoIter`. + /// + /// # Panics + /// + /// This panics if the first yielded item does not have a `HeaderName`. + /// + /// # Examples + /// + /// ``` + /// # use http::header::*; + /// let mut map = HeaderMap::new(); + /// + /// map.insert(ACCEPT, "text/plain".parse().unwrap()); + /// map.insert(HOST, "hello.world".parse().unwrap()); + /// + /// let mut extra = HeaderMap::new(); + /// + /// extra.insert(HOST, "foo.bar".parse().unwrap()); + /// extra.insert(COOKIE, "hello".parse().unwrap()); + /// extra.append(COOKIE, "world".parse().unwrap()); + /// + /// map.extend(extra); + /// + /// assert_eq!(map["host"], "foo.bar"); + /// assert_eq!(map["accept"], "text/plain"); + /// assert_eq!(map["cookie"], "hello"); + /// + /// let v = map.get_all("host"); + /// assert_eq!(1, v.iter().count()); + /// + /// let v = map.get_all("cookie"); + /// assert_eq!(2, v.iter().count()); + /// ``` + fn extend<I: IntoIterator<Item = (Option<HeaderName>, T)>>(&mut self, iter: I) { + let mut iter = iter.into_iter(); + + // The structure of this is a bit weird, but it is mostly to make the + // borrow checker happy. + let (mut key, mut val) = match iter.next() { + Some((Some(key), val)) => (key, val), + Some((None, _)) => panic!("expected a header name, but got None"), + None => return, + }; + + 'outer: loop { + let mut entry = match self.entry2(key) { + Entry::Occupied(mut e) => { + // Replace all previous values while maintaining a handle to + // the entry. + e.insert(val); + e + } + Entry::Vacant(e) => e.insert_entry(val), + }; + + // As long as `HeaderName` is none, keep inserting the value into + // the current entry + loop { + match iter.next() { + Some((Some(k), v)) => { + key = k; + val = v; + continue 'outer; + } + Some((None, v)) => { + entry.append(v); + } + None => { + return; + } + } + } + } + } +} + +impl<T> Extend<(HeaderName, T)> for HeaderMap<T> { + fn extend<I: IntoIterator<Item = (HeaderName, T)>>(&mut self, iter: I) { + // Keys may be already present or show multiple times in the iterator. + // Reserve the entire hint lower bound if the map is empty. + // Otherwise reserve half the hint (rounded up), so the map + // will only resize twice in the worst case. + let iter = iter.into_iter(); + + let reserve = if self.is_empty() { + iter.size_hint().0 + } else { + (iter.size_hint().0 + 1) / 2 + }; + + self.reserve(reserve); + + for (k, v) in iter { + self.append(k, v); + } + } +} + +impl<T: PartialEq> PartialEq for HeaderMap<T> { + fn eq(&self, other: &HeaderMap<T>) -> bool { + if self.len() != other.len() { + return false; + } + + self.keys() + .all(|key| self.get_all(key) == other.get_all(key)) + } +} + +impl<T: Eq> Eq for HeaderMap<T> {} + +impl<T: fmt::Debug> fmt::Debug for HeaderMap<T> { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + f.debug_map().entries(self.iter()).finish() + } +} + +impl<T> Default for HeaderMap<T> { + fn default() -> Self { + HeaderMap::with_capacity(0) + } +} + +impl<'a, K, T> ops::Index<K> for HeaderMap<T> +where + K: AsHeaderName, +{ + type Output = T; + + /// # Panics + /// Using the index operator will cause a panic if the header you're querying isn't set. + #[inline] + fn index(&self, index: K) -> &T { + match self.get2(&index) { + Some(val) => val, + None => panic!("no entry found for key {:?}", index.as_str()), + } + } +} + +/// phase 2 is post-insert where we forward-shift `Pos` in the indices. +/// +/// returns the number of displaced elements +#[inline] +fn do_insert_phase_two(indices: &mut [Pos], mut probe: usize, mut old_pos: Pos) -> usize { + let mut num_displaced = 0; + + probe_loop!(probe < indices.len(), { + let pos = &mut indices[probe]; + + if pos.is_none() { + *pos = old_pos; + break; + } else { + num_displaced += 1; + old_pos = mem::replace(pos, old_pos); + } + }); + + num_displaced +} + +#[inline] +fn append_value<T>( + entry_idx: usize, + entry: &mut Bucket<T>, + extra: &mut Vec<ExtraValue<T>>, + value: T, +) { + match entry.links { + Some(links) => { + let idx = extra.len(); + extra.push(ExtraValue { + value: value, + prev: Link::Extra(links.tail), + next: Link::Entry(entry_idx), + }); + + extra[links.tail].next = Link::Extra(idx); + + entry.links = Some(Links { tail: idx, ..links }); + } + None => { + let idx = extra.len(); + extra.push(ExtraValue { + value: value, + prev: Link::Entry(entry_idx), + next: Link::Entry(entry_idx), + }); + + entry.links = Some(Links { + next: idx, + tail: idx, + }); + } + } +} + +// ===== impl Iter ===== + +impl<'a, T> Iterator for Iter<'a, T> { + type Item = (&'a HeaderName, &'a T); + + fn next(&mut self) -> Option<Self::Item> { + self.inner + .next_unsafe() + .map(|(key, ptr)| (key, unsafe { &*ptr })) + } + + fn size_hint(&self) -> (usize, Option<usize>) { + self.inner.size_hint() + } +} + +impl<'a, T> FusedIterator for Iter<'a, T> {} + +unsafe impl<'a, T: Sync> Sync for Iter<'a, T> {} +unsafe impl<'a, T: Sync> Send for Iter<'a, T> {} + +// ===== impl IterMut ===== + +impl<'a, T> IterMut<'a, T> { + fn next_unsafe(&mut self) -> Option<(&'a HeaderName, *mut T)> { + use self::Cursor::*; + + if self.cursor.is_none() { + if (self.entry + 1) >= unsafe { &*self.map }.entries.len() { + return None; + } + + self.entry += 1; + self.cursor = Some(Cursor::Head); + } + + let entry = unsafe { &(*self.map).entries[self.entry] }; + + match self.cursor.unwrap() { + Head => { + self.cursor = entry.links.map(|l| Values(l.next)); + Some((&entry.key, &entry.value as *const _ as *mut _)) + } + Values(idx) => { + let extra = unsafe { &(*self.map).extra_values[idx] }; + + match extra.next { + Link::Entry(_) => self.cursor = None, + Link::Extra(i) => self.cursor = Some(Values(i)), + } + + Some((&entry.key, &extra.value as *const _ as *mut _)) + } + } + } +} + +impl<'a, T> Iterator for IterMut<'a, T> { + type Item = (&'a HeaderName, &'a mut T); + + fn next(&mut self) -> Option<Self::Item> { + self.next_unsafe() + .map(|(key, ptr)| (key, unsafe { &mut *ptr })) + } + + fn size_hint(&self) -> (usize, Option<usize>) { + let map = unsafe { &*self.map }; + debug_assert!(map.entries.len() >= self.entry); + + let lower = map.entries.len() - self.entry; + // We could pessimistically guess at the upper bound, saying + // that its lower + map.extra_values.len(). That could be + // way over though, such as if we're near the end, and have + // already gone through several extra values... + (lower, None) + } +} + +impl<'a, T> FusedIterator for IterMut<'a, T> {} + +unsafe impl<'a, T: Sync> Sync for IterMut<'a, T> {} +unsafe impl<'a, T: Send> Send for IterMut<'a, T> {} + +// ===== impl Keys ===== + +impl<'a, T> Iterator for Keys<'a, T> { + type Item = &'a HeaderName; + + fn next(&mut self) -> Option<Self::Item> { + self.inner.next().map(|b| &b.key) + } + + fn size_hint(&self) -> (usize, Option<usize>) { + self.inner.size_hint() + } +} + +impl<'a, T> ExactSizeIterator for Keys<'a, T> {} +impl<'a, T> FusedIterator for Keys<'a, T> {} + +// ===== impl Values ==== + +impl<'a, T> Iterator for Values<'a, T> { + type Item = &'a T; + + fn next(&mut self) -> Option<Self::Item> { + self.inner.next().map(|(_, v)| v) + } + + fn size_hint(&self) -> (usize, Option<usize>) { + self.inner.size_hint() + } +} + +impl<'a, T> FusedIterator for Values<'a, T> {} + +// ===== impl ValuesMut ==== + +impl<'a, T> Iterator for ValuesMut<'a, T> { + type Item = &'a mut T; + + fn next(&mut self) -> Option<Self::Item> { + self.inner.next().map(|(_, v)| v) + } + + fn size_hint(&self) -> (usize, Option<usize>) { + self.inner.size_hint() + } +} + +impl<'a, T> FusedIterator for ValuesMut<'a, T> {} + +// ===== impl Drain ===== + +impl<'a, T> Iterator for Drain<'a, T> { + type Item = (Option<HeaderName>, T); + + fn next(&mut self) -> Option<Self::Item> { + if let Some(next) = self.next { + // Remove the extra value + + let raw_links = RawLinks(self.entries); + let extra = unsafe { + remove_extra_value(raw_links, &mut *self.extra_values, next) + }; + + match extra.next { + Link::Extra(idx) => self.next = Some(idx), + Link::Entry(_) => self.next = None, + } + + return Some((None, extra.value)); + } + + let idx = self.idx; + + if idx == self.len { + return None; + } + + self.idx += 1; + + unsafe { + let entry = &(*self.entries)[idx]; + + // Read the header name + let key = ptr::read(&entry.key as *const _); + let value = ptr::read(&entry.value as *const _); + self.next = entry.links.map(|l| l.next); + + Some((Some(key), value)) + } + } + + fn size_hint(&self) -> (usize, Option<usize>) { + // At least this many names... It's unknown if the user wants + // to count the extra_values on top. + // + // For instance, extending a new `HeaderMap` wouldn't need to + // reserve the upper-bound in `entries`, only the lower-bound. + let lower = self.len - self.idx; + let upper = unsafe { (*self.extra_values).len() } + lower; + (lower, Some(upper)) + } +} + +impl<'a, T> FusedIterator for Drain<'a, T> {} + +impl<'a, T> Drop for Drain<'a, T> { + fn drop(&mut self) { + for _ in self {} + } +} + +unsafe impl<'a, T: Sync> Sync for Drain<'a, T> {} +unsafe impl<'a, T: Send> Send for Drain<'a, T> {} + +// ===== impl Entry ===== + +impl<'a, T> Entry<'a, T> { + /// Ensures a value is in the entry by inserting the default if empty. + /// + /// Returns a mutable reference to the **first** value in the entry. + /// + /// # Examples + /// + /// ``` + /// # use http::HeaderMap; + /// let mut map: HeaderMap<u32> = HeaderMap::default(); + /// + /// let headers = &[ + /// "content-length", + /// "x-hello", + /// "Content-Length", + /// "x-world", + /// ]; + /// + /// for &header in headers { + /// let counter = map.entry(header) + /// .or_insert(0); + /// *counter += 1; + /// } + /// + /// assert_eq!(map["content-length"], 2); + /// assert_eq!(map["x-hello"], 1); + /// ``` + pub fn or_insert(self, default: T) -> &'a mut T { + use self::Entry::*; + + match self { + Occupied(e) => e.into_mut(), + Vacant(e) => e.insert(default), + } + } + + /// Ensures a value is in the entry by inserting the result of the default + /// function if empty. + /// + /// The default function is not called if the entry exists in the map. + /// Returns a mutable reference to the **first** value in the entry. + /// + /// # Examples + /// + /// Basic usage. + /// + /// ``` + /// # use http::HeaderMap; + /// let mut map = HeaderMap::new(); + /// + /// let res = map.entry("x-hello") + /// .or_insert_with(|| "world".parse().unwrap()); + /// + /// assert_eq!(res, "world"); + /// ``` + /// + /// The default function is not called if the entry exists in the map. + /// + /// ``` + /// # use http::HeaderMap; + /// # use http::header::HOST; + /// let mut map = HeaderMap::new(); + /// map.insert(HOST, "world".parse().unwrap()); + /// + /// let res = map.entry("host") + /// .or_insert_with(|| unreachable!()); + /// + /// + /// assert_eq!(res, "world"); + /// ``` + pub fn or_insert_with<F: FnOnce() -> T>(self, default: F) -> &'a mut T { + use self::Entry::*; + + match self { + Occupied(e) => e.into_mut(), + Vacant(e) => e.insert(default()), + } + } + + /// Returns a reference to the entry's key + /// + /// # Examples + /// + /// ``` + /// # use http::HeaderMap; + /// let mut map = HeaderMap::new(); + /// + /// assert_eq!(map.entry("x-hello").key(), "x-hello"); + /// ``` + pub fn key(&self) -> &HeaderName { + use self::Entry::*; + + match *self { + Vacant(ref e) => e.key(), + Occupied(ref e) => e.key(), + } + } +} + +// ===== impl VacantEntry ===== + +impl<'a, T> VacantEntry<'a, T> { + /// Returns a reference to the entry's key + /// + /// # Examples + /// + /// ``` + /// # use http::HeaderMap; + /// let mut map = HeaderMap::new(); + /// + /// assert_eq!(map.entry("x-hello").key().as_str(), "x-hello"); + /// ``` + pub fn key(&self) -> &HeaderName { + &self.key + } + + /// Take ownership of the key + /// + /// # Examples + /// + /// ``` + /// # use http::header::{HeaderMap, Entry}; + /// let mut map = HeaderMap::new(); + /// + /// if let Entry::Vacant(v) = map.entry("x-hello") { + /// assert_eq!(v.into_key().as_str(), "x-hello"); + /// } + /// ``` + pub fn into_key(self) -> HeaderName { + self.key + } + + /// Insert the value into the entry. + /// + /// The value will be associated with this entry's key. A mutable reference + /// to the inserted value will be returned. + /// + /// # Examples + /// + /// ``` + /// # use http::header::{HeaderMap, Entry}; + /// let mut map = HeaderMap::new(); + /// + /// if let Entry::Vacant(v) = map.entry("x-hello") { + /// v.insert("world".parse().unwrap()); + /// } + /// + /// assert_eq!(map["x-hello"], "world"); + /// ``` + pub fn insert(self, value: T) -> &'a mut T { + // Ensure that there is space in the map + let index = + self.map + .insert_phase_two(self.key, value.into(), self.hash, self.probe, self.danger); + + &mut self.map.entries[index].value + } + + /// Insert the value into the entry. + /// + /// The value will be associated with this entry's key. The new + /// `OccupiedEntry` is returned, allowing for further manipulation. + /// + /// # Examples + /// + /// ``` + /// # use http::header::*; + /// let mut map = HeaderMap::new(); + /// + /// if let Entry::Vacant(v) = map.entry("x-hello") { + /// let mut e = v.insert_entry("world".parse().unwrap()); + /// e.insert("world2".parse().unwrap()); + /// } + /// + /// assert_eq!(map["x-hello"], "world2"); + /// ``` + pub fn insert_entry(self, value: T) -> OccupiedEntry<'a, T> { + // Ensure that there is space in the map + let index = + self.map + .insert_phase_two(self.key, value.into(), self.hash, self.probe, self.danger); + + OccupiedEntry { + map: self.map, + index: index, + probe: self.probe, + } + } +} + +// ===== impl GetAll ===== + +impl<'a, T: 'a> GetAll<'a, T> { + /// Returns an iterator visiting all values associated with the entry. + /// + /// Values are iterated in insertion order. + /// + /// # Examples + /// + /// ``` + /// # use http::HeaderMap; + /// # use http::header::HOST; + /// let mut map = HeaderMap::new(); + /// map.insert(HOST, "hello.world".parse().unwrap()); + /// map.append(HOST, "hello.earth".parse().unwrap()); + /// + /// let values = map.get_all("host"); + /// let mut iter = values.iter(); + /// assert_eq!(&"hello.world", iter.next().unwrap()); + /// assert_eq!(&"hello.earth", iter.next().unwrap()); + /// assert!(iter.next().is_none()); + /// ``` + pub fn iter(&self) -> ValueIter<'a, T> { + // This creates a new GetAll struct so that the lifetime + // isn't bound to &self. + GetAll { + map: self.map, + index: self.index, + } + .into_iter() + } +} + +impl<'a, T: PartialEq> PartialEq for GetAll<'a, T> { + fn eq(&self, other: &Self) -> bool { + self.iter().eq(other.iter()) + } +} + +impl<'a, T> IntoIterator for GetAll<'a, T> { + type Item = &'a T; + type IntoIter = ValueIter<'a, T>; + + fn into_iter(self) -> ValueIter<'a, T> { + self.map.value_iter(self.index) + } +} + +impl<'a, 'b: 'a, T> IntoIterator for &'b GetAll<'a, T> { + type Item = &'a T; + type IntoIter = ValueIter<'a, T>; + + fn into_iter(self) -> ValueIter<'a, T> { + self.map.value_iter(self.index) + } +} + +// ===== impl ValueIter ===== + +impl<'a, T: 'a> Iterator for ValueIter<'a, T> { + type Item = &'a T; + + fn next(&mut self) -> Option<Self::Item> { + use self::Cursor::*; + + match self.front { + Some(Head) => { + let entry = &self.map.entries[self.index]; + + if self.back == Some(Head) { + self.front = None; + self.back = None; + } else { + // Update the iterator state + match entry.links { + Some(links) => { + self.front = Some(Values(links.next)); + } + None => unreachable!(), + } + } + + Some(&entry.value) + } + Some(Values(idx)) => { + let extra = &self.map.extra_values[idx]; + + if self.front == self.back { + self.front = None; + self.back = None; + } else { + match extra.next { + Link::Entry(_) => self.front = None, + Link::Extra(i) => self.front = Some(Values(i)), + } + } + + Some(&extra.value) + } + None => None, + } + } + + fn size_hint(&self) -> (usize, Option<usize>) { + match (self.front, self.back) { + // Exactly 1 value... + (Some(Cursor::Head), Some(Cursor::Head)) => (1, Some(1)), + // At least 1... + (Some(_), _) => (1, None), + // No more values... + (None, _) => (0, Some(0)), + } + } +} + +impl<'a, T: 'a> DoubleEndedIterator for ValueIter<'a, T> { + fn next_back(&mut self) -> Option<Self::Item> { + use self::Cursor::*; + + match self.back { + Some(Head) => { + self.front = None; + self.back = None; + Some(&self.map.entries[self.index].value) + } + Some(Values(idx)) => { + let extra = &self.map.extra_values[idx]; + + if self.front == self.back { + self.front = None; + self.back = None; + } else { + match extra.prev { + Link::Entry(_) => self.back = Some(Head), + Link::Extra(idx) => self.back = Some(Values(idx)), + } + } + + Some(&extra.value) + } + None => None, + } + } +} + +impl<'a, T> FusedIterator for ValueIter<'a, T> {} + +// ===== impl ValueIterMut ===== + +impl<'a, T: 'a> Iterator for ValueIterMut<'a, T> { + type Item = &'a mut T; + + fn next(&mut self) -> Option<Self::Item> { + use self::Cursor::*; + + let entry = unsafe { &mut (*self.map).entries[self.index] }; + + match self.front { + Some(Head) => { + if self.back == Some(Head) { + self.front = None; + self.back = None; + } else { + // Update the iterator state + match entry.links { + Some(links) => { + self.front = Some(Values(links.next)); + } + None => unreachable!(), + } + } + + Some(&mut entry.value) + } + Some(Values(idx)) => { + let extra = unsafe { &mut (*self.map).extra_values[idx] }; + + if self.front == self.back { + self.front = None; + self.back = None; + } else { + match extra.next { + Link::Entry(_) => self.front = None, + Link::Extra(i) => self.front = Some(Values(i)), + } + } + + Some(&mut extra.value) + } + None => None, + } + } +} + +impl<'a, T: 'a> DoubleEndedIterator for ValueIterMut<'a, T> { + fn next_back(&mut self) -> Option<Self::Item> { + use self::Cursor::*; + + let entry = unsafe { &mut (*self.map).entries[self.index] }; + + match self.back { + Some(Head) => { + self.front = None; + self.back = None; + Some(&mut entry.value) + } + Some(Values(idx)) => { + let extra = unsafe { &mut (*self.map).extra_values[idx] }; + + if self.front == self.back { + self.front = None; + self.back = None; + } else { + match extra.prev { + Link::Entry(_) => self.back = Some(Head), + Link::Extra(idx) => self.back = Some(Values(idx)), + } + } + + Some(&mut extra.value) + } + None => None, + } + } +} + +impl<'a, T> FusedIterator for ValueIterMut<'a, T> {} + +unsafe impl<'a, T: Sync> Sync for ValueIterMut<'a, T> {} +unsafe impl<'a, T: Send> Send for ValueIterMut<'a, T> {} + +// ===== impl IntoIter ===== + +impl<T> Iterator for IntoIter<T> { + type Item = (Option<HeaderName>, T); + + fn next(&mut self) -> Option<Self::Item> { + if let Some(next) = self.next { + self.next = match self.extra_values[next].next { + Link::Entry(_) => None, + Link::Extra(v) => Some(v), + }; + + let value = unsafe { ptr::read(&self.extra_values[next].value) }; + + return Some((None, value)); + } + + if let Some(bucket) = self.entries.next() { + self.next = bucket.links.map(|l| l.next); + let name = Some(bucket.key); + let value = bucket.value; + + return Some((name, value)); + } + + None + } + + fn size_hint(&self) -> (usize, Option<usize>) { + let (lower, _) = self.entries.size_hint(); + // There could be more than just the entries upper, as there + // could be items in the `extra_values`. We could guess, saying + // `upper + extra_values.len()`, but that could overestimate by a lot. + (lower, None) + } +} + +impl<T> FusedIterator for IntoIter<T> {} + +impl<T> Drop for IntoIter<T> { + fn drop(&mut self) { + // Ensure the iterator is consumed + for _ in self.by_ref() {} + + // All the values have already been yielded out. + unsafe { + self.extra_values.set_len(0); + } + } +} + +// ===== impl OccupiedEntry ===== + +impl<'a, T> OccupiedEntry<'a, T> { + /// Returns a reference to the entry's key. + /// + /// # Examples + /// + /// ``` + /// # use http::header::{HeaderMap, Entry, HOST}; + /// let mut map = HeaderMap::new(); + /// map.insert(HOST, "world".parse().unwrap()); + /// + /// if let Entry::Occupied(e) = map.entry("host") { + /// assert_eq!("host", e.key()); + /// } + /// ``` + pub fn key(&self) -> &HeaderName { + &self.map.entries[self.index].key + } + + /// Get a reference to the first value in the entry. + /// + /// Values are stored in insertion order. + /// + /// # Panics + /// + /// `get` panics if there are no values associated with the entry. + /// + /// # Examples + /// + /// ``` + /// # use http::header::{HeaderMap, Entry, HOST}; + /// let mut map = HeaderMap::new(); + /// map.insert(HOST, "hello.world".parse().unwrap()); + /// + /// if let Entry::Occupied(mut e) = map.entry("host") { + /// assert_eq!(e.get(), &"hello.world"); + /// + /// e.append("hello.earth".parse().unwrap()); + /// + /// assert_eq!(e.get(), &"hello.world"); + /// } + /// ``` + pub fn get(&self) -> &T { + &self.map.entries[self.index].value + } + + /// Get a mutable reference to the first value in the entry. + /// + /// Values are stored in insertion order. + /// + /// # Panics + /// + /// `get_mut` panics if there are no values associated with the entry. + /// + /// # Examples + /// + /// ``` + /// # use http::header::{HeaderMap, Entry, HOST}; + /// let mut map = HeaderMap::default(); + /// map.insert(HOST, "hello.world".to_string()); + /// + /// if let Entry::Occupied(mut e) = map.entry("host") { + /// e.get_mut().push_str("-2"); + /// assert_eq!(e.get(), &"hello.world-2"); + /// } + /// ``` + pub fn get_mut(&mut self) -> &mut T { + &mut self.map.entries[self.index].value + } + + /// Converts the `OccupiedEntry` into a mutable reference to the **first** + /// value. + /// + /// The lifetime of the returned reference is bound to the original map. + /// + /// # Panics + /// + /// `into_mut` panics if there are no values associated with the entry. + /// + /// # Examples + /// + /// ``` + /// # use http::header::{HeaderMap, Entry, HOST}; + /// let mut map = HeaderMap::default(); + /// map.insert(HOST, "hello.world".to_string()); + /// map.append(HOST, "hello.earth".to_string()); + /// + /// if let Entry::Occupied(e) = map.entry("host") { + /// e.into_mut().push_str("-2"); + /// } + /// + /// assert_eq!("hello.world-2", map["host"]); + /// ``` + pub fn into_mut(self) -> &'a mut T { + &mut self.map.entries[self.index].value + } + + /// Sets the value of the entry. + /// + /// All previous values associated with the entry are removed and the first + /// one is returned. See `insert_mult` for an API that returns all values. + /// + /// # Examples + /// + /// ``` + /// # use http::header::{HeaderMap, Entry, HOST}; + /// let mut map = HeaderMap::new(); + /// map.insert(HOST, "hello.world".parse().unwrap()); + /// + /// if let Entry::Occupied(mut e) = map.entry("host") { + /// let mut prev = e.insert("earth".parse().unwrap()); + /// assert_eq!("hello.world", prev); + /// } + /// + /// assert_eq!("earth", map["host"]); + /// ``` + pub fn insert(&mut self, value: T) -> T { + self.map.insert_occupied(self.index, value.into()) + } + + /// Sets the value of the entry. + /// + /// This function does the same as `insert` except it returns an iterator + /// that yields all values previously associated with the key. + /// + /// # Examples + /// + /// ``` + /// # use http::header::{HeaderMap, Entry, HOST}; + /// let mut map = HeaderMap::new(); + /// map.insert(HOST, "world".parse().unwrap()); + /// map.append(HOST, "world2".parse().unwrap()); + /// + /// if let Entry::Occupied(mut e) = map.entry("host") { + /// let mut prev = e.insert_mult("earth".parse().unwrap()); + /// assert_eq!("world", prev.next().unwrap()); + /// assert_eq!("world2", prev.next().unwrap()); + /// assert!(prev.next().is_none()); + /// } + /// + /// assert_eq!("earth", map["host"]); + /// ``` + pub fn insert_mult(&mut self, value: T) -> ValueDrain<'_, T> { + self.map.insert_occupied_mult(self.index, value.into()) + } + + /// Insert the value into the entry. + /// + /// The new value is appended to the end of the entry's value list. All + /// previous values associated with the entry are retained. + /// + /// # Examples + /// + /// ``` + /// # use http::header::{HeaderMap, Entry, HOST}; + /// let mut map = HeaderMap::new(); + /// map.insert(HOST, "world".parse().unwrap()); + /// + /// if let Entry::Occupied(mut e) = map.entry("host") { + /// e.append("earth".parse().unwrap()); + /// } + /// + /// let values = map.get_all("host"); + /// let mut i = values.iter(); + /// assert_eq!("world", *i.next().unwrap()); + /// assert_eq!("earth", *i.next().unwrap()); + /// ``` + pub fn append(&mut self, value: T) { + let idx = self.index; + let entry = &mut self.map.entries[idx]; + append_value(idx, entry, &mut self.map.extra_values, value.into()); + } + + /// Remove the entry from the map. + /// + /// All values associated with the entry are removed and the first one is + /// returned. See `remove_entry_mult` for an API that returns all values. + /// + /// # Examples + /// + /// ``` + /// # use http::header::{HeaderMap, Entry, HOST}; + /// let mut map = HeaderMap::new(); + /// map.insert(HOST, "world".parse().unwrap()); + /// + /// if let Entry::Occupied(e) = map.entry("host") { + /// let mut prev = e.remove(); + /// assert_eq!("world", prev); + /// } + /// + /// assert!(!map.contains_key("host")); + /// ``` + pub fn remove(self) -> T { + self.remove_entry().1 + } + + /// Remove the entry from the map. + /// + /// The key and all values associated with the entry are removed and the + /// first one is returned. See `remove_entry_mult` for an API that returns + /// all values. + /// + /// # Examples + /// + /// ``` + /// # use http::header::{HeaderMap, Entry, HOST}; + /// let mut map = HeaderMap::new(); + /// map.insert(HOST, "world".parse().unwrap()); + /// + /// if let Entry::Occupied(e) = map.entry("host") { + /// let (key, mut prev) = e.remove_entry(); + /// assert_eq!("host", key.as_str()); + /// assert_eq!("world", prev); + /// } + /// + /// assert!(!map.contains_key("host")); + /// ``` + pub fn remove_entry(self) -> (HeaderName, T) { + if let Some(links) = self.map.entries[self.index].links { + self.map.remove_all_extra_values(links.next); + } + + let entry = self.map.remove_found(self.probe, self.index); + + (entry.key, entry.value) + } + + /// Remove the entry from the map. + /// + /// The key and all values associated with the entry are removed and + /// returned. + pub fn remove_entry_mult(self) -> (HeaderName, ValueDrain<'a, T>) { + let raw_links = self.map.raw_links(); + let extra_values = &mut self.map.extra_values; + + let next = self.map.entries[self.index].links.map(|l| { + drain_all_extra_values(raw_links, extra_values, l.next) + .into_iter() + }); + + let entry = self.map.remove_found(self.probe, self.index); + + let drain = ValueDrain { + first: Some(entry.value), + next, + lt: PhantomData, + }; + (entry.key, drain) + } + + /// Returns an iterator visiting all values associated with the entry. + /// + /// Values are iterated in insertion order. + /// + /// # Examples + /// + /// ``` + /// # use http::header::{HeaderMap, Entry, HOST}; + /// let mut map = HeaderMap::new(); + /// map.insert(HOST, "world".parse().unwrap()); + /// map.append(HOST, "earth".parse().unwrap()); + /// + /// if let Entry::Occupied(e) = map.entry("host") { + /// let mut iter = e.iter(); + /// assert_eq!(&"world", iter.next().unwrap()); + /// assert_eq!(&"earth", iter.next().unwrap()); + /// assert!(iter.next().is_none()); + /// } + /// ``` + pub fn iter(&self) -> ValueIter<'_, T> { + self.map.value_iter(Some(self.index)) + } + + /// Returns an iterator mutably visiting all values associated with the + /// entry. + /// + /// Values are iterated in insertion order. + /// + /// # Examples + /// + /// ``` + /// # use http::header::{HeaderMap, Entry, HOST}; + /// let mut map = HeaderMap::default(); + /// map.insert(HOST, "world".to_string()); + /// map.append(HOST, "earth".to_string()); + /// + /// if let Entry::Occupied(mut e) = map.entry("host") { + /// for e in e.iter_mut() { + /// e.push_str("-boop"); + /// } + /// } + /// + /// let mut values = map.get_all("host"); + /// let mut i = values.iter(); + /// assert_eq!(&"world-boop", i.next().unwrap()); + /// assert_eq!(&"earth-boop", i.next().unwrap()); + /// ``` + pub fn iter_mut(&mut self) -> ValueIterMut<'_, T> { + self.map.value_iter_mut(self.index) + } +} + +impl<'a, T> IntoIterator for OccupiedEntry<'a, T> { + type Item = &'a mut T; + type IntoIter = ValueIterMut<'a, T>; + + fn into_iter(self) -> ValueIterMut<'a, T> { + self.map.value_iter_mut(self.index) + } +} + +impl<'a, 'b: 'a, T> IntoIterator for &'b OccupiedEntry<'a, T> { + type Item = &'a T; + type IntoIter = ValueIter<'a, T>; + + fn into_iter(self) -> ValueIter<'a, T> { + self.iter() + } +} + +impl<'a, 'b: 'a, T> IntoIterator for &'b mut OccupiedEntry<'a, T> { + type Item = &'a mut T; + type IntoIter = ValueIterMut<'a, T>; + + fn into_iter(self) -> ValueIterMut<'a, T> { + self.iter_mut() + } +} + +// ===== impl ValueDrain ===== + +impl<'a, T> Iterator for ValueDrain<'a, T> { + type Item = T; + + fn next(&mut self) -> Option<T> { + if self.first.is_some() { + self.first.take() + } else if let Some(ref mut extras) = self.next { + extras.next() + } else { + None + } + } + + fn size_hint(&self) -> (usize, Option<usize>) { + match (&self.first, &self.next) { + // Exactly 1 + (&Some(_), &None) => (1, Some(1)), + // 1 + extras + (&Some(_), &Some(ref extras)) => { + let (l, u) = extras.size_hint(); + (l + 1, u.map(|u| u + 1)) + }, + // Extras only + (&None, &Some(ref extras)) => extras.size_hint(), + // No more + (&None, &None) => (0, Some(0)), + } + } +} + +impl<'a, T> FusedIterator for ValueDrain<'a, T> {} + +impl<'a, T> Drop for ValueDrain<'a, T> { + fn drop(&mut self) { + while let Some(_) = self.next() {} + } +} + +unsafe impl<'a, T: Sync> Sync for ValueDrain<'a, T> {} +unsafe impl<'a, T: Send> Send for ValueDrain<'a, T> {} + +// ===== impl RawLinks ===== + +impl<T> Clone for RawLinks<T> { + fn clone(&self) -> RawLinks<T> { + *self + } +} + +impl<T> Copy for RawLinks<T> {} + +impl<T> ops::Index<usize> for RawLinks<T> { + type Output = Option<Links>; + + fn index(&self, idx: usize) -> &Self::Output { + unsafe { + &(*self.0)[idx].links + } + } +} + +impl<T> ops::IndexMut<usize> for RawLinks<T> { + fn index_mut(&mut self, idx: usize) -> &mut Self::Output { + unsafe { + &mut (*self.0)[idx].links + } + } +} + +// ===== impl Pos ===== + +impl Pos { + #[inline] + fn new(index: usize, hash: HashValue) -> Self { + debug_assert!(index < MAX_SIZE); + Pos { + index: index as Size, + hash: hash, + } + } + + #[inline] + fn none() -> Self { + Pos { + index: !0, + hash: HashValue(0), + } + } + + #[inline] + fn is_some(&self) -> bool { + !self.is_none() + } + + #[inline] + fn is_none(&self) -> bool { + self.index == !0 + } + + #[inline] + fn resolve(&self) -> Option<(usize, HashValue)> { + if self.is_some() { + Some((self.index as usize, self.hash)) + } else { + None + } + } +} + +impl Danger { + fn is_red(&self) -> bool { + match *self { + Danger::Red(_) => true, + _ => false, + } + } + + fn to_red(&mut self) { + debug_assert!(self.is_yellow()); + *self = Danger::Red(RandomState::new()); + } + + fn is_yellow(&self) -> bool { + match *self { + Danger::Yellow => true, + _ => false, + } + } + + fn to_yellow(&mut self) { + match *self { + Danger::Green => { + *self = Danger::Yellow; + } + _ => {} + } + } + + fn to_green(&mut self) { + debug_assert!(self.is_yellow()); + *self = Danger::Green; + } +} + +// ===== impl Utils ===== + +#[inline] +fn usable_capacity(cap: usize) -> usize { + cap - cap / 4 +} + +#[inline] +fn to_raw_capacity(n: usize) -> usize { + n + n / 3 +} + +#[inline] +fn desired_pos(mask: Size, hash: HashValue) -> usize { + (hash.0 & mask) as usize +} + +/// The number of steps that `current` is forward of the desired position for hash +#[inline] +fn probe_distance(mask: Size, hash: HashValue, current: usize) -> usize { + current.wrapping_sub(desired_pos(mask, hash)) & mask as usize +} + +fn hash_elem_using<K: ?Sized>(danger: &Danger, k: &K) -> HashValue +where + K: Hash, +{ + use fnv::FnvHasher; + + const MASK: u64 = (MAX_SIZE as u64) - 1; + + let hash = match *danger { + // Safe hash + Danger::Red(ref hasher) => { + let mut h = hasher.build_hasher(); + k.hash(&mut h); + h.finish() + } + // Fast hash + _ => { + let mut h = FnvHasher::default(); + k.hash(&mut h); + h.finish() + } + }; + + HashValue((hash & MASK) as u16) +} + +/* + * + * ===== impl IntoHeaderName / AsHeaderName ===== + * + */ + +mod into_header_name { + use super::{Entry, HdrName, HeaderMap, HeaderName}; + + /// A marker trait used to identify values that can be used as insert keys + /// to a `HeaderMap`. + pub trait IntoHeaderName: Sealed {} + + // All methods are on this pub(super) trait, instead of `IntoHeaderName`, + // so that they aren't publicly exposed to the world. + // + // Being on the `IntoHeaderName` trait would mean users could call + // `"host".insert(&mut map, "localhost")`. + // + // Ultimately, this allows us to adjust the signatures of these methods + // without breaking any external crate. + pub trait Sealed { + #[doc(hidden)] + fn insert<T>(self, map: &mut HeaderMap<T>, val: T) -> Option<T>; + + #[doc(hidden)] + fn append<T>(self, map: &mut HeaderMap<T>, val: T) -> bool; + + #[doc(hidden)] + fn entry<T>(self, map: &mut HeaderMap<T>) -> Entry<'_, T>; + } + + // ==== impls ==== + + impl Sealed for HeaderName { + #[inline] + fn insert<T>(self, map: &mut HeaderMap<T>, val: T) -> Option<T> { + map.insert2(self, val) + } + + #[inline] + fn append<T>(self, map: &mut HeaderMap<T>, val: T) -> bool { + map.append2(self, val) + } + + #[inline] + fn entry<T>(self, map: &mut HeaderMap<T>) -> Entry<'_, T> { + map.entry2(self) + } + } + + impl IntoHeaderName for HeaderName {} + + impl<'a> Sealed for &'a HeaderName { + #[inline] + fn insert<T>(self, map: &mut HeaderMap<T>, val: T) -> Option<T> { + map.insert2(self, val) + } + #[inline] + fn append<T>(self, map: &mut HeaderMap<T>, val: T) -> bool { + map.append2(self, val) + } + + #[inline] + fn entry<T>(self, map: &mut HeaderMap<T>) -> Entry<'_, T> { + map.entry2(self) + } + } + + impl<'a> IntoHeaderName for &'a HeaderName {} + + impl Sealed for &'static str { + #[inline] + fn insert<T>(self, map: &mut HeaderMap<T>, val: T) -> Option<T> { + HdrName::from_static(self, move |hdr| map.insert2(hdr, val)) + } + #[inline] + fn append<T>(self, map: &mut HeaderMap<T>, val: T) -> bool { + HdrName::from_static(self, move |hdr| map.append2(hdr, val)) + } + + #[inline] + fn entry<T>(self, map: &mut HeaderMap<T>) -> Entry<'_, T> { + HdrName::from_static(self, move |hdr| map.entry2(hdr)) + } + } + + impl IntoHeaderName for &'static str {} +} + +mod as_header_name { + use super::{Entry, HdrName, HeaderMap, HeaderName, InvalidHeaderName}; + + /// A marker trait used to identify values that can be used as search keys + /// to a `HeaderMap`. + pub trait AsHeaderName: Sealed {} + + // All methods are on this pub(super) trait, instead of `AsHeaderName`, + // so that they aren't publicly exposed to the world. + // + // Being on the `AsHeaderName` trait would mean users could call + // `"host".find(&map)`. + // + // Ultimately, this allows us to adjust the signatures of these methods + // without breaking any external crate. + pub trait Sealed { + #[doc(hidden)] + fn try_entry<T>(self, map: &mut HeaderMap<T>) -> Result<Entry<'_, T>, InvalidHeaderName>; + + #[doc(hidden)] + fn find<T>(&self, map: &HeaderMap<T>) -> Option<(usize, usize)>; + + #[doc(hidden)] + fn as_str(&self) -> &str; + } + + // ==== impls ==== + + impl Sealed for HeaderName { + #[inline] + fn try_entry<T>(self, map: &mut HeaderMap<T>) -> Result<Entry<'_, T>, InvalidHeaderName> { + Ok(map.entry2(self)) + } + + #[inline] + fn find<T>(&self, map: &HeaderMap<T>) -> Option<(usize, usize)> { + map.find(self) + } + + fn as_str(&self) -> &str { + <HeaderName>::as_str(self) + } + } + + impl AsHeaderName for HeaderName {} + + impl<'a> Sealed for &'a HeaderName { + #[inline] + fn try_entry<T>(self, map: &mut HeaderMap<T>) -> Result<Entry<'_, T>, InvalidHeaderName> { + Ok(map.entry2(self)) + } + + #[inline] + fn find<T>(&self, map: &HeaderMap<T>) -> Option<(usize, usize)> { + map.find(*self) + } + + fn as_str(&self) -> &str { + <HeaderName>::as_str(*self) + } + } + + impl<'a> AsHeaderName for &'a HeaderName {} + + impl<'a> Sealed for &'a str { + #[inline] + fn try_entry<T>(self, map: &mut HeaderMap<T>) -> Result<Entry<'_, T>, InvalidHeaderName> { + HdrName::from_bytes(self.as_bytes(), move |hdr| map.entry2(hdr)) + } + + #[inline] + fn find<T>(&self, map: &HeaderMap<T>) -> Option<(usize, usize)> { + HdrName::from_bytes(self.as_bytes(), move |hdr| map.find(&hdr)).unwrap_or(None) + } + + fn as_str(&self) -> &str { + self + } + } + + impl<'a> AsHeaderName for &'a str {} + + impl Sealed for String { + #[inline] + fn try_entry<T>(self, map: &mut HeaderMap<T>) -> Result<Entry<'_, T>, InvalidHeaderName> { + self.as_str().try_entry(map) + } + + #[inline] + fn find<T>(&self, map: &HeaderMap<T>) -> Option<(usize, usize)> { + Sealed::find(&self.as_str(), map) + } + + fn as_str(&self) -> &str { + self + } + } + + impl AsHeaderName for String {} + + impl<'a> Sealed for &'a String { + #[inline] + fn try_entry<T>(self, map: &mut HeaderMap<T>) -> Result<Entry<'_, T>, InvalidHeaderName> { + self.as_str().try_entry(map) + } + + #[inline] + fn find<T>(&self, map: &HeaderMap<T>) -> Option<(usize, usize)> { + Sealed::find(*self, map) + } + + fn as_str(&self) -> &str { + *self + } + } + + impl<'a> AsHeaderName for &'a String {} +} + +#[test] +fn test_bounds() { + fn check_bounds<T: Send + Send>() {} + + check_bounds::<HeaderMap<()>>(); + check_bounds::<Iter<'static, ()>>(); + check_bounds::<IterMut<'static, ()>>(); + check_bounds::<Keys<'static, ()>>(); + check_bounds::<Values<'static, ()>>(); + check_bounds::<ValuesMut<'static, ()>>(); + check_bounds::<Drain<'static, ()>>(); + check_bounds::<GetAll<'static, ()>>(); + check_bounds::<Entry<'static, ()>>(); + check_bounds::<VacantEntry<'static, ()>>(); + check_bounds::<OccupiedEntry<'static, ()>>(); + check_bounds::<ValueIter<'static, ()>>(); + check_bounds::<ValueIterMut<'static, ()>>(); + check_bounds::<ValueDrain<'static, ()>>(); +} + +#[test] +fn skip_duplicates_during_key_iteration() { + let mut map = HeaderMap::new(); + map.append("a", HeaderValue::from_static("a")); + map.append("a", HeaderValue::from_static("b")); + assert_eq!(map.keys().count(), map.keys_len()); +} diff --git a/third_party/rust/http/src/header/mod.rs b/third_party/rust/http/src/header/mod.rs new file mode 100644 index 0000000000..1ca4945094 --- /dev/null +++ b/third_party/rust/http/src/header/mod.rs @@ -0,0 +1,174 @@ +//! HTTP header types +//! +//! The module provides [`HeaderName`], [`HeaderMap`], and a number of types +//! used for interacting with `HeaderMap`. These types allow representing both +//! HTTP/1 and HTTP/2 headers. +//! +//! # `HeaderName` +//! +//! The `HeaderName` type represents both standard header names as well as +//! custom header names. The type handles the case insensitive nature of header +//! names and is used as the key portion of `HeaderMap`. Header names are +//! normalized to lower case. In other words, when creating a `HeaderName` with +//! a string, even if upper case characters are included, when getting a string +//! representation of the `HeaderName`, it will be all lower case. This allows +//! for faster `HeaderMap` comparison operations. +//! +//! The internal representation is optimized to efficiently handle the cases +//! most commonly encountered when working with HTTP. Standard header names are +//! special cased and are represented internally as an enum. Short custom +//! headers will be stored directly in the `HeaderName` struct and will not +//! incur any allocation overhead, however longer strings will require an +//! allocation for storage. +//! +//! ## Limitations +//! +//! `HeaderName` has a max length of 32,768 for header names. Attempting to +//! parse longer names will result in a panic. +//! +//! # `HeaderMap` +//! +//! `HeaderMap` is a map structure of header names highly optimized for use +//! cases common with HTTP. It is a [multimap] structure, where each header name +//! may have multiple associated header values. Given this, some of the APIs +//! diverge from [`HashMap`]. +//! +//! ## Overview +//! +//! Just like `HashMap` in Rust's stdlib, `HeaderMap` is based on [Robin Hood +//! hashing]. This algorithm tends to reduce the worst case search times in the +//! table and enables high load factors without seriously affecting performance. +//! Internally, keys and values are stored in vectors. As such, each insertion +//! will not incur allocation overhead. However, once the underlying vector +//! storage is full, a larger vector must be allocated and all values copied. +//! +//! ## Deterministic ordering +//! +//! Unlike Rust's `HashMap`, values in `HeaderMap` are deterministically +//! ordered. Roughly, values are ordered by insertion. This means that a +//! function that deterministically operates on a header map can rely on the +//! iteration order to remain consistent across processes and platforms. +//! +//! ## Adaptive hashing +//! +//! `HeaderMap` uses an adaptive hashing strategy in order to efficiently handle +//! most common cases. All standard headers have statically computed hash values +//! which removes the need to perform any hashing of these headers at runtime. +//! The default hash function emphasizes performance over robustness. However, +//! `HeaderMap` detects high collision rates and switches to a secure hash +//! function in those events. The threshold is set such that only denial of +//! service attacks should trigger it. +//! +//! ## Limitations +//! +//! `HeaderMap` can store a maximum of 32,768 headers (header name / value +//! pairs). Attempting to insert more will result in a panic. +//! +//! [`HeaderName`]: struct.HeaderName.html +//! [`HeaderMap`]: struct.HeaderMap.html +//! [multimap]: https://en.wikipedia.org/wiki/Multimap +//! [`HashMap`]: https://doc.rust-lang.org/std/collections/struct.HashMap.html +//! [Robin Hood hashing]: https://en.wikipedia.org/wiki/Hash_table#Robin_Hood_hashing + +mod map; +mod name; +mod value; + +pub use self::map::{ + AsHeaderName, Drain, Entry, GetAll, HeaderMap, IntoHeaderName, IntoIter, Iter, IterMut, Keys, + OccupiedEntry, VacantEntry, ValueDrain, ValueIter, ValueIterMut, Values, ValuesMut, +}; +pub use self::name::{HeaderName, InvalidHeaderName}; +pub use self::value::{HeaderValue, InvalidHeaderValue, ToStrError}; + +// Use header name constants +pub use self::name::{ + ACCEPT, + ACCEPT_CHARSET, + ACCEPT_ENCODING, + ACCEPT_LANGUAGE, + ACCEPT_RANGES, + ACCESS_CONTROL_ALLOW_CREDENTIALS, + ACCESS_CONTROL_ALLOW_HEADERS, + ACCESS_CONTROL_ALLOW_METHODS, + ACCESS_CONTROL_ALLOW_ORIGIN, + ACCESS_CONTROL_EXPOSE_HEADERS, + ACCESS_CONTROL_MAX_AGE, + ACCESS_CONTROL_REQUEST_HEADERS, + ACCESS_CONTROL_REQUEST_METHOD, + AGE, + ALLOW, + ALT_SVC, + AUTHORIZATION, + CACHE_CONTROL, + CACHE_STATUS, + CDN_CACHE_CONTROL, + CONNECTION, + CONTENT_DISPOSITION, + CONTENT_ENCODING, + CONTENT_LANGUAGE, + CONTENT_LENGTH, + CONTENT_LOCATION, + CONTENT_RANGE, + CONTENT_SECURITY_POLICY, + CONTENT_SECURITY_POLICY_REPORT_ONLY, + CONTENT_TYPE, + COOKIE, + DNT, + DATE, + ETAG, + EXPECT, + EXPIRES, + FORWARDED, + FROM, + HOST, + IF_MATCH, + IF_MODIFIED_SINCE, + IF_NONE_MATCH, + IF_RANGE, + IF_UNMODIFIED_SINCE, + LAST_MODIFIED, + LINK, + LOCATION, + MAX_FORWARDS, + ORIGIN, + PRAGMA, + PROXY_AUTHENTICATE, + PROXY_AUTHORIZATION, + PUBLIC_KEY_PINS, + PUBLIC_KEY_PINS_REPORT_ONLY, + RANGE, + REFERER, + REFERRER_POLICY, + REFRESH, + RETRY_AFTER, + SEC_WEBSOCKET_ACCEPT, + SEC_WEBSOCKET_EXTENSIONS, + SEC_WEBSOCKET_KEY, + SEC_WEBSOCKET_PROTOCOL, + SEC_WEBSOCKET_VERSION, + SERVER, + SET_COOKIE, + STRICT_TRANSPORT_SECURITY, + TE, + TRAILER, + TRANSFER_ENCODING, + UPGRADE, + UPGRADE_INSECURE_REQUESTS, + USER_AGENT, + VARY, + VIA, + WARNING, + WWW_AUTHENTICATE, + X_CONTENT_TYPE_OPTIONS, + X_DNS_PREFETCH_CONTROL, + X_FRAME_OPTIONS, + X_XSS_PROTECTION, +}; + +/// Maximum length of a header name +/// +/// Generally, 64kb for a header name is WAY too much than would ever be needed +/// in practice. Restricting it to this size enables using `u16` values to +/// represent offsets when dealing with header names. +const MAX_HEADER_NAME_LEN: usize = (1 << 16) - 1; diff --git a/third_party/rust/http/src/header/name.rs b/third_party/rust/http/src/header/name.rs new file mode 100644 index 0000000000..6080cf088b --- /dev/null +++ b/third_party/rust/http/src/header/name.rs @@ -0,0 +1,1868 @@ +use crate::byte_str::ByteStr; +use bytes::{Bytes, BytesMut}; + +use std::borrow::Borrow; +use std::error::Error; +use std::convert::{TryFrom}; +use std::hash::{Hash, Hasher}; +use std::mem::MaybeUninit; +use std::str::FromStr; +use std::fmt; + +/// Represents an HTTP header field name +/// +/// Header field names identify the header. Header sets may include multiple +/// headers with the same name. The HTTP specification defines a number of +/// standard headers, but HTTP messages may include non-standard header names as +/// well as long as they adhere to the specification. +/// +/// `HeaderName` is used as the [`HeaderMap`] key. Constants are available for +/// all standard header names in the [`header`] module. +/// +/// # Representation +/// +/// `HeaderName` represents standard header names using an `enum`, as such they +/// will not require an allocation for storage. All custom header names are +/// lower cased upon conversion to a `HeaderName` value. This avoids the +/// overhead of dynamically doing lower case conversion during the hash code +/// computation and the comparison operation. +/// +/// [`HeaderMap`]: struct.HeaderMap.html +/// [`header`]: index.html +#[derive(Clone, Eq, PartialEq, Hash)] +pub struct HeaderName { + inner: Repr<Custom>, +} + +// Almost a full `HeaderName` +#[derive(Debug, Hash)] +pub struct HdrName<'a> { + inner: Repr<MaybeLower<'a>>, +} + +#[derive(Debug, Clone, Eq, PartialEq, Hash)] +enum Repr<T> { + Standard(StandardHeader), + Custom(T), +} + +// Used to hijack the Hash impl +#[derive(Debug, Clone, Eq, PartialEq)] +struct Custom(ByteStr); + +#[derive(Debug, Clone)] +// Invariant: If lower then buf is valid UTF-8. +struct MaybeLower<'a> { + buf: &'a [u8], + lower: bool, +} + +/// A possible error when converting a `HeaderName` from another type. +pub struct InvalidHeaderName { + _priv: (), +} + +macro_rules! standard_headers { + ( + $( + $(#[$docs:meta])* + ($konst:ident, $upcase:ident, $name_bytes:literal); + )+ + ) => { + #[derive(Debug, Clone, Copy, Eq, PartialEq, Hash)] + enum StandardHeader { + $( + $konst, + )+ + } + + $( + $(#[$docs])* + pub const $upcase: HeaderName = HeaderName { + inner: Repr::Standard(StandardHeader::$konst), + }; + )+ + + impl StandardHeader { + #[inline] + fn as_str(&self) -> &'static str { + match *self { + // Safety: test_parse_standard_headers ensures these &[u8]s are &str-safe. + $( + StandardHeader::$konst => unsafe { std::str::from_utf8_unchecked( $name_bytes ) }, + )+ + } + } + + const fn from_bytes(name_bytes: &[u8]) -> Option<StandardHeader> { + match name_bytes { + $( + $name_bytes => Some(StandardHeader::$konst), + )+ + _ => None, + } + } + } + + #[cfg(test)] + const TEST_HEADERS: &'static [(StandardHeader, &'static [u8])] = &[ + $( + (StandardHeader::$konst, $name_bytes), + )+ + ]; + + #[test] + fn test_parse_standard_headers() { + for &(std, name_bytes) in TEST_HEADERS { + // Test lower case + assert_eq!(HeaderName::from_bytes(name_bytes).unwrap(), HeaderName::from(std)); + + // Test upper case + let upper = std::str::from_utf8(name_bytes).expect("byte string constants are all utf-8").to_uppercase(); + assert_eq!(HeaderName::from_bytes(upper.as_bytes()).unwrap(), HeaderName::from(std)); + } + } + + #[test] + fn test_standard_headers_into_bytes() { + for &(std, name_bytes) in TEST_HEADERS { + let name = std::str::from_utf8(name_bytes).unwrap(); + let std = HeaderName::from(std); + // Test lower case + let bytes: Bytes = + HeaderName::from_bytes(name_bytes).unwrap().inner.into(); + assert_eq!(bytes, name); + assert_eq!(HeaderName::from_bytes(name_bytes).unwrap(), std); + + // Test upper case + let upper = name.to_uppercase(); + let bytes: Bytes = + HeaderName::from_bytes(upper.as_bytes()).unwrap().inner.into(); + assert_eq!(bytes, name_bytes); + assert_eq!(HeaderName::from_bytes(upper.as_bytes()).unwrap(), + std); + } + + } + } +} + +// Generate constants for all standard HTTP headers. This includes a static hash +// code for the "fast hash" path. The hash code for static headers *do not* have +// to match the text representation of those headers. This is because header +// strings are always converted to the static values (when they match) before +// being hashed. This means that it is impossible to compare the static hash +// code of CONTENT_LENGTH with "content-length". +standard_headers! { + /// Advertises which content types the client is able to understand. + /// + /// The Accept request HTTP header advertises which content types, expressed + /// as MIME types, the client is able to understand. Using content + /// negotiation, the server then selects one of the proposals, uses it and + /// informs the client of its choice with the Content-Type response header. + /// Browsers set adequate values for this header depending of the context + /// where the request is done: when fetching a CSS stylesheet a different + /// value is set for the request than when fetching an image, video or a + /// script. + (Accept, ACCEPT, b"accept"); + + /// Advertises which character set the client is able to understand. + /// + /// The Accept-Charset request HTTP header advertises which character set + /// the client is able to understand. Using content negotiation, the server + /// then selects one of the proposals, uses it and informs the client of its + /// choice within the Content-Type response header. Browsers usually don't + /// set this header as the default value for each content type is usually + /// correct and transmitting it would allow easier fingerprinting. + /// + /// If the server cannot serve any matching character set, it can + /// theoretically send back a 406 (Not Acceptable) error code. But, for a + /// better user experience, this is rarely done and the more common way is + /// to ignore the Accept-Charset header in this case. + (AcceptCharset, ACCEPT_CHARSET, b"accept-charset"); + + /// Advertises which content encoding the client is able to understand. + /// + /// The Accept-Encoding request HTTP header advertises which content + /// encoding, usually a compression algorithm, the client is able to + /// understand. Using content negotiation, the server selects one of the + /// proposals, uses it and informs the client of its choice with the + /// Content-Encoding response header. + /// + /// Even if both the client and the server supports the same compression + /// algorithms, the server may choose not to compress the body of a + /// response, if the identity value is also acceptable. Two common cases + /// lead to this: + /// + /// * The data to be sent is already compressed and a second compression + /// won't lead to smaller data to be transmitted. This may the case with + /// some image formats; + /// + /// * The server is overloaded and cannot afford the computational overhead + /// induced by the compression requirement. Typically, Microsoft recommends + /// not to compress if a server use more than 80 % of its computational + /// power. + /// + /// As long as the identity value, meaning no encryption, is not explicitly + /// forbidden, by an identity;q=0 or a *;q=0 without another explicitly set + /// value for identity, the server must never send back a 406 Not Acceptable + /// error. + (AcceptEncoding, ACCEPT_ENCODING, b"accept-encoding"); + + /// Advertises which languages the client is able to understand. + /// + /// The Accept-Language request HTTP header advertises which languages the + /// client is able to understand, and which locale variant is preferred. + /// Using content negotiation, the server then selects one of the proposals, + /// uses it and informs the client of its choice with the Content-Language + /// response header. Browsers set adequate values for this header according + /// their user interface language and even if a user can change it, this + /// happens rarely (and is frown upon as it leads to fingerprinting). + /// + /// This header is a hint to be used when the server has no way of + /// determining the language via another way, like a specific URL, that is + /// controlled by an explicit user decision. It is recommended that the + /// server never overrides an explicit decision. The content of the + /// Accept-Language is often out of the control of the user (like when + /// traveling and using an Internet Cafe in a different country); the user + /// may also want to visit a page in another language than the locale of + /// their user interface. + /// + /// If the server cannot serve any matching language, it can theoretically + /// send back a 406 (Not Acceptable) error code. But, for a better user + /// experience, this is rarely done and more common way is to ignore the + /// Accept-Language header in this case. + (AcceptLanguage, ACCEPT_LANGUAGE, b"accept-language"); + + /// Marker used by the server to advertise partial request support. + /// + /// The Accept-Ranges response HTTP header is a marker used by the server to + /// advertise its support of partial requests. The value of this field + /// indicates the unit that can be used to define a range. + /// + /// In presence of an Accept-Ranges header, the browser may try to resume an + /// interrupted download, rather than to start it from the start again. + (AcceptRanges, ACCEPT_RANGES, b"accept-ranges"); + + /// Preflight response indicating if the response to the request can be + /// exposed to the page. + /// + /// The Access-Control-Allow-Credentials response header indicates whether + /// or not the response to the request can be exposed to the page. It can be + /// exposed when the true value is returned; it can't in other cases. + /// + /// Credentials are cookies, authorization headers or TLS client + /// certificates. + /// + /// When used as part of a response to a preflight request, this indicates + /// whether or not the actual request can be made using credentials. Note + /// that simple GET requests are not preflighted, and so if a request is + /// made for a resource with credentials, if this header is not returned + /// with the resource, the response is ignored by the browser and not + /// returned to web content. + /// + /// The Access-Control-Allow-Credentials header works in conjunction with + /// the XMLHttpRequest.withCredentials property or with the credentials + /// option in the Request() constructor of the Fetch API. Credentials must + /// be set on both sides (the Access-Control-Allow-Credentials header and in + /// the XHR or Fetch request) in order for the CORS request with credentials + /// to succeed. + (AccessControlAllowCredentials, ACCESS_CONTROL_ALLOW_CREDENTIALS, b"access-control-allow-credentials"); + + /// Preflight response indicating permitted HTTP headers. + /// + /// The Access-Control-Allow-Headers response header is used in response to + /// a preflight request to indicate which HTTP headers will be available via + /// Access-Control-Expose-Headers when making the actual request. + /// + /// The simple headers, Accept, Accept-Language, Content-Language, + /// Content-Type (but only with a MIME type of its parsed value (ignoring + /// parameters) of either application/x-www-form-urlencoded, + /// multipart/form-data, or text/plain), are always available and don't need + /// to be listed by this header. + /// + /// This header is required if the request has an + /// Access-Control-Request-Headers header. + (AccessControlAllowHeaders, ACCESS_CONTROL_ALLOW_HEADERS, b"access-control-allow-headers"); + + /// Preflight header response indicating permitted access methods. + /// + /// The Access-Control-Allow-Methods response header specifies the method or + /// methods allowed when accessing the resource in response to a preflight + /// request. + (AccessControlAllowMethods, ACCESS_CONTROL_ALLOW_METHODS, b"access-control-allow-methods"); + + /// Indicates whether the response can be shared with resources with the + /// given origin. + (AccessControlAllowOrigin, ACCESS_CONTROL_ALLOW_ORIGIN, b"access-control-allow-origin"); + + /// Indicates which headers can be exposed as part of the response by + /// listing their names. + (AccessControlExposeHeaders, ACCESS_CONTROL_EXPOSE_HEADERS, b"access-control-expose-headers"); + + /// Indicates how long the results of a preflight request can be cached. + (AccessControlMaxAge, ACCESS_CONTROL_MAX_AGE, b"access-control-max-age"); + + /// Informs the server which HTTP headers will be used when an actual + /// request is made. + (AccessControlRequestHeaders, ACCESS_CONTROL_REQUEST_HEADERS, b"access-control-request-headers"); + + /// Informs the server know which HTTP method will be used when the actual + /// request is made. + (AccessControlRequestMethod, ACCESS_CONTROL_REQUEST_METHOD, b"access-control-request-method"); + + /// Indicates the time in seconds the object has been in a proxy cache. + /// + /// The Age header is usually close to zero. If it is Age: 0, it was + /// probably just fetched from the origin server; otherwise It is usually + /// calculated as a difference between the proxy's current date and the Date + /// general header included in the HTTP response. + (Age, AGE, b"age"); + + /// Lists the set of methods support by a resource. + /// + /// This header must be sent if the server responds with a 405 Method Not + /// Allowed status code to indicate which request methods can be used. An + /// empty Allow header indicates that the resource allows no request + /// methods, which might occur temporarily for a given resource, for + /// example. + (Allow, ALLOW, b"allow"); + + /// Advertises the availability of alternate services to clients. + (AltSvc, ALT_SVC, b"alt-svc"); + + /// Contains the credentials to authenticate a user agent with a server. + /// + /// Usually this header is included after the server has responded with a + /// 401 Unauthorized status and the WWW-Authenticate header. + (Authorization, AUTHORIZATION, b"authorization"); + + /// Specifies directives for caching mechanisms in both requests and + /// responses. + /// + /// Caching directives are unidirectional, meaning that a given directive in + /// a request is not implying that the same directive is to be given in the + /// response. + (CacheControl, CACHE_CONTROL, b"cache-control"); + + /// Indicates how caches have handled a response and its corresponding request. + /// + /// See [RFC 9211](https://www.rfc-editor.org/rfc/rfc9211.html). + (CacheStatus, CACHE_STATUS, b"cache-status"); + + /// Specifies directives that allow origin servers to control the behavior of CDN caches + /// interposed between them and clients separately from other caches that might handle the + /// response. + /// + /// See [RFC 9213](https://www.rfc-editor.org/rfc/rfc9213.html). + (CdnCacheControl, CDN_CACHE_CONTROL, b"cdn-cache-control"); + + /// Controls whether or not the network connection stays open after the + /// current transaction finishes. + /// + /// If the value sent is keep-alive, the connection is persistent and not + /// closed, allowing for subsequent requests to the same server to be done. + /// + /// Except for the standard hop-by-hop headers (Keep-Alive, + /// Transfer-Encoding, TE, Connection, Trailer, Upgrade, Proxy-Authorization + /// and Proxy-Authenticate), any hop-by-hop headers used by the message must + /// be listed in the Connection header, so that the first proxy knows he has + /// to consume them and not to forward them further. Standard hop-by-hop + /// headers can be listed too (it is often the case of Keep-Alive, but this + /// is not mandatory. + (Connection, CONNECTION, b"connection"); + + /// Indicates if the content is expected to be displayed inline. + /// + /// In a regular HTTP response, the Content-Disposition response header is a + /// header indicating if the content is expected to be displayed inline in + /// the browser, that is, as a Web page or as part of a Web page, or as an + /// attachment, that is downloaded and saved locally. + /// + /// In a multipart/form-data body, the HTTP Content-Disposition general + /// header is a header that can be used on the subpart of a multipart body + /// to give information about the field it applies to. The subpart is + /// delimited by the boundary defined in the Content-Type header. Used on + /// the body itself, Content-Disposition has no effect. + /// + /// The Content-Disposition header is defined in the larger context of MIME + /// messages for e-mail, but only a subset of the possible parameters apply + /// to HTTP forms and POST requests. Only the value form-data, as well as + /// the optional directive name and filename, can be used in the HTTP + /// context. + (ContentDisposition, CONTENT_DISPOSITION, b"content-disposition"); + + /// Used to compress the media-type. + /// + /// When present, its value indicates what additional content encoding has + /// been applied to the entity-body. It lets the client know, how to decode + /// in order to obtain the media-type referenced by the Content-Type header. + /// + /// It is recommended to compress data as much as possible and therefore to + /// use this field, but some types of resources, like jpeg images, are + /// already compressed. Sometimes using additional compression doesn't + /// reduce payload size and can even make the payload longer. + (ContentEncoding, CONTENT_ENCODING, b"content-encoding"); + + /// Used to describe the languages intended for the audience. + /// + /// This header allows a user to differentiate according to the users' own + /// preferred language. For example, if "Content-Language: de-DE" is set, it + /// says that the document is intended for German language speakers + /// (however, it doesn't indicate the document is written in German. For + /// example, it might be written in English as part of a language course for + /// German speakers). + /// + /// If no Content-Language is specified, the default is that the content is + /// intended for all language audiences. Multiple language tags are also + /// possible, as well as applying the Content-Language header to various + /// media types and not only to textual documents. + (ContentLanguage, CONTENT_LANGUAGE, b"content-language"); + + /// Indicates the size of the entity-body. + /// + /// The header value must be a decimal indicating the number of octets sent + /// to the recipient. + (ContentLength, CONTENT_LENGTH, b"content-length"); + + /// Indicates an alternate location for the returned data. + /// + /// The principal use case is to indicate the URL of the resource + /// transmitted as the result of content negotiation. + /// + /// Location and Content-Location are different: Location indicates the + /// target of a redirection (or the URL of a newly created document), while + /// Content-Location indicates the direct URL to use to access the resource, + /// without the need of further content negotiation. Location is a header + /// associated with the response, while Content-Location is associated with + /// the entity returned. + (ContentLocation, CONTENT_LOCATION, b"content-location"); + + /// Indicates where in a full body message a partial message belongs. + (ContentRange, CONTENT_RANGE, b"content-range"); + + /// Allows controlling resources the user agent is allowed to load for a + /// given page. + /// + /// With a few exceptions, policies mostly involve specifying server origins + /// and script endpoints. This helps guard against cross-site scripting + /// attacks (XSS). + (ContentSecurityPolicy, CONTENT_SECURITY_POLICY, b"content-security-policy"); + + /// Allows experimenting with policies by monitoring their effects. + /// + /// The HTTP Content-Security-Policy-Report-Only response header allows web + /// developers to experiment with policies by monitoring (but not enforcing) + /// their effects. These violation reports consist of JSON documents sent + /// via an HTTP POST request to the specified URI. + (ContentSecurityPolicyReportOnly, CONTENT_SECURITY_POLICY_REPORT_ONLY, b"content-security-policy-report-only"); + + /// Used to indicate the media type of the resource. + /// + /// In responses, a Content-Type header tells the client what the content + /// type of the returned content actually is. Browsers will do MIME sniffing + /// in some cases and will not necessarily follow the value of this header; + /// to prevent this behavior, the header X-Content-Type-Options can be set + /// to nosniff. + /// + /// In requests, (such as POST or PUT), the client tells the server what + /// type of data is actually sent. + (ContentType, CONTENT_TYPE, b"content-type"); + + /// Contains stored HTTP cookies previously sent by the server with the + /// Set-Cookie header. + /// + /// The Cookie header might be omitted entirely, if the privacy setting of + /// the browser are set to block them, for example. + (Cookie, COOKIE, b"cookie"); + + /// Indicates the client's tracking preference. + /// + /// This header lets users indicate whether they would prefer privacy rather + /// than personalized content. + (Dnt, DNT, b"dnt"); + + /// Contains the date and time at which the message was originated. + (Date, DATE, b"date"); + + /// Identifier for a specific version of a resource. + /// + /// This header allows caches to be more efficient, and saves bandwidth, as + /// a web server does not need to send a full response if the content has + /// not changed. On the other side, if the content has changed, etags are + /// useful to help prevent simultaneous updates of a resource from + /// overwriting each other ("mid-air collisions"). + /// + /// If the resource at a given URL changes, a new Etag value must be + /// generated. Etags are therefore similar to fingerprints and might also be + /// used for tracking purposes by some servers. A comparison of them allows + /// to quickly determine whether two representations of a resource are the + /// same, but they might also be set to persist indefinitely by a tracking + /// server. + (Etag, ETAG, b"etag"); + + /// Indicates expectations that need to be fulfilled by the server in order + /// to properly handle the request. + /// + /// The only expectation defined in the specification is Expect: + /// 100-continue, to which the server shall respond with: + /// + /// * 100 if the information contained in the header is sufficient to cause + /// an immediate success, + /// + /// * 417 (Expectation Failed) if it cannot meet the expectation; or any + /// other 4xx status otherwise. + /// + /// For example, the server may reject a request if its Content-Length is + /// too large. + /// + /// No common browsers send the Expect header, but some other clients such + /// as cURL do so by default. + (Expect, EXPECT, b"expect"); + + /// Contains the date/time after which the response is considered stale. + /// + /// Invalid dates, like the value 0, represent a date in the past and mean + /// that the resource is already expired. + /// + /// If there is a Cache-Control header with the "max-age" or "s-max-age" + /// directive in the response, the Expires header is ignored. + (Expires, EXPIRES, b"expires"); + + /// Contains information from the client-facing side of proxy servers that + /// is altered or lost when a proxy is involved in the path of the request. + /// + /// The alternative and de-facto standard versions of this header are the + /// X-Forwarded-For, X-Forwarded-Host and X-Forwarded-Proto headers. + /// + /// This header is used for debugging, statistics, and generating + /// location-dependent content and by design it exposes privacy sensitive + /// information, such as the IP address of the client. Therefore the user's + /// privacy must be kept in mind when deploying this header. + (Forwarded, FORWARDED, b"forwarded"); + + /// Contains an Internet email address for a human user who controls the + /// requesting user agent. + /// + /// If you are running a robotic user agent (e.g. a crawler), the From + /// header should be sent, so you can be contacted if problems occur on + /// servers, such as if the robot is sending excessive, unwanted, or invalid + /// requests. + (From, FROM, b"from"); + + /// Specifies the domain name of the server and (optionally) the TCP port + /// number on which the server is listening. + /// + /// If no port is given, the default port for the service requested (e.g., + /// "80" for an HTTP URL) is implied. + /// + /// A Host header field must be sent in all HTTP/1.1 request messages. A 400 + /// (Bad Request) status code will be sent to any HTTP/1.1 request message + /// that lacks a Host header field or contains more than one. + (Host, HOST, b"host"); + + /// Makes a request conditional based on the E-Tag. + /// + /// For GET and HEAD methods, the server will send back the requested + /// resource only if it matches one of the listed ETags. For PUT and other + /// non-safe methods, it will only upload the resource in this case. + /// + /// The comparison with the stored ETag uses the strong comparison + /// algorithm, meaning two files are considered identical byte to byte only. + /// This is weakened when the W/ prefix is used in front of the ETag. + /// + /// There are two common use cases: + /// + /// * For GET and HEAD methods, used in combination with an Range header, it + /// can guarantee that the new ranges requested comes from the same resource + /// than the previous one. If it doesn't match, then a 416 (Range Not + /// Satisfiable) response is returned. + /// + /// * For other methods, and in particular for PUT, If-Match can be used to + /// prevent the lost update problem. It can check if the modification of a + /// resource that the user wants to upload will not override another change + /// that has been done since the original resource was fetched. If the + /// request cannot be fulfilled, the 412 (Precondition Failed) response is + /// returned. + (IfMatch, IF_MATCH, b"if-match"); + + /// Makes a request conditional based on the modification date. + /// + /// The If-Modified-Since request HTTP header makes the request conditional: + /// the server will send back the requested resource, with a 200 status, + /// only if it has been last modified after the given date. If the request + /// has not been modified since, the response will be a 304 without any + /// body; the Last-Modified header will contain the date of last + /// modification. Unlike If-Unmodified-Since, If-Modified-Since can only be + /// used with a GET or HEAD. + /// + /// When used in combination with If-None-Match, it is ignored, unless the + /// server doesn't support If-None-Match. + /// + /// The most common use case is to update a cached entity that has no + /// associated ETag. + (IfModifiedSince, IF_MODIFIED_SINCE, b"if-modified-since"); + + /// Makes a request conditional based on the E-Tag. + /// + /// The If-None-Match HTTP request header makes the request conditional. For + /// GET and HEAD methods, the server will send back the requested resource, + /// with a 200 status, only if it doesn't have an ETag matching the given + /// ones. For other methods, the request will be processed only if the + /// eventually existing resource's ETag doesn't match any of the values + /// listed. + /// + /// When the condition fails for GET and HEAD methods, then the server must + /// return HTTP status code 304 (Not Modified). For methods that apply + /// server-side changes, the status code 412 (Precondition Failed) is used. + /// Note that the server generating a 304 response MUST generate any of the + /// following header fields that would have been sent in a 200 (OK) response + /// to the same request: Cache-Control, Content-Location, Date, ETag, + /// Expires, and Vary. + /// + /// The comparison with the stored ETag uses the weak comparison algorithm, + /// meaning two files are considered identical not only if they are + /// identical byte to byte, but if the content is equivalent. For example, + /// two pages that would differ only by the date of generation in the footer + /// would be considered as identical. + /// + /// When used in combination with If-Modified-Since, it has precedence (if + /// the server supports it). + /// + /// There are two common use cases: + /// + /// * For `GET` and `HEAD` methods, to update a cached entity that has an associated ETag. + /// * For other methods, and in particular for `PUT`, `If-None-Match` used with + /// the `*` value can be used to save a file not known to exist, + /// guaranteeing that another upload didn't happen before, losing the data + /// of the previous put; this problems is the variation of the lost update + /// problem. + (IfNoneMatch, IF_NONE_MATCH, b"if-none-match"); + + /// Makes a request conditional based on range. + /// + /// The If-Range HTTP request header makes a range request conditional: if + /// the condition is fulfilled, the range request will be issued and the + /// server sends back a 206 Partial Content answer with the appropriate + /// body. If the condition is not fulfilled, the full resource is sent back, + /// with a 200 OK status. + /// + /// This header can be used either with a Last-Modified validator, or with + /// an ETag, but not with both. + /// + /// The most common use case is to resume a download, to guarantee that the + /// stored resource has not been modified since the last fragment has been + /// received. + (IfRange, IF_RANGE, b"if-range"); + + /// Makes the request conditional based on the last modification date. + /// + /// The If-Unmodified-Since request HTTP header makes the request + /// conditional: the server will send back the requested resource, or accept + /// it in the case of a POST or another non-safe method, only if it has not + /// been last modified after the given date. If the request has been + /// modified after the given date, the response will be a 412 (Precondition + /// Failed) error. + /// + /// There are two common use cases: + /// + /// * In conjunction non-safe methods, like POST, it can be used to + /// implement an optimistic concurrency control, like done by some wikis: + /// editions are rejected if the stored document has been modified since the + /// original has been retrieved. + /// + /// * In conjunction with a range request with a If-Range header, it can be + /// used to ensure that the new fragment requested comes from an unmodified + /// document. + (IfUnmodifiedSince, IF_UNMODIFIED_SINCE, b"if-unmodified-since"); + + /// Content-Types that are acceptable for the response. + (LastModified, LAST_MODIFIED, b"last-modified"); + + /// Allows the server to point an interested client to another resource + /// containing metadata about the requested resource. + (Link, LINK, b"link"); + + /// Indicates the URL to redirect a page to. + /// + /// The Location response header indicates the URL to redirect a page to. It + /// only provides a meaning when served with a 3xx status response. + /// + /// The HTTP method used to make the new request to fetch the page pointed + /// to by Location depends of the original method and of the kind of + /// redirection: + /// + /// * If 303 (See Also) responses always lead to the use of a GET method, + /// 307 (Temporary Redirect) and 308 (Permanent Redirect) don't change the + /// method used in the original request; + /// + /// * 301 (Permanent Redirect) and 302 (Found) doesn't change the method + /// most of the time, though older user-agents may (so you basically don't + /// know). + /// + /// All responses with one of these status codes send a Location header. + /// + /// Beside redirect response, messages with 201 (Created) status also + /// include the Location header. It indicates the URL to the newly created + /// resource. + /// + /// Location and Content-Location are different: Location indicates the + /// target of a redirection (or the URL of a newly created resource), while + /// Content-Location indicates the direct URL to use to access the resource + /// when content negotiation happened, without the need of further content + /// negotiation. Location is a header associated with the response, while + /// Content-Location is associated with the entity returned. + (Location, LOCATION, b"location"); + + /// Indicates the max number of intermediaries the request should be sent + /// through. + (MaxForwards, MAX_FORWARDS, b"max-forwards"); + + /// Indicates where a fetch originates from. + /// + /// It doesn't include any path information, but only the server name. It is + /// sent with CORS requests, as well as with POST requests. It is similar to + /// the Referer header, but, unlike this header, it doesn't disclose the + /// whole path. + (Origin, ORIGIN, b"origin"); + + /// HTTP/1.0 header usually used for backwards compatibility. + /// + /// The Pragma HTTP/1.0 general header is an implementation-specific header + /// that may have various effects along the request-response chain. It is + /// used for backwards compatibility with HTTP/1.0 caches where the + /// Cache-Control HTTP/1.1 header is not yet present. + (Pragma, PRAGMA, b"pragma"); + + /// Defines the authentication method that should be used to gain access to + /// a proxy. + /// + /// Unlike `www-authenticate`, the `proxy-authenticate` header field applies + /// only to the next outbound client on the response chain. This is because + /// only the client that chose a given proxy is likely to have the + /// credentials necessary for authentication. However, when multiple proxies + /// are used within the same administrative domain, such as office and + /// regional caching proxies within a large corporate network, it is common + /// for credentials to be generated by the user agent and passed through the + /// hierarchy until consumed. Hence, in such a configuration, it will appear + /// as if Proxy-Authenticate is being forwarded because each proxy will send + /// the same challenge set. + /// + /// The `proxy-authenticate` header is sent along with a `407 Proxy + /// Authentication Required`. + (ProxyAuthenticate, PROXY_AUTHENTICATE, b"proxy-authenticate"); + + /// Contains the credentials to authenticate a user agent to a proxy server. + /// + /// This header is usually included after the server has responded with a + /// 407 Proxy Authentication Required status and the Proxy-Authenticate + /// header. + (ProxyAuthorization, PROXY_AUTHORIZATION, b"proxy-authorization"); + + /// Associates a specific cryptographic public key with a certain server. + /// + /// This decreases the risk of MITM attacks with forged certificates. If one + /// or several keys are pinned and none of them are used by the server, the + /// browser will not accept the response as legitimate, and will not display + /// it. + (PublicKeyPins, PUBLIC_KEY_PINS, b"public-key-pins"); + + /// Sends reports of pinning violation to the report-uri specified in the + /// header. + /// + /// Unlike `Public-Key-Pins`, this header still allows browsers to connect + /// to the server if the pinning is violated. + (PublicKeyPinsReportOnly, PUBLIC_KEY_PINS_REPORT_ONLY, b"public-key-pins-report-only"); + + /// Indicates the part of a document that the server should return. + /// + /// Several parts can be requested with one Range header at once, and the + /// server may send back these ranges in a multipart document. If the server + /// sends back ranges, it uses the 206 Partial Content for the response. If + /// the ranges are invalid, the server returns the 416 Range Not Satisfiable + /// error. The server can also ignore the Range header and return the whole + /// document with a 200 status code. + (Range, RANGE, b"range"); + + /// Contains the address of the previous web page from which a link to the + /// currently requested page was followed. + /// + /// The Referer header allows servers to identify where people are visiting + /// them from and may use that data for analytics, logging, or optimized + /// caching, for example. + (Referer, REFERER, b"referer"); + + /// Governs which referrer information should be included with requests + /// made. + (ReferrerPolicy, REFERRER_POLICY, b"referrer-policy"); + + /// Informs the web browser that the current page or frame should be + /// refreshed. + (Refresh, REFRESH, b"refresh"); + + /// The Retry-After response HTTP header indicates how long the user agent + /// should wait before making a follow-up request. There are two main cases + /// this header is used: + /// + /// * When sent with a 503 (Service Unavailable) response, it indicates how + /// long the service is expected to be unavailable. + /// + /// * When sent with a redirect response, such as 301 (Moved Permanently), + /// it indicates the minimum time that the user agent is asked to wait + /// before issuing the redirected request. + (RetryAfter, RETRY_AFTER, b"retry-after"); + + /// The |Sec-WebSocket-Accept| header field is used in the WebSocket + /// opening handshake. It is sent from the server to the client to + /// confirm that the server is willing to initiate the WebSocket + /// connection. + (SecWebSocketAccept, SEC_WEBSOCKET_ACCEPT, b"sec-websocket-accept"); + + /// The |Sec-WebSocket-Extensions| header field is used in the WebSocket + /// opening handshake. It is initially sent from the client to the + /// server, and then subsequently sent from the server to the client, to + /// agree on a set of protocol-level extensions to use for the duration + /// of the connection. + (SecWebSocketExtensions, SEC_WEBSOCKET_EXTENSIONS, b"sec-websocket-extensions"); + + /// The |Sec-WebSocket-Key| header field is used in the WebSocket opening + /// handshake. It is sent from the client to the server to provide part + /// of the information used by the server to prove that it received a + /// valid WebSocket opening handshake. This helps ensure that the server + /// does not accept connections from non-WebSocket clients (e.g., HTTP + /// clients) that are being abused to send data to unsuspecting WebSocket + /// servers. + (SecWebSocketKey, SEC_WEBSOCKET_KEY, b"sec-websocket-key"); + + /// The |Sec-WebSocket-Protocol| header field is used in the WebSocket + /// opening handshake. It is sent from the client to the server and back + /// from the server to the client to confirm the subprotocol of the + /// connection. This enables scripts to both select a subprotocol and be + /// sure that the server agreed to serve that subprotocol. + (SecWebSocketProtocol, SEC_WEBSOCKET_PROTOCOL, b"sec-websocket-protocol"); + + /// The |Sec-WebSocket-Version| header field is used in the WebSocket + /// opening handshake. It is sent from the client to the server to + /// indicate the protocol version of the connection. This enables + /// servers to correctly interpret the opening handshake and subsequent + /// data being sent from the data, and close the connection if the server + /// cannot interpret that data in a safe manner. + (SecWebSocketVersion, SEC_WEBSOCKET_VERSION, b"sec-websocket-version"); + + /// Contains information about the software used by the origin server to + /// handle the request. + /// + /// Overly long and detailed Server values should be avoided as they + /// potentially reveal internal implementation details that might make it + /// (slightly) easier for attackers to find and exploit known security + /// holes. + (Server, SERVER, b"server"); + + /// Used to send cookies from the server to the user agent. + (SetCookie, SET_COOKIE, b"set-cookie"); + + /// Tells the client to communicate with HTTPS instead of using HTTP. + (StrictTransportSecurity, STRICT_TRANSPORT_SECURITY, b"strict-transport-security"); + + /// Informs the server of transfer encodings willing to be accepted as part + /// of the response. + /// + /// See also the Transfer-Encoding response header for more details on + /// transfer encodings. Note that chunked is always acceptable for HTTP/1.1 + /// recipients and you that don't have to specify "chunked" using the TE + /// header. However, it is useful for setting if the client is accepting + /// trailer fields in a chunked transfer coding using the "trailers" value. + (Te, TE, b"te"); + + /// Allows the sender to include additional fields at the end of chunked + /// messages. + (Trailer, TRAILER, b"trailer"); + + /// Specifies the form of encoding used to safely transfer the entity to the + /// client. + /// + /// `transfer-encoding` is a hop-by-hop header, that is applying to a + /// message between two nodes, not to a resource itself. Each segment of a + /// multi-node connection can use different `transfer-encoding` values. If + /// you want to compress data over the whole connection, use the end-to-end + /// header `content-encoding` header instead. + /// + /// When present on a response to a `HEAD` request that has no body, it + /// indicates the value that would have applied to the corresponding `GET` + /// message. + (TransferEncoding, TRANSFER_ENCODING, b"transfer-encoding"); + + /// Contains a string that allows identifying the requesting client's + /// software. + (UserAgent, USER_AGENT, b"user-agent"); + + /// Used as part of the exchange to upgrade the protocol. + (Upgrade, UPGRADE, b"upgrade"); + + /// Sends a signal to the server expressing the client’s preference for an + /// encrypted and authenticated response. + (UpgradeInsecureRequests, UPGRADE_INSECURE_REQUESTS, b"upgrade-insecure-requests"); + + /// Determines how to match future requests with cached responses. + /// + /// The `vary` HTTP response header determines how to match future request + /// headers to decide whether a cached response can be used rather than + /// requesting a fresh one from the origin server. It is used by the server + /// to indicate which headers it used when selecting a representation of a + /// resource in a content negotiation algorithm. + /// + /// The `vary` header should be set on a 304 Not Modified response exactly + /// like it would have been set on an equivalent 200 OK response. + (Vary, VARY, b"vary"); + + /// Added by proxies to track routing. + /// + /// The `via` general header is added by proxies, both forward and reverse + /// proxies, and can appear in the request headers and the response headers. + /// It is used for tracking message forwards, avoiding request loops, and + /// identifying the protocol capabilities of senders along the + /// request/response chain. + (Via, VIA, b"via"); + + /// General HTTP header contains information about possible problems with + /// the status of the message. + /// + /// More than one `warning` header may appear in a response. Warning header + /// fields can in general be applied to any message, however some warn-codes + /// are specific to caches and can only be applied to response messages. + (Warning, WARNING, b"warning"); + + /// Defines the authentication method that should be used to gain access to + /// a resource. + (WwwAuthenticate, WWW_AUTHENTICATE, b"www-authenticate"); + + /// Marker used by the server to indicate that the MIME types advertised in + /// the `content-type` headers should not be changed and be followed. + /// + /// This allows to opt-out of MIME type sniffing, or, in other words, it is + /// a way to say that the webmasters knew what they were doing. + /// + /// This header was introduced by Microsoft in IE 8 as a way for webmasters + /// to block content sniffing that was happening and could transform + /// non-executable MIME types into executable MIME types. Since then, other + /// browsers have introduced it, even if their MIME sniffing algorithms were + /// less aggressive. + /// + /// Site security testers usually expect this header to be set. + (XContentTypeOptions, X_CONTENT_TYPE_OPTIONS, b"x-content-type-options"); + + /// Controls DNS prefetching. + /// + /// The `x-dns-prefetch-control` HTTP response header controls DNS + /// prefetching, a feature by which browsers proactively perform domain name + /// resolution on both links that the user may choose to follow as well as + /// URLs for items referenced by the document, including images, CSS, + /// JavaScript, and so forth. + /// + /// This prefetching is performed in the background, so that the DNS is + /// likely to have been resolved by the time the referenced items are + /// needed. This reduces latency when the user clicks a link. + (XDnsPrefetchControl, X_DNS_PREFETCH_CONTROL, b"x-dns-prefetch-control"); + + /// Indicates whether or not a browser should be allowed to render a page in + /// a frame. + /// + /// Sites can use this to avoid clickjacking attacks, by ensuring that their + /// content is not embedded into other sites. + /// + /// The added security is only provided if the user accessing the document + /// is using a browser supporting `x-frame-options`. + (XFrameOptions, X_FRAME_OPTIONS, b"x-frame-options"); + + /// Stop pages from loading when an XSS attack is detected. + /// + /// The HTTP X-XSS-Protection response header is a feature of Internet + /// Explorer, Chrome and Safari that stops pages from loading when they + /// detect reflected cross-site scripting (XSS) attacks. Although these + /// protections are largely unnecessary in modern browsers when sites + /// implement a strong Content-Security-Policy that disables the use of + /// inline JavaScript ('unsafe-inline'), they can still provide protections + /// for users of older web browsers that don't yet support CSP. + (XXssProtection, X_XSS_PROTECTION, b"x-xss-protection"); +} + +/// Valid header name characters +/// +/// ```not_rust +/// field-name = token +/// separators = "(" | ")" | "<" | ">" | "@" +/// | "," | ";" | ":" | "\" | <"> +/// | "/" | "[" | "]" | "?" | "=" +/// | "{" | "}" | SP | HT +/// token = 1*tchar +/// tchar = "!" / "#" / "$" / "%" / "&" / "'" / "*" +/// / "+" / "-" / "." / "^" / "_" / "`" / "|" / "~" +/// / DIGIT / ALPHA +/// ; any VCHAR, except delimiters +/// ``` +// HEADER_CHARS maps every byte that is 128 or larger to 0 so everything that is +// mapped by HEADER_CHARS, maps to a valid single-byte UTF-8 codepoint. +const HEADER_CHARS: [u8; 256] = [ + // 0 1 2 3 4 5 6 7 8 9 + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // x + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 1x + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 2x + 0, 0, 0, b'!', b'"', b'#', b'$', b'%', b'&', b'\'', // 3x + 0, 0, b'*', b'+', 0, b'-', b'.', 0, b'0', b'1', // 4x + b'2', b'3', b'4', b'5', b'6', b'7', b'8', b'9', 0, 0, // 5x + 0, 0, 0, 0, 0, b'a', b'b', b'c', b'd', b'e', // 6x + b'f', b'g', b'h', b'i', b'j', b'k', b'l', b'm', b'n', b'o', // 7x + b'p', b'q', b'r', b's', b't', b'u', b'v', b'w', b'x', b'y', // 8x + b'z', 0, 0, 0, b'^', b'_', b'`', b'a', b'b', b'c', // 9x + b'd', b'e', b'f', b'g', b'h', b'i', b'j', b'k', b'l', b'm', // 10x + b'n', b'o', b'p', b'q', b'r', b's', b't', b'u', b'v', b'w', // 11x + b'x', b'y', b'z', 0, b'|', 0, b'~', 0, 0, 0, // 12x + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 13x + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 14x + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 15x + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 16x + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 17x + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 18x + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 19x + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 20x + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 21x + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 22x + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 23x + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 24x + 0, 0, 0, 0, 0, 0 // 25x +]; + +/// Valid header name characters for HTTP/2.0 and HTTP/3.0 +// HEADER_CHARS_H2 maps every byte that is 128 or larger to 0 so everything that is +// mapped by HEADER_CHARS_H2, maps to a valid single-byte UTF-8 codepoint. +const HEADER_CHARS_H2: [u8; 256] = [ + // 0 1 2 3 4 5 6 7 8 9 + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // x + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 1x + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 2x + 0, 0, 0, b'!', b'"', b'#', b'$', b'%', b'&', b'\'', // 3x + 0, 0, b'*', b'+', 0, b'-', b'.', 0, b'0', b'1', // 4x + b'2', b'3', b'4', b'5', b'6', b'7', b'8', b'9', 0, 0, // 5x + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 6x + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 7x + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 8x + 0, 0, 0, 0, b'^', b'_', b'`', b'a', b'b', b'c', // 9x + b'd', b'e', b'f', b'g', b'h', b'i', b'j', b'k', b'l', b'm', // 10x + b'n', b'o', b'p', b'q', b'r', b's', b't', b'u', b'v', b'w', // 11x + b'x', b'y', b'z', 0, b'|', 0, b'~', 0, 0, 0, // 12x + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 13x + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 14x + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 15x + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 16x + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 17x + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 18x + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 19x + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 20x + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 21x + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 22x + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 23x + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 24x + 0, 0, 0, 0, 0, 0 // 25x +]; + +fn parse_hdr<'a>( + data: &'a [u8], + b: &'a mut [MaybeUninit<u8>; SCRATCH_BUF_SIZE], + table: &[u8; 256], +) -> Result<HdrName<'a>, InvalidHeaderName> { + match data.len() { + 0 => Err(InvalidHeaderName::new()), + len @ 1..=SCRATCH_BUF_SIZE => { + // Read from data into the buffer - transforming using `table` as we go + data.iter() + .zip(b.iter_mut()) + .for_each(|(index, out)| *out = MaybeUninit::new(table[*index as usize])); + // Safety: len bytes of b were just initialized. + let name: &'a [u8] = unsafe { slice_assume_init(&b[0..len]) }; + match StandardHeader::from_bytes(name) { + Some(sh) => Ok(sh.into()), + None => { + if name.contains(&0) { + Err(InvalidHeaderName::new()) + } else { + Ok(HdrName::custom(name, true)) + } + } + } + } + SCRATCH_BUF_OVERFLOW..=super::MAX_HEADER_NAME_LEN => Ok(HdrName::custom(data, false)), + _ => Err(InvalidHeaderName::new()), + } +} + + + +impl<'a> From<StandardHeader> for HdrName<'a> { + fn from(hdr: StandardHeader) -> HdrName<'a> { + HdrName { inner: Repr::Standard(hdr) } + } +} + +impl HeaderName { + /// Converts a slice of bytes to an HTTP header name. + /// + /// This function normalizes the input. + pub fn from_bytes(src: &[u8]) -> Result<HeaderName, InvalidHeaderName> { + let mut buf = uninit_u8_array(); + // Precondition: HEADER_CHARS is a valid table for parse_hdr(). + match parse_hdr(src, &mut buf, &HEADER_CHARS)?.inner { + Repr::Standard(std) => Ok(std.into()), + Repr::Custom(MaybeLower { buf, lower: true }) => { + let buf = Bytes::copy_from_slice(buf); + // Safety: the invariant on MaybeLower ensures buf is valid UTF-8. + let val = unsafe { ByteStr::from_utf8_unchecked(buf) }; + Ok(Custom(val).into()) + } + Repr::Custom(MaybeLower { buf, lower: false }) => { + use bytes::{BufMut}; + let mut dst = BytesMut::with_capacity(buf.len()); + + for b in buf.iter() { + // HEADER_CHARS maps all bytes to valid single-byte UTF-8 + let b = HEADER_CHARS[*b as usize]; + + if b == 0 { + return Err(InvalidHeaderName::new()); + } + + dst.put_u8(b); + } + + // Safety: the loop above maps all bytes in buf to valid single byte + // UTF-8 before copying them into dst. This means that dst (and hence + // dst.freeze()) is valid UTF-8. + let val = unsafe { ByteStr::from_utf8_unchecked(dst.freeze()) }; + + Ok(Custom(val).into()) + } + } + } + + /// Converts a slice of bytes to an HTTP header name. + /// + /// This function expects the input to only contain lowercase characters. + /// This is useful when decoding HTTP/2.0 or HTTP/3.0 headers. Both + /// require that all headers be represented in lower case. + /// + /// # Examples + /// + /// ``` + /// # use http::header::*; + /// + /// // Parsing a lower case header + /// let hdr = HeaderName::from_lowercase(b"content-length").unwrap(); + /// assert_eq!(CONTENT_LENGTH, hdr); + /// + /// // Parsing a header that contains uppercase characters + /// assert!(HeaderName::from_lowercase(b"Content-Length").is_err()); + /// ``` + pub fn from_lowercase(src: &[u8]) -> Result<HeaderName, InvalidHeaderName> { + let mut buf = uninit_u8_array(); + // Precondition: HEADER_CHARS_H2 is a valid table for parse_hdr() + match parse_hdr(src, &mut buf, &HEADER_CHARS_H2)?.inner { + Repr::Standard(std) => Ok(std.into()), + Repr::Custom(MaybeLower { buf, lower: true }) => { + let buf = Bytes::copy_from_slice(buf); + // Safety: the invariant on MaybeLower ensures buf is valid UTF-8. + let val = unsafe { ByteStr::from_utf8_unchecked(buf) }; + Ok(Custom(val).into()) + } + Repr::Custom(MaybeLower { buf, lower: false }) => { + for &b in buf.iter() { + // HEADER_CHARS maps all bytes that are not valid single-byte + // UTF-8 to 0 so this check returns an error for invalid UTF-8. + if b != HEADER_CHARS[b as usize] { + return Err(InvalidHeaderName::new()); + } + } + + let buf = Bytes::copy_from_slice(buf); + // Safety: the loop above checks that each byte of buf (either + // version) is valid UTF-8. + let val = unsafe { ByteStr::from_utf8_unchecked(buf) }; + Ok(Custom(val).into()) + } + } + } + + /// Converts a static string to a HTTP header name. + /// + /// This function requires the static string to only contain lowercase + /// characters, numerals and symbols, as per the HTTP/2.0 specification + /// and header names internal representation within this library. + /// + /// # Panics + /// + /// This function panics when the static string is a invalid header. + /// + /// Until [Allow panicking in constants](https://github.com/rust-lang/rfcs/pull/2345) + /// makes its way into stable, the panic message at compile-time is + /// going to look cryptic, but should at least point at your header value: + /// + /// ```text + /// error: any use of this value will cause an error + /// --> http/src/header/name.rs:1241:13 + /// | + /// 1241 | ([] as [u8; 0])[0]; // Invalid header name + /// | ^^^^^^^^^^^^^^^^^^ + /// | | + /// | index out of bounds: the length is 0 but the index is 0 + /// | inside `http::HeaderName::from_static` at http/src/header/name.rs:1241:13 + /// | inside `INVALID_NAME` at src/main.rs:3:34 + /// | + /// ::: src/main.rs:3:1 + /// | + /// 3 | const INVALID_NAME: HeaderName = HeaderName::from_static("Capitalized"); + /// | ------------------------------------------------------------------------ + /// ``` + /// + /// # Examples + /// + /// ``` + /// # use http::header::*; + /// // Parsing a standard header + /// let hdr = HeaderName::from_static("content-length"); + /// assert_eq!(CONTENT_LENGTH, hdr); + /// + /// // Parsing a custom header + /// let CUSTOM_HEADER: &'static str = "custom-header"; + /// + /// let a = HeaderName::from_lowercase(b"custom-header").unwrap(); + /// let b = HeaderName::from_static(CUSTOM_HEADER); + /// assert_eq!(a, b); + /// ``` + /// + /// ```should_panic + /// # use http::header::*; + /// # + /// // Parsing a header that contains invalid symbols(s): + /// HeaderName::from_static("content{}{}length"); // This line panics! + /// + /// // Parsing a header that contains invalid uppercase characters. + /// let a = HeaderName::from_static("foobar"); + /// let b = HeaderName::from_static("FOOBAR"); // This line panics! + /// ``` + #[allow(unconditional_panic)] // required for the panic circumvention + pub const fn from_static(src: &'static str) -> HeaderName { + let name_bytes = src.as_bytes(); + if let Some(standard) = StandardHeader::from_bytes(name_bytes) { + return HeaderName{ + inner: Repr::Standard(standard), + }; + } + + if name_bytes.len() == 0 || name_bytes.len() > super::MAX_HEADER_NAME_LEN || { + let mut i = 0; + loop { + if i >= name_bytes.len() { + break false; + } else if HEADER_CHARS_H2[name_bytes[i] as usize] == 0 { + break true; + } + i += 1; + } + } { + ([] as [u8; 0])[0]; // Invalid header name + } + + HeaderName { + inner: Repr::Custom(Custom(ByteStr::from_static(src))) + } + } + + /// Returns a `str` representation of the header. + /// + /// The returned string will always be lower case. + #[inline] + pub fn as_str(&self) -> &str { + match self.inner { + Repr::Standard(v) => v.as_str(), + Repr::Custom(ref v) => &*v.0, + } + } + + pub(super) fn into_bytes(self) -> Bytes { + self.inner.into() + } +} + +impl FromStr for HeaderName { + type Err = InvalidHeaderName; + + fn from_str(s: &str) -> Result<HeaderName, InvalidHeaderName> { + HeaderName::from_bytes(s.as_bytes()).map_err(|_| InvalidHeaderName { _priv: () }) + } +} + +impl AsRef<str> for HeaderName { + fn as_ref(&self) -> &str { + self.as_str() + } +} + +impl AsRef<[u8]> for HeaderName { + fn as_ref(&self) -> &[u8] { + self.as_str().as_bytes() + } +} + +impl Borrow<str> for HeaderName { + fn borrow(&self) -> &str { + self.as_str() + } +} + +impl fmt::Debug for HeaderName { + fn fmt(&self, fmt: &mut fmt::Formatter<'_>) -> fmt::Result { + fmt::Debug::fmt(self.as_str(), fmt) + } +} + +impl fmt::Display for HeaderName { + fn fmt(&self, fmt: &mut fmt::Formatter<'_>) -> fmt::Result { + fmt::Display::fmt(self.as_str(), fmt) + } +} + +impl InvalidHeaderName { + fn new() -> InvalidHeaderName { + InvalidHeaderName { _priv: () } + } +} + +impl<'a> From<&'a HeaderName> for HeaderName { + fn from(src: &'a HeaderName) -> HeaderName { + src.clone() + } +} + +#[doc(hidden)] +impl<T> From<Repr<T>> for Bytes +where + T: Into<Bytes>, +{ + fn from(repr: Repr<T>) -> Bytes { + match repr { + Repr::Standard(header) => Bytes::from_static(header.as_str().as_bytes()), + Repr::Custom(header) => header.into(), + } + } +} + +impl From<Custom> for Bytes { + #[inline] + fn from(Custom(inner): Custom) -> Bytes { + Bytes::from(inner) + } +} + +impl<'a> TryFrom<&'a str> for HeaderName { + type Error = InvalidHeaderName; + #[inline] + fn try_from(s: &'a str) -> Result<Self, Self::Error> { + Self::from_bytes(s.as_bytes()) + } +} + +impl<'a> TryFrom<&'a String> for HeaderName { + type Error = InvalidHeaderName; + #[inline] + fn try_from(s: &'a String) -> Result<Self, Self::Error> { + Self::from_bytes(s.as_bytes()) + } +} + +impl<'a> TryFrom<&'a [u8]> for HeaderName { + type Error = InvalidHeaderName; + #[inline] + fn try_from(s: &'a [u8]) -> Result<Self, Self::Error> { + Self::from_bytes(s) + } +} + +impl TryFrom<String> for HeaderName { + type Error = InvalidHeaderName; + + #[inline] + fn try_from(s: String) -> Result<Self, Self::Error> { + Self::from_bytes(s.as_bytes()) + } +} + +impl TryFrom<Vec<u8>> for HeaderName { + type Error = InvalidHeaderName; + + #[inline] + fn try_from(vec: Vec<u8>) -> Result<Self, Self::Error> { + Self::from_bytes(&vec) + } +} + +#[doc(hidden)] +impl From<StandardHeader> for HeaderName { + fn from(src: StandardHeader) -> HeaderName { + HeaderName { + inner: Repr::Standard(src), + } + } +} + +#[doc(hidden)] +impl From<Custom> for HeaderName { + fn from(src: Custom) -> HeaderName { + HeaderName { + inner: Repr::Custom(src), + } + } +} + +impl<'a> PartialEq<&'a HeaderName> for HeaderName { + #[inline] + fn eq(&self, other: &&'a HeaderName) -> bool { + *self == **other + } +} + +impl<'a> PartialEq<HeaderName> for &'a HeaderName { + #[inline] + fn eq(&self, other: &HeaderName) -> bool { + *other == *self + } +} + +impl PartialEq<str> for HeaderName { + /// Performs a case-insensitive comparison of the string against the header + /// name + /// + /// # Examples + /// + /// ``` + /// use http::header::CONTENT_LENGTH; + /// + /// assert_eq!(CONTENT_LENGTH, "content-length"); + /// assert_eq!(CONTENT_LENGTH, "Content-Length"); + /// assert_ne!(CONTENT_LENGTH, "content length"); + /// ``` + #[inline] + fn eq(&self, other: &str) -> bool { + eq_ignore_ascii_case(self.as_ref(), other.as_bytes()) + } +} + +impl PartialEq<HeaderName> for str { + /// Performs a case-insensitive comparison of the string against the header + /// name + /// + /// # Examples + /// + /// ``` + /// use http::header::CONTENT_LENGTH; + /// + /// assert_eq!(CONTENT_LENGTH, "content-length"); + /// assert_eq!(CONTENT_LENGTH, "Content-Length"); + /// assert_ne!(CONTENT_LENGTH, "content length"); + /// ``` + #[inline] + fn eq(&self, other: &HeaderName) -> bool { + *other == *self + } +} + +impl<'a> PartialEq<&'a str> for HeaderName { + /// Performs a case-insensitive comparison of the string against the header + /// name + #[inline] + fn eq(&self, other: &&'a str) -> bool { + *self == **other + } +} + +impl<'a> PartialEq<HeaderName> for &'a str { + /// Performs a case-insensitive comparison of the string against the header + /// name + #[inline] + fn eq(&self, other: &HeaderName) -> bool { + *other == *self + } +} + +impl fmt::Debug for InvalidHeaderName { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + f.debug_struct("InvalidHeaderName") + // skip _priv noise + .finish() + } +} + +impl fmt::Display for InvalidHeaderName { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + f.write_str("invalid HTTP header name") + } +} + +impl Error for InvalidHeaderName {} + +// ===== HdrName ===== + +impl<'a> HdrName<'a> { + // Precondition: if lower then buf is valid UTF-8 + fn custom(buf: &'a [u8], lower: bool) -> HdrName<'a> { + HdrName { + // Invariant (on MaybeLower): follows from the precondition + inner: Repr::Custom(MaybeLower { + buf: buf, + lower: lower, + }), + } + } + + pub fn from_bytes<F, U>(hdr: &[u8], f: F) -> Result<U, InvalidHeaderName> + where F: FnOnce(HdrName<'_>) -> U, + { + let mut buf = uninit_u8_array(); + // Precondition: HEADER_CHARS is a valid table for parse_hdr(). + let hdr = parse_hdr(hdr, &mut buf, &HEADER_CHARS)?; + Ok(f(hdr)) + } + + pub fn from_static<F, U>(hdr: &'static str, f: F) -> U + where + F: FnOnce(HdrName<'_>) -> U, + { + let mut buf = uninit_u8_array(); + let hdr = + // Precondition: HEADER_CHARS is a valid table for parse_hdr(). + parse_hdr(hdr.as_bytes(), &mut buf, &HEADER_CHARS).expect("static str is invalid name"); + f(hdr) + } +} + +#[doc(hidden)] +impl<'a> From<HdrName<'a>> for HeaderName { + fn from(src: HdrName<'a>) -> HeaderName { + match src.inner { + Repr::Standard(s) => HeaderName { + inner: Repr::Standard(s), + }, + Repr::Custom(maybe_lower) => { + if maybe_lower.lower { + let buf = Bytes::copy_from_slice(&maybe_lower.buf[..]); + // Safety: the invariant on MaybeLower ensures buf is valid UTF-8. + let byte_str = unsafe { ByteStr::from_utf8_unchecked(buf) }; + + HeaderName { + inner: Repr::Custom(Custom(byte_str)), + } + } else { + use bytes::BufMut; + let mut dst = BytesMut::with_capacity(maybe_lower.buf.len()); + + for b in maybe_lower.buf.iter() { + // HEADER_CHARS maps each byte to a valid single-byte UTF-8 + // codepoint. + dst.put_u8(HEADER_CHARS[*b as usize]); + } + + // Safety: the loop above maps each byte of maybe_lower.buf to a + // valid single-byte UTF-8 codepoint before copying it into dst. + // dst (and hence dst.freeze()) is thus valid UTF-8. + let buf = unsafe { ByteStr::from_utf8_unchecked(dst.freeze()) }; + + HeaderName { + inner: Repr::Custom(Custom(buf)), + } + } + } + } + } +} + +#[doc(hidden)] +impl<'a> PartialEq<HdrName<'a>> for HeaderName { + #[inline] + fn eq(&self, other: &HdrName<'a>) -> bool { + match self.inner { + Repr::Standard(a) => match other.inner { + Repr::Standard(b) => a == b, + _ => false, + }, + Repr::Custom(Custom(ref a)) => match other.inner { + Repr::Custom(ref b) => { + if b.lower { + a.as_bytes() == b.buf + } else { + eq_ignore_ascii_case(a.as_bytes(), b.buf) + } + } + _ => false, + }, + } + } +} + +// ===== Custom ===== + +impl Hash for Custom { + #[inline] + fn hash<H: Hasher>(&self, hasher: &mut H) { + hasher.write(self.0.as_bytes()) + } +} + +// ===== MaybeLower ===== + +impl<'a> Hash for MaybeLower<'a> { + #[inline] + fn hash<H: Hasher>(&self, hasher: &mut H) { + if self.lower { + hasher.write(self.buf); + } else { + for &b in self.buf { + hasher.write(&[HEADER_CHARS[b as usize]]); + } + } + } +} + +// Assumes that the left hand side is already lower case +#[inline] +fn eq_ignore_ascii_case(lower: &[u8], s: &[u8]) -> bool { + if lower.len() != s.len() { + return false; + } + + lower.iter().zip(s).all(|(a, b)| { + *a == HEADER_CHARS[*b as usize] + }) +} + +// Utility functions for MaybeUninit<>. These are drawn from unstable API's on +// MaybeUninit<> itself. +const SCRATCH_BUF_SIZE: usize = 64; +const SCRATCH_BUF_OVERFLOW: usize = SCRATCH_BUF_SIZE + 1; + +fn uninit_u8_array() -> [MaybeUninit<u8>; SCRATCH_BUF_SIZE] { + let arr = MaybeUninit::<[MaybeUninit<u8>; SCRATCH_BUF_SIZE]>::uninit(); + // Safety: assume_init() is claiming that an array of MaybeUninit<> + // has been initilized, but MaybeUninit<>'s do not require initilizaton. + unsafe { arr.assume_init() } +} + +// Assuming all the elements are initilized, get a slice of them. +// +// Safety: All elements of `slice` must be initilized to prevent +// undefined behavior. +unsafe fn slice_assume_init<T>(slice: &[MaybeUninit<T>]) -> &[T] { + &*(slice as *const [MaybeUninit<T>] as *const [T]) +} + +#[cfg(test)] +mod tests { + use super::*; + use self::StandardHeader::Vary; + + #[test] + fn test_bounds() { + fn check_bounds<T: Sync + Send>() {} + check_bounds::<HeaderName>(); + } + + #[test] + fn test_parse_invalid_headers() { + for i in 0..128 { + let hdr = vec![1u8; i]; + assert!(HeaderName::from_bytes(&hdr).is_err(), "{} invalid header chars did not fail", i); + } + } + + const ONE_TOO_LONG: &[u8] = &[b'a'; super::super::MAX_HEADER_NAME_LEN+1]; + + #[test] + fn test_invalid_name_lengths() { + assert!( + HeaderName::from_bytes(&[]).is_err(), + "zero-length header name is an error", + ); + + let long = &ONE_TOO_LONG[0..super::super::MAX_HEADER_NAME_LEN]; + + let long_str = std::str::from_utf8(long).unwrap(); + assert_eq!(HeaderName::from_static(long_str), long_str); // shouldn't panic! + + assert!( + HeaderName::from_bytes(long).is_ok(), + "max header name length is ok", + ); + assert!( + HeaderName::from_bytes(ONE_TOO_LONG).is_err(), + "longer than max header name length is an error", + ); + } + + #[test] + #[should_panic] + fn test_static_invalid_name_lengths() { + // Safety: ONE_TOO_LONG contains only the UTF-8 safe, single-byte codepoint b'a'. + let _ = HeaderName::from_static(unsafe { std::str::from_utf8_unchecked(ONE_TOO_LONG) }); + } + + #[test] + fn test_from_hdr_name() { + use self::StandardHeader::Vary; + + let name = HeaderName::from(HdrName { + inner: Repr::Standard(Vary), + }); + + assert_eq!(name.inner, Repr::Standard(Vary)); + + let name = HeaderName::from(HdrName { + inner: Repr::Custom(MaybeLower { + buf: b"hello-world", + lower: true, + }), + }); + + assert_eq!(name.inner, Repr::Custom(Custom(ByteStr::from_static("hello-world")))); + + let name = HeaderName::from(HdrName { + inner: Repr::Custom(MaybeLower { + buf: b"Hello-World", + lower: false, + }), + }); + + assert_eq!(name.inner, Repr::Custom(Custom(ByteStr::from_static("hello-world")))); + } + + #[test] + fn test_eq_hdr_name() { + use self::StandardHeader::Vary; + + let a = HeaderName { inner: Repr::Standard(Vary) }; + let b = HdrName { inner: Repr::Standard(Vary) }; + + assert_eq!(a, b); + + let a = HeaderName { inner: Repr::Custom(Custom(ByteStr::from_static("vaary"))) }; + assert_ne!(a, b); + + let b = HdrName { inner: Repr::Custom(MaybeLower { + buf: b"vaary", + lower: true, + })}; + + assert_eq!(a, b); + + let b = HdrName { inner: Repr::Custom(MaybeLower { + buf: b"vaary", + lower: false, + })}; + + assert_eq!(a, b); + + let b = HdrName { inner: Repr::Custom(MaybeLower { + buf: b"VAARY", + lower: false, + })}; + + assert_eq!(a, b); + + let a = HeaderName { inner: Repr::Standard(Vary) }; + assert_ne!(a, b); + } + + #[test] + fn test_from_static_std() { + let a = HeaderName { inner: Repr::Standard(Vary) }; + + let b = HeaderName::from_static("vary"); + assert_eq!(a, b); + + let b = HeaderName::from_static("vaary"); + assert_ne!(a, b); + } + + #[test] + #[should_panic] + fn test_from_static_std_uppercase() { + HeaderName::from_static("Vary"); + } + + #[test] + #[should_panic] + fn test_from_static_std_symbol() { + HeaderName::from_static("vary{}"); + } + + // MaybeLower { lower: true } + #[test] + fn test_from_static_custom_short() { + let a = HeaderName { inner: Repr::Custom(Custom(ByteStr::from_static("customheader"))) }; + let b = HeaderName::from_static("customheader"); + assert_eq!(a, b); + } + + #[test] + #[should_panic] + fn test_from_static_custom_short_uppercase() { + HeaderName::from_static("custom header"); + } + + #[test] + #[should_panic] + fn test_from_static_custom_short_symbol() { + HeaderName::from_static("CustomHeader"); + } + + // MaybeLower { lower: false } + #[test] + fn test_from_static_custom_long() { + let a = HeaderName { inner: Repr::Custom(Custom(ByteStr::from_static( + "longer-than-63--thisheaderislongerthansixtythreecharactersandthushandleddifferent" + ))) }; + let b = HeaderName::from_static( + "longer-than-63--thisheaderislongerthansixtythreecharactersandthushandleddifferent" + ); + assert_eq!(a, b); + } + + #[test] + #[should_panic] + fn test_from_static_custom_long_uppercase() { + HeaderName::from_static( + "Longer-Than-63--ThisHeaderIsLongerThanSixtyThreeCharactersAndThusHandledDifferent" + ); + } + + #[test] + #[should_panic] + fn test_from_static_custom_long_symbol() { + HeaderName::from_static( + "longer-than-63--thisheader{}{}{}{}islongerthansixtythreecharactersandthushandleddifferent" + ); + } + + #[test] + fn test_from_static_custom_single_char() { + let a = HeaderName { inner: Repr::Custom(Custom(ByteStr::from_static("a"))) }; + let b = HeaderName::from_static("a"); + assert_eq!(a, b); + } + + #[test] + #[should_panic] + fn test_from_static_empty() { + HeaderName::from_static(""); + } + + #[test] + fn test_all_tokens() { + HeaderName::from_static("!#$%&'*+-.^_`|~0123456789abcdefghijklmnopqrstuvwxyz"); + } +} diff --git a/third_party/rust/http/src/header/value.rs b/third_party/rust/http/src/header/value.rs new file mode 100644 index 0000000000..bf05f16f4e --- /dev/null +++ b/third_party/rust/http/src/header/value.rs @@ -0,0 +1,795 @@ +use bytes::{Bytes, BytesMut}; + +use std::convert::TryFrom; +use std::error::Error; +use std::fmt::Write; +use std::str::FromStr; +use std::{cmp, fmt, mem, str}; + +use crate::header::name::HeaderName; + +/// Represents an HTTP header field value. +/// +/// In practice, HTTP header field values are usually valid ASCII. However, the +/// HTTP spec allows for a header value to contain opaque bytes as well. In this +/// case, the header field value is not able to be represented as a string. +/// +/// To handle this, the `HeaderValue` is useable as a type and can be compared +/// with strings and implements `Debug`. A `to_str` fn is provided that returns +/// an `Err` if the header value contains non visible ascii characters. +#[derive(Clone, Hash)] +pub struct HeaderValue { + inner: Bytes, + is_sensitive: bool, +} + +/// A possible error when converting a `HeaderValue` from a string or byte +/// slice. +pub struct InvalidHeaderValue { + _priv: (), +} + +/// A possible error when converting a `HeaderValue` to a string representation. +/// +/// Header field values may contain opaque bytes, in which case it is not +/// possible to represent the value as a string. +#[derive(Debug)] +pub struct ToStrError { + _priv: (), +} + +impl HeaderValue { + /// Convert a static string to a `HeaderValue`. + /// + /// This function will not perform any copying, however the string is + /// checked to ensure that no invalid characters are present. Only visible + /// ASCII characters (32-127) are permitted. + /// + /// # Panics + /// + /// This function panics if the argument contains invalid header value + /// characters. + /// + /// Until [Allow panicking in constants](https://github.com/rust-lang/rfcs/pull/2345) + /// makes its way into stable, the panic message at compile-time is + /// going to look cryptic, but should at least point at your header value: + /// + /// ```text + /// error: any use of this value will cause an error + /// --> http/src/header/value.rs:67:17 + /// | + /// 67 | ([] as [u8; 0])[0]; // Invalid header value + /// | ^^^^^^^^^^^^^^^^^^ + /// | | + /// | index out of bounds: the length is 0 but the index is 0 + /// | inside `HeaderValue::from_static` at http/src/header/value.rs:67:17 + /// | inside `INVALID_HEADER` at src/main.rs:73:33 + /// | + /// ::: src/main.rs:73:1 + /// | + /// 73 | const INVALID_HEADER: HeaderValue = HeaderValue::from_static("жsome value"); + /// | ---------------------------------------------------------------------------- + /// ``` + /// + /// # Examples + /// + /// ``` + /// # use http::header::HeaderValue; + /// let val = HeaderValue::from_static("hello"); + /// assert_eq!(val, "hello"); + /// ``` + #[inline] + #[allow(unconditional_panic)] // required for the panic circumvention + pub const fn from_static(src: &'static str) -> HeaderValue { + let bytes = src.as_bytes(); + let mut i = 0; + while i < bytes.len() { + if !is_visible_ascii(bytes[i]) { + ([] as [u8; 0])[0]; // Invalid header value + } + i += 1; + } + + HeaderValue { + inner: Bytes::from_static(bytes), + is_sensitive: false, + } + } + + /// Attempt to convert a string to a `HeaderValue`. + /// + /// If the argument contains invalid header value characters, an error is + /// returned. Only visible ASCII characters (32-127) are permitted. Use + /// `from_bytes` to create a `HeaderValue` that includes opaque octets + /// (128-255). + /// + /// This function is intended to be replaced in the future by a `TryFrom` + /// implementation once the trait is stabilized in std. + /// + /// # Examples + /// + /// ``` + /// # use http::header::HeaderValue; + /// let val = HeaderValue::from_str("hello").unwrap(); + /// assert_eq!(val, "hello"); + /// ``` + /// + /// An invalid value + /// + /// ``` + /// # use http::header::HeaderValue; + /// let val = HeaderValue::from_str("\n"); + /// assert!(val.is_err()); + /// ``` + #[inline] + pub fn from_str(src: &str) -> Result<HeaderValue, InvalidHeaderValue> { + HeaderValue::try_from_generic(src, |s| Bytes::copy_from_slice(s.as_bytes())) + } + + /// Converts a HeaderName into a HeaderValue + /// + /// Since every valid HeaderName is a valid HeaderValue this is done infallibly. + /// + /// # Examples + /// + /// ``` + /// # use http::header::{HeaderValue, HeaderName}; + /// # use http::header::ACCEPT; + /// let val = HeaderValue::from_name(ACCEPT); + /// assert_eq!(val, HeaderValue::from_bytes(b"accept").unwrap()); + /// ``` + #[inline] + pub fn from_name(name: HeaderName) -> HeaderValue { + name.into() + } + + /// Attempt to convert a byte slice to a `HeaderValue`. + /// + /// If the argument contains invalid header value bytes, an error is + /// returned. Only byte values between 32 and 255 (inclusive) are permitted, + /// excluding byte 127 (DEL). + /// + /// This function is intended to be replaced in the future by a `TryFrom` + /// implementation once the trait is stabilized in std. + /// + /// # Examples + /// + /// ``` + /// # use http::header::HeaderValue; + /// let val = HeaderValue::from_bytes(b"hello\xfa").unwrap(); + /// assert_eq!(val, &b"hello\xfa"[..]); + /// ``` + /// + /// An invalid value + /// + /// ``` + /// # use http::header::HeaderValue; + /// let val = HeaderValue::from_bytes(b"\n"); + /// assert!(val.is_err()); + /// ``` + #[inline] + pub fn from_bytes(src: &[u8]) -> Result<HeaderValue, InvalidHeaderValue> { + HeaderValue::try_from_generic(src, Bytes::copy_from_slice) + } + + /// Attempt to convert a `Bytes` buffer to a `HeaderValue`. + /// + /// This will try to prevent a copy if the type passed is the type used + /// internally, and will copy the data if it is not. + pub fn from_maybe_shared<T>(src: T) -> Result<HeaderValue, InvalidHeaderValue> + where + T: AsRef<[u8]> + 'static, + { + if_downcast_into!(T, Bytes, src, { + return HeaderValue::from_shared(src); + }); + + HeaderValue::from_bytes(src.as_ref()) + } + + /// Convert a `Bytes` directly into a `HeaderValue` without validating. + /// + /// This function does NOT validate that illegal bytes are not contained + /// within the buffer. + pub unsafe fn from_maybe_shared_unchecked<T>(src: T) -> HeaderValue + where + T: AsRef<[u8]> + 'static, + { + if cfg!(debug_assertions) { + match HeaderValue::from_maybe_shared(src) { + Ok(val) => val, + Err(_err) => { + panic!("HeaderValue::from_maybe_shared_unchecked() with invalid bytes"); + } + } + } else { + + if_downcast_into!(T, Bytes, src, { + return HeaderValue { + inner: src, + is_sensitive: false, + }; + }); + + let src = Bytes::copy_from_slice(src.as_ref()); + HeaderValue { + inner: src, + is_sensitive: false, + } + } + } + + fn from_shared(src: Bytes) -> Result<HeaderValue, InvalidHeaderValue> { + HeaderValue::try_from_generic(src, std::convert::identity) + } + + fn try_from_generic<T: AsRef<[u8]>, F: FnOnce(T) -> Bytes>(src: T, into: F) -> Result<HeaderValue, InvalidHeaderValue> { + for &b in src.as_ref() { + if !is_valid(b) { + return Err(InvalidHeaderValue { _priv: () }); + } + } + Ok(HeaderValue { + inner: into(src), + is_sensitive: false, + }) + } + + /// Yields a `&str` slice if the `HeaderValue` only contains visible ASCII + /// chars. + /// + /// This function will perform a scan of the header value, checking all the + /// characters. + /// + /// # Examples + /// + /// ``` + /// # use http::header::HeaderValue; + /// let val = HeaderValue::from_static("hello"); + /// assert_eq!(val.to_str().unwrap(), "hello"); + /// ``` + pub fn to_str(&self) -> Result<&str, ToStrError> { + let bytes = self.as_ref(); + + for &b in bytes { + if !is_visible_ascii(b) { + return Err(ToStrError { _priv: () }); + } + } + + unsafe { Ok(str::from_utf8_unchecked(bytes)) } + } + + /// Returns the length of `self`. + /// + /// This length is in bytes. + /// + /// # Examples + /// + /// ``` + /// # use http::header::HeaderValue; + /// let val = HeaderValue::from_static("hello"); + /// assert_eq!(val.len(), 5); + /// ``` + #[inline] + pub fn len(&self) -> usize { + self.as_ref().len() + } + + /// Returns true if the `HeaderValue` has a length of zero bytes. + /// + /// # Examples + /// + /// ``` + /// # use http::header::HeaderValue; + /// let val = HeaderValue::from_static(""); + /// assert!(val.is_empty()); + /// + /// let val = HeaderValue::from_static("hello"); + /// assert!(!val.is_empty()); + /// ``` + #[inline] + pub fn is_empty(&self) -> bool { + self.len() == 0 + } + + /// Converts a `HeaderValue` to a byte slice. + /// + /// # Examples + /// + /// ``` + /// # use http::header::HeaderValue; + /// let val = HeaderValue::from_static("hello"); + /// assert_eq!(val.as_bytes(), b"hello"); + /// ``` + #[inline] + pub fn as_bytes(&self) -> &[u8] { + self.as_ref() + } + + /// Mark that the header value represents sensitive information. + /// + /// # Examples + /// + /// ``` + /// # use http::header::HeaderValue; + /// let mut val = HeaderValue::from_static("my secret"); + /// + /// val.set_sensitive(true); + /// assert!(val.is_sensitive()); + /// + /// val.set_sensitive(false); + /// assert!(!val.is_sensitive()); + /// ``` + #[inline] + pub fn set_sensitive(&mut self, val: bool) { + self.is_sensitive = val; + } + + /// Returns `true` if the value represents sensitive data. + /// + /// Sensitive data could represent passwords or other data that should not + /// be stored on disk or in memory. By marking header values as sensitive, + /// components using this crate can be instructed to treat them with special + /// care for security reasons. For example, caches can avoid storing + /// sensitive values, and HPACK encoders used by HTTP/2.0 implementations + /// can choose not to compress them. + /// + /// Additionally, sensitive values will be masked by the `Debug` + /// implementation of `HeaderValue`. + /// + /// Note that sensitivity is not factored into equality or ordering. + /// + /// # Examples + /// + /// ``` + /// # use http::header::HeaderValue; + /// let mut val = HeaderValue::from_static("my secret"); + /// + /// val.set_sensitive(true); + /// assert!(val.is_sensitive()); + /// + /// val.set_sensitive(false); + /// assert!(!val.is_sensitive()); + /// ``` + #[inline] + pub fn is_sensitive(&self) -> bool { + self.is_sensitive + } +} + +impl AsRef<[u8]> for HeaderValue { + #[inline] + fn as_ref(&self) -> &[u8] { + self.inner.as_ref() + } +} + +impl fmt::Debug for HeaderValue { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + if self.is_sensitive { + f.write_str("Sensitive") + } else { + f.write_str("\"")?; + let mut from = 0; + let bytes = self.as_bytes(); + for (i, &b) in bytes.iter().enumerate() { + if !is_visible_ascii(b) || b == b'"' { + if from != i { + f.write_str(unsafe { str::from_utf8_unchecked(&bytes[from..i]) })?; + } + if b == b'"' { + f.write_str("\\\"")?; + } else { + write!(f, "\\x{:x}", b)?; + } + from = i + 1; + } + } + + f.write_str(unsafe { str::from_utf8_unchecked(&bytes[from..]) })?; + f.write_str("\"") + } + } +} + +impl From<HeaderName> for HeaderValue { + #[inline] + fn from(h: HeaderName) -> HeaderValue { + HeaderValue { + inner: h.into_bytes(), + is_sensitive: false, + } + } +} + +macro_rules! from_integers { + ($($name:ident: $t:ident => $max_len:expr),*) => {$( + impl From<$t> for HeaderValue { + fn from(num: $t) -> HeaderValue { + let mut buf = if mem::size_of::<BytesMut>() - 1 < $max_len { + // On 32bit platforms, BytesMut max inline size + // is 15 bytes, but the $max_len could be bigger. + // + // The likelihood of the number *actually* being + // that big is very small, so only allocate + // if the number needs that space. + // + // The largest decimal number in 15 digits: + // It wold be 10.pow(15) - 1, but this is a constant + // version. + if num as u64 > 999_999_999_999_999_999 { + BytesMut::with_capacity($max_len) + } else { + // fits inline... + BytesMut::new() + } + } else { + // full value fits inline, so don't allocate! + BytesMut::new() + }; + let _ = buf.write_str(::itoa::Buffer::new().format(num)); + HeaderValue { + inner: buf.freeze(), + is_sensitive: false, + } + } + } + + #[test] + fn $name() { + let n: $t = 55; + let val = HeaderValue::from(n); + assert_eq!(val, &n.to_string()); + + let n = ::std::$t::MAX; + let val = HeaderValue::from(n); + assert_eq!(val, &n.to_string()); + } + )*}; +} + +from_integers! { + // integer type => maximum decimal length + + // u8 purposely left off... HeaderValue::from(b'3') could be confusing + from_u16: u16 => 5, + from_i16: i16 => 6, + from_u32: u32 => 10, + from_i32: i32 => 11, + from_u64: u64 => 20, + from_i64: i64 => 20 +} + +#[cfg(target_pointer_width = "16")] +from_integers! { + from_usize: usize => 5, + from_isize: isize => 6 +} + +#[cfg(target_pointer_width = "32")] +from_integers! { + from_usize: usize => 10, + from_isize: isize => 11 +} + +#[cfg(target_pointer_width = "64")] +from_integers! { + from_usize: usize => 20, + from_isize: isize => 20 +} + +#[cfg(test)] +mod from_header_name_tests { + use super::*; + use crate::header::map::HeaderMap; + use crate::header::name; + + #[test] + fn it_can_insert_header_name_as_header_value() { + let mut map = HeaderMap::new(); + map.insert(name::UPGRADE, name::SEC_WEBSOCKET_PROTOCOL.into()); + map.insert( + name::ACCEPT, + name::HeaderName::from_bytes(b"hello-world").unwrap().into(), + ); + + assert_eq!( + map.get(name::UPGRADE).unwrap(), + HeaderValue::from_bytes(b"sec-websocket-protocol").unwrap() + ); + + assert_eq!( + map.get(name::ACCEPT).unwrap(), + HeaderValue::from_bytes(b"hello-world").unwrap() + ); + } +} + +impl FromStr for HeaderValue { + type Err = InvalidHeaderValue; + + #[inline] + fn from_str(s: &str) -> Result<HeaderValue, Self::Err> { + HeaderValue::from_str(s) + } +} + +impl<'a> From<&'a HeaderValue> for HeaderValue { + #[inline] + fn from(t: &'a HeaderValue) -> Self { + t.clone() + } +} + +impl<'a> TryFrom<&'a str> for HeaderValue { + type Error = InvalidHeaderValue; + + #[inline] + fn try_from(t: &'a str) -> Result<Self, Self::Error> { + t.parse() + } +} + +impl<'a> TryFrom<&'a String> for HeaderValue { + type Error = InvalidHeaderValue; + #[inline] + fn try_from(s: &'a String) -> Result<Self, Self::Error> { + Self::from_bytes(s.as_bytes()) + } +} + +impl<'a> TryFrom<&'a [u8]> for HeaderValue { + type Error = InvalidHeaderValue; + + #[inline] + fn try_from(t: &'a [u8]) -> Result<Self, Self::Error> { + HeaderValue::from_bytes(t) + } +} + +impl TryFrom<String> for HeaderValue { + type Error = InvalidHeaderValue; + + #[inline] + fn try_from(t: String) -> Result<Self, Self::Error> { + HeaderValue::from_shared(t.into()) + } +} + +impl TryFrom<Vec<u8>> for HeaderValue { + type Error = InvalidHeaderValue; + + #[inline] + fn try_from(vec: Vec<u8>) -> Result<Self, Self::Error> { + HeaderValue::from_shared(vec.into()) + } +} + +#[cfg(test)] +mod try_from_header_name_tests { + use super::*; + use crate::header::name; + + #[test] + fn it_converts_using_try_from() { + assert_eq!( + HeaderValue::try_from(name::UPGRADE).unwrap(), + HeaderValue::from_bytes(b"upgrade").unwrap() + ); + } +} + +const fn is_visible_ascii(b: u8) -> bool { + b >= 32 && b < 127 || b == b'\t' +} + +#[inline] +fn is_valid(b: u8) -> bool { + b >= 32 && b != 127 || b == b'\t' +} + +impl fmt::Debug for InvalidHeaderValue { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + f.debug_struct("InvalidHeaderValue") + // skip _priv noise + .finish() + } +} + +impl fmt::Display for InvalidHeaderValue { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + f.write_str("failed to parse header value") + } +} + +impl Error for InvalidHeaderValue {} + +impl fmt::Display for ToStrError { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + f.write_str("failed to convert header to a str") + } +} + +impl Error for ToStrError {} + +// ===== PartialEq / PartialOrd ===== + +impl PartialEq for HeaderValue { + #[inline] + fn eq(&self, other: &HeaderValue) -> bool { + self.inner == other.inner + } +} + +impl Eq for HeaderValue {} + +impl PartialOrd for HeaderValue { + #[inline] + fn partial_cmp(&self, other: &HeaderValue) -> Option<cmp::Ordering> { + self.inner.partial_cmp(&other.inner) + } +} + +impl Ord for HeaderValue { + #[inline] + fn cmp(&self, other: &Self) -> cmp::Ordering { + self.inner.cmp(&other.inner) + } +} + +impl PartialEq<str> for HeaderValue { + #[inline] + fn eq(&self, other: &str) -> bool { + self.inner == other.as_bytes() + } +} + +impl PartialEq<[u8]> for HeaderValue { + #[inline] + fn eq(&self, other: &[u8]) -> bool { + self.inner == other + } +} + +impl PartialOrd<str> for HeaderValue { + #[inline] + fn partial_cmp(&self, other: &str) -> Option<cmp::Ordering> { + (*self.inner).partial_cmp(other.as_bytes()) + } +} + +impl PartialOrd<[u8]> for HeaderValue { + #[inline] + fn partial_cmp(&self, other: &[u8]) -> Option<cmp::Ordering> { + (*self.inner).partial_cmp(other) + } +} + +impl PartialEq<HeaderValue> for str { + #[inline] + fn eq(&self, other: &HeaderValue) -> bool { + *other == *self + } +} + +impl PartialEq<HeaderValue> for [u8] { + #[inline] + fn eq(&self, other: &HeaderValue) -> bool { + *other == *self + } +} + +impl PartialOrd<HeaderValue> for str { + #[inline] + fn partial_cmp(&self, other: &HeaderValue) -> Option<cmp::Ordering> { + self.as_bytes().partial_cmp(other.as_bytes()) + } +} + +impl PartialOrd<HeaderValue> for [u8] { + #[inline] + fn partial_cmp(&self, other: &HeaderValue) -> Option<cmp::Ordering> { + self.partial_cmp(other.as_bytes()) + } +} + +impl PartialEq<String> for HeaderValue { + #[inline] + fn eq(&self, other: &String) -> bool { + *self == &other[..] + } +} + +impl PartialOrd<String> for HeaderValue { + #[inline] + fn partial_cmp(&self, other: &String) -> Option<cmp::Ordering> { + self.inner.partial_cmp(other.as_bytes()) + } +} + +impl PartialEq<HeaderValue> for String { + #[inline] + fn eq(&self, other: &HeaderValue) -> bool { + *other == *self + } +} + +impl PartialOrd<HeaderValue> for String { + #[inline] + fn partial_cmp(&self, other: &HeaderValue) -> Option<cmp::Ordering> { + self.as_bytes().partial_cmp(other.as_bytes()) + } +} + +impl<'a> PartialEq<HeaderValue> for &'a HeaderValue { + #[inline] + fn eq(&self, other: &HeaderValue) -> bool { + **self == *other + } +} + +impl<'a> PartialOrd<HeaderValue> for &'a HeaderValue { + #[inline] + fn partial_cmp(&self, other: &HeaderValue) -> Option<cmp::Ordering> { + (**self).partial_cmp(other) + } +} + +impl<'a, T: ?Sized> PartialEq<&'a T> for HeaderValue +where + HeaderValue: PartialEq<T>, +{ + #[inline] + fn eq(&self, other: &&'a T) -> bool { + *self == **other + } +} + +impl<'a, T: ?Sized> PartialOrd<&'a T> for HeaderValue +where + HeaderValue: PartialOrd<T>, +{ + #[inline] + fn partial_cmp(&self, other: &&'a T) -> Option<cmp::Ordering> { + self.partial_cmp(*other) + } +} + +impl<'a> PartialEq<HeaderValue> for &'a str { + #[inline] + fn eq(&self, other: &HeaderValue) -> bool { + *other == *self + } +} + +impl<'a> PartialOrd<HeaderValue> for &'a str { + #[inline] + fn partial_cmp(&self, other: &HeaderValue) -> Option<cmp::Ordering> { + self.as_bytes().partial_cmp(other.as_bytes()) + } +} + +#[test] +fn test_try_from() { + HeaderValue::try_from(vec![127]).unwrap_err(); +} + +#[test] +fn test_debug() { + let cases = &[ + ("hello", "\"hello\""), + ("hello \"world\"", "\"hello \\\"world\\\"\""), + ("\u{7FFF}hello", "\"\\xe7\\xbf\\xbfhello\""), + ]; + + for &(value, expected) in cases { + let val = HeaderValue::from_bytes(value.as_bytes()).unwrap(); + let actual = format!("{:?}", val); + assert_eq!(expected, actual); + } + + let mut sensitive = HeaderValue::from_static("password"); + sensitive.set_sensitive(true); + assert_eq!("Sensitive", format!("{:?}", sensitive)); +} |