michael@0: /* -*- Mode: C++; tab-width: 2; indent-tabs-mode: nil; c-basic-offset: 2 -*- */
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: /**
michael@0:  * Lexical analyzer for XPath expressions
michael@0:  */
michael@0: 
michael@0: #include "txExprLexer.h"
michael@0: #include "nsGkAtoms.h"
michael@0: #include "nsString.h"
michael@0: #include "nsError.h"
michael@0: #include "txXMLUtils.h"
michael@0: 
michael@0: /**
michael@0:  * Creates a new ExprLexer
michael@0:  */
michael@0: txExprLexer::txExprLexer()
michael@0:   : mCurrentItem(nullptr),
michael@0:     mFirstItem(nullptr),
michael@0:     mLastItem(nullptr),
michael@0:     mTokenCount(0)
michael@0: {
michael@0: }
michael@0: 
michael@0: /**
michael@0:  * Destroys this instance of an txExprLexer
michael@0:  */
michael@0: txExprLexer::~txExprLexer()
michael@0: {
michael@0:   //-- delete tokens
michael@0:   Token* tok = mFirstItem;
michael@0:   while (tok) {
michael@0:     Token* temp = tok->mNext;
michael@0:     delete tok;
michael@0:     tok = temp;
michael@0:   }
michael@0:   mCurrentItem = nullptr;
michael@0: }
michael@0: 
michael@0: Token*
michael@0: txExprLexer::nextToken()
michael@0: {
michael@0:   if (!mCurrentItem) {
michael@0:     NS_NOTREACHED("nextToken called on uninitialized lexer");
michael@0:     return nullptr;
michael@0:   }
michael@0: 
michael@0:   if (mCurrentItem->mType == Token::END) {
michael@0:     // Do not progress beyond the end token
michael@0:     return mCurrentItem;
michael@0:   }
michael@0: 
michael@0:   Token* token = mCurrentItem;
michael@0:   mCurrentItem = mCurrentItem->mNext;
michael@0:   return token;
michael@0: }
michael@0: 
michael@0: void
michael@0: txExprLexer::addToken(Token* aToken)
michael@0: {
michael@0:   if (mLastItem) {
michael@0:     mLastItem->mNext = aToken;
michael@0:   }
michael@0:   if (!mFirstItem) {
michael@0:     mFirstItem = aToken;
michael@0:     mCurrentItem = aToken;
michael@0:   }
michael@0:   mLastItem = aToken;
michael@0:   ++mTokenCount;
michael@0: }
michael@0: 
michael@0: /**
michael@0:  * Returns true if the following Token should be an operator.
michael@0:  * This is a helper for the first bullet of [XPath 3.7]
michael@0:  *  Lexical Structure
michael@0:  */
michael@0: bool
michael@0: txExprLexer::nextIsOperatorToken(Token* aToken)
michael@0: {
michael@0:   if (!aToken || aToken->mType == Token::NULL_TOKEN) {
michael@0:     return false;
michael@0:   }
michael@0:   /* This relies on the tokens having the right order in txExprLexer.h */
michael@0:   return aToken->mType < Token::COMMA ||
michael@0:     aToken->mType > Token::UNION_OP;
michael@0: 
michael@0: }
michael@0: 
michael@0: /**
michael@0:  * Parses the given string into a sequence of Tokens
michael@0:  */
michael@0: nsresult
michael@0: txExprLexer::parse(const nsASingleFragmentString& aPattern)
michael@0: {
michael@0:   iterator start, end;
michael@0:   start = aPattern.BeginReading(mPosition);
michael@0:   aPattern.EndReading(end);
michael@0: 
michael@0:   //-- initialize previous token, this will automatically get
michael@0:   //-- deleted when it goes out of scope
michael@0:   Token nullToken(nullptr, nullptr, Token::NULL_TOKEN);
michael@0: 
michael@0:   Token::Type defType;
michael@0:   Token* newToken = nullptr;
michael@0:   Token* prevToken = &nullToken;
michael@0:   bool isToken;
michael@0: 
michael@0:   while (mPosition < end) {
michael@0: 
michael@0:     defType = Token::CNAME;
michael@0:     isToken = true;
michael@0: 
michael@0:     if (*mPosition == DOLLAR_SIGN) {
michael@0:       if (++mPosition == end || !XMLUtils::isLetter(*mPosition)) {
michael@0:         return NS_ERROR_XPATH_INVALID_VAR_NAME;
michael@0:       }
michael@0:       defType = Token::VAR_REFERENCE;
michael@0:     } 
michael@0:     // just reuse the QName parsing, which will use defType 
michael@0:     // the token to construct
michael@0: 
michael@0:     if (XMLUtils::isLetter(*mPosition)) {
michael@0:       // NCName, can get QName or OperatorName;
michael@0:       //  FunctionName, NodeName, and AxisSpecifier may want whitespace,
michael@0:       //  and are dealt with below
michael@0:       start = mPosition;
michael@0:       while (++mPosition < end && XMLUtils::isNCNameChar(*mPosition)) {
michael@0:         /* just go */
michael@0:       }
michael@0:       if (mPosition < end && *mPosition == COLON) {
michael@0:         // try QName or wildcard, might need to step back for axis
michael@0:         if (++mPosition == end) {
michael@0:           return NS_ERROR_XPATH_UNEXPECTED_END;
michael@0:         }
michael@0:         if (XMLUtils::isLetter(*mPosition)) {
michael@0:           while (++mPosition < end && XMLUtils::isNCNameChar(*mPosition)) {
michael@0:             /* just go */
michael@0:           }
michael@0:         }
michael@0:         else if (*mPosition == '*' && defType != Token::VAR_REFERENCE) {
michael@0:           // eat wildcard for NameTest, bail for var ref at COLON
michael@0:           ++mPosition;
michael@0:         }
michael@0:         else {
michael@0:           --mPosition; // step back
michael@0:         }
michael@0:       }
michael@0:       if (nextIsOperatorToken(prevToken)) {
michael@0:         nsDependentSubstring op(Substring(start, mPosition));
michael@0:         if (nsGkAtoms::_and->Equals(op)) {
michael@0:           defType = Token::AND_OP;
michael@0:         }
michael@0:         else if (nsGkAtoms::_or->Equals(op)) {
michael@0:           defType = Token::OR_OP;
michael@0:         }
michael@0:         else if (nsGkAtoms::mod->Equals(op)) {
michael@0:           defType = Token::MODULUS_OP;
michael@0:         }
michael@0:         else if (nsGkAtoms::div->Equals(op)) {
michael@0:           defType = Token::DIVIDE_OP;
michael@0:         }
michael@0:         else {
michael@0:           // XXX QUESTION: spec is not too precise
michael@0:           // badops is sure an error, but is bad:ops, too? We say yes!
michael@0:           return NS_ERROR_XPATH_OPERATOR_EXPECTED;
michael@0:         }
michael@0:       }
michael@0:       newToken = new Token(start, mPosition, defType);
michael@0:     }
michael@0:     else if (isXPathDigit(*mPosition)) {
michael@0:       start = mPosition;
michael@0:       while (++mPosition < end && isXPathDigit(*mPosition)) {
michael@0:         /* just go */
michael@0:       }
michael@0:       if (mPosition < end && *mPosition == '.') {
michael@0:         while (++mPosition < end && isXPathDigit(*mPosition)) {
michael@0:           /* just go */
michael@0:         }
michael@0:       }
michael@0:       newToken = new Token(start, mPosition, Token::NUMBER);
michael@0:     }
michael@0:     else {
michael@0:       switch (*mPosition) {
michael@0:         //-- ignore whitespace
michael@0:       case SPACE:
michael@0:       case TX_TAB:
michael@0:       case TX_CR:
michael@0:       case TX_LF:
michael@0:         ++mPosition;
michael@0:         isToken = false;
michael@0:         break;
michael@0:       case S_QUOTE :
michael@0:       case D_QUOTE :
michael@0:         start = mPosition;
michael@0:         while (++mPosition < end && *mPosition != *start) {
michael@0:           // eat literal
michael@0:         }
michael@0:         if (mPosition == end) {
michael@0:           mPosition = start;
michael@0:           return NS_ERROR_XPATH_UNCLOSED_LITERAL;
michael@0:         }
michael@0:         newToken = new Token(start + 1, mPosition, Token::LITERAL);
michael@0:         ++mPosition;
michael@0:         break;
michael@0:       case PERIOD:
michael@0:         // period can be .., .(DIGITS)+ or ., check next
michael@0:         if (++mPosition == end) {
michael@0:           newToken = new Token(mPosition - 1, Token::SELF_NODE);
michael@0:         }
michael@0:         else if (isXPathDigit(*mPosition)) {
michael@0:           start = mPosition - 1;
michael@0:           while (++mPosition < end && isXPathDigit(*mPosition)) {
michael@0:             /* just go */
michael@0:           }
michael@0:           newToken = new Token(start, mPosition, Token::NUMBER);
michael@0:         }
michael@0:         else if (*mPosition == PERIOD) {
michael@0:           ++mPosition;
michael@0:           newToken = new Token(mPosition - 2, mPosition, Token::PARENT_NODE);
michael@0:         }
michael@0:         else {
michael@0:           newToken = new Token(mPosition - 1, Token::SELF_NODE);
michael@0:         }
michael@0:         break;
michael@0:       case COLON: // QNames are dealt above, must be axis ident
michael@0:         if (++mPosition >= end || *mPosition != COLON ||
michael@0:             prevToken->mType != Token::CNAME) {
michael@0:           return NS_ERROR_XPATH_BAD_COLON;
michael@0:         }
michael@0:         prevToken->mType = Token::AXIS_IDENTIFIER;
michael@0:         ++mPosition;
michael@0:         isToken = false;
michael@0:         break;
michael@0:       case FORWARD_SLASH :
michael@0:         if (++mPosition < end && *mPosition == FORWARD_SLASH) {
michael@0:           ++mPosition;
michael@0:           newToken = new Token(mPosition - 2, mPosition, Token::ANCESTOR_OP);
michael@0:         }
michael@0:         else {
michael@0:           newToken = new Token(mPosition - 1, Token::PARENT_OP);
michael@0:         }
michael@0:         break;
michael@0:       case BANG : // can only be !=
michael@0:         if (++mPosition < end && *mPosition == EQUAL) {
michael@0:           ++mPosition;
michael@0:           newToken = new Token(mPosition - 2, mPosition, Token::NOT_EQUAL_OP);
michael@0:           break;
michael@0:         }
michael@0:         // Error ! is not not()
michael@0:         return NS_ERROR_XPATH_BAD_BANG;
michael@0:       case EQUAL:
michael@0:         newToken = new Token(mPosition, Token::EQUAL_OP);
michael@0:         ++mPosition;
michael@0:         break;
michael@0:       case L_ANGLE:
michael@0:         if (++mPosition == end) {
michael@0:           return NS_ERROR_XPATH_UNEXPECTED_END;
michael@0:         }
michael@0:         if (*mPosition == EQUAL) {
michael@0:           ++mPosition;
michael@0:           newToken = new Token(mPosition - 2, mPosition,
michael@0:                                Token::LESS_OR_EQUAL_OP);
michael@0:         }
michael@0:         else {
michael@0:           newToken = new Token(mPosition - 1, Token::LESS_THAN_OP);
michael@0:         }
michael@0:         break;
michael@0:       case R_ANGLE:
michael@0:         if (++mPosition == end) {
michael@0:           return NS_ERROR_XPATH_UNEXPECTED_END;
michael@0:         }
michael@0:         if (*mPosition == EQUAL) {
michael@0:           ++mPosition;
michael@0:           newToken = new Token(mPosition - 2, mPosition,
michael@0:                                Token::GREATER_OR_EQUAL_OP);
michael@0:         }
michael@0:         else {
michael@0:           newToken = new Token(mPosition - 1, Token::GREATER_THAN_OP);
michael@0:         }
michael@0:         break;
michael@0:       case HYPHEN :
michael@0:         newToken = new Token(mPosition, Token::SUBTRACTION_OP);
michael@0:         ++mPosition;
michael@0:         break;
michael@0:       case ASTERIX:
michael@0:         if (nextIsOperatorToken(prevToken)) {
michael@0:           newToken = new Token(mPosition, Token::MULTIPLY_OP);
michael@0:         }
michael@0:         else {
michael@0:           newToken = new Token(mPosition, Token::CNAME);
michael@0:         }
michael@0:         ++mPosition;
michael@0:         break;
michael@0:       case L_PAREN:
michael@0:         if (prevToken->mType == Token::CNAME) {
michael@0:           const nsDependentSubstring& val = prevToken->Value();
michael@0:           if (val.EqualsLiteral("comment")) {
michael@0:             prevToken->mType = Token::COMMENT_AND_PAREN;
michael@0:           }
michael@0:           else if (val.EqualsLiteral("node")) {
michael@0:             prevToken->mType = Token::NODE_AND_PAREN;
michael@0:           }
michael@0:           else if (val.EqualsLiteral("processing-instruction")) {
michael@0:             prevToken->mType = Token::PROC_INST_AND_PAREN;
michael@0:           }
michael@0:           else if (val.EqualsLiteral("text")) {
michael@0:             prevToken->mType = Token::TEXT_AND_PAREN;
michael@0:           }
michael@0:           else {
michael@0:             prevToken->mType = Token::FUNCTION_NAME_AND_PAREN;
michael@0:           }
michael@0:           isToken = false;
michael@0:         }
michael@0:         else {
michael@0:           newToken = new Token(mPosition, Token::L_PAREN);
michael@0:         }
michael@0:         ++mPosition;
michael@0:         break;
michael@0:       case R_PAREN:
michael@0:         newToken = new Token(mPosition, Token::R_PAREN);
michael@0:         ++mPosition;
michael@0:         break;
michael@0:       case L_BRACKET:
michael@0:         newToken = new Token(mPosition, Token::L_BRACKET);
michael@0:         ++mPosition;
michael@0:         break;
michael@0:       case R_BRACKET:
michael@0:         newToken = new Token(mPosition, Token::R_BRACKET);
michael@0:         ++mPosition;
michael@0:         break;
michael@0:       case COMMA:
michael@0:         newToken = new Token(mPosition, Token::COMMA);
michael@0:         ++mPosition;
michael@0:         break;
michael@0:       case AT_SIGN :
michael@0:         newToken = new Token(mPosition, Token::AT_SIGN);
michael@0:         ++mPosition;
michael@0:         break;
michael@0:       case PLUS:
michael@0:         newToken = new Token(mPosition, Token::ADDITION_OP);
michael@0:         ++mPosition;
michael@0:         break;
michael@0:       case VERT_BAR:
michael@0:         newToken = new Token(mPosition, Token::UNION_OP);
michael@0:         ++mPosition;
michael@0:         break;
michael@0:       default:
michael@0:         // Error, don't grok character :-(
michael@0:         return NS_ERROR_XPATH_ILLEGAL_CHAR;
michael@0:       }
michael@0:     }
michael@0:     if (isToken) {
michael@0:       NS_ENSURE_TRUE(newToken, NS_ERROR_OUT_OF_MEMORY);
michael@0:       NS_ENSURE_TRUE(newToken != mLastItem, NS_ERROR_FAILURE);
michael@0:       prevToken = newToken;
michael@0:       addToken(newToken);
michael@0:     }
michael@0:   }
michael@0: 
michael@0:   // add a endToken to the list
michael@0:   newToken = new Token(end, end, Token::END);
michael@0:   if (!newToken) {
michael@0:     return NS_ERROR_OUT_OF_MEMORY;
michael@0:   }
michael@0:   addToken(newToken);
michael@0: 
michael@0:   return NS_OK;
michael@0: }