michael@0: /* This Source Code Form is subject to the terms of the Mozilla Public
michael@0:  * License, v. 2.0. If a copy of the MPL was not distributed with this
michael@0:  * file, You can obtain one at http://mozilla.org/MPL/2.0/. */
michael@0: 
michael@0: #include "nsSecurityHeaderParser.h"
michael@0: #include "prlog.h"
michael@0: 
michael@0: // The character classes in this file are informed by [RFC2616], Section 2.2.
michael@0: // signed char is a signed data type one byte (8 bits) wide, so its value can
michael@0: // never be greater than 127. The following implicitly makes use of this.
michael@0: 
michael@0: // A token is one or more CHAR except CTLs or separators.
michael@0: // A CHAR is any US-ASCII character (octets 0 - 127).
michael@0: // A CTL is any US-ASCII control character (octets 0 - 31) and DEL (127).
michael@0: // A separator is one of ()<>@,;:\"/[]?={} as well as space and
michael@0: // horizontal-tab (32 and 9, respectively).
michael@0: // So, this returns true if chr is any octet 33-126 except ()<>@,;:\"/[]?={}
michael@0: bool
michael@0: IsTokenSymbol(signed char chr) {
michael@0:   if (chr < 33 || chr == 127 ||
michael@0:       chr == '(' || chr == ')' || chr == '<' || chr == '>' ||
michael@0:       chr == '@' || chr == ',' || chr == ';' || chr == ':' ||
michael@0:       chr == '"' || chr == '/' || chr == '[' || chr == ']' ||
michael@0:       chr == '?' || chr == '=' || chr == '{' || chr == '}' || chr == '\\') {
michael@0:     return false;
michael@0:   }
michael@0:   return true;
michael@0: }
michael@0: 
michael@0: // A quoted-string consists of a quote (") followed by any amount of
michael@0: // qdtext or quoted-pair, followed by a quote.
michael@0: // qdtext is any TEXT except a quote.
michael@0: // TEXT is any 8-bit octet except CTLs, but including LWS.
michael@0: // quoted-pair is a backslash (\) followed by a CHAR.
michael@0: // So, it turns out, \ can't really be a qdtext symbol for our purposes.
michael@0: // This returns true if chr is any octet 9,10,13,32-126 except <"> or "\"
michael@0: bool
michael@0: IsQuotedTextSymbol(signed char chr) {
michael@0:   return ((chr >= 32 && chr != '"' && chr != '\\' && chr != 127) ||
michael@0:           chr == 0x9 || chr == 0xa || chr == 0xd);
michael@0: }
michael@0: 
michael@0: // The octet following the "\" in a quoted pair can be anything 0-127.
michael@0: bool
michael@0: IsQuotedPairSymbol(signed char chr) {
michael@0:   return (chr >= 0);
michael@0: }
michael@0: 
michael@0: #if defined(PR_LOGGING)
michael@0: static PRLogModuleInfo *
michael@0: GetSHParserLog()
michael@0: {
michael@0:   static PRLogModuleInfo *sSHParserLog;
michael@0:   if (!sSHParserLog) {
michael@0:     sSHParserLog = PR_NewLogModule("nsSecurityHeaderParser");
michael@0:   }
michael@0:   return sSHParserLog;
michael@0: }
michael@0: #endif
michael@0: 
michael@0: #define SHPARSERLOG(args) PR_LOG(GetSHParserLog(), PR_LOG_DEBUG, args)
michael@0: 
michael@0: nsSecurityHeaderParser::nsSecurityHeaderParser(const char *aHeader)
michael@0:   : mCursor(aHeader)
michael@0:   , mError(false)
michael@0: {
michael@0: }
michael@0: 
michael@0: nsSecurityHeaderParser::~nsSecurityHeaderParser() {
michael@0:   nsSecurityHeaderDirective *directive;
michael@0:   while ((directive = mDirectives.popFirst())) {
michael@0:     delete directive;
michael@0:   }
michael@0: }
michael@0: 
michael@0: mozilla::LinkedList<nsSecurityHeaderDirective> *
michael@0: nsSecurityHeaderParser::GetDirectives() {
michael@0:   return &mDirectives;
michael@0: }
michael@0: 
michael@0: nsresult
michael@0: nsSecurityHeaderParser::Parse() {
michael@0:   MOZ_ASSERT(mDirectives.isEmpty());
michael@0:   SHPARSERLOG(("trying to parse '%s'", mCursor));
michael@0: 
michael@0:   Header();
michael@0: 
michael@0:   // if we didn't consume the entire input, we were unable to parse it => error
michael@0:   if (mError || *mCursor) {
michael@0:     return NS_ERROR_FAILURE;
michael@0:   } else {
michael@0:     return NS_OK;
michael@0:   }
michael@0: }
michael@0: 
michael@0: bool
michael@0: nsSecurityHeaderParser::Accept(char aChr)
michael@0: {
michael@0:   if (*mCursor == aChr) {
michael@0:     Advance();
michael@0:     return true;
michael@0:   }
michael@0: 
michael@0:   return false;
michael@0: }
michael@0: 
michael@0: bool
michael@0: nsSecurityHeaderParser::Accept(bool (*aClassifier) (signed char))
michael@0: {
michael@0:   if (aClassifier(*mCursor)) {
michael@0:     Advance();
michael@0:     return true;
michael@0:   }
michael@0: 
michael@0:   return false;
michael@0: }
michael@0: 
michael@0: void
michael@0: nsSecurityHeaderParser::Expect(char aChr)
michael@0: {
michael@0:   if (*mCursor != aChr) {
michael@0:     mError = true;
michael@0:   } else {
michael@0:     Advance();
michael@0:   }
michael@0: }
michael@0: 
michael@0: void
michael@0: nsSecurityHeaderParser::Advance()
michael@0: {
michael@0:   // Technically, 0 is valid in quoted-pair, but we were handed a
michael@0:   // null-terminated const char *, so this doesn't handle that.
michael@0:   if (*mCursor) {
michael@0:     mOutput.Append(*mCursor);
michael@0:     mCursor++;
michael@0:   } else {
michael@0:     mError = true;
michael@0:   }
michael@0: }
michael@0: 
michael@0: void
michael@0: nsSecurityHeaderParser::Header()
michael@0: {
michael@0:   Directive();
michael@0:   while (Accept(';')) {
michael@0:     Directive();
michael@0:   }
michael@0: }
michael@0: 
michael@0: void
michael@0: nsSecurityHeaderParser::Directive()
michael@0: {
michael@0:   mDirective = new nsSecurityHeaderDirective();
michael@0:   LWSMultiple();
michael@0:   DirectiveName();
michael@0:   LWSMultiple();
michael@0:   if (Accept('=')) {
michael@0:     LWSMultiple();
michael@0:     DirectiveValue();
michael@0:     LWSMultiple();
michael@0:   }
michael@0:   mDirectives.insertBack(mDirective);
michael@0:   SHPARSERLOG(("read directive name '%s', value '%s'",
michael@0:                mDirective->mName.Data(), mDirective->mValue.Data()));
michael@0: }
michael@0: 
michael@0: void
michael@0: nsSecurityHeaderParser::DirectiveName()
michael@0: {
michael@0:   mOutput.Truncate(0);
michael@0:   Token();
michael@0:   mDirective->mName.Assign(mOutput);
michael@0: }
michael@0: 
michael@0: void
michael@0: nsSecurityHeaderParser::DirectiveValue()
michael@0: {
michael@0:   mOutput.Truncate(0);
michael@0:   if (Accept(IsTokenSymbol)) {
michael@0:     Token();
michael@0:     mDirective->mValue.Assign(mOutput);
michael@0:   } else if (Accept('"')) {
michael@0:     // Accept advances the cursor if successful, which appends a character to
michael@0:     // mOutput. The " is not part of what we want to capture, so truncate
michael@0:     // mOutput again.
michael@0:     mOutput.Truncate(0);
michael@0:     QuotedString();
michael@0:     mDirective->mValue.Assign(mOutput);
michael@0:     Expect('"');
michael@0:   }
michael@0: }
michael@0: 
michael@0: void
michael@0: nsSecurityHeaderParser::Token()
michael@0: {
michael@0:   while (Accept(IsTokenSymbol));
michael@0: }
michael@0: 
michael@0: void
michael@0: nsSecurityHeaderParser::QuotedString()
michael@0: {
michael@0:   while (true) {
michael@0:     if (Accept(IsQuotedTextSymbol)) {
michael@0:       QuotedText();
michael@0:     } else if (Accept('\\')) {
michael@0:       QuotedPair();
michael@0:     } else {
michael@0:       break;
michael@0:     }
michael@0:   }
michael@0: }
michael@0: 
michael@0: void
michael@0: nsSecurityHeaderParser::QuotedText()
michael@0: {
michael@0:   while (Accept(IsQuotedTextSymbol));
michael@0: }
michael@0: 
michael@0: void
michael@0: nsSecurityHeaderParser::QuotedPair()
michael@0: {
michael@0:   Accept(IsQuotedPairSymbol);
michael@0: }
michael@0: 
michael@0: void
michael@0: nsSecurityHeaderParser::LWSMultiple()
michael@0: {
michael@0:   while (true) {
michael@0:     if (Accept('\r')) {
michael@0:       LWSCRLF();
michael@0:     } else if (Accept(' ') || Accept('\t')) {
michael@0:       LWS();
michael@0:     } else {
michael@0:       break;
michael@0:     }
michael@0:   }
michael@0: }
michael@0: 
michael@0: void
michael@0: nsSecurityHeaderParser::LWSCRLF() {
michael@0:   Expect('\n');
michael@0:   if (!(Accept(' ') || Accept('\t'))) {
michael@0:     mError = true;
michael@0:   }
michael@0:   LWS();
michael@0: }
michael@0: 
michael@0: void
michael@0: nsSecurityHeaderParser::LWS()
michael@0: {
michael@0:   // Note that becaue of how we're called, we don't have to check for
michael@0:   // the mandatory presense of at least one of SP or HT.
michael@0:   while (Accept(' ') || Accept('\t'));
michael@0: }