AST refactoring 4/6: Rewrite compilation to not extend the AST nodes

redux
David Majda 15 years ago
parent 5885a34dde
commit 41abb7ad92

@ -47,8 +47,8 @@ PEG.ArrayUtils = {
},
/*
* The code needs to be in sync with a code template in
* PEG.Grammar.Action.prototype.compile.
* The code needs to be in sync with the code template in the compilation
* function for "action" nodes.
*/
contains: function(array, value) {
/*
@ -90,8 +90,8 @@ PEG.StringUtils = {
* Surrounds the string with quotes and escapes characters inside so that the
* result is a valid JavaScript string.
*
* The code needs to be in sync with a code template in
* PEG.Grammar.Action.prototype.compile.
* The code needs to be in sync with th code template in the compilation
* function for "action" nodes.
*/
quote: function(s) {
/*
@ -447,6 +447,397 @@ PEG.Compiler = {
}
],
_compileFunctions: {
rule: function(node) {
var resultVar = PEG.Compiler.generateUniqueIdentifier("result");
if (node.displayName !== null) {
var setReportMatchFailuresCode = PEG.Compiler.formatCode(
"var savedReportMatchFailures = context.reportMatchFailures;",
"context.reportMatchFailures = false;"
);
var restoreReportMatchFailuresCode = PEG.Compiler.formatCode(
"context.reportMatchFailures = savedReportMatchFailures;"
);
var reportMatchFailureCode = PEG.Compiler.formatCode(
"if (context.reportMatchFailures && ${resultVar} === null) {",
" this._matchFailed(${displayName|string});",
"}",
{
displayName: node.displayName,
resultVar: resultVar
}
);
} else {
var setReportMatchFailuresCode = "";
var restoreReportMatchFailuresCode = "";
var reportMatchFailureCode = "";
}
return PEG.Compiler.formatCode(
"_parse_${name}: function(context) {",
" var cacheKey = ${name|string} + '@' + this._pos;",
" var cachedResult = this._cache[cacheKey];",
" if (cachedResult !== undefined) {",
" this._pos = cachedResult.nextPos;",
" return cachedResult.result;",
" }",
" ",
" var pos = this._pos;",
" ",
" ${setReportMatchFailuresCode}",
" ${code}",
" ${restoreReportMatchFailuresCode}",
" ${reportMatchFailureCode}",
" ",
" this._cache[cacheKey] = {",
" nextPos: this._pos,",
" result: ${resultVar}",
" };",
" return ${resultVar};",
"},",
{
name: node.name,
setReportMatchFailuresCode: setReportMatchFailuresCode,
restoreReportMatchFailuresCode: restoreReportMatchFailuresCode,
reportMatchFailureCode: reportMatchFailureCode,
code: PEG.Compiler.compileNode(node.expression, resultVar),
resultVar: resultVar
}
);
},
/*
* The contract for all code fragments generated by the following functions
* is as follows:
*
* * The code fragment should try to match a part of the input starting with
* the position indicated in |this._pos|. That position may point past the
* end of the input.
*
* * If the code fragment matches the input, it advances |this._pos| after
* the matched part of the input and sets variable with a name stored in
* |resultVar| to appropriate value, which is always non-null.
*
* * If the code fragment does not match the input, it does not change
* |this._pos| and it sets a variable with a name stored in |resultVar| to
* |null|.
*/
choice: function(node, resultVar) {
var code = PEG.Compiler.formatCode(
"var ${resultVar} = null;",
{ resultVar: resultVar }
);
for (var i = node.alternatives.length - 1; i >= 0; i--) {
var alternativeResultVar = PEG.Compiler.generateUniqueIdentifier("result");
code = PEG.Compiler.formatCode(
"${alternativeCode}",
"if (${alternativeResultVar} !== null) {",
" var ${resultVar} = ${alternativeResultVar};",
"} else {",
" ${code};",
"}",
{
alternativeCode: PEG.Compiler.compileNode(node.alternatives[i], alternativeResultVar),
alternativeResultVar: alternativeResultVar,
code: code,
resultVar: resultVar
}
);
}
return code;
},
sequence: function(node, resultVar) {
var savedPosVar = PEG.Compiler.generateUniqueIdentifier("savedPos");
var elementResultVars = PEG.ArrayUtils.map(node.elements, function() {
return PEG.Compiler.generateUniqueIdentifier("result")
});
var code = PEG.Compiler.formatCode(
"var ${resultVar} = ${elementResultVarArray};",
{
resultVar: resultVar,
elementResultVarArray: "[" + elementResultVars.join(", ") + "]"
}
);
for (var i = node.elements.length - 1; i >= 0; i--) {
code = PEG.Compiler.formatCode(
"${elementCode}",
"if (${elementResultVar} !== null) {",
" ${code}",
"} else {",
" var ${resultVar} = null;",
" this._pos = ${savedPosVar};",
"}",
{
elementCode: PEG.Compiler.compileNode(node.elements[i], elementResultVars[i]),
elementResultVar: elementResultVars[i],
code: code,
savedPosVar: savedPosVar,
resultVar: resultVar
}
);
}
return PEG.Compiler.formatCode(
"var ${savedPosVar} = this._pos;",
"${code}",
{
code: code,
savedPosVar: savedPosVar
}
);
},
and_predicate: function(node, resultVar) {
var savedPosVar = PEG.Compiler.generateUniqueIdentifier("savedPos");
var savedReportMatchFailuresVar = PEG.Compiler.generateUniqueIdentifier("savedReportMatchFailuresVar");
var expressionResultVar = PEG.Compiler.generateUniqueIdentifier("result");
return PEG.Compiler.formatCode(
"var ${savedPosVar} = this._pos;",
"var ${savedReportMatchFailuresVar} = context.reportMatchFailures;",
"context.reportMatchFailures = false;",
"${expressionCode}",
"context.reportMatchFailures = ${savedReportMatchFailuresVar};",
"if (${expressionResultVar} !== null) {",
" var ${resultVar} = '';",
" this._pos = ${savedPosVar};",
"} else {",
" var ${resultVar} = null;",
"}",
{
expressionCode: PEG.Compiler.compileNode(node.expression, expressionResultVar),
expressionResultVar: expressionResultVar,
savedPosVar: savedPosVar,
savedReportMatchFailuresVar: savedReportMatchFailuresVar,
resultVar: resultVar
}
);
},
not_predicate: function(node, resultVar) {
var savedPosVar = PEG.Compiler.generateUniqueIdentifier("savedPos");
var savedReportMatchFailuresVar = PEG.Compiler.generateUniqueIdentifier("savedReportMatchFailuresVar");
var expressionResultVar = PEG.Compiler.generateUniqueIdentifier("result");
return PEG.Compiler.formatCode(
"var ${savedPosVar} = this._pos;",
"var ${savedReportMatchFailuresVar} = context.reportMatchFailures;",
"context.reportMatchFailures = false;",
"${expressionCode}",
"context.reportMatchFailures = ${savedReportMatchFailuresVar};",
"if (${expressionResultVar} === null) {",
" var ${resultVar} = '';",
"} else {",
" var ${resultVar} = null;",
" this._pos = ${savedPosVar};",
"}",
{
expressionCode: PEG.Compiler.compileNode(node.expression, expressionResultVar),
expressionResultVar: expressionResultVar,
savedPosVar: savedPosVar,
savedReportMatchFailuresVar: savedReportMatchFailuresVar,
resultVar: resultVar
}
);
},
optional: function(node, resultVar) {
var expressionResultVar = PEG.Compiler.generateUniqueIdentifier("result");
return PEG.Compiler.formatCode(
"${expressionCode}",
"var ${resultVar} = ${expressionResultVar} !== null ? ${expressionResultVar} : '';",
{
expressionCode: PEG.Compiler.compileNode(node.expression, expressionResultVar),
expressionResultVar: expressionResultVar,
resultVar: resultVar
}
);
},
zero_or_more: function(node, resultVar) {
var expressionResultVar = PEG.Compiler.generateUniqueIdentifier("result");
return PEG.Compiler.formatCode(
"var ${resultVar} = [];",
"${expressionCode}",
"while (${expressionResultVar} !== null) {",
" ${resultVar}.push(${expressionResultVar});",
" ${expressionCode}",
"}",
{
expressionCode: PEG.Compiler.compileNode(node.expression, expressionResultVar),
expressionResultVar: expressionResultVar,
resultVar: resultVar
}
);
},
one_or_more: function(node, resultVar) {
var expressionResultVar = PEG.Compiler.generateUniqueIdentifier("result");
return PEG.Compiler.formatCode(
"${expressionCode}",
"if (${expressionResultVar} !== null) {",
" var ${resultVar} = [];",
" while (${expressionResultVar} !== null) {",
" ${resultVar}.push(${expressionResultVar});",
" ${expressionCode}",
" }",
"} else {",
" var ${resultVar} = null;",
"}",
{
expressionCode: PEG.Compiler.compileNode(node.expression, expressionResultVar),
expressionResultVar: expressionResultVar,
resultVar: resultVar
}
);
},
action: function(node, resultVar) {
/*
* In case of sequences, we splat their elements into function arguments
* one by one. Example:
*
* start: "a" "b" "c" { alert(arguments.length) } // => 3
*
* This behavior is reflected in this function.
*/
var expressionResultVar = PEG.Compiler.generateUniqueIdentifier("result");
if (node.expression instanceof PEG.Grammar.Sequence) {
var params = PEG.ArrayUtils.map(
PEG.ArrayUtils.range(1, node.expression.elements.length + 1),
function(n) { return "$" + n; }
).join(", ");
var invocationCode = PEG.Compiler.formatCode(
"(function(${params}) { ${action} }).apply(null, ${expressionResultVar})",
{
params: params,
action: node.action,
expressionResultVar: expressionResultVar
}
);
} else {
var invocationCode = PEG.Compiler.formatCode(
"(function($1) { ${action} })(${expressionResultVar})",
{
action: node.action,
expressionResultVar: expressionResultVar
}
);
}
return PEG.Compiler.formatCode(
"${expressionCode}",
"var ${resultVar} = ${expressionResultVar} !== null",
" ? ${invocationCode}",
" : null;",
{
expressionCode: PEG.Compiler.compileNode(node.expression, expressionResultVar),
expressionResultVar: expressionResultVar,
invocationCode: invocationCode,
resultVar: resultVar
}
);
},
rule_ref: function(node, resultVar) {
return PEG.Compiler.formatCode(
"var ${resultVar} = this.${ruleMethod}(context);",
{
ruleMethod: "_parse_" + node.name,
resultVar: resultVar
}
);
},
literal: function(node, resultVar) {
return PEG.Compiler.formatCode(
"if (this._input.substr(this._pos, ${length}) === ${value|string}) {",
" var ${resultVar} = ${value|string};",
" this._pos += ${length};",
"} else {",
" var ${resultVar} = null;",
" if (context.reportMatchFailures) {",
" this._matchFailed(this._quoteString(${value|string}));",
" }",
"}",
{
value: node.value,
length: node.value.length,
resultVar: resultVar
}
);
},
any: function(node, resultVar) {
return PEG.Compiler.formatCode(
"if (this._input.length > this._pos) {",
" var ${resultVar} = this._input.charAt(this._pos);",
" this._pos++;",
"} else {",
" var ${resultVar} = null;",
" if (context.reportMatchFailures) {",
" this._matchFailed('any character');",
" }",
"}",
{ resultVar: resultVar }
);
},
"class": function(node, resultVar) {
/*
* Stupid IE considers regexps /[]/ and /[^]/ syntactically invalid, so we
* translate them into euqivalents it can handle.
*/
if (node.characters === "") {
var regexp = "/^(?!)/";
} else if (node.characters === "^") {
var regexp = "/^[\\S\\s]/";
} else {
var regexp = "/^[" + node.characters + "]/";
}
return PEG.Compiler.formatCode(
"if (this._input.substr(this._pos).match(${regexp}) !== null) {",
" var ${resultVar} = this._input.charAt(this._pos);",
" this._pos++;",
"} else {",
" var ${resultVar} = null;",
" if (context.reportMatchFailures) {",
" this._matchFailed('[' + ${characters|string} + ']');",
" }",
"}",
{
characters: node.characters,
regexp: regexp,
resultVar: resultVar
}
);
}
},
/*
* Compiles an AST node and returns the generated code. The |resultVar|
* parameter contains a name of variable in which the match result will be
* stored in the generated code.
*/
compileNode: function(node, resultVar) {
return this._compileFunctions[node.type](node, resultVar);
},
/*
* Generates a parser from a specified grammar AST and start rule. Throws
* |PEG.Grammar.GrammarError| if the AST contains a semantic error. Note that
@ -466,7 +857,7 @@ PEG.Compiler = {
var parseFunctionDefinitions = [];
for (var rule in ast) {
parseFunctionDefinitions.push(ast[rule].compile());
parseFunctionDefinitions.push(this.compileNode(ast[rule]));
}
var source = this.formatCode(
@ -671,385 +1062,4 @@ PEG.Compiler = {
}
};
PEG.Grammar.Rule.prototype.compile = function() {
var resultVar = PEG.Compiler.generateUniqueIdentifier("result");
if (this.displayName !== null) {
var setReportMatchFailuresCode = PEG.Compiler.formatCode(
"var savedReportMatchFailures = context.reportMatchFailures;",
"context.reportMatchFailures = false;"
);
var restoreReportMatchFailuresCode = PEG.Compiler.formatCode(
"context.reportMatchFailures = savedReportMatchFailures;"
);
var reportMatchFailureCode = PEG.Compiler.formatCode(
"if (context.reportMatchFailures && ${resultVar} === null) {",
" this._matchFailed(${displayName|string});",
"}",
{
displayName: this.displayName,
resultVar: resultVar
}
);
} else {
var setReportMatchFailuresCode = "";
var restoreReportMatchFailuresCode = "";
var reportMatchFailureCode = "";
}
return PEG.Compiler.formatCode(
"_parse_${name}: function(context) {",
" var cacheKey = ${name|string} + '@' + this._pos;",
" var cachedResult = this._cache[cacheKey];",
" if (cachedResult !== undefined) {",
" this._pos = cachedResult.nextPos;",
" return cachedResult.result;",
" }",
" ",
" var pos = this._pos;",
" ",
" ${setReportMatchFailuresCode}",
" ${code}",
" ${restoreReportMatchFailuresCode}",
" ${reportMatchFailureCode}",
" ",
" this._cache[cacheKey] = {",
" nextPos: this._pos,",
" result: ${resultVar}",
" };",
" return ${resultVar};",
"},",
{
name: this.name,
setReportMatchFailuresCode: setReportMatchFailuresCode,
restoreReportMatchFailuresCode: restoreReportMatchFailuresCode,
reportMatchFailureCode: reportMatchFailureCode,
code: this.expression.compile(resultVar),
resultVar: resultVar
}
);
};
/*
* The contract for all code fragments generated by the following |compile|
* methods is as follows:
*
* * The code fragment should try to match a part of the input starting with the
* position indicated in |this._pos|. That position may point past the end of
* the input.
*
* * If the code fragment matches the input, it advances |this._pos| after the
* matched part of the input and sets variable with a name stored in
* |resultVar| to appropriate value, which is always non-null.
*
* * If the code fragment does not match the input, it does not change
* |this._pos| and it sets a variable with a name stored in |resultVar| to
* |null|.
*/
PEG.Grammar.Choice.prototype.compile = function(resultVar) {
var code = PEG.Compiler.formatCode(
"var ${resultVar} = null;",
{ resultVar: resultVar }
);
for (var i = this.alternatives.length - 1; i >= 0; i--) {
var alternativeResultVar = PEG.Compiler.generateUniqueIdentifier("result");
code = PEG.Compiler.formatCode(
"${alternativeCode}",
"if (${alternativeResultVar} !== null) {",
" var ${resultVar} = ${alternativeResultVar};",
"} else {",
" ${code};",
"}",
{
alternativeCode: this.alternatives[i].compile(alternativeResultVar),
alternativeResultVar: alternativeResultVar,
code: code,
resultVar: resultVar
}
);
}
return code;
};
PEG.Grammar.Sequence.prototype.compile = function(resultVar) {
var savedPosVar = PEG.Compiler.generateUniqueIdentifier("savedPos");
var elementResultVars = PEG.ArrayUtils.map(this.elements, function() {
return PEG.Compiler.generateUniqueIdentifier("result")
});
var code = PEG.Compiler.formatCode(
"var ${resultVar} = ${elementResultVarArray};",
{
resultVar: resultVar,
elementResultVarArray: "[" + elementResultVars.join(", ") + "]"
}
);
for (var i = this.elements.length - 1; i >= 0; i--) {
code = PEG.Compiler.formatCode(
"${elementCode}",
"if (${elementResultVar} !== null) {",
" ${code}",
"} else {",
" var ${resultVar} = null;",
" this._pos = ${savedPosVar};",
"}",
{
elementCode: this.elements[i].compile(elementResultVars[i]),
elementResultVar: elementResultVars[i],
code: code,
savedPosVar: savedPosVar,
resultVar: resultVar
}
);
}
return PEG.Compiler.formatCode(
"var ${savedPosVar} = this._pos;",
"${code}",
{
code: code,
savedPosVar: savedPosVar
}
);
};
PEG.Grammar.AndPredicate.prototype.compile = function(resultVar) {
var savedPosVar = PEG.Compiler.generateUniqueIdentifier("savedPos");
var savedReportMatchFailuresVar = PEG.Compiler.generateUniqueIdentifier("savedReportMatchFailuresVar");
var expressionResultVar = PEG.Compiler.generateUniqueIdentifier("result");
return PEG.Compiler.formatCode(
"var ${savedPosVar} = this._pos;",
"var ${savedReportMatchFailuresVar} = context.reportMatchFailures;",
"context.reportMatchFailures = false;",
"${expressionCode}",
"context.reportMatchFailures = ${savedReportMatchFailuresVar};",
"if (${expressionResultVar} !== null) {",
" var ${resultVar} = '';",
" this._pos = ${savedPosVar};",
"} else {",
" var ${resultVar} = null;",
"}",
{
expressionCode: this.expression.compile(expressionResultVar),
expressionResultVar: expressionResultVar,
savedPosVar: savedPosVar,
savedReportMatchFailuresVar: savedReportMatchFailuresVar,
resultVar: resultVar
}
);
};
PEG.Grammar.NotPredicate.prototype.compile = function(resultVar) {
var savedPosVar = PEG.Compiler.generateUniqueIdentifier("savedPos");
var savedReportMatchFailuresVar = PEG.Compiler.generateUniqueIdentifier("savedReportMatchFailuresVar");
var expressionResultVar = PEG.Compiler.generateUniqueIdentifier("result");
return PEG.Compiler.formatCode(
"var ${savedPosVar} = this._pos;",
"var ${savedReportMatchFailuresVar} = context.reportMatchFailures;",
"context.reportMatchFailures = false;",
"${expressionCode}",
"context.reportMatchFailures = ${savedReportMatchFailuresVar};",
"if (${expressionResultVar} === null) {",
" var ${resultVar} = '';",
"} else {",
" var ${resultVar} = null;",
" this._pos = ${savedPosVar};",
"}",
{
expressionCode: this.expression.compile(expressionResultVar),
expressionResultVar: expressionResultVar,
savedPosVar: savedPosVar,
savedReportMatchFailuresVar: savedReportMatchFailuresVar,
resultVar: resultVar
}
);
};
PEG.Grammar.Optional.prototype.compile = function(resultVar) {
var expressionResultVar = PEG.Compiler.generateUniqueIdentifier("result");
return PEG.Compiler.formatCode(
"${expressionCode}",
"var ${resultVar} = ${expressionResultVar} !== null ? ${expressionResultVar} : '';",
{
expressionCode: this.expression.compile(expressionResultVar),
expressionResultVar: expressionResultVar,
resultVar: resultVar
}
);
};
PEG.Grammar.ZeroOrMore.prototype.compile = function(resultVar) {
var expressionResultVar = PEG.Compiler.generateUniqueIdentifier("result");
return PEG.Compiler.formatCode(
"var ${resultVar} = [];",
"${expressionCode}",
"while (${expressionResultVar} !== null) {",
" ${resultVar}.push(${expressionResultVar});",
" ${expressionCode}",
"}",
{
expressionCode: this.expression.compile(expressionResultVar),
expressionResultVar: expressionResultVar,
resultVar: resultVar
}
);
};
PEG.Grammar.OneOrMore.prototype.compile = function(resultVar) {
var expressionResultVar = PEG.Compiler.generateUniqueIdentifier("result");
return PEG.Compiler.formatCode(
"${expressionCode}",
"if (${expressionResultVar} !== null) {",
" var ${resultVar} = [];",
" while (${expressionResultVar} !== null) {",
" ${resultVar}.push(${expressionResultVar});",
" ${expressionCode}",
" }",
"} else {",
" var ${resultVar} = null;",
"}",
{
expressionCode: this.expression.compile(expressionResultVar),
expressionResultVar: expressionResultVar,
resultVar: resultVar
}
);
};
PEG.Grammar.Action.prototype.compile = function(resultVar) {
/*
* In case of sequences, we splat their elements into function arguments one
* by one. Example:
*
* start: "a" "b" "c" { alert(arguments.length) } // => 3
*
* This behavior is reflected in this function.
*/
var expressionResultVar = PEG.Compiler.generateUniqueIdentifier("result");
if (this.expression.type === "sequence") {
var params = PEG.ArrayUtils.map(
PEG.ArrayUtils.range(1, this.expression.elements.length + 1),
function(n) { return "$" + n; }
).join(", ");
var invocationCode = PEG.Compiler.formatCode(
"(function(${params}) { ${action} }).apply(null, ${expressionResultVar})",
{
params: params,
action: this.action,
expressionResultVar: expressionResultVar
}
);
} else {
var invocationCode = PEG.Compiler.formatCode(
"(function($1) { ${action} })(${expressionResultVar})",
{
action: this.action,
expressionResultVar: expressionResultVar
}
);
}
return PEG.Compiler.formatCode(
"${expressionCode}",
"var ${resultVar} = ${expressionResultVar} !== null",
" ? ${invocationCode}",
" : null;",
{
expressionCode: this.expression.compile(expressionResultVar),
expressionResultVar: expressionResultVar,
invocationCode: invocationCode,
resultVar: resultVar
}
);
};
PEG.Grammar.RuleRef.prototype.compile = function(resultVar) {
return PEG.Compiler.formatCode(
"var ${resultVar} = this.${ruleMethod}(context);",
{
ruleMethod: "_parse_" + this.name,
resultVar: resultVar
}
);
};
PEG.Grammar.Literal.prototype.compile = function(resultVar) {
return PEG.Compiler.formatCode(
"if (this._input.substr(this._pos, ${length}) === ${value|string}) {",
" var ${resultVar} = ${value|string};",
" this._pos += ${length};",
"} else {",
" var ${resultVar} = null;",
" if (context.reportMatchFailures) {",
" this._matchFailed(this._quoteString(${value|string}));",
" }",
"}",
{
value: this.value,
length: this.value.length,
resultVar: resultVar
}
);
};
PEG.Grammar.Any.prototype.compile = function(resultVar) {
return PEG.Compiler.formatCode(
"if (this._input.length > this._pos) {",
" var ${resultVar} = this._input.charAt(this._pos);",
" this._pos++;",
"} else {",
" var ${resultVar} = null;",
" if (context.reportMatchFailures) {",
" this._matchFailed('any character');",
" }",
"}",
{ resultVar: resultVar }
);
};
PEG.Grammar.Class.prototype.compile = function(resultVar) {
/*
* Stupid IE considers regexps /[]/ and /[^]/ syntactically invalid, so we
* translate them into euqivalents it can handle.
*/
if (this.characters === "") {
var regexp = "/^(?!)/";
} else if (this.characters === "^") {
var regexp = "/^[\\S\\s]/";
} else {
var regexp = "/^[" + this.characters + "]/";
}
return PEG.Compiler.formatCode(
"if (this._input.substr(this._pos).match(${regexp}) !== null) {",
" var ${resultVar} = this._input.charAt(this._pos);",
" this._pos++;",
"} else {",
" var ${resultVar} = null;",
" if (context.reportMatchFailures) {",
" this._matchFailed('[' + ${characters|string} + ']');",
" }",
"}",
{
characters: this.characters,
regexp: regexp,
resultVar: resultVar
}
);
};
})();

Loading…
Cancel
Save