Merge branch 'mmun-refactor-parser'

This commit is contained in:
kpdecker
2014-08-23 16:29:50 -05:00
9 changed files with 229 additions and 257 deletions
+6 -172
View File
@@ -1,6 +1,6 @@
import Exception from "../exception";
function LocationInfo(locInfo){
function LocationInfo(locInfo) {
locInfo = locInfo || {};
this.firstLine = locInfo.first_line;
this.firstColumn = locInfo.first_column;
@@ -9,41 +9,11 @@ function LocationInfo(locInfo){
}
var AST = {
ProgramNode: function(isRoot, statements, inverseStrip, inverse, locInfo) {
var inverseLocationInfo, firstInverseNode;
if (arguments.length === 4) {
locInfo = inverse;
inverse = null;
} else if (arguments.length === 3) {
locInfo = inverseStrip;
inverseStrip = null;
}
ProgramNode: function(statements, strip, locInfo) {
LocationInfo.call(this, locInfo);
this.type = "program";
this.statements = statements;
this.strip = {};
if(inverse) {
firstInverseNode = inverse[0];
if (firstInverseNode) {
inverseLocationInfo = {
first_line: firstInverseNode.firstLine,
last_line: firstInverseNode.lastLine,
last_column: firstInverseNode.lastColumn,
first_column: firstInverseNode.firstColumn
};
this.inverse = new AST.ProgramNode(isRoot, inverse, inverseStrip, inverseLocationInfo);
} else {
this.inverse = new AST.ProgramNode(isRoot, inverse, inverseStrip);
}
this.strip.right = inverseStrip.left;
} else if (inverseStrip) {
this.strip.left = inverseStrip.right;
}
// Scan all children to complete the standalone analysis
checkStandalone(this, isRoot, statements);
this.strip = strip;
},
MustacheNode: function(rawParams, hash, open, strip, locInfo) {
@@ -111,43 +81,14 @@ var AST = {
this.strip.inlineStandalone = true;
},
BlockNode: function(mustache, program, inverse, close, locInfo) {
BlockNode: function(mustache, program, inverse, strip, locInfo) {
LocationInfo.call(this, locInfo);
if(mustache.sexpr.id.original !== close.path.original) {
throw new Exception(mustache.sexpr.id.original + " doesn't match " + close.path.original, this);
}
this.type = 'block';
this.mustache = mustache;
this.program = program;
this.inverse = inverse;
var firstChild = program || inverse,
lastChild = inverse || program;
this.strip = {
left: mustache.strip.left,
right: close.strip.right,
// Determine the standalone candiacy. Basically flag our content as being possibly standalone
// so our parent can determine if we actually are standalone
openStandalone: isNextWhitespace(firstChild),
closeStandalone: isPrevWhitespace(lastChild)
};
// Calculate stripping for any else statements
firstChild.strip.left = mustache.strip.right;
lastChild.strip.right = close.strip.left;
// Find standalone else statments
if (program && inverse
&& isPrevWhitespace(program)
&& isNextWhitespace(inverse)) {
omitLeft(program);
omitRight(inverse);
}
this.strip = strip;
if (inverse && !program) {
this.isInverse = true;
@@ -165,7 +106,7 @@ var AST = {
this.type = 'block';
this.mustache = mustache;
this.program = new AST.ProgramNode(false, [content], locInfo);
this.program = new AST.ProgramNode([content], {}, locInfo);
},
ContentNode: function(string, locInfo) {
@@ -269,113 +210,6 @@ var AST = {
};
function checkStandalone(program, isRoot, statements) {
for (var i = 0, l = statements.length; i < l; i++) {
var current = statements[i],
strip = current.strip;
if (!strip) {
continue;
}
var _isPrevWhitespace = isPrevWhitespace(program, i, isRoot, current.type === 'partial'),
_isNextWhitespace = isNextWhitespace(program, i, isRoot);
strip.openStandalone = strip.openStandalone && _isPrevWhitespace;
strip.closeStandalone = strip.closeStandalone && _isNextWhitespace;
strip.inlineStandalone = strip.inlineStandalone && _isPrevWhitespace && _isNextWhitespace;
if (strip.inlineStandalone) {
omitRight(program, i);
if (omitLeft(program, i)) {
// If we are on a standalone node, save the indent info for partials
if (current.type === 'partial') {
current.indent = statements[i-1].string;
}
}
}
if (strip.openStandalone) {
omitRight(current.program || current.inverse);
// Strip out the previous content node if it's whitespace only
omitLeft(program, i);
}
if (strip.closeStandalone) {
// Always strip the next node
omitRight(program, i);
omitLeft(current.inverse || current.program);
}
}
}
function isPrevWhitespace(parent, i, isRoot, disallowIndent) {
var statements = parent.statements;
if (i === undefined) {
i = statements.length;
}
// Nodes that end with newlines are considered whitespace (but are special
// cased for strip operations)
var prev = statements[i-1];
if (prev && /\n$/.test(prev.string)) {
return true;
}
return checkWhitespace(isRoot, prev, statements[i-2]);
}
function isNextWhitespace(parent, i, isRoot) {
var statements = parent.statements;
if (i === undefined) {
i = -1;
}
return checkWhitespace(isRoot, statements[i+1], statements[i+2]);
}
function checkWhitespace(isRoot, next1, next2, disallowIndent) {
if (!next1) {
return isRoot;
} else if (next1.type === 'content') {
// Check if the previous node is empty or whitespace only
if (disallowIndent ? !next1.string : /^[\s]*$/.test(next1.string)) {
if (next2) {
return next2.type === 'content' || /\n$/.test(next1.string);
} else {
return isRoot || (next1.string.indexOf('\n') >= 0);
}
}
}
}
// Marks the node to the right of the position as omitted.
// I.e. " "{{foo}} will mark the " " node as omitted.
//
// If i is undefined, then the first child will be marked as such.
function omitRight(program, i) {
var first = program.statements[i == null ? 0 : i + 1];
if (first) {
first.omit = true;
}
}
// Marks the node to the left of the position as omitted.
// I.e. " "{{foo}} will mark the " " node as omitted.
//
// If i is undefined then the last child will be marked as such.
function omitLeft(program, i) {
var statements = program.statements;
if (i === undefined) {
i = statements.length;
}
var last = statements[i-1],
prev = statements[i-2];
// We omit the last node if it's whitespace only and not preceeded by a non-content node.
if (last && /^[\s]*$/.test(last.string) && (!prev || prev.type === 'content')) {
return last.omit = true;
}
}
// Must be exported as an object rather than the root of the module as the jison lexer
// most modify the object to operate properly.
export default AST;
+8 -2
View File
@@ -1,12 +1,18 @@
import parser from "./parser";
import AST from "./ast";
module Helpers from "./helpers";
import { extend } from "../utils";
export { parser };
var yy = {};
extend(yy, Helpers, AST);
export function parse(input) {
// Just return if an already-compile AST was passed in.
if(input.constructor === AST.ProgramNode) { return input; }
if (input.constructor === AST.ProgramNode) { return input; }
parser.yy = yy;
parser.yy = AST;
return parser.parse(input);
}
+163
View File
@@ -0,0 +1,163 @@
import Exception from "../exception";
export function stripFlags(open, close) {
return {
left: open.charAt(2) === '~',
right: close.charAt(close.length-3) === '~'
};
}
export function prepareBlock(mustache, program, inverseAndProgram, close, inverted, locInfo) {
/*jshint -W040 */
if (mustache.sexpr.id.original !== close.path.original) {
throw new Exception(mustache.sexpr.id.original + ' doesn\'t match ' + close.path.original, mustache);
}
var inverse = inverseAndProgram && inverseAndProgram.program;
var strip = {
left: mustache.strip.left,
right: close.strip.right,
// Determine the standalone candiacy. Basically flag our content as being possibly standalone
// so our parent can determine if we actually are standalone
openStandalone: isNextWhitespace(program.statements),
closeStandalone: isPrevWhitespace((inverse || program).statements)
};
if (inverse) {
var inverseStrip = inverseAndProgram.strip;
program.strip.left = mustache.strip.right;
program.strip.right = inverseStrip.left;
inverse.strip.left = inverseStrip.right;
inverse.strip.right = close.strip.left;
// Find standalone else statments
if (isPrevWhitespace(program.statements)
&& isNextWhitespace(inverse.statements)) {
omitLeft(program.statements);
omitRight(inverse.statements);
}
} else {
program.strip.left = mustache.strip.right;
program.strip.right = close.strip.left;
}
if (inverted) {
return new this.BlockNode(mustache, inverse, program, strip, locInfo);
} else {
return new this.BlockNode(mustache, program, inverse, strip, locInfo);
}
}
export function prepareProgram(statements, isRoot) {
for (var i = 0, l = statements.length; i < l; i++) {
var current = statements[i],
strip = current.strip;
if (!strip) {
continue;
}
var _isPrevWhitespace = isPrevWhitespace(statements, i, isRoot, current.type === 'partial'),
_isNextWhitespace = isNextWhitespace(statements, i, isRoot),
openStandalone = strip.openStandalone && _isPrevWhitespace,
closeStandalone = strip.closeStandalone && _isNextWhitespace,
inlineStandalone = strip.inlineStandalone && _isPrevWhitespace && _isNextWhitespace;
if (inlineStandalone) {
omitRight(statements, i);
if (omitLeft(statements, i)) {
// If we are on a standalone node, save the indent info for partials
if (current.type === 'partial') {
current.indent = statements[i-1].original;
}
}
}
if (openStandalone) {
omitRight((current.program || current.inverse).statements);
// Strip out the previous content node if it's whitespace only
omitLeft(statements, i);
}
if (closeStandalone) {
// Always strip the next node
omitRight(statements, i);
omitLeft((current.inverse || current.program).statements);
}
}
return statements;
}
function isPrevWhitespace(statements, i, isRoot, disallowIndent) {
if (i === undefined) {
i = statements.length;
}
// Nodes that end with newlines are considered whitespace (but are special
// cased for strip operations)
var prev = statements[i-1];
if (prev && /\n$/.test(prev.string)) {
return true;
}
return checkWhitespace(isRoot, prev, statements[i-2]);
}
function isNextWhitespace(statements, i, isRoot) {
if (i === undefined) {
i = -1;
}
return checkWhitespace(isRoot, statements[i+1], statements[i+2]);
}
function checkWhitespace(isRoot, next1, next2, disallowIndent) {
if (!next1) {
return isRoot;
} else if (next1.type === 'content') {
// Check if the previous node is empty or whitespace only
if (disallowIndent ? !next1.string : /^[\s]*$/.test(next1.string)) {
if (next2) {
return next2.type === 'content' || /\n$/.test(next1.string);
} else {
return isRoot || (next1.string.indexOf('\n') >= 0);
}
}
}
}
// Marks the node to the right of the position as omitted.
// I.e. {{foo}}' ' will mark the ' ' node as omitted.
//
// If i is undefined, then the first child will be marked as such.
function omitRight(statements, i) {
var first = statements[i == null ? 0 : i + 1];
if (first) {
first.omit = true;
}
}
// Marks the node to the left of the position as omitted.
// I.e. ' '{{foo}} will mark the ' ' node as omitted.
//
// If i is undefined then the last child will be marked as such.
function omitLeft(statements, i) {
if (i === undefined) {
i = statements.length;
}
var last = statements[i-1],
prev = statements[i-2];
// We omit the last node if it's whitespace only and not preceeded by a non-content node.
if (last && /^[\s]*$/.test(last.string) && (!prev || prev.type === 'content')) {
return last.omit = true;
}
}
+13 -35
View File
@@ -68,17 +68,9 @@ describe('ast', function() {
});
});
describe('BlockNode', function() {
it('should throw on mustache mismatch (old sexpr-less version)', function() {
shouldThrow(function() {
var mustacheNode = new handlebarsEnv.AST.MustacheNode([{ original: 'foo'}], null, '{{', {});
new handlebarsEnv.AST.BlockNode(mustacheNode, {}, {}, {path: {original: 'bar'}});
}, Handlebars.Exception, "foo doesn't match bar");
});
it('should throw on mustache mismatch', function() {
shouldThrow(function() {
var sexprNode = new handlebarsEnv.AST.SexprNode([{ original: 'foo'}], null);
var mustacheNode = new handlebarsEnv.AST.MustacheNode(sexprNode, null, '{{', {});
new handlebarsEnv.AST.BlockNode(mustacheNode, {}, {}, {path: {original: 'bar'}}, {first_line: 2, first_column: 2});
handlebarsEnv.parse("\n {{#foo}}{{/bar}}")
}, Handlebars.Exception, "foo doesn't match bar - 2:2");
});
@@ -197,32 +189,11 @@ describe('ast', function() {
testLocationInfoStorage(pn);
});
});
describe("ProgramNode", function(){
describe("storing location info", function(){
it("stores when `inverse` argument isn't passed", function(){
var pn = new handlebarsEnv.AST.ProgramNode(false, [], LOCATION_INFO);
testLocationInfoStorage(pn);
});
it("stores when `inverse` or `stripInverse` arguments passed", function(){
var pn = new handlebarsEnv.AST.ProgramNode(false, [], {strip: {}}, undefined, LOCATION_INFO);
testLocationInfoStorage(pn);
var clone = {
strip: {},
firstLine: 0,
lastLine: 0,
firstColumn: 0,
lastColumn: 0
};
pn = new handlebarsEnv.AST.ProgramNode(false, [], {strip: {}}, [ clone ], LOCATION_INFO);
testLocationInfoStorage(pn);
// Assert that the newly created ProgramNode has the same location
// information as the inverse
testLocationInfoStorage(pn.inverse);
});
describe('ProgramNode', function(){
it('storing location info', function(){
var pn = new handlebarsEnv.AST.ProgramNode([], {}, LOCATION_INFO);
testLocationInfoStorage(pn);
});
});
@@ -265,11 +236,18 @@ describe('ast', function() {
testColumns(blockHelperNode, 3, 7, 8, 23);
});
it('correctly records the line numbers the program of a block helper', function(){
var blockHelperNode = statements[7],
program = blockHelperNode.program;
testColumns(program, 3, 5, 8, 5);
});
it('correctly records the line numbers of an inverse of a block helper', function(){
var blockHelperNode = statements[7],
inverse = blockHelperNode.inverse;
testColumns(inverse, 5, 6, 13, 0);
testColumns(inverse, 5, 7, 5, 0);
});
});
+2 -2
View File
@@ -41,7 +41,7 @@ describe('compiler', function() {
});
it('can utilize AST instance', function() {
equal(Handlebars.compile(new Handlebars.AST.ProgramNode(true, [ new Handlebars.AST.ContentNode("Hello")]))(), 'Hello');
equal(Handlebars.compile(new Handlebars.AST.ProgramNode([ new Handlebars.AST.ContentNode("Hello")], {}))(), 'Hello');
});
it("can pass through an empty string", function() {
@@ -60,7 +60,7 @@ describe('compiler', function() {
});
it('can utilize AST instance', function() {
equal(/return "Hello"/.test(Handlebars.precompile(new Handlebars.AST.ProgramNode(true, [ new Handlebars.AST.ContentNode("Hello")]))), true);
equal(/return "Hello"/.test(Handlebars.precompile(new Handlebars.AST.ProgramNode([ new Handlebars.AST.ContentNode("Hello")]), {})), true);
});
it("can pass through an empty string", function() {
+6 -3
View File
@@ -121,11 +121,11 @@ describe('parser', function() {
});
it('parses empty blocks with empty inverse section', function() {
equals(ast_for("{{#foo}}{{^}}{{/foo}}"), "BLOCK:\n {{ ID:foo [] }}\n PROGRAM:\n");
equals(ast_for("{{#foo}}{{^}}{{/foo}}"), "BLOCK:\n {{ ID:foo [] }}\n PROGRAM:\n {{^}}\n");
});
it('parses empty blocks with empty inverse (else-style) section', function() {
equals(ast_for("{{#foo}}{{else}}{{/foo}}"), "BLOCK:\n {{ ID:foo [] }}\n PROGRAM:\n");
equals(ast_for("{{#foo}}{{else}}{{/foo}}"), "BLOCK:\n {{ ID:foo [] }}\n PROGRAM:\n {{^}}\n");
});
it('parses non-empty blocks with empty inverse section', function() {
@@ -147,6 +147,9 @@ describe('parser', function() {
it('parses a standalone inverse section', function() {
equals(ast_for("{{^foo}}bar{{/foo}}"), "BLOCK:\n {{ ID:foo [] }}\n {{^}}\n CONTENT[ 'bar' ]\n");
});
it('parses a standalone inverse section', function() {
equals(ast_for("{{else foo}}bar{{/foo}}"), "BLOCK:\n {{ ID:foo [] }}\n {{^}}\n CONTENT[ 'bar' ]\n");
});
it("raises if there's a Parse error", function() {
shouldThrow(function() {
@@ -184,7 +187,7 @@ describe('parser', function() {
describe('externally compiled AST', function() {
it('can pass through an already-compiled AST', function() {
equals(ast_for(new Handlebars.AST.ProgramNode(false, [ new Handlebars.AST.ContentNode("Hello")])), "CONTENT[ \'Hello\' ]\n");
equals(ast_for(new Handlebars.AST.ProgramNode([ new Handlebars.AST.ContentNode("Hello")])), "CONTENT[ \'Hello\' ]\n");
});
});
});
+4 -4
View File
@@ -237,10 +237,10 @@ describe('Tokenizer', function() {
shouldMatchTokens(result, ['OPEN_BLOCK', 'ID', 'CLOSE', 'CONTENT', 'OPEN_ENDBLOCK', 'ID', 'CLOSE']);
});
it('tokenizes inverse sections as "OPEN_INVERSE CLOSE"', function() {
shouldMatchTokens(tokenize("{{^}}"), ['OPEN_INVERSE', 'CLOSE']);
shouldMatchTokens(tokenize("{{else}}"), ['OPEN_INVERSE', 'CLOSE']);
shouldMatchTokens(tokenize("{{ else }}"), ['OPEN_INVERSE', 'CLOSE']);
it('tokenizes inverse sections as "INVERSE"', function() {
shouldMatchTokens(tokenize("{{^}}"), ['INVERSE']);
shouldMatchTokens(tokenize("{{else}}"), ['INVERSE']);
shouldMatchTokens(tokenize("{{ else }}"), ['INVERSE']);
});
it('tokenizes inverse sections with ID as "OPEN_INVERSE ID CLOSE"', function() {
+2
View File
@@ -70,6 +70,8 @@ ID [^\s!"#%-,\.\/;->@\[-\^`\{-~]+/{LOOKAHEAD}
<mu>"{{"{LEFT_STRIP}?">" return 'OPEN_PARTIAL';
<mu>"{{"{LEFT_STRIP}?"#" return 'OPEN_BLOCK';
<mu>"{{"{LEFT_STRIP}?"/" return 'OPEN_ENDBLOCK';
<mu>"{{"{LEFT_STRIP}?"^"\s*{RIGHT_STRIP}?"}}" this.popState(); return 'INVERSE';
<mu>"{{"{LEFT_STRIP}?\s*"else"\s*{RIGHT_STRIP}?"}}" this.popState(); return 'INVERSE';
<mu>"{{"{LEFT_STRIP}?"^" return 'OPEN_INVERSE';
<mu>"{{"{LEFT_STRIP}?\s*"else" return 'OPEN_INVERSE';
<mu>"{{"{LEFT_STRIP}?"{" return 'OPEN_UNESCAPED';
+25 -39
View File
@@ -2,78 +2,64 @@
%ebnf
%{
function stripFlags(open, close) {
return {
left: open.charAt(2) === '~',
right: close.charAt(0) === '~' || close.charAt(1) === '~'
};
}
%}
%%
root
: statements EOF { return new yy.ProgramNode(true, $1, @$); }
| EOF { return new yy.ProgramNode(true, [], @$); }
: program EOF { yy.prepareProgram($1.statements, true); return $1; }
;
program
: simpleInverse statements -> new yy.ProgramNode(false, [], $1, $2, @$)
| statements simpleInverse statements -> new yy.ProgramNode(false, $1, $2, $3, @$)
| statements simpleInverse -> new yy.ProgramNode(false, $1, $2, [], @$)
| statements -> new yy.ProgramNode(false, $1, @$)
| simpleInverse -> new yy.ProgramNode(false, [], @$)
| "" -> new yy.ProgramNode(false, [], @$)
;
statements
: statement -> [$1]
| statements statement { $1.push($2); $$ = $1; }
: statement* -> new yy.ProgramNode(yy.prepareProgram($1), {}, @$)
;
statement
: openRawBlock CONTENT END_RAW_BLOCK -> new yy.RawBlockNode($1, $2, $3, @$)
| openInverse program closeBlock -> new yy.BlockNode($1, $2.inverse, $2, $3, @$)
| openBlock program closeBlock -> new yy.BlockNode($1, $2, $2.inverse, $3, @$)
| mustache -> $1
: mustache -> $1
| block -> $1
| rawBlock -> $1
| partial -> $1
| CONTENT -> new yy.ContentNode($1, @$)
| COMMENT -> new yy.CommentNode($1, @$)
;
rawBlock
: openRawBlock CONTENT END_RAW_BLOCK -> new yy.RawBlockNode($1, $2, $3, @$)
;
openRawBlock
: OPEN_RAW_BLOCK sexpr CLOSE_RAW_BLOCK -> new yy.MustacheNode($2, null, '', '', @$)
;
block
: openBlock program inverseAndProgram? closeBlock -> yy.prepareBlock($1, $2, $3, $4, false, @$)
| openInverse program inverseAndProgram? closeBlock -> yy.prepareBlock($1, $2, $3, $4, true, @$)
;
openBlock
: OPEN_BLOCK sexpr CLOSE -> new yy.MustacheNode($2, null, $1, stripFlags($1, $3), @$)
: OPEN_BLOCK sexpr CLOSE -> new yy.MustacheNode($2, null, $1, yy.stripFlags($1, $3), @$)
;
openInverse
: OPEN_INVERSE sexpr CLOSE -> new yy.MustacheNode($2, null, $1, stripFlags($1, $3), @$)
: OPEN_INVERSE sexpr CLOSE -> new yy.MustacheNode($2, null, $1, yy.stripFlags($1, $3), @$)
;
inverseAndProgram
: INVERSE program -> { strip: yy.stripFlags($1, $1), program: $2 }
;
closeBlock
: OPEN_ENDBLOCK path CLOSE -> {path: $2, strip: stripFlags($1, $3)}
: OPEN_ENDBLOCK path CLOSE -> {path: $2, strip: yy.stripFlags($1, $3)}
;
mustache
// Parsing out the '&' escape token at AST level saves ~500 bytes after min due to the removal of one parser node.
// This also allows for handler unification as all mustache node instances can utilize the same handler
: OPEN sexpr CLOSE -> new yy.MustacheNode($2, null, $1, stripFlags($1, $3), @$)
| OPEN_UNESCAPED sexpr CLOSE_UNESCAPED -> new yy.MustacheNode($2, null, $1, stripFlags($1, $3), @$)
: OPEN sexpr CLOSE -> new yy.MustacheNode($2, null, $1, yy.stripFlags($1, $3), @$)
| OPEN_UNESCAPED sexpr CLOSE_UNESCAPED -> new yy.MustacheNode($2, null, $1, yy.stripFlags($1, $3), @$)
;
partial
: OPEN_PARTIAL partialName param hash? CLOSE -> new yy.PartialNode($2, $3, $4, stripFlags($1, $5), @$)
| OPEN_PARTIAL partialName hash? CLOSE -> new yy.PartialNode($2, undefined, $3, stripFlags($1, $4), @$)
;
simpleInverse
: OPEN_INVERSE CLOSE -> stripFlags($1, $2)
: OPEN_PARTIAL partialName param hash? CLOSE -> new yy.PartialNode($2, $3, $4, yy.stripFlags($1, $5), @$)
| OPEN_PARTIAL partialName hash? CLOSE -> new yy.PartialNode($2, undefined, $3, yy.stripFlags($1, $4), @$)
;
sexpr