631 lines
19 KiB
JavaScript
631 lines
19 KiB
JavaScript
import { MATCH, MISMATCH, DISALLOW_EMPTY } from './match-graph.js';
|
||
import * as TYPE from '../tokenizer/types.js';
|
||
|
||
const { hasOwnProperty } = Object.prototype;
|
||
const STUB = 0;
|
||
const TOKEN = 1;
|
||
const OPEN_SYNTAX = 2;
|
||
const CLOSE_SYNTAX = 3;
|
||
|
||
const EXIT_REASON_MATCH = 'Match';
|
||
const EXIT_REASON_MISMATCH = 'Mismatch';
|
||
const EXIT_REASON_ITERATION_LIMIT = 'Maximum iteration number exceeded (please fill an issue on https://github.com/csstree/csstree/issues)';
|
||
|
||
const ITERATION_LIMIT = 15000;
|
||
export let totalIterationCount = 0;
|
||
|
||
function reverseList(list) {
|
||
let prev = null;
|
||
let next = null;
|
||
let item = list;
|
||
|
||
while (item !== null) {
|
||
next = item.prev;
|
||
item.prev = prev;
|
||
prev = item;
|
||
item = next;
|
||
}
|
||
|
||
return prev;
|
||
}
|
||
|
||
function areStringsEqualCaseInsensitive(testStr, referenceStr) {
|
||
if (testStr.length !== referenceStr.length) {
|
||
return false;
|
||
}
|
||
|
||
for (let i = 0; i < testStr.length; i++) {
|
||
const referenceCode = referenceStr.charCodeAt(i);
|
||
let testCode = testStr.charCodeAt(i);
|
||
|
||
// testCode.toLowerCase() for U+0041 LATIN CAPITAL LETTER A (A) .. U+005A LATIN CAPITAL LETTER Z (Z).
|
||
if (testCode >= 0x0041 && testCode <= 0x005A) {
|
||
testCode = testCode | 32;
|
||
}
|
||
|
||
if (testCode !== referenceCode) {
|
||
return false;
|
||
}
|
||
}
|
||
|
||
return true;
|
||
}
|
||
|
||
function isContextEdgeDelim(token) {
|
||
if (token.type !== TYPE.Delim) {
|
||
return false;
|
||
}
|
||
|
||
// Fix matching for unicode-range: U+30??, U+FF00-FF9F
|
||
// Probably we need to check out previous match instead
|
||
return token.value !== '?';
|
||
}
|
||
|
||
function isCommaContextStart(token) {
|
||
if (token === null) {
|
||
return true;
|
||
}
|
||
|
||
return (
|
||
token.type === TYPE.Comma ||
|
||
token.type === TYPE.Function ||
|
||
token.type === TYPE.LeftParenthesis ||
|
||
token.type === TYPE.LeftSquareBracket ||
|
||
token.type === TYPE.LeftCurlyBracket ||
|
||
isContextEdgeDelim(token)
|
||
);
|
||
}
|
||
|
||
function isCommaContextEnd(token) {
|
||
if (token === null) {
|
||
return true;
|
||
}
|
||
|
||
return (
|
||
token.type === TYPE.RightParenthesis ||
|
||
token.type === TYPE.RightSquareBracket ||
|
||
token.type === TYPE.RightCurlyBracket ||
|
||
(token.type === TYPE.Delim && token.value === '/')
|
||
);
|
||
}
|
||
|
||
function internalMatch(tokens, state, syntaxes) {
|
||
function moveToNextToken() {
|
||
do {
|
||
tokenIndex++;
|
||
token = tokenIndex < tokens.length ? tokens[tokenIndex] : null;
|
||
} while (token !== null && (token.type === TYPE.WhiteSpace || token.type === TYPE.Comment));
|
||
}
|
||
|
||
function getNextToken(offset) {
|
||
const nextIndex = tokenIndex + offset;
|
||
|
||
return nextIndex < tokens.length ? tokens[nextIndex] : null;
|
||
}
|
||
|
||
function stateSnapshotFromSyntax(nextState, prev) {
|
||
return {
|
||
nextState,
|
||
matchStack,
|
||
syntaxStack,
|
||
thenStack,
|
||
tokenIndex,
|
||
prev
|
||
};
|
||
}
|
||
|
||
function pushThenStack(nextState) {
|
||
thenStack = {
|
||
nextState,
|
||
matchStack,
|
||
syntaxStack,
|
||
prev: thenStack
|
||
};
|
||
}
|
||
|
||
function pushElseStack(nextState) {
|
||
elseStack = stateSnapshotFromSyntax(nextState, elseStack);
|
||
}
|
||
|
||
function addTokenToMatch() {
|
||
matchStack = {
|
||
type: TOKEN,
|
||
syntax: state.syntax,
|
||
token,
|
||
prev: matchStack
|
||
};
|
||
|
||
moveToNextToken();
|
||
syntaxStash = null;
|
||
|
||
if (tokenIndex > longestMatch) {
|
||
longestMatch = tokenIndex;
|
||
}
|
||
}
|
||
|
||
function openSyntax() {
|
||
syntaxStack = {
|
||
syntax: state.syntax,
|
||
opts: state.syntax.opts || (syntaxStack !== null && syntaxStack.opts) || null,
|
||
prev: syntaxStack
|
||
};
|
||
|
||
matchStack = {
|
||
type: OPEN_SYNTAX,
|
||
syntax: state.syntax,
|
||
token: matchStack.token,
|
||
prev: matchStack
|
||
};
|
||
}
|
||
|
||
function closeSyntax() {
|
||
if (matchStack.type === OPEN_SYNTAX) {
|
||
matchStack = matchStack.prev;
|
||
} else {
|
||
matchStack = {
|
||
type: CLOSE_SYNTAX,
|
||
syntax: syntaxStack.syntax,
|
||
token: matchStack.token,
|
||
prev: matchStack
|
||
};
|
||
}
|
||
|
||
syntaxStack = syntaxStack.prev;
|
||
}
|
||
|
||
let syntaxStack = null;
|
||
let thenStack = null;
|
||
let elseStack = null;
|
||
|
||
// null – stashing allowed, nothing stashed
|
||
// false – stashing disabled, nothing stashed
|
||
// anithing else – fail stashable syntaxes, some syntax stashed
|
||
let syntaxStash = null;
|
||
|
||
let iterationCount = 0; // count iterations and prevent infinite loop
|
||
let exitReason = null;
|
||
|
||
let token = null;
|
||
let tokenIndex = -1;
|
||
let longestMatch = 0;
|
||
let matchStack = {
|
||
type: STUB,
|
||
syntax: null,
|
||
token: null,
|
||
prev: null
|
||
};
|
||
|
||
moveToNextToken();
|
||
|
||
while (exitReason === null && ++iterationCount < ITERATION_LIMIT) {
|
||
// function mapList(list, fn) {
|
||
// const result = [];
|
||
// while (list) {
|
||
// result.unshift(fn(list));
|
||
// list = list.prev;
|
||
// }
|
||
// return result;
|
||
// }
|
||
// console.log('--\n',
|
||
// '#' + iterationCount,
|
||
// require('util').inspect({
|
||
// match: mapList(matchStack, x => x.type === TOKEN ? x.token && x.token.value : x.syntax ? ({ [OPEN_SYNTAX]: '<', [CLOSE_SYNTAX]: '</' }[x.type] || x.type) + '!' + x.syntax.name : null),
|
||
// token: token && token.value,
|
||
// tokenIndex,
|
||
// syntax: syntax.type + (syntax.id ? ' #' + syntax.id : '')
|
||
// }, { depth: null })
|
||
// );
|
||
switch (state.type) {
|
||
case 'Match':
|
||
if (thenStack === null) {
|
||
// turn to MISMATCH when some tokens left unmatched
|
||
if (token !== null) {
|
||
// doesn't mismatch if just one token left and it's an IE hack
|
||
if (tokenIndex !== tokens.length - 1 || (token.value !== '\\0' && token.value !== '\\9')) {
|
||
state = MISMATCH;
|
||
break;
|
||
}
|
||
}
|
||
|
||
// break the main loop, return a result - MATCH
|
||
exitReason = EXIT_REASON_MATCH;
|
||
break;
|
||
}
|
||
|
||
// go to next syntax (`then` branch)
|
||
state = thenStack.nextState;
|
||
|
||
// check match is not empty
|
||
if (state === DISALLOW_EMPTY) {
|
||
if (thenStack.matchStack === matchStack) {
|
||
state = MISMATCH;
|
||
break;
|
||
} else {
|
||
state = MATCH;
|
||
}
|
||
}
|
||
|
||
// close syntax if needed
|
||
while (thenStack.syntaxStack !== syntaxStack) {
|
||
closeSyntax();
|
||
}
|
||
|
||
// pop stack
|
||
thenStack = thenStack.prev;
|
||
break;
|
||
|
||
case 'Mismatch':
|
||
// when some syntax is stashed
|
||
if (syntaxStash !== null && syntaxStash !== false) {
|
||
// there is no else branches or a branch reduce match stack
|
||
if (elseStack === null || tokenIndex > elseStack.tokenIndex) {
|
||
// restore state from the stash
|
||
elseStack = syntaxStash;
|
||
syntaxStash = false; // disable stashing
|
||
}
|
||
} else if (elseStack === null) {
|
||
// no else branches -> break the main loop
|
||
// return a result - MISMATCH
|
||
exitReason = EXIT_REASON_MISMATCH;
|
||
break;
|
||
}
|
||
|
||
// go to next syntax (`else` branch)
|
||
state = elseStack.nextState;
|
||
|
||
// restore all the rest stack states
|
||
thenStack = elseStack.thenStack;
|
||
syntaxStack = elseStack.syntaxStack;
|
||
matchStack = elseStack.matchStack;
|
||
tokenIndex = elseStack.tokenIndex;
|
||
token = tokenIndex < tokens.length ? tokens[tokenIndex] : null;
|
||
|
||
// pop stack
|
||
elseStack = elseStack.prev;
|
||
break;
|
||
|
||
case 'MatchGraph':
|
||
state = state.match;
|
||
break;
|
||
|
||
case 'If':
|
||
// IMPORTANT: else stack push must go first,
|
||
// since it stores the state of thenStack before changes
|
||
if (state.else !== MISMATCH) {
|
||
pushElseStack(state.else);
|
||
}
|
||
|
||
if (state.then !== MATCH) {
|
||
pushThenStack(state.then);
|
||
}
|
||
|
||
state = state.match;
|
||
break;
|
||
|
||
case 'MatchOnce':
|
||
state = {
|
||
type: 'MatchOnceBuffer',
|
||
syntax: state,
|
||
index: 0,
|
||
mask: 0
|
||
};
|
||
break;
|
||
|
||
case 'MatchOnceBuffer': {
|
||
const terms = state.syntax.terms;
|
||
|
||
if (state.index === terms.length) {
|
||
// no matches at all or it's required all terms to be matched
|
||
if (state.mask === 0 || state.syntax.all) {
|
||
state = MISMATCH;
|
||
break;
|
||
}
|
||
|
||
// a partial match is ok
|
||
state = MATCH;
|
||
break;
|
||
}
|
||
|
||
// all terms are matched
|
||
if (state.mask === (1 << terms.length) - 1) {
|
||
state = MATCH;
|
||
break;
|
||
}
|
||
|
||
for (; state.index < terms.length; state.index++) {
|
||
const matchFlag = 1 << state.index;
|
||
|
||
if ((state.mask & matchFlag) === 0) {
|
||
// IMPORTANT: else stack push must go first,
|
||
// since it stores the state of thenStack before changes
|
||
pushElseStack(state);
|
||
pushThenStack({
|
||
type: 'AddMatchOnce',
|
||
syntax: state.syntax,
|
||
mask: state.mask | matchFlag
|
||
});
|
||
|
||
// match
|
||
state = terms[state.index++];
|
||
break;
|
||
}
|
||
}
|
||
break;
|
||
}
|
||
|
||
case 'AddMatchOnce':
|
||
state = {
|
||
type: 'MatchOnceBuffer',
|
||
syntax: state.syntax,
|
||
index: 0,
|
||
mask: state.mask
|
||
};
|
||
break;
|
||
|
||
case 'Enum':
|
||
if (token !== null) {
|
||
let name = token.value.toLowerCase();
|
||
|
||
// drop \0 and \9 hack from keyword name
|
||
if (name.indexOf('\\') !== -1) {
|
||
name = name.replace(/\\[09].*$/, '');
|
||
}
|
||
|
||
if (hasOwnProperty.call(state.map, name)) {
|
||
state = state.map[name];
|
||
break;
|
||
}
|
||
}
|
||
|
||
state = MISMATCH;
|
||
break;
|
||
|
||
case 'Generic': {
|
||
const opts = syntaxStack !== null ? syntaxStack.opts : null;
|
||
const lastTokenIndex = tokenIndex + Math.floor(state.fn(token, getNextToken, opts));
|
||
|
||
if (!isNaN(lastTokenIndex) && lastTokenIndex > tokenIndex) {
|
||
while (tokenIndex < lastTokenIndex) {
|
||
addTokenToMatch();
|
||
}
|
||
|
||
state = MATCH;
|
||
} else {
|
||
state = MISMATCH;
|
||
}
|
||
|
||
break;
|
||
}
|
||
|
||
case 'Type':
|
||
case 'Property': {
|
||
const syntaxDict = state.type === 'Type' ? 'types' : 'properties';
|
||
const dictSyntax = hasOwnProperty.call(syntaxes, syntaxDict) ? syntaxes[syntaxDict][state.name] : null;
|
||
|
||
if (!dictSyntax || !dictSyntax.match) {
|
||
throw new Error(
|
||
'Bad syntax reference: ' +
|
||
(state.type === 'Type'
|
||
? '<' + state.name + '>'
|
||
: '<\'' + state.name + '\'>')
|
||
);
|
||
}
|
||
|
||
// stash a syntax for types with low priority
|
||
if (syntaxStash !== false && token !== null && state.type === 'Type') {
|
||
const lowPriorityMatching =
|
||
// https://drafts.csswg.org/css-values-4/#custom-idents
|
||
// When parsing positionally-ambiguous keywords in a property value, a <custom-ident> production
|
||
// can only claim the keyword if no other unfulfilled production can claim it.
|
||
(state.name === 'custom-ident' && token.type === TYPE.Ident) ||
|
||
|
||
// https://drafts.csswg.org/css-values-4/#lengths
|
||
// ... if a `0` could be parsed as either a <number> or a <length> in a property (such as line-height),
|
||
// it must parse as a <number>
|
||
(state.name === 'length' && token.value === '0');
|
||
|
||
if (lowPriorityMatching) {
|
||
if (syntaxStash === null) {
|
||
syntaxStash = stateSnapshotFromSyntax(state, elseStack);
|
||
}
|
||
|
||
state = MISMATCH;
|
||
break;
|
||
}
|
||
}
|
||
|
||
openSyntax();
|
||
state = dictSyntax.match;
|
||
break;
|
||
}
|
||
|
||
case 'Keyword': {
|
||
const name = state.name;
|
||
|
||
if (token !== null) {
|
||
let keywordName = token.value;
|
||
|
||
// drop \0 and \9 hack from keyword name
|
||
if (keywordName.indexOf('\\') !== -1) {
|
||
keywordName = keywordName.replace(/\\[09].*$/, '');
|
||
}
|
||
|
||
if (areStringsEqualCaseInsensitive(keywordName, name)) {
|
||
addTokenToMatch();
|
||
state = MATCH;
|
||
break;
|
||
}
|
||
}
|
||
|
||
state = MISMATCH;
|
||
break;
|
||
}
|
||
|
||
case 'AtKeyword':
|
||
case 'Function':
|
||
if (token !== null && areStringsEqualCaseInsensitive(token.value, state.name)) {
|
||
addTokenToMatch();
|
||
state = MATCH;
|
||
break;
|
||
}
|
||
|
||
state = MISMATCH;
|
||
break;
|
||
|
||
case 'Token':
|
||
if (token !== null && token.value === state.value) {
|
||
addTokenToMatch();
|
||
state = MATCH;
|
||
break;
|
||
}
|
||
|
||
state = MISMATCH;
|
||
break;
|
||
|
||
case 'Comma':
|
||
if (token !== null && token.type === TYPE.Comma) {
|
||
if (isCommaContextStart(matchStack.token)) {
|
||
state = MISMATCH;
|
||
} else {
|
||
addTokenToMatch();
|
||
state = isCommaContextEnd(token) ? MISMATCH : MATCH;
|
||
}
|
||
} else {
|
||
state = isCommaContextStart(matchStack.token) || isCommaContextEnd(token) ? MATCH : MISMATCH;
|
||
}
|
||
|
||
break;
|
||
|
||
case 'String':
|
||
let string = '';
|
||
let lastTokenIndex = tokenIndex;
|
||
|
||
for (; lastTokenIndex < tokens.length && string.length < state.value.length; lastTokenIndex++) {
|
||
string += tokens[lastTokenIndex].value;
|
||
}
|
||
|
||
if (areStringsEqualCaseInsensitive(string, state.value)) {
|
||
while (tokenIndex < lastTokenIndex) {
|
||
addTokenToMatch();
|
||
}
|
||
|
||
state = MATCH;
|
||
} else {
|
||
state = MISMATCH;
|
||
}
|
||
|
||
break;
|
||
|
||
default:
|
||
throw new Error('Unknown node type: ' + state.type);
|
||
}
|
||
}
|
||
|
||
totalIterationCount += iterationCount;
|
||
|
||
switch (exitReason) {
|
||
case null:
|
||
console.warn('[csstree-match] BREAK after ' + ITERATION_LIMIT + ' iterations');
|
||
exitReason = EXIT_REASON_ITERATION_LIMIT;
|
||
matchStack = null;
|
||
break;
|
||
|
||
case EXIT_REASON_MATCH:
|
||
while (syntaxStack !== null) {
|
||
closeSyntax();
|
||
}
|
||
break;
|
||
|
||
default:
|
||
matchStack = null;
|
||
}
|
||
|
||
return {
|
||
tokens,
|
||
reason: exitReason,
|
||
iterations: iterationCount,
|
||
match: matchStack,
|
||
longestMatch
|
||
};
|
||
}
|
||
|
||
export function matchAsList(tokens, matchGraph, syntaxes) {
|
||
const matchResult = internalMatch(tokens, matchGraph, syntaxes || {});
|
||
|
||
if (matchResult.match !== null) {
|
||
let item = reverseList(matchResult.match).prev;
|
||
|
||
matchResult.match = [];
|
||
|
||
while (item !== null) {
|
||
switch (item.type) {
|
||
case OPEN_SYNTAX:
|
||
case CLOSE_SYNTAX:
|
||
matchResult.match.push({
|
||
type: item.type,
|
||
syntax: item.syntax
|
||
});
|
||
break;
|
||
|
||
default:
|
||
matchResult.match.push({
|
||
token: item.token.value,
|
||
node: item.token.node
|
||
});
|
||
break;
|
||
}
|
||
|
||
item = item.prev;
|
||
}
|
||
}
|
||
|
||
return matchResult;
|
||
}
|
||
|
||
export function matchAsTree(tokens, matchGraph, syntaxes) {
|
||
const matchResult = internalMatch(tokens, matchGraph, syntaxes || {});
|
||
|
||
if (matchResult.match === null) {
|
||
return matchResult;
|
||
}
|
||
|
||
let item = matchResult.match;
|
||
let host = matchResult.match = {
|
||
syntax: matchGraph.syntax || null,
|
||
match: []
|
||
};
|
||
const hostStack = [host];
|
||
|
||
// revert a list and start with 2nd item since 1st is a stub item
|
||
item = reverseList(item).prev;
|
||
|
||
// build a tree
|
||
while (item !== null) {
|
||
switch (item.type) {
|
||
case OPEN_SYNTAX:
|
||
host.match.push(host = {
|
||
syntax: item.syntax,
|
||
match: []
|
||
});
|
||
hostStack.push(host);
|
||
break;
|
||
|
||
case CLOSE_SYNTAX:
|
||
hostStack.pop();
|
||
host = hostStack[hostStack.length - 1];
|
||
break;
|
||
|
||
default:
|
||
host.match.push({
|
||
syntax: item.syntax || null,
|
||
token: item.token.value,
|
||
node: item.token.node
|
||
});
|
||
}
|
||
|
||
item = item.prev;
|
||
}
|
||
|
||
return matchResult;
|
||
}
|