From 1682a25b0d969dd25626eb0055bdccc5f6fb3173 Mon Sep 17 00:00:00 2001 From: David Majda Date: Sun, 25 Jul 2010 14:12:56 +0200 Subject: [PATCH] Move emitter utility functions out of |PEG.Compiler| --- lib/compiler.js | 278 ++++++++++++++++++++---------------------- test/compiler-test.js | 62 ---------- 2 files changed, 129 insertions(+), 211 deletions(-) diff --git a/lib/compiler.js b/lib/compiler.js index 3ec6bad..a97944e 100644 --- a/lib/compiler.js +++ b/lib/compiler.js @@ -130,116 +130,6 @@ PEG.RegExpUtils = { /* ===== PEG.Compiler ===== */ PEG.Compiler = { - /* - * Takes parts of code, interpolates variables inside them and joins them with - * a newline. - * - * Variables are delimited with "${" and "}" and their names must be valid - * identifiers (i.e. they must match [a-zA-Z_][a-zA-Z0-9_]*). Variable values - * are specified as properties of the last parameter (if this is an object, - * otherwise empty variable set is assumed). Undefined variables result in - * throwing |Error|. - * - * There can be a filter specified after the variable name, prefixed with "|". - * The filter name must be a valid identifier. The only recognized filter - * right now is "string", which quotes the variable value as a JavaScript - * string. Unrecognized filters result in throwing |Error|. - * - * If any part has multiple lines and the first line is indented by some - * amount of whitespace (as defined by the /\s+/ JavaScript regular - * expression), second to last lines are indented by the same amount of - * whitespace. This results in nicely indented multiline code in variables - * without making the templates look ugly. - * - * Examples: - * - * PEG.Compiler.formatCode("foo", "bar"); // "foo\nbar" - * PEG.Compiler.formatCode( - * "foo", "${bar}", - * { bar: "baz" } - * ); // "foo\nbaz" - * PEG.Compiler.formatCode("foo", "${bar}"); // throws Error - * PEG.Compiler.formatCode( - * "foo", "${bar|string}", - * { bar: "baz" } - * ); // "foo\n\"baz\"" - * PEG.Compiler.formatCode( - * "foo", "${bar|eeek}", - * { bar: "baz" } - * ); // throws Error - * PEG.Compiler.formatCode( - * "foo", "${bar}", - * { bar: " baz\nqux" } - * ); // "foo\n baz\n qux" - */ - formatCode: function() { - function interpolateVariablesInParts(parts) { - return PEG.ArrayUtils.map(parts, function(part) { - return part.replace( - /\$\{([a-zA-Z_][a-zA-Z0-9_]*)(\|([a-zA-Z_][a-zA-Z0-9_]*))?\}/g, - function(match, name, dummy, filter) { - var value = vars[name]; - if (value === undefined) { - throw new Error("Undefined variable: \"" + name + "\"."); - } - - if (filter !== undefined && filter != "") { // JavaScript engines differ here. - if (filter === "string") { - return PEG.StringUtils.quote(value); - } else { - throw new Error("Unrecognized filter: \"" + filter + "\"."); - } - } else { - return value; - } - } - ); - }); - } - - function indentMultilineParts(parts) { - return PEG.ArrayUtils.map(parts, function(part) { - if (!/\n/.test(part)) { return part; } - - var firstLineWhitespacePrefix = part.match(/^\s*/)[0]; - var lines = part.split("\n"); - var linesIndented = [lines[0]].concat( - PEG.ArrayUtils.map(lines.slice(1), function(line) { - return firstLineWhitespacePrefix + line; - }) - ); - return linesIndented.join("\n"); - }); - } - - var args = Array.prototype.slice.call(arguments); - var vars = args[args.length - 1] instanceof Object ? args.pop() : {}; - - return indentMultilineParts(interpolateVariablesInParts(args)).join("\n"); - }, - - _uniqueIdentifierCounters: {}, - - /* - * Generates a unique identifier with specified prefix. The sequence of - * generated identifiers with given prefix is repeatable and will be the same - * within different language runtimes. - */ - generateUniqueIdentifier: function(prefix) { - this._uniqueIdentifierCounters[prefix] - = this._uniqueIdentifierCounters[prefix] || 0; - return prefix + this._uniqueIdentifierCounters[prefix]++; - }, - - /* - * Resets internal counters of the unique identifier generator. The sequence - * of identifiers with given prefix generated by |generateUniqueIdentifier| - * will start from the beginning. - */ - resetUniqueIdentifierCounters: function() { - this._uniqueIdentifierCounters = {}; - }, - /* * Generates a parser from a specified grammar AST. Throws |PEG.GrammarError| * if the AST contains a semantic error. Note that not all errors are detected @@ -477,6 +367,96 @@ PEG.Compiler.passes = [ /* Emits the generated code for the AST. */ PEG.Compiler.emitter = function(ast) { + /* + * Takes parts of code, interpolates variables inside them and joins them with + * a newline. + * + * Variables are delimited with "${" and "}" and their names must be valid + * identifiers (i.e. they must match [a-zA-Z_][a-zA-Z0-9_]*). Variable values + * are specified as properties of the last parameter (if this is an object, + * otherwise empty variable set is assumed). Undefined variables result in + * throwing |Error|. + * + * There can be a filter specified after the variable name, prefixed with "|". + * The filter name must be a valid identifier. The only recognized filter + * right now is "string", which quotes the variable value as a JavaScript + * string. Unrecognized filters result in throwing |Error|. + * + * If any part has multiple lines and the first line is indented by some + * amount of whitespace (as defined by the /\s+/ JavaScript regular + * expression), second to last lines are indented by the same amount of + * whitespace. This results in nicely indented multiline code in variables + * without making the templates look ugly. + * + * Examples: + * + * formatCode("foo", "bar"); // "foo\nbar" + * formatCode("foo", "${bar}", { bar: "baz" }); // "foo\nbaz" + * formatCode("foo", "${bar}"); // throws Error + * formatCode("foo", "${bar|string}", { bar: "baz" }); // "foo\n\"baz\"" + * formatCode("foo", "${bar|eeek}", { bar: "baz" }); // throws Error + * formatCode("foo", "${bar}", { bar: " baz\nqux" }); // "foo\n baz\n qux" + */ + function formatCode() { + function interpolateVariablesInParts(parts) { + return PEG.ArrayUtils.map(parts, function(part) { + return part.replace( + /\$\{([a-zA-Z_][a-zA-Z0-9_]*)(\|([a-zA-Z_][a-zA-Z0-9_]*))?\}/g, + function(match, name, dummy, filter) { + var value = vars[name]; + if (value === undefined) { + throw new Error("Undefined variable: \"" + name + "\"."); + } + + if (filter !== undefined && filter != "") { // JavaScript engines differ here. + if (filter === "string") { + return PEG.StringUtils.quote(value); + } else { + throw new Error("Unrecognized filter: \"" + filter + "\"."); + } + } else { + return value; + } + } + ); + }); + } + + function indentMultilineParts(parts) { + return PEG.ArrayUtils.map(parts, function(part) { + if (!/\n/.test(part)) { return part; } + + var firstLineWhitespacePrefix = part.match(/^\s*/)[0]; + var lines = part.split("\n"); + var linesIndented = [lines[0]].concat( + PEG.ArrayUtils.map(lines.slice(1), function(line) { + return firstLineWhitespacePrefix + line; + }) + ); + return linesIndented.join("\n"); + }); + } + + var args = Array.prototype.slice.call(arguments); + var vars = args[args.length - 1] instanceof Object ? args.pop() : {}; + + return indentMultilineParts(interpolateVariablesInParts(args)).join("\n"); + }; + + /* Unique ID generator. */ + var UID = { + _counters: {}, + + next: function(prefix) { + this._counters[prefix] = this._counters[prefix] || 0; + return prefix + this._counters[prefix]++; + }, + + reset: function() { + this._counters = {}; + } + }; + var emitFunctions = { grammar: function(node) { var initializerCode = node.initializer !== null @@ -488,7 +468,7 @@ PEG.Compiler.emitter = function(ast) { parseFunctionDefinitions.push(emit(node.rules[name])); } - return PEG.Compiler.formatCode( + return formatCode( "(function(){", " /* Generated by PEG.js (http://pegjs.majda.cz/). */", " ", @@ -687,19 +667,19 @@ PEG.Compiler.emitter = function(ast) { * generated grammar stored in a VCS (this is true e.g. for our * metagrammar). */ - PEG.Compiler.resetUniqueIdentifierCounters(); + UID.reset(); - var resultVar = PEG.Compiler.generateUniqueIdentifier("result"); + var resultVar = UID.next("result"); if (node.displayName !== null) { - var setReportMatchFailuresCode = PEG.Compiler.formatCode( + var setReportMatchFailuresCode = formatCode( "var savedReportMatchFailures = context.reportMatchFailures;", "context.reportMatchFailures = false;" ); - var restoreReportMatchFailuresCode = PEG.Compiler.formatCode( + var restoreReportMatchFailuresCode = formatCode( "context.reportMatchFailures = savedReportMatchFailures;" ); - var reportMatchFailureCode = PEG.Compiler.formatCode( + var reportMatchFailureCode = formatCode( "if (context.reportMatchFailures && ${resultVar} === null) {", " matchFailed(${displayName|string});", "}", @@ -714,7 +694,7 @@ PEG.Compiler.emitter = function(ast) { var reportMatchFailureCode = ""; } - return PEG.Compiler.formatCode( + return formatCode( "function parse_${name}(context) {", " var cacheKey = ${name|string} + '@' + pos;", " var cachedResult = cache[cacheKey];", @@ -762,14 +742,14 @@ PEG.Compiler.emitter = function(ast) { */ choice: function(node, resultVar) { - var code = PEG.Compiler.formatCode( + var code = formatCode( "var ${resultVar} = null;", { resultVar: resultVar } ); for (var i = node.alternatives.length - 1; i >= 0; i--) { - var alternativeResultVar = PEG.Compiler.generateUniqueIdentifier("result"); - code = PEG.Compiler.formatCode( + var alternativeResultVar = UID.next("result"); + code = formatCode( "${alternativeCode}", "if (${alternativeResultVar} !== null) {", " var ${resultVar} = ${alternativeResultVar};", @@ -789,13 +769,13 @@ PEG.Compiler.emitter = function(ast) { }, sequence: function(node, resultVar) { - var savedPosVar = PEG.Compiler.generateUniqueIdentifier("savedPos"); + var savedPosVar = UID.next("savedPos"); var elementResultVars = PEG.ArrayUtils.map(node.elements, function() { - return PEG.Compiler.generateUniqueIdentifier("result") + return UID.next("result") }); - var code = PEG.Compiler.formatCode( + var code = formatCode( "var ${resultVar} = ${elementResultVarArray};", { resultVar: resultVar, @@ -804,7 +784,7 @@ PEG.Compiler.emitter = function(ast) { ); for (var i = node.elements.length - 1; i >= 0; i--) { - code = PEG.Compiler.formatCode( + code = formatCode( "${elementCode}", "if (${elementResultVar} !== null) {", " ${code}", @@ -822,7 +802,7 @@ PEG.Compiler.emitter = function(ast) { ); } - return PEG.Compiler.formatCode( + return formatCode( "var ${savedPosVar} = pos;", "${code}", { @@ -837,11 +817,11 @@ PEG.Compiler.emitter = function(ast) { }, simple_and: function(node, resultVar) { - var savedPosVar = PEG.Compiler.generateUniqueIdentifier("savedPos"); - var savedReportMatchFailuresVar = PEG.Compiler.generateUniqueIdentifier("savedReportMatchFailuresVar"); - var expressionResultVar = PEG.Compiler.generateUniqueIdentifier("result"); + var savedPosVar = UID.next("savedPos"); + var savedReportMatchFailuresVar = UID.next("savedReportMatchFailuresVar"); + var expressionResultVar = UID.next("result"); - return PEG.Compiler.formatCode( + return formatCode( "var ${savedPosVar} = pos;", "var ${savedReportMatchFailuresVar} = context.reportMatchFailures;", "context.reportMatchFailures = false;", @@ -864,11 +844,11 @@ PEG.Compiler.emitter = function(ast) { }, simple_not: function(node, resultVar) { - var savedPosVar = PEG.Compiler.generateUniqueIdentifier("savedPos"); - var savedReportMatchFailuresVar = PEG.Compiler.generateUniqueIdentifier("savedReportMatchFailuresVar"); - var expressionResultVar = PEG.Compiler.generateUniqueIdentifier("result"); + var savedPosVar = UID.next("savedPos"); + var savedReportMatchFailuresVar = UID.next("savedReportMatchFailuresVar"); + var expressionResultVar = UID.next("result"); - return PEG.Compiler.formatCode( + return formatCode( "var ${savedPosVar} = pos;", "var ${savedReportMatchFailuresVar} = context.reportMatchFailures;", "context.reportMatchFailures = false;", @@ -891,9 +871,9 @@ PEG.Compiler.emitter = function(ast) { }, semantic_and: function(node, resultVar) { - var savedPosVar = PEG.Compiler.generateUniqueIdentifier("savedPos"); + var savedPosVar = UID.next("savedPos"); - return PEG.Compiler.formatCode( + return formatCode( "var ${resultVar} = (function() {${actionCode}})() ? '' : null;", { actionCode: node.code, @@ -903,9 +883,9 @@ PEG.Compiler.emitter = function(ast) { }, semantic_not: function(node, resultVar) { - var savedPosVar = PEG.Compiler.generateUniqueIdentifier("savedPos"); + var savedPosVar = UID.next("savedPos"); - return PEG.Compiler.formatCode( + return formatCode( "var ${resultVar} = (function() {${actionCode}})() ? null : '';", { actionCode: node.code, @@ -915,9 +895,9 @@ PEG.Compiler.emitter = function(ast) { }, optional: function(node, resultVar) { - var expressionResultVar = PEG.Compiler.generateUniqueIdentifier("result"); + var expressionResultVar = UID.next("result"); - return PEG.Compiler.formatCode( + return formatCode( "${expressionCode}", "var ${resultVar} = ${expressionResultVar} !== null ? ${expressionResultVar} : '';", { @@ -929,9 +909,9 @@ PEG.Compiler.emitter = function(ast) { }, zero_or_more: function(node, resultVar) { - var expressionResultVar = PEG.Compiler.generateUniqueIdentifier("result"); + var expressionResultVar = UID.next("result"); - return PEG.Compiler.formatCode( + return formatCode( "var ${resultVar} = [];", "${expressionCode}", "while (${expressionResultVar} !== null) {", @@ -947,9 +927,9 @@ PEG.Compiler.emitter = function(ast) { }, one_or_more: function(node, resultVar) { - var expressionResultVar = PEG.Compiler.generateUniqueIdentifier("result"); + var expressionResultVar = UID.next("result"); - return PEG.Compiler.formatCode( + return formatCode( "${expressionCode}", "if (${expressionResultVar} !== null) {", " var ${resultVar} = [];", @@ -978,7 +958,7 @@ PEG.Compiler.emitter = function(ast) { * This behavior is reflected in this function. */ - var expressionResultVar = PEG.Compiler.generateUniqueIdentifier("result"); + var expressionResultVar = UID.next("result"); if (node.expression.type === "sequence") { var formalParams = []; @@ -1000,7 +980,7 @@ PEG.Compiler.emitter = function(ast) { var actualParams = []; } - return PEG.Compiler.formatCode( + return formatCode( "${expressionCode}", "var ${resultVar} = ${expressionResultVar} !== null", " ? (function(${formalParams}) {${actionCode}})(${actualParams})", @@ -1017,7 +997,7 @@ PEG.Compiler.emitter = function(ast) { }, rule_ref: function(node, resultVar) { - return PEG.Compiler.formatCode( + return formatCode( "var ${resultVar} = ${ruleMethod}(context);", { ruleMethod: "parse_" + node.name, @@ -1027,7 +1007,7 @@ PEG.Compiler.emitter = function(ast) { }, literal: function(node, resultVar) { - return PEG.Compiler.formatCode( + return formatCode( "if (input.substr(pos, ${length}) === ${value|string}) {", " var ${resultVar} = ${value|string};", " pos += ${length};", @@ -1046,7 +1026,7 @@ PEG.Compiler.emitter = function(ast) { }, any: function(node, resultVar) { - return PEG.Compiler.formatCode( + return formatCode( "if (input.length > pos) {", " var ${resultVar} = input.charAt(pos);", " pos++;", @@ -1080,7 +1060,7 @@ PEG.Compiler.emitter = function(ast) { var regexp = node.inverted ? "/^[\\S\\s]/" : "/^(?!)/"; } - return PEG.Compiler.formatCode( + return formatCode( "if (input.substr(pos).match(${regexp}) !== null) {", " var ${resultVar} = input.charAt(pos);", " pos++;", diff --git a/test/compiler-test.js b/test/compiler-test.js index 56d0815..f3eca93 100644 --- a/test/compiler-test.js +++ b/test/compiler-test.js @@ -117,68 +117,6 @@ test("quoteForClass", function() { ); }); -/* ===== PEG.Compiler ===== */ - -module("PEG.Compiler"); - -test("formatCode joins parts", function() { - strictEqual(PEG.Compiler.formatCode("foo", "bar"), "foo\nbar"); -}); - -test("formatCode interpolates variables", function() { - strictEqual( - PEG.Compiler.formatCode("foo", "${bar}", { bar: "baz" }), - "foo\nbaz" - ); - - throws( - function() { PEG.Compiler.formatCode("foo", "${bar}"); }, - Error, - { message: "Undefined variable: \"bar\"." } - ); -}); - -test("formatCode filters variables", function() { - strictEqual( - PEG.Compiler.formatCode("foo", "${bar|string}", { bar: "baz" }), - "foo\n\"baz\"" - ); - - throws( - function() { PEG.Compiler.formatCode("foo", "${bar|eeek}", { bar: "baz" }); }, - Error, - { message: "Unrecognized filter: \"eeek\"." } - ); -}); - -test("formatCode indents multiline parts", function() { - strictEqual( - PEG.Compiler.formatCode("foo", "${bar}", { bar: " baz\nqux" }), - "foo\n baz\n qux" - ); -}); - -test("generateUniqueIdentifier", function() { - notStrictEqual( - PEG.Compiler.generateUniqueIdentifier("prefix"), - PEG.Compiler.generateUniqueIdentifier("prefix") - ); -}); - -test("resetUniqueIdentifierCounters", function() { - var ida1 = PEG.Compiler.generateUniqueIdentifier("a"); - var ida2 = PEG.Compiler.generateUniqueIdentifier("a"); - var idb1 = PEG.Compiler.generateUniqueIdentifier("b"); - var idb2 = PEG.Compiler.generateUniqueIdentifier("b"); - - PEG.Compiler.resetUniqueIdentifierCounters(); - - strictEqual(PEG.Compiler.generateUniqueIdentifier("a"), ida1); - strictEqual(PEG.Compiler.generateUniqueIdentifier("a"), ida2); - strictEqual(PEG.Compiler.generateUniqueIdentifier("b"), idb1); - strictEqual(PEG.Compiler.generateUniqueIdentifier("b"), idb2); -}); - /* ===== PEG ===== */ module("PEG");