chevrotain
Version:
Chevrotain is a high performance fault tolerant javascript parsing DSL for building recursive decent parsers
489 lines • 22.7 kB
JavaScript
;
var __extends = (this && this.__extends) || (function () {
var extendStatics = function (d, b) {
extendStatics = Object.setPrototypeOf ||
({ __proto__: [] } instanceof Array && function (d, b) { d.__proto__ = b; }) ||
function (d, b) { for (var p in b) if (Object.prototype.hasOwnProperty.call(b, p)) d[p] = b[p]; };
return extendStatics(d, b);
};
return function (d, b) {
extendStatics(d, b);
function __() { this.constructor = d; }
d.prototype = b === null ? Object.create(b) : (__.prototype = b.prototype, new __());
};
})();
Object.defineProperty(exports, "__esModule", { value: true });
exports.checkPrefixAlternativesAmbiguities = exports.validateSomeNonEmptyLookaheadPath = exports.validateTooManyAlts = exports.RepetionCollector = exports.validateAmbiguousAlternationAlternatives = exports.validateEmptyOrAlternative = exports.getFirstNoneTerminal = exports.validateNoLeftRecursion = exports.validateRuleIsOverridden = exports.validateRuleDoesNotAlreadyExist = exports.OccurrenceValidationCollector = exports.identifyProductionForDuplicates = exports.validateGrammar = void 0;
var utils = require("../../utils/utils");
var utils_1 = require("../../utils/utils");
var parser_1 = require("../parser/parser");
var gast_1 = require("./gast/gast");
var lookahead_1 = require("./lookahead");
var interpreter_1 = require("./interpreter");
var gast_public_1 = require("./gast/gast_public");
var gast_visitor_public_1 = require("./gast/gast_visitor_public");
function validateGrammar(topLevels, globalMaxLookahead, tokenTypes, errMsgProvider, grammarName) {
var duplicateErrors = utils.map(topLevels, function (currTopLevel) {
return validateDuplicateProductions(currTopLevel, errMsgProvider);
});
var leftRecursionErrors = utils.map(topLevels, function (currTopRule) {
return validateNoLeftRecursion(currTopRule, currTopRule, errMsgProvider);
});
var emptyAltErrors = [];
var ambiguousAltsErrors = [];
var emptyRepetitionErrors = [];
// left recursion could cause infinite loops in the following validations.
// It is safest to first have the user fix the left recursion errors first and only then examine Further issues.
if (utils_1.every(leftRecursionErrors, utils_1.isEmpty)) {
emptyAltErrors = utils_1.map(topLevels, function (currTopRule) {
return validateEmptyOrAlternative(currTopRule, errMsgProvider);
});
ambiguousAltsErrors = utils_1.map(topLevels, function (currTopRule) {
return validateAmbiguousAlternationAlternatives(currTopRule, globalMaxLookahead, errMsgProvider);
});
emptyRepetitionErrors = validateSomeNonEmptyLookaheadPath(topLevels, globalMaxLookahead, errMsgProvider);
}
var termsNamespaceConflictErrors = checkTerminalAndNoneTerminalsNameSpace(topLevels, tokenTypes, errMsgProvider);
var tooManyAltsErrors = utils_1.map(topLevels, function (curRule) {
return validateTooManyAlts(curRule, errMsgProvider);
});
var duplicateRulesError = utils_1.map(topLevels, function (curRule) {
return validateRuleDoesNotAlreadyExist(curRule, topLevels, grammarName, errMsgProvider);
});
return (utils.flatten(duplicateErrors.concat(emptyRepetitionErrors, leftRecursionErrors, emptyAltErrors, ambiguousAltsErrors, termsNamespaceConflictErrors, tooManyAltsErrors, duplicateRulesError)));
}
exports.validateGrammar = validateGrammar;
function validateDuplicateProductions(topLevelRule, errMsgProvider) {
var collectorVisitor = new OccurrenceValidationCollector();
topLevelRule.accept(collectorVisitor);
var allRuleProductions = collectorVisitor.allProductions;
var productionGroups = utils.groupBy(allRuleProductions, identifyProductionForDuplicates);
var duplicates = utils.pick(productionGroups, function (currGroup) {
return currGroup.length > 1;
});
var errors = utils.map(utils.values(duplicates), function (currDuplicates) {
var firstProd = utils.first(currDuplicates);
var msg = errMsgProvider.buildDuplicateFoundError(topLevelRule, currDuplicates);
var dslName = gast_1.getProductionDslName(firstProd);
var defError = {
message: msg,
type: parser_1.ParserDefinitionErrorType.DUPLICATE_PRODUCTIONS,
ruleName: topLevelRule.name,
dslName: dslName,
occurrence: firstProd.idx
};
var param = getExtraProductionArgument(firstProd);
if (param) {
defError.parameter = param;
}
return defError;
});
return errors;
}
function identifyProductionForDuplicates(prod) {
return gast_1.getProductionDslName(prod) + "_#_" + prod.idx + "_#_" + getExtraProductionArgument(prod);
}
exports.identifyProductionForDuplicates = identifyProductionForDuplicates;
function getExtraProductionArgument(prod) {
if (prod instanceof gast_public_1.Terminal) {
return prod.terminalType.name;
}
else if (prod instanceof gast_public_1.NonTerminal) {
return prod.nonTerminalName;
}
else {
return "";
}
}
var OccurrenceValidationCollector = /** @class */ (function (_super) {
__extends(OccurrenceValidationCollector, _super);
function OccurrenceValidationCollector() {
var _this = _super !== null && _super.apply(this, arguments) || this;
_this.allProductions = [];
return _this;
}
OccurrenceValidationCollector.prototype.visitNonTerminal = function (subrule) {
this.allProductions.push(subrule);
};
OccurrenceValidationCollector.prototype.visitOption = function (option) {
this.allProductions.push(option);
};
OccurrenceValidationCollector.prototype.visitRepetitionWithSeparator = function (manySep) {
this.allProductions.push(manySep);
};
OccurrenceValidationCollector.prototype.visitRepetitionMandatory = function (atLeastOne) {
this.allProductions.push(atLeastOne);
};
OccurrenceValidationCollector.prototype.visitRepetitionMandatoryWithSeparator = function (atLeastOneSep) {
this.allProductions.push(atLeastOneSep);
};
OccurrenceValidationCollector.prototype.visitRepetition = function (many) {
this.allProductions.push(many);
};
OccurrenceValidationCollector.prototype.visitAlternation = function (or) {
this.allProductions.push(or);
};
OccurrenceValidationCollector.prototype.visitTerminal = function (terminal) {
this.allProductions.push(terminal);
};
return OccurrenceValidationCollector;
}(gast_visitor_public_1.GAstVisitor));
exports.OccurrenceValidationCollector = OccurrenceValidationCollector;
function validateRuleDoesNotAlreadyExist(rule, allRules, className, errMsgProvider) {
var errors = [];
var occurrences = utils_1.reduce(allRules, function (result, curRule) {
if (curRule.name === rule.name) {
return result + 1;
}
return result;
}, 0);
if (occurrences > 1) {
var errMsg = errMsgProvider.buildDuplicateRuleNameError({
topLevelRule: rule,
grammarName: className
});
errors.push({
message: errMsg,
type: parser_1.ParserDefinitionErrorType.DUPLICATE_RULE_NAME,
ruleName: rule.name
});
}
return errors;
}
exports.validateRuleDoesNotAlreadyExist = validateRuleDoesNotAlreadyExist;
// TODO: is there anyway to get only the rule names of rules inherited from the super grammars?
// This is not part of the IGrammarErrorProvider because the validation cannot be performed on
// The grammar structure, only at runtime.
function validateRuleIsOverridden(ruleName, definedRulesNames, className) {
var errors = [];
var errMsg;
if (!utils.contains(definedRulesNames, ruleName)) {
errMsg =
"Invalid rule override, rule: ->" + ruleName + "<- cannot be overridden in the grammar: ->" + className + "<-" +
"as it is not defined in any of the super grammars ";
errors.push({
message: errMsg,
type: parser_1.ParserDefinitionErrorType.INVALID_RULE_OVERRIDE,
ruleName: ruleName
});
}
return errors;
}
exports.validateRuleIsOverridden = validateRuleIsOverridden;
function validateNoLeftRecursion(topRule, currRule, errMsgProvider, path) {
if (path === void 0) { path = []; }
var errors = [];
var nextNonTerminals = getFirstNoneTerminal(currRule.definition);
if (utils.isEmpty(nextNonTerminals)) {
return [];
}
else {
var ruleName = topRule.name;
var foundLeftRecursion = utils.contains(nextNonTerminals, topRule);
if (foundLeftRecursion) {
errors.push({
message: errMsgProvider.buildLeftRecursionError({
topLevelRule: topRule,
leftRecursionPath: path
}),
type: parser_1.ParserDefinitionErrorType.LEFT_RECURSION,
ruleName: ruleName
});
}
// we are only looking for cyclic paths leading back to the specific topRule
// other cyclic paths are ignored, we still need this difference to avoid infinite loops...
var validNextSteps = utils.difference(nextNonTerminals, path.concat([topRule]));
var errorsFromNextSteps = utils.map(validNextSteps, function (currRefRule) {
var newPath = utils.cloneArr(path);
newPath.push(currRefRule);
return validateNoLeftRecursion(topRule, currRefRule, errMsgProvider, newPath);
});
return errors.concat(utils.flatten(errorsFromNextSteps));
}
}
exports.validateNoLeftRecursion = validateNoLeftRecursion;
function getFirstNoneTerminal(definition) {
var result = [];
if (utils.isEmpty(definition)) {
return result;
}
var firstProd = utils.first(definition);
/* istanbul ignore else */
if (firstProd instanceof gast_public_1.NonTerminal) {
result.push(firstProd.referencedRule);
}
else if (firstProd instanceof gast_public_1.Alternative ||
firstProd instanceof gast_public_1.Option ||
firstProd instanceof gast_public_1.RepetitionMandatory ||
firstProd instanceof gast_public_1.RepetitionMandatoryWithSeparator ||
firstProd instanceof gast_public_1.RepetitionWithSeparator ||
firstProd instanceof gast_public_1.Repetition) {
result = result.concat(getFirstNoneTerminal(firstProd.definition));
}
else if (firstProd instanceof gast_public_1.Alternation) {
// each sub definition in alternation is a FLAT
result = utils.flatten(utils.map(firstProd.definition, function (currSubDef) {
return getFirstNoneTerminal(currSubDef.definition);
}));
}
else if (firstProd instanceof gast_public_1.Terminal) {
// nothing to see, move along
}
else {
throw Error("non exhaustive match");
}
var isFirstOptional = gast_1.isOptionalProd(firstProd);
var hasMore = definition.length > 1;
if (isFirstOptional && hasMore) {
var rest = utils.drop(definition);
return result.concat(getFirstNoneTerminal(rest));
}
else {
return result;
}
}
exports.getFirstNoneTerminal = getFirstNoneTerminal;
var OrCollector = /** @class */ (function (_super) {
__extends(OrCollector, _super);
function OrCollector() {
var _this = _super !== null && _super.apply(this, arguments) || this;
_this.alternations = [];
return _this;
}
OrCollector.prototype.visitAlternation = function (node) {
this.alternations.push(node);
};
return OrCollector;
}(gast_visitor_public_1.GAstVisitor));
function validateEmptyOrAlternative(topLevelRule, errMsgProvider) {
var orCollector = new OrCollector();
topLevelRule.accept(orCollector);
var ors = orCollector.alternations;
var errors = utils.reduce(ors, function (errors, currOr) {
var exceptLast = utils.dropRight(currOr.definition);
var currErrors = utils.map(exceptLast, function (currAlternative, currAltIdx) {
var possibleFirstInAlt = interpreter_1.nextPossibleTokensAfter([currAlternative], [], null, 1);
if (utils.isEmpty(possibleFirstInAlt)) {
return {
message: errMsgProvider.buildEmptyAlternationError({
topLevelRule: topLevelRule,
alternation: currOr,
emptyChoiceIdx: currAltIdx
}),
type: parser_1.ParserDefinitionErrorType.NONE_LAST_EMPTY_ALT,
ruleName: topLevelRule.name,
occurrence: currOr.idx,
alternative: currAltIdx + 1
};
}
else {
return null;
}
});
return errors.concat(utils.compact(currErrors));
}, []);
return errors;
}
exports.validateEmptyOrAlternative = validateEmptyOrAlternative;
function validateAmbiguousAlternationAlternatives(topLevelRule, globalMaxLookahead, errMsgProvider) {
var orCollector = new OrCollector();
topLevelRule.accept(orCollector);
var ors = orCollector.alternations;
// New Handling of ignoring ambiguities
// - https://github.com/chevrotain/chevrotain/issues/869
ors = utils_1.reject(ors, function (currOr) { return currOr.ignoreAmbiguities === true; });
var errors = utils.reduce(ors, function (result, currOr) {
var currOccurrence = currOr.idx;
var actualMaxLookahead = currOr.maxLookahead || globalMaxLookahead;
var alternatives = lookahead_1.getLookaheadPathsForOr(currOccurrence, topLevelRule, actualMaxLookahead, currOr);
var altsAmbiguityErrors = checkAlternativesAmbiguities(alternatives, currOr, topLevelRule, errMsgProvider);
var altsPrefixAmbiguityErrors = checkPrefixAlternativesAmbiguities(alternatives, currOr, topLevelRule, errMsgProvider);
return result.concat(altsAmbiguityErrors, altsPrefixAmbiguityErrors);
}, []);
return errors;
}
exports.validateAmbiguousAlternationAlternatives = validateAmbiguousAlternationAlternatives;
var RepetionCollector = /** @class */ (function (_super) {
__extends(RepetionCollector, _super);
function RepetionCollector() {
var _this = _super !== null && _super.apply(this, arguments) || this;
_this.allProductions = [];
return _this;
}
RepetionCollector.prototype.visitRepetitionWithSeparator = function (manySep) {
this.allProductions.push(manySep);
};
RepetionCollector.prototype.visitRepetitionMandatory = function (atLeastOne) {
this.allProductions.push(atLeastOne);
};
RepetionCollector.prototype.visitRepetitionMandatoryWithSeparator = function (atLeastOneSep) {
this.allProductions.push(atLeastOneSep);
};
RepetionCollector.prototype.visitRepetition = function (many) {
this.allProductions.push(many);
};
return RepetionCollector;
}(gast_visitor_public_1.GAstVisitor));
exports.RepetionCollector = RepetionCollector;
function validateTooManyAlts(topLevelRule, errMsgProvider) {
var orCollector = new OrCollector();
topLevelRule.accept(orCollector);
var ors = orCollector.alternations;
var errors = utils.reduce(ors, function (errors, currOr) {
if (currOr.definition.length > 255) {
errors.push({
message: errMsgProvider.buildTooManyAlternativesError({
topLevelRule: topLevelRule,
alternation: currOr
}),
type: parser_1.ParserDefinitionErrorType.TOO_MANY_ALTS,
ruleName: topLevelRule.name,
occurrence: currOr.idx
});
}
return errors;
}, []);
return errors;
}
exports.validateTooManyAlts = validateTooManyAlts;
function validateSomeNonEmptyLookaheadPath(topLevelRules, maxLookahead, errMsgProvider) {
var errors = [];
utils_1.forEach(topLevelRules, function (currTopRule) {
var collectorVisitor = new RepetionCollector();
currTopRule.accept(collectorVisitor);
var allRuleProductions = collectorVisitor.allProductions;
utils_1.forEach(allRuleProductions, function (currProd) {
var prodType = lookahead_1.getProdType(currProd);
var actualMaxLookahead = currProd.maxLookahead || maxLookahead;
var currOccurrence = currProd.idx;
var paths = lookahead_1.getLookaheadPathsForOptionalProd(currOccurrence, currTopRule, prodType, actualMaxLookahead);
var pathsInsideProduction = paths[0];
if (utils_1.isEmpty(utils_1.flatten(pathsInsideProduction))) {
var errMsg = errMsgProvider.buildEmptyRepetitionError({
topLevelRule: currTopRule,
repetition: currProd
});
errors.push({
message: errMsg,
type: parser_1.ParserDefinitionErrorType.NO_NON_EMPTY_LOOKAHEAD,
ruleName: currTopRule.name
});
}
});
});
return errors;
}
exports.validateSomeNonEmptyLookaheadPath = validateSomeNonEmptyLookaheadPath;
function checkAlternativesAmbiguities(alternatives, alternation, rule, errMsgProvider) {
var foundAmbiguousPaths = [];
var identicalAmbiguities = utils_1.reduce(alternatives, function (result, currAlt, currAltIdx) {
// ignore (skip) ambiguities with this alternative
if (alternation.definition[currAltIdx].ignoreAmbiguities === true) {
return result;
}
utils_1.forEach(currAlt, function (currPath) {
var altsCurrPathAppearsIn = [currAltIdx];
utils_1.forEach(alternatives, function (currOtherAlt, currOtherAltIdx) {
if (currAltIdx !== currOtherAltIdx &&
lookahead_1.containsPath(currOtherAlt, currPath) &&
// ignore (skip) ambiguities with this "other" alternative
alternation.definition[currOtherAltIdx].ignoreAmbiguities !== true) {
altsCurrPathAppearsIn.push(currOtherAltIdx);
}
});
if (altsCurrPathAppearsIn.length > 1 &&
!lookahead_1.containsPath(foundAmbiguousPaths, currPath)) {
foundAmbiguousPaths.push(currPath);
result.push({
alts: altsCurrPathAppearsIn,
path: currPath
});
}
});
return result;
}, []);
var currErrors = utils.map(identicalAmbiguities, function (currAmbDescriptor) {
var ambgIndices = utils_1.map(currAmbDescriptor.alts, function (currAltIdx) { return currAltIdx + 1; });
var currMessage = errMsgProvider.buildAlternationAmbiguityError({
topLevelRule: rule,
alternation: alternation,
ambiguityIndices: ambgIndices,
prefixPath: currAmbDescriptor.path
});
return {
message: currMessage,
type: parser_1.ParserDefinitionErrorType.AMBIGUOUS_ALTS,
ruleName: rule.name,
occurrence: alternation.idx,
alternatives: [currAmbDescriptor.alts]
};
});
return currErrors;
}
function checkPrefixAlternativesAmbiguities(alternatives, alternation, rule, errMsgProvider) {
var errors = [];
// flatten
var pathsAndIndices = utils_1.reduce(alternatives, function (result, currAlt, idx) {
var currPathsAndIdx = utils_1.map(currAlt, function (currPath) {
return { idx: idx, path: currPath };
});
return result.concat(currPathsAndIdx);
}, []);
utils_1.forEach(pathsAndIndices, function (currPathAndIdx) {
var alternativeGast = alternation.definition[currPathAndIdx.idx];
// ignore (skip) ambiguities with this alternative
if (alternativeGast.ignoreAmbiguities === true) {
return;
}
var targetIdx = currPathAndIdx.idx;
var targetPath = currPathAndIdx.path;
var prefixAmbiguitiesPathsAndIndices = utils_1.findAll(pathsAndIndices, function (searchPathAndIdx) {
// prefix ambiguity can only be created from lower idx (higher priority) path
return (
// ignore (skip) ambiguities with this "other" alternative
alternation.definition[searchPathAndIdx.idx].ignoreAmbiguities !==
true &&
searchPathAndIdx.idx < targetIdx &&
// checking for strict prefix because identical lookaheads
// will be be detected using a different validation.
lookahead_1.isStrictPrefixOfPath(searchPathAndIdx.path, targetPath));
});
var currPathPrefixErrors = utils_1.map(prefixAmbiguitiesPathsAndIndices, function (currAmbPathAndIdx) {
var ambgIndices = [currAmbPathAndIdx.idx + 1, targetIdx + 1];
var occurrence = alternation.idx === 0 ? "" : alternation.idx;
var message = errMsgProvider.buildAlternationPrefixAmbiguityError({
topLevelRule: rule,
alternation: alternation,
ambiguityIndices: ambgIndices,
prefixPath: currAmbPathAndIdx.path
});
return {
message: message,
type: parser_1.ParserDefinitionErrorType.AMBIGUOUS_PREFIX_ALTS,
ruleName: rule.name,
occurrence: occurrence,
alternatives: ambgIndices
};
});
errors = errors.concat(currPathPrefixErrors);
});
return errors;
}
exports.checkPrefixAlternativesAmbiguities = checkPrefixAlternativesAmbiguities;
function checkTerminalAndNoneTerminalsNameSpace(topLevels, tokenTypes, errMsgProvider) {
var errors = [];
var tokenNames = utils_1.map(tokenTypes, function (currToken) { return currToken.name; });
utils_1.forEach(topLevels, function (currRule) {
var currRuleName = currRule.name;
if (utils_1.contains(tokenNames, currRuleName)) {
var errMsg = errMsgProvider.buildNamespaceConflictError(currRule);
errors.push({
message: errMsg,
type: parser_1.ParserDefinitionErrorType.CONFLICT_TOKENS_RULES_NAMESPACE,
ruleName: currRuleName
});
}
});
return errors;
}
//# sourceMappingURL=checks.js.map