From ca67b09c015d4af3ae3cce12aa72e60941dbb8b5 Mon Sep 17 00:00:00 2001 From: Daniel Baumann Date: Sun, 7 Apr 2024 18:29:52 +0200 Subject: Adding debian version 2.06-13+deb12u1. Signed-off-by: Daniel Baumann --- debian/grub-extras/disabled/gpxe/src/core/uri.c | 463 ++++++++++++++++++++++++ 1 file changed, 463 insertions(+) create mode 100644 debian/grub-extras/disabled/gpxe/src/core/uri.c (limited to 'debian/grub-extras/disabled/gpxe/src/core/uri.c') diff --git a/debian/grub-extras/disabled/gpxe/src/core/uri.c b/debian/grub-extras/disabled/gpxe/src/core/uri.c new file mode 100644 index 0000000..f87ec1b --- /dev/null +++ b/debian/grub-extras/disabled/gpxe/src/core/uri.c @@ -0,0 +1,463 @@ +/* + * Copyright (C) 2007 Michael Brown . + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License as + * published by the Free Software Foundation; either version 2 of the + * License, or any later version. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. + */ + +FILE_LICENCE ( GPL2_OR_LATER ); + +/** @file + * + * Uniform Resource Identifiers + * + */ + +#include +#include +#include +#include +#include +#include +#include + +/** + * Dump URI for debugging + * + * @v uri URI + */ +static void dump_uri ( struct uri *uri ) { + if ( ! uri ) + return; + if ( uri->scheme ) + DBG ( " scheme \"%s\"", uri->scheme ); + if ( uri->opaque ) + DBG ( " opaque \"%s\"", uri->opaque ); + if ( uri->user ) + DBG ( " user \"%s\"", uri->user ); + if ( uri->password ) + DBG ( " password \"%s\"", uri->password ); + if ( uri->host ) + DBG ( " host \"%s\"", uri->host ); + if ( uri->port ) + DBG ( " port \"%s\"", uri->port ); + if ( uri->path ) + DBG ( " path \"%s\"", uri->path ); + if ( uri->query ) + DBG ( " query \"%s\"", uri->query ); + if ( uri->fragment ) + DBG ( " fragment \"%s\"", uri->fragment ); +} + +/** + * Parse URI + * + * @v uri_string URI as a string + * @ret uri URI + * + * Splits a URI into its component parts. The return URI structure is + * dynamically allocated and must eventually be freed by calling + * uri_put(). + */ +struct uri * parse_uri ( const char *uri_string ) { + struct uri *uri; + char *raw; + char *tmp; + char *path = NULL; + char *authority = NULL; + size_t raw_len; + + /* Allocate space for URI struct and a copy of the string */ + raw_len = ( strlen ( uri_string ) + 1 /* NUL */ ); + uri = zalloc ( sizeof ( *uri ) + raw_len ); + if ( ! uri ) + return NULL; + raw = ( ( ( char * ) uri ) + sizeof ( *uri ) ); + + /* Zero URI struct and copy in the raw string */ + memcpy ( raw, uri_string, raw_len ); + + /* Start by chopping off the fragment, if it exists */ + if ( ( tmp = strchr ( raw, '#' ) ) ) { + *(tmp++) = '\0'; + uri->fragment = tmp; + } + + /* Identify absolute/relative URI. We ignore schemes that are + * apparently only a single character long, since otherwise we + * misinterpret a DOS-style path name ("C:\path\to\file") as a + * URI with scheme="C",opaque="\path\to\file". + */ + if ( ( tmp = strchr ( raw, ':' ) ) && ( tmp > ( raw + 1 ) ) ) { + /* Absolute URI: identify hierarchical/opaque */ + uri->scheme = raw; + *(tmp++) = '\0'; + if ( *tmp == '/' ) { + /* Absolute URI with hierarchical part */ + path = tmp; + } else { + /* Absolute URI with opaque part */ + uri->opaque = tmp; + } + } else { + /* Relative URI */ + path = raw; + } + + /* If we don't have a path (i.e. we have an absolute URI with + * an opaque portion, we're already finished processing + */ + if ( ! path ) + goto done; + + /* Chop off the query, if it exists */ + if ( ( tmp = strchr ( path, '?' ) ) ) { + *(tmp++) = '\0'; + uri->query = tmp; + } + + /* Identify net/absolute/relative path */ + if ( strncmp ( path, "//", 2 ) == 0 ) { + /* Net path. If this is terminated by the first '/' + * of an absolute path, then we have no space for a + * terminator after the authority field, so shuffle + * the authority down by one byte, overwriting one of + * the two slashes. + */ + authority = ( path + 2 ); + if ( ( tmp = strchr ( authority, '/' ) ) ) { + /* Shuffle down */ + uri->path = tmp; + memmove ( ( authority - 1 ), authority, + ( tmp - authority ) ); + authority--; + *(--tmp) = '\0'; + } + } else { + /* Absolute/relative path */ + uri->path = path; + } + + /* Split authority into user[:password] and host[:port] portions */ + if ( ( tmp = strchr ( authority, '@' ) ) ) { + /* Has user[:password] */ + *(tmp++) = '\0'; + uri->host = tmp; + uri->user = authority; + if ( ( tmp = strchr ( authority, ':' ) ) ) { + /* Has password */ + *(tmp++) = '\0'; + uri->password = tmp; + } + } else { + /* No user:password */ + uri->host = authority; + } + + /* Split host into host[:port] */ + if ( ( tmp = strchr ( uri->host, ':' ) ) ) { + *(tmp++) = '\0'; + uri->port = tmp; + } + + done: + DBG ( "URI \"%s\" split into", uri_string ); + dump_uri ( uri ); + DBG ( "\n" ); + + return uri; +} + +/** + * Get port from URI + * + * @v uri URI, or NULL + * @v default_port Default port to use if none specified in URI + * @ret port Port + */ +unsigned int uri_port ( struct uri *uri, unsigned int default_port ) { + if ( ( ! uri ) || ( ! uri->port ) ) + return default_port; + return ( strtoul ( uri->port, NULL, 0 ) ); +} + +/** + * Unparse URI + * + * @v buf Buffer to fill with URI string + * @v size Size of buffer + * @v uri URI to write into buffer, or NULL + * @ret len Length of URI string + */ +int unparse_uri ( char *buf, size_t size, struct uri *uri ) { + int used = 0; + + DBG ( "URI unparsing" ); + dump_uri ( uri ); + DBG ( "\n" ); + + /* Special-case NULL URI */ + if ( ! uri ) { + if ( size ) + buf[0] = '\0'; + return 0; + } + + /* Special-case opaque URIs */ + if ( uri->opaque ) { + return ssnprintf ( ( buf + used ), ( size - used ), + "%s:%s", uri->scheme, uri->opaque ); + } + + /* scheme:// */ + if ( uri->scheme ) { + used += ssnprintf ( ( buf + used ), ( size - used ), + "%s://", uri->scheme ); + } + + /* [user[:password]@]host[:port] */ + if ( uri->host ) { + if ( uri->user ) { + used += ssnprintf ( ( buf + used ), ( size - used ), + "%s", uri->user ); + if ( uri->password ) { + used += ssnprintf ( ( buf + used ), + ( size - used ), + ":%s", uri->password ); + } + used += ssnprintf ( ( buf + used ), ( size - used ), + "@" ); + } + used += ssnprintf ( ( buf + used ), ( size - used ), "%s", + uri->host ); + if ( uri->port ) { + used += ssnprintf ( ( buf + used ), ( size - used ), + ":%s", uri->port ); + } + } + + /* /path */ + if ( uri->path ) { + used += ssnprintf ( ( buf + used ), ( size - used ), + "%s", uri->path ); + } + + /* ?query */ + if ( uri->query ) { + used += ssnprintf ( ( buf + used ), ( size - used ), + "?%s", uri->query ); + } + + /* #fragment */ + if ( uri->fragment ) { + used += ssnprintf ( ( buf + used ), ( size - used ), + "#%s", uri->fragment ); + } + + return used; +} + +/** + * Duplicate URI + * + * @v uri URI + * @ret uri Duplicate URI + * + * Creates a modifiable copy of a URI. + */ +struct uri * uri_dup ( struct uri *uri ) { + size_t len = ( unparse_uri ( NULL, 0, uri ) + 1 ); + char buf[len]; + + unparse_uri ( buf, len, uri ); + return parse_uri ( buf ); +} + +/** + * Resolve base+relative path + * + * @v base_uri Base path + * @v relative_uri Relative path + * @ret resolved_uri Resolved path + * + * Takes a base path (e.g. "/var/lib/tftpboot/vmlinuz" and a relative + * path (e.g. "initrd.gz") and produces a new path + * (e.g. "/var/lib/tftpboot/initrd.gz"). Note that any non-directory + * portion of the base path will automatically be stripped; this + * matches the semantics used when resolving the path component of + * URIs. + */ +char * resolve_path ( const char *base_path, + const char *relative_path ) { + size_t base_len = ( strlen ( base_path ) + 1 ); + char base_path_copy[base_len]; + char *base_tmp = base_path_copy; + + /* If relative path is absolute, just re-use it */ + if ( relative_path[0] == '/' ) + return strdup ( relative_path ); + + /* Create modifiable copy of path for dirname() */ + memcpy ( base_tmp, base_path, base_len ); + base_tmp = dirname ( base_tmp ); + + /* Process "./" and "../" elements */ + while ( *relative_path == '.' ) { + relative_path++; + if ( *relative_path == 0 ) { + /* Do nothing */ + } else if ( *relative_path == '/' ) { + relative_path++; + } else if ( *relative_path == '.' ) { + relative_path++; + if ( *relative_path == 0 ) { + base_tmp = dirname ( base_tmp ); + } else if ( *relative_path == '/' ) { + base_tmp = dirname ( base_tmp ); + relative_path++; + } else { + relative_path -= 2; + break; + } + } else { + relative_path--; + break; + } + } + + /* Create and return new path */ + return grub_xasprintf ( "%s%s%s", base_tmp, + ( ( base_tmp[ strlen ( base_tmp ) - 1 ] == '/' ) ? + "" : "/" ), relative_path ); +} + +/** + * Resolve base+relative URI + * + * @v base_uri Base URI, or NULL + * @v relative_uri Relative URI + * @ret resolved_uri Resolved URI + * + * Takes a base URI (e.g. "http://etherboot.org/kernels/vmlinuz" and a + * relative URI (e.g. "../initrds/initrd.gz") and produces a new URI + * (e.g. "http://etherboot.org/initrds/initrd.gz"). + */ +struct uri * resolve_uri ( struct uri *base_uri, + struct uri *relative_uri ) { + struct uri tmp_uri; + char *tmp_path = NULL; + struct uri *new_uri; + + /* If relative URI is absolute, just re-use it */ + if ( uri_is_absolute ( relative_uri ) || ( ! base_uri ) ) + return uri_get ( relative_uri ); + + /* Mangle URI */ + memcpy ( &tmp_uri, base_uri, sizeof ( tmp_uri ) ); + if ( relative_uri->path ) { + tmp_path = resolve_path ( ( base_uri->path ? + base_uri->path : "/" ), + relative_uri->path ); + tmp_uri.path = tmp_path; + tmp_uri.query = relative_uri->query; + tmp_uri.fragment = relative_uri->fragment; + } else if ( relative_uri->query ) { + tmp_uri.query = relative_uri->query; + tmp_uri.fragment = relative_uri->fragment; + } else if ( relative_uri->fragment ) { + tmp_uri.fragment = relative_uri->fragment; + } + + /* Create demangled URI */ + new_uri = uri_dup ( &tmp_uri ); + free ( tmp_path ); + return new_uri; +} + +/** + * Test for unreserved URI characters + * + * @v c Character to test + * @ret is_unreserved Character is an unreserved character + */ +static int is_unreserved_uri_char ( int c ) { + /* According to RFC3986, the unreserved character set is + * + * A-Z a-z 0-9 - _ . ~ + */ + return ( isupper ( c ) || islower ( c ) || isdigit ( c ) || + ( c == '-' ) || ( c == '_' ) || + ( c == '.' ) || ( c == '~' ) ); +} + +/** + * URI-encode string + * + * @v raw_string String to be URI-encoded + * @v buf Buffer to contain encoded string + * @v len Length of buffer + * @ret len Length of encoded string (excluding NUL) + */ +size_t uri_encode ( const char *raw_string, char *buf, size_t len ) { + ssize_t remaining = len; + size_t used; + unsigned char c; + + if ( len ) + buf[0] = '\0'; + + while ( ( c = *(raw_string++) ) ) { + if ( is_unreserved_uri_char ( c ) ) { + used = ssnprintf ( buf, remaining, "%c", c ); + } else { + used = ssnprintf ( buf, remaining, "%%%02X", c ); + } + buf += used; + remaining -= used; + } + + return ( len - remaining ); +} + +/** + * Decode URI-encoded string + * + * @v encoded_string URI-encoded string + * @v buf Buffer to contain decoded string + * @v len Length of buffer + * @ret len Length of decoded string (excluding NUL) + */ +size_t uri_decode ( const char *encoded_string, char *buf, size_t len ) { + ssize_t remaining = len; + char hexbuf[3]; + char *hexbuf_end; + unsigned char c; + + if ( len ) + buf[0] = '\0'; + + while ( *encoded_string ) { + if ( *encoded_string == '%' ) { + encoded_string++; + snprintf ( hexbuf, sizeof ( hexbuf ), "%s", + encoded_string ); + c = strtoul ( hexbuf, &hexbuf_end, 16 ); + encoded_string += ( hexbuf_end - hexbuf ); + } else { + c = *(encoded_string++); + } + ssnprintf ( buf++, remaining--, "%c", c ); + } + return ( len - remaining ); +} -- cgit v1.2.3