diff options
Diffstat (limited to 'third_party/rust/debugid/src/lib.rs')
| -rw-r--r-- | third_party/rust/debugid/src/lib.rs | 543 |
1 files changed, 543 insertions, 0 deletions
diff --git a/third_party/rust/debugid/src/lib.rs b/third_party/rust/debugid/src/lib.rs new file mode 100644 index 0000000000..7022051a41 --- /dev/null +++ b/third_party/rust/debugid/src/lib.rs @@ -0,0 +1,543 @@ +//! This crate provides types for identifiers of object files, such as executables, dynamic +//! libraries or debug companion files. The concept originates in Google Breakpad and defines two +//! types: +//! +//! - [`CodeId`]: Identifies the file containing source code, i.e. the actual library or +//! executable. The identifier is platform dependent and implementation defined. Thus, there is +//! no canonical representation. +//! - [`DebugId`]: Identifies a debug information file, which may or may not use information from +//! the Code ID. The contents are also implementation defined, but as opposed to `CodeId`, the +//! structure is streamlined across platforms. It is also guaranteed to be 32 bytes in size. +//! +//! [`CodeId`]: struct.CodeId.html +//! [`DebugId`]: struct.DebugId.html + +#![warn(missing_docs)] + +use std::error; +use std::fmt; +use std::fmt::Write; +use std::str; + +use uuid::{Bytes, Uuid}; + +/// Indicates an error parsing a [`DebugId`](struct.DebugId.html). +#[derive(Clone, Copy, Debug, Eq, PartialEq)] +pub struct ParseDebugIdError; + +impl error::Error for ParseDebugIdError {} + +impl fmt::Display for ParseDebugIdError { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + write!(f, "invalid debug identifier") + } +} + +#[derive(Clone, Copy, Debug)] +struct ParseOptions { + allow_hyphens: bool, + require_appendix: bool, + allow_tail: bool, +} + +/// Unique identifier for debug information files and their debug information. +/// +/// This type is analogous to [`CodeId`], except that it identifies a debug file instead of the +/// actual library or executable. One some platforms, a `DebugId` is an alias for a `CodeId` but the +/// exact rules around this are complex. On Windows, the identifiers are completely different and +/// refer to separate files. +/// +/// The string representation must be between 33 and 40 characters long and consist of: +/// +/// 1. 36 character hyphenated hex representation of the UUID field +/// 2. 1-16 character lowercase hex representation of the u32 appendix +/// +/// The debug identifier is compatible to Google Breakpad. Use [`DebugId::breakpad`] to get a +/// breakpad string representation of this debug identifier. +/// +/// There is one exception to this: for the old PDB 2.0 format the debug identifier consists +/// of only a 32-bit integer + age resulting in a string representation of between 9 and 16 +/// hex characters. +/// +/// # Example +/// +/// ``` +/// # extern crate debugid; +/// use std::str::FromStr; +/// use debugid::DebugId; +/// +/// # fn foo() -> Result<(), ::debugid::ParseDebugIdError> { +/// let id = DebugId::from_str("dfb8e43a-f242-3d73-a453-aeb6a777ef75-a")?; +/// assert_eq!("dfb8e43a-f242-3d73-a453-aeb6a777ef75-a".to_string(), id.to_string()); +/// # Ok(()) +/// # } +/// +/// # fn main() { foo().unwrap() } +/// ``` +/// +/// # In-memory representation +/// +/// The in-memory representation takes up 32 bytes and can be directly written to storage +/// and mapped back into an object reference. +/// +/// ``` +/// use std::str::FromStr; +/// use debugid::DebugId; +/// +/// let debug_id = DebugId::from_str("dfb8e43a-f242-3d73-a453-aeb6a777ef75-a").unwrap(); +/// +/// let slice = &[debug_id]; +/// let ptr = slice.as_ptr() as *const u8; +/// let len = std::mem::size_of_val(slice); +/// let buf: &[u8] = unsafe { std::slice::from_raw_parts(ptr, len) }; +/// +/// let mut new_buf: Vec<u8> = Vec::new(); +/// std::io::copy(&mut std::io::Cursor::new(buf), &mut new_buf).unwrap(); +/// +/// let ptr = new_buf.as_ptr() as *const DebugId; +/// let new_debug_id = unsafe { &*ptr }; +/// +/// assert_eq!(*new_debug_id, debug_id); +/// ``` +/// +/// As long the bytes were written using the same major version of this crate you will be +/// able to read it again like this. +/// +/// [`CodeId`]: struct.CodeId.html +/// [`DebugId::breakpad`]: struct.DebugId.html#method.breakpad +// This needs to be backwards compatible also in its exact in-memory byte-layout since this +// struct is directly mapped from disk in e.g. Symbolic SymCache formats. The first version +// of this struct was defined as: +// +// ```rust +// struct DebugId { +// uuid: Uuid, +// appendix: u32, +// _padding: [u8; 12], +// } +// ``` +// +// For this reason the current `typ` byte represents the type of `DebugId` stored in the +// `Bytes`: +// +// - `0u8`: The `bytes` field contains a UUID. +// - `1u8`: The first 4 bytes of the `bytes` field contain a big-endian u32, the remaining +// bytes are 0. +#[repr(C, packed)] +#[derive(Default, Eq, PartialEq, Ord, PartialOrd, Hash, Clone, Copy)] +pub struct DebugId { + bytes: Bytes, + appendix: u32, + _padding: [u8; 11], + typ: u8, +} + +impl DebugId { + /// Constructs an empty debug identifier, containing only zeros. + pub fn nil() -> Self { + Self::default() + } + + /// Constructs a `DebugId` from its `uuid`. + pub fn from_uuid(uuid: Uuid) -> Self { + Self::from_parts(uuid, 0) + } + + /// Constructs a `DebugId` from its `uuid` and `appendix` parts. + pub fn from_parts(uuid: Uuid, appendix: u32) -> Self { + DebugId { + bytes: *uuid.as_bytes(), + appendix, + typ: 0, + _padding: [0; 11], + } + } + + /// Constructs a `DebugId` from a Microsoft little-endian GUID and age. + pub fn from_guid_age(guid: &[u8], age: u32) -> Result<Self, ParseDebugIdError> { + if guid.len() != 16 { + return Err(ParseDebugIdError); + } + + let uuid = Uuid::from_bytes([ + guid[3], guid[2], guid[1], guid[0], guid[5], guid[4], guid[7], guid[6], guid[8], + guid[9], guid[10], guid[11], guid[12], guid[13], guid[14], guid[15], + ]); + + Ok(DebugId::from_parts(uuid, age)) + } + + /// Constructs a `DebugId` from a PDB 2.0 timestamp and age. + pub fn from_pdb20(timestamp: u32, age: u32) -> Self { + // The big-endian byte-order here has to match the one used to read this number in + // the DebugId::timestamp method. + DebugId { + bytes: [ + (timestamp >> 24) as u8, + (timestamp >> 16) as u8, + (timestamp >> 8) as u8, + timestamp as u8, + 0u8, + 0u8, + 0u8, + 0u8, + 0u8, + 0u8, + 0u8, + 0u8, + 0u8, + 0u8, + 0u8, + 0u8, + ], + appendix: age, + _padding: [0u8; 11], + typ: 1u8, + } + } + + /// Parses a breakpad identifier from a string. + pub fn from_breakpad(string: &str) -> Result<Self, ParseDebugIdError> { + let options = ParseOptions { + allow_hyphens: false, + require_appendix: true, + allow_tail: false, + }; + Self::parse_str(string, options).ok_or(ParseDebugIdError) + } + + /// Returns the UUID part of the code module's debug_identifier. + /// + /// If this is a debug identifier for the PDB 2.0 format an invalid UUID is returned + /// where only the first 4 bytes are filled in and the remainder of the bytes are 0. + /// This means the UUID has variant [`uuid::Variant::NCS`] and an unknown version, + /// [`Uuid::get_version`] will return `None`, which is not a valid UUID. + /// + /// This may seem odd however does seem reasonable: + /// + /// - Every [`DebugId`] can be represented as [`Uuid`] and will still mostly look + /// reasonable e.g. in comparisons etc. + /// - The PDB 2.0 format is very old and very unlikely to appear practically. + pub fn uuid(&self) -> Uuid { + Uuid::from_bytes(self.bytes) + } + + /// Returns the appendix part of the code module's debug identifier. + /// + /// On Windows, this is an incrementing counter to identify the build. + /// On all other platforms, this value will always be zero. + pub fn appendix(&self) -> u32 { + self.appendix + } + + /// Returns whether this identifier is nil, i.e. it consists only of zeros. + pub fn is_nil(&self) -> bool { + self.bytes == [0u8; 16] && self.appendix == 0 + } + + /// Returns whether this identifier is from the PDB 2.0 format. + pub fn is_pdb20(&self) -> bool { + self.typ == 1 + } + + /// Returns a wrapper which when formatted via `fmt::Display` will format a + /// a breakpad identifier. + pub fn breakpad(&self) -> BreakpadFormat<'_> { + BreakpadFormat { inner: self } + } + + fn parse_str(string: &str, options: ParseOptions) -> Option<Self> { + let is_hyphenated = string.get(8..9) == Some("-"); + if is_hyphenated && !options.allow_hyphens || !string.is_ascii() { + return None; + } + + // Can the PDB 2.0 format match? This can never be true for a valid UUID. + let min_len = if is_hyphenated { 10 } else { 9 }; + let max_len = if is_hyphenated { 17 } else { 16 }; + if min_len <= string.len() && string.len() <= max_len { + let timestamp_str = string.get(..8)?; + let timestamp = u32::from_str_radix(timestamp_str, 16).ok()?; + let appendix_str = match is_hyphenated { + true => string.get(9..)?, + false => string.get(8..)?, + }; + let appendix = u32::from_str_radix(appendix_str, 16).ok()?; + return Some(Self::from_pdb20(timestamp, appendix)); + } + + let uuid_len = if is_hyphenated { 36 } else { 32 }; + let uuid = string.get(..uuid_len)?.parse().ok()?; + if !options.require_appendix && string.len() == uuid_len { + return Some(Self::from_parts(uuid, 0)); + } + + let mut appendix_str = &string[uuid_len..]; + if is_hyphenated ^ appendix_str.starts_with('-') { + return None; // Require a hyphen if and only if we're hyphenated. + } else if is_hyphenated { + appendix_str = &appendix_str[1..]; // Skip the hyphen for parsing. + } + + if options.allow_tail && appendix_str.len() > 8 { + appendix_str = &appendix_str[..8]; + } + + // Parse the appendix, which fails on empty strings. + let appendix = u32::from_str_radix(appendix_str, 16).ok()?; + Some(Self::from_parts(uuid, appendix)) + } + + /// Returns the PDB 2.0 timestamp. + /// + /// Only valid if you know this is a PDB 2.0 debug identifier. + fn timestamp(&self) -> u32 { + u32::from_be_bytes([self.bytes[0], self.bytes[1], self.bytes[2], self.bytes[3]]) + } +} + +impl fmt::Debug for DebugId { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + let uuid = self.uuid(); + f.debug_struct("DebugId") + .field("uuid", &uuid.hyphenated().to_string()) + .field("appendix", &self.appendix()) + .finish() + } +} + +impl fmt::Display for DebugId { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + match self.is_pdb20() { + true => { + let timestamp = self.timestamp(); + write!(f, "{:08X}", timestamp)?; + } + false => { + let uuid = self.uuid(); + uuid.fmt(f)?; + } + } + if self.appendix > 0 { + write!(f, "-{:x}", { self.appendix })?; + } + Ok(()) + } +} + +impl str::FromStr for DebugId { + type Err = ParseDebugIdError; + + fn from_str(string: &str) -> Result<Self, ParseDebugIdError> { + let options = ParseOptions { + allow_hyphens: true, + require_appendix: false, + allow_tail: true, + }; + Self::parse_str(string, options).ok_or(ParseDebugIdError) + } +} + +impl From<Uuid> for DebugId { + fn from(uuid: Uuid) -> Self { + DebugId::from_uuid(uuid) + } +} + +impl From<(Uuid, u32)> for DebugId { + fn from(tuple: (Uuid, u32)) -> Self { + let (uuid, appendix) = tuple; + DebugId::from_parts(uuid, appendix) + } +} + +/// Wrapper around [`DebugId`] for Breakpad formatting. +/// +/// **Example:** +/// +/// ``` +/// # extern crate debugid; +/// use std::str::FromStr; +/// use debugid::DebugId; +/// +/// # fn foo() -> Result<(), debugid::ParseDebugIdError> { +/// let id = DebugId::from_breakpad("DFB8E43AF2423D73A453AEB6A777EF75a")?; +/// assert_eq!("DFB8E43AF2423D73A453AEB6A777EF75a".to_string(), id.breakpad().to_string()); +/// # Ok(()) +/// # } +/// +/// # fn main() { foo().unwrap() } +/// ``` +/// +/// [`DebugId`]: struct.DebugId.html +#[derive(Debug)] +pub struct BreakpadFormat<'a> { + inner: &'a DebugId, +} + +impl<'a> fmt::Display for BreakpadFormat<'a> { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + match self.inner.is_pdb20() { + true => { + let timestamp = self.inner.timestamp(); + write!(f, "{:08X}{:x}", timestamp, self.inner.appendix()) + } + false => { + let uuid = self.inner.uuid(); + write!(f, "{:X}{:x}", uuid.simple(), self.inner.appendix()) + } + } + } +} + +/// Indicates an error parsing a [`CodeId`](struct.CodeId.html). +#[derive(Clone, Copy, Debug, Eq, PartialEq)] +pub struct ParseCodeIdError; + +impl error::Error for ParseCodeIdError {} + +impl fmt::Display for ParseCodeIdError { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + write!(f, "invalid code identifier") + } +} + +/// Unique platform-dependent identifier of code files. +/// +/// This identifier assumes a string representation that depends on the platform and compiler used. +/// The representation only retains hex characters and canonically stores lower case. +/// +/// There are the following known formats: +/// +/// - **MachO UUID**: The unique identifier of a Mach binary, specified in the `LC_UUID` load +/// command header. +/// - **GNU Build ID**: Contents of the `.gnu.build-id` note or section contents formatted as +/// lowercase hex string. +/// - **PE Timestamp**: Timestamp and size of image values from a Windows PE header. The size of +/// image value is truncated, so the length of the `CodeId` might not be a multiple of 2. +#[derive(Clone, Default, Eq, Hash, Ord, PartialEq, PartialOrd)] +pub struct CodeId { + inner: String, +} + +impl CodeId { + /// Constructs an empty code identifier. + pub fn nil() -> Self { + Self::default() + } + + /// Constructs a `CodeId` from its string representation. + pub fn new(mut string: String) -> Self { + string.retain(|c| c.is_ascii_hexdigit()); + string.make_ascii_lowercase(); + CodeId { inner: string } + } + + /// Constructs a `CodeId` from a binary slice. + pub fn from_binary(slice: &[u8]) -> Self { + let mut string = String::with_capacity(slice.len() * 2); + + for byte in slice { + write!(&mut string, "{:02x}", byte).expect(""); + } + + Self::new(string) + } + + /// Returns whether this identifier is nil, i.e. it is empty. + pub fn is_nil(&self) -> bool { + self.inner.is_empty() + } + + /// Returns the string representation of this code identifier. + pub fn as_str(&self) -> &str { + self.inner.as_str() + } +} + +impl fmt::Display for CodeId { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + f.write_str(&self.inner) + } +} + +impl fmt::Debug for CodeId { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + write!(f, "CodeId({})", self) + } +} + +impl From<String> for CodeId { + fn from(string: String) -> Self { + Self::new(string) + } +} + +impl From<&'_ str> for CodeId { + fn from(string: &str) -> Self { + Self::new(string.into()) + } +} + +impl AsRef<str> for CodeId { + fn as_ref(&self) -> &str { + self.as_str() + } +} + +impl str::FromStr for CodeId { + type Err = ParseCodeIdError; + + fn from_str(string: &str) -> Result<Self, ParseCodeIdError> { + Ok(Self::new(string.into())) + } +} + +#[cfg(feature = "serde")] +mod serde_support { + use serde::de::{self, Deserialize, Deserializer, Unexpected, Visitor}; + use serde::ser::{Serialize, Serializer}; + + use super::*; + + impl Serialize for CodeId { + fn serialize<S: Serializer>(&self, serializer: S) -> Result<S::Ok, S::Error> { + serializer.serialize_str(self.as_str()) + } + } + + impl<'de> Deserialize<'de> for CodeId { + fn deserialize<D: Deserializer<'de>>(deserializer: D) -> Result<Self, D::Error> { + let string = String::deserialize(deserializer)?; + Ok(CodeId::new(string)) + } + } + + impl<'de> Deserialize<'de> for DebugId { + fn deserialize<D: Deserializer<'de>>(deserializer: D) -> Result<Self, D::Error> { + struct V; + + impl<'de> Visitor<'de> for V { + type Value = DebugId; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str("DebugId") + } + + fn visit_str<E: de::Error>(self, value: &str) -> Result<DebugId, E> { + value + .parse() + .map_err(|_| de::Error::invalid_value(Unexpected::Str(value), &self)) + } + } + + deserializer.deserialize_str(V) + } + } + + impl Serialize for DebugId { + fn serialize<S: Serializer>(&self, serializer: S) -> Result<S::Ok, S::Error> { + serializer.serialize_str(&self.to_string()) + } + } +} |