parse5 3.0.3 → 5.1.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.md +14 -16
- package/lib/common/doctype.js +125 -108
- package/lib/common/error-codes.js +65 -0
- package/lib/common/foreign-content.js +265 -0
- package/lib/common/html.js +102 -102
- package/lib/common/unicode.js +81 -19
- package/lib/extensions/error-reporting/mixin-base.js +43 -0
- package/lib/extensions/error-reporting/parser-mixin.js +52 -0
- package/lib/extensions/error-reporting/preprocessor-mixin.js +24 -0
- package/lib/extensions/error-reporting/tokenizer-mixin.js +17 -0
- package/lib/extensions/location-info/open-element-stack-mixin.js +35 -0
- package/lib/extensions/location-info/parser-mixin.js +222 -0
- package/lib/extensions/location-info/tokenizer-mixin.js +146 -0
- package/lib/extensions/position-tracking/preprocessor-mixin.js +64 -0
- package/lib/index.js +6 -21
- package/lib/parser/formatting-element-list.js +181 -0
- package/lib/parser/index.js +1469 -1332
- package/lib/parser/open-element-stack.js +482 -0
- package/lib/serializer/index.js +136 -122
- package/lib/tokenizer/index.js +1856 -1804
- package/lib/tokenizer/{named_entity_data.js → named-entity-data.js} +1 -1
- package/lib/tokenizer/preprocessor.js +115 -103
- package/lib/tree-adapters/default.js +217 -0
- package/lib/utils/{merge_options.js → merge-options.js} +2 -2
- package/lib/utils/mixin.js +33 -12
- package/package.json +32 -61
- package/LICENSE +0 -19
- package/lib/common/foreign_content.js +0 -260
- package/lib/extensions/location_info/open_element_stack_mixin.js +0 -34
- package/lib/extensions/location_info/parser_mixin.js +0 -213
- package/lib/extensions/location_info/tokenizer_mixin.js +0 -117
- package/lib/extensions/position_tracking/preprocessor_mixin.js +0 -72
- package/lib/index.d.ts +0 -1077
- package/lib/parser/formatting_element_list.js +0 -167
- package/lib/parser/open_element_stack.js +0 -395
- package/lib/parser/parser_stream.js +0 -76
- package/lib/parser/plain_text_conversion_stream.js +0 -20
- package/lib/sax/dev_null_stream.js +0 -14
- package/lib/sax/index.js +0 -118
- package/lib/sax/parser_feedback_simulator.js +0 -153
- package/lib/serializer/serializer_stream.js +0 -28
- package/lib/tree_adapters/default.js +0 -209
- package/lib/tree_adapters/htmlparser2.js +0 -338
|
@@ -1,338 +0,0 @@
|
|
|
1
|
-
'use strict';
|
|
2
|
-
|
|
3
|
-
var doctype = require('../common/doctype'),
|
|
4
|
-
DOCUMENT_MODE = require('../common/html').DOCUMENT_MODE;
|
|
5
|
-
|
|
6
|
-
|
|
7
|
-
//Conversion tables for DOM Level1 structure emulation
|
|
8
|
-
var nodeTypes = {
|
|
9
|
-
element: 1,
|
|
10
|
-
text: 3,
|
|
11
|
-
cdata: 4,
|
|
12
|
-
comment: 8
|
|
13
|
-
};
|
|
14
|
-
|
|
15
|
-
var nodePropertyShorthands = {
|
|
16
|
-
tagName: 'name',
|
|
17
|
-
childNodes: 'children',
|
|
18
|
-
parentNode: 'parent',
|
|
19
|
-
previousSibling: 'prev',
|
|
20
|
-
nextSibling: 'next',
|
|
21
|
-
nodeValue: 'data'
|
|
22
|
-
};
|
|
23
|
-
|
|
24
|
-
//Node
|
|
25
|
-
var Node = function (props) {
|
|
26
|
-
for (var key in props) {
|
|
27
|
-
if (props.hasOwnProperty(key))
|
|
28
|
-
this[key] = props[key];
|
|
29
|
-
}
|
|
30
|
-
};
|
|
31
|
-
|
|
32
|
-
Node.prototype = {
|
|
33
|
-
get firstChild() {
|
|
34
|
-
var children = this.children;
|
|
35
|
-
|
|
36
|
-
return children && children[0] || null;
|
|
37
|
-
},
|
|
38
|
-
|
|
39
|
-
get lastChild() {
|
|
40
|
-
var children = this.children;
|
|
41
|
-
|
|
42
|
-
return children && children[children.length - 1] || null;
|
|
43
|
-
},
|
|
44
|
-
|
|
45
|
-
get nodeType() {
|
|
46
|
-
return nodeTypes[this.type] || nodeTypes.element;
|
|
47
|
-
}
|
|
48
|
-
};
|
|
49
|
-
|
|
50
|
-
Object.keys(nodePropertyShorthands).forEach(function (key) {
|
|
51
|
-
var shorthand = nodePropertyShorthands[key];
|
|
52
|
-
|
|
53
|
-
Object.defineProperty(Node.prototype, key, {
|
|
54
|
-
get: function () {
|
|
55
|
-
return this[shorthand] || null;
|
|
56
|
-
},
|
|
57
|
-
set: function (val) {
|
|
58
|
-
this[shorthand] = val;
|
|
59
|
-
return val;
|
|
60
|
-
}
|
|
61
|
-
});
|
|
62
|
-
});
|
|
63
|
-
|
|
64
|
-
|
|
65
|
-
//Node construction
|
|
66
|
-
exports.createDocument = function () {
|
|
67
|
-
return new Node({
|
|
68
|
-
type: 'root',
|
|
69
|
-
name: 'root',
|
|
70
|
-
parent: null,
|
|
71
|
-
prev: null,
|
|
72
|
-
next: null,
|
|
73
|
-
children: [],
|
|
74
|
-
'x-mode': DOCUMENT_MODE.NO_QUIRKS
|
|
75
|
-
});
|
|
76
|
-
};
|
|
77
|
-
|
|
78
|
-
exports.createDocumentFragment = function () {
|
|
79
|
-
return new Node({
|
|
80
|
-
type: 'root',
|
|
81
|
-
name: 'root',
|
|
82
|
-
parent: null,
|
|
83
|
-
prev: null,
|
|
84
|
-
next: null,
|
|
85
|
-
children: []
|
|
86
|
-
});
|
|
87
|
-
};
|
|
88
|
-
|
|
89
|
-
exports.createElement = function (tagName, namespaceURI, attrs) {
|
|
90
|
-
var attribs = Object.create(null),
|
|
91
|
-
attribsNamespace = Object.create(null),
|
|
92
|
-
attribsPrefix = Object.create(null);
|
|
93
|
-
|
|
94
|
-
for (var i = 0; i < attrs.length; i++) {
|
|
95
|
-
var attrName = attrs[i].name;
|
|
96
|
-
|
|
97
|
-
attribs[attrName] = attrs[i].value;
|
|
98
|
-
attribsNamespace[attrName] = attrs[i].namespace;
|
|
99
|
-
attribsPrefix[attrName] = attrs[i].prefix;
|
|
100
|
-
}
|
|
101
|
-
|
|
102
|
-
return new Node({
|
|
103
|
-
type: tagName === 'script' || tagName === 'style' ? tagName : 'tag',
|
|
104
|
-
name: tagName,
|
|
105
|
-
namespace: namespaceURI,
|
|
106
|
-
attribs: attribs,
|
|
107
|
-
'x-attribsNamespace': attribsNamespace,
|
|
108
|
-
'x-attribsPrefix': attribsPrefix,
|
|
109
|
-
children: [],
|
|
110
|
-
parent: null,
|
|
111
|
-
prev: null,
|
|
112
|
-
next: null
|
|
113
|
-
});
|
|
114
|
-
};
|
|
115
|
-
|
|
116
|
-
exports.createCommentNode = function (data) {
|
|
117
|
-
return new Node({
|
|
118
|
-
type: 'comment',
|
|
119
|
-
data: data,
|
|
120
|
-
parent: null,
|
|
121
|
-
prev: null,
|
|
122
|
-
next: null
|
|
123
|
-
});
|
|
124
|
-
};
|
|
125
|
-
|
|
126
|
-
var createTextNode = function (value) {
|
|
127
|
-
return new Node({
|
|
128
|
-
type: 'text',
|
|
129
|
-
data: value,
|
|
130
|
-
parent: null,
|
|
131
|
-
prev: null,
|
|
132
|
-
next: null
|
|
133
|
-
});
|
|
134
|
-
};
|
|
135
|
-
|
|
136
|
-
|
|
137
|
-
//Tree mutation
|
|
138
|
-
var appendChild = exports.appendChild = function (parentNode, newNode) {
|
|
139
|
-
var prev = parentNode.children[parentNode.children.length - 1];
|
|
140
|
-
|
|
141
|
-
if (prev) {
|
|
142
|
-
prev.next = newNode;
|
|
143
|
-
newNode.prev = prev;
|
|
144
|
-
}
|
|
145
|
-
|
|
146
|
-
parentNode.children.push(newNode);
|
|
147
|
-
newNode.parent = parentNode;
|
|
148
|
-
};
|
|
149
|
-
|
|
150
|
-
var insertBefore = exports.insertBefore = function (parentNode, newNode, referenceNode) {
|
|
151
|
-
var insertionIdx = parentNode.children.indexOf(referenceNode),
|
|
152
|
-
prev = referenceNode.prev;
|
|
153
|
-
|
|
154
|
-
if (prev) {
|
|
155
|
-
prev.next = newNode;
|
|
156
|
-
newNode.prev = prev;
|
|
157
|
-
}
|
|
158
|
-
|
|
159
|
-
referenceNode.prev = newNode;
|
|
160
|
-
newNode.next = referenceNode;
|
|
161
|
-
|
|
162
|
-
parentNode.children.splice(insertionIdx, 0, newNode);
|
|
163
|
-
newNode.parent = parentNode;
|
|
164
|
-
};
|
|
165
|
-
|
|
166
|
-
exports.setTemplateContent = function (templateElement, contentElement) {
|
|
167
|
-
appendChild(templateElement, contentElement);
|
|
168
|
-
};
|
|
169
|
-
|
|
170
|
-
exports.getTemplateContent = function (templateElement) {
|
|
171
|
-
return templateElement.children[0];
|
|
172
|
-
};
|
|
173
|
-
|
|
174
|
-
exports.setDocumentType = function (document, name, publicId, systemId) {
|
|
175
|
-
var data = doctype.serializeContent(name, publicId, systemId),
|
|
176
|
-
doctypeNode = null;
|
|
177
|
-
|
|
178
|
-
for (var i = 0; i < document.children.length; i++) {
|
|
179
|
-
if (document.children[i].type === 'directive' && document.children[i].name === '!doctype') {
|
|
180
|
-
doctypeNode = document.children[i];
|
|
181
|
-
break;
|
|
182
|
-
}
|
|
183
|
-
}
|
|
184
|
-
|
|
185
|
-
if (doctypeNode) {
|
|
186
|
-
doctypeNode.data = data;
|
|
187
|
-
doctypeNode['x-name'] = name;
|
|
188
|
-
doctypeNode['x-publicId'] = publicId;
|
|
189
|
-
doctypeNode['x-systemId'] = systemId;
|
|
190
|
-
}
|
|
191
|
-
|
|
192
|
-
else {
|
|
193
|
-
appendChild(document, new Node({
|
|
194
|
-
type: 'directive',
|
|
195
|
-
name: '!doctype',
|
|
196
|
-
data: data,
|
|
197
|
-
'x-name': name,
|
|
198
|
-
'x-publicId': publicId,
|
|
199
|
-
'x-systemId': systemId
|
|
200
|
-
}));
|
|
201
|
-
}
|
|
202
|
-
|
|
203
|
-
};
|
|
204
|
-
|
|
205
|
-
exports.setDocumentMode = function (document, mode) {
|
|
206
|
-
document['x-mode'] = mode;
|
|
207
|
-
};
|
|
208
|
-
|
|
209
|
-
exports.getDocumentMode = function (document) {
|
|
210
|
-
return document['x-mode'];
|
|
211
|
-
};
|
|
212
|
-
|
|
213
|
-
exports.detachNode = function (node) {
|
|
214
|
-
if (node.parent) {
|
|
215
|
-
var idx = node.parent.children.indexOf(node),
|
|
216
|
-
prev = node.prev,
|
|
217
|
-
next = node.next;
|
|
218
|
-
|
|
219
|
-
node.prev = null;
|
|
220
|
-
node.next = null;
|
|
221
|
-
|
|
222
|
-
if (prev)
|
|
223
|
-
prev.next = next;
|
|
224
|
-
|
|
225
|
-
if (next)
|
|
226
|
-
next.prev = prev;
|
|
227
|
-
|
|
228
|
-
node.parent.children.splice(idx, 1);
|
|
229
|
-
node.parent = null;
|
|
230
|
-
}
|
|
231
|
-
};
|
|
232
|
-
|
|
233
|
-
exports.insertText = function (parentNode, text) {
|
|
234
|
-
var lastChild = parentNode.children[parentNode.children.length - 1];
|
|
235
|
-
|
|
236
|
-
if (lastChild && lastChild.type === 'text')
|
|
237
|
-
lastChild.data += text;
|
|
238
|
-
else
|
|
239
|
-
appendChild(parentNode, createTextNode(text));
|
|
240
|
-
};
|
|
241
|
-
|
|
242
|
-
exports.insertTextBefore = function (parentNode, text, referenceNode) {
|
|
243
|
-
var prevNode = parentNode.children[parentNode.children.indexOf(referenceNode) - 1];
|
|
244
|
-
|
|
245
|
-
if (prevNode && prevNode.type === 'text')
|
|
246
|
-
prevNode.data += text;
|
|
247
|
-
else
|
|
248
|
-
insertBefore(parentNode, createTextNode(text), referenceNode);
|
|
249
|
-
};
|
|
250
|
-
|
|
251
|
-
exports.adoptAttributes = function (recipient, attrs) {
|
|
252
|
-
for (var i = 0; i < attrs.length; i++) {
|
|
253
|
-
var attrName = attrs[i].name;
|
|
254
|
-
|
|
255
|
-
if (typeof recipient.attribs[attrName] === 'undefined') {
|
|
256
|
-
recipient.attribs[attrName] = attrs[i].value;
|
|
257
|
-
recipient['x-attribsNamespace'][attrName] = attrs[i].namespace;
|
|
258
|
-
recipient['x-attribsPrefix'][attrName] = attrs[i].prefix;
|
|
259
|
-
}
|
|
260
|
-
}
|
|
261
|
-
};
|
|
262
|
-
|
|
263
|
-
|
|
264
|
-
//Tree traversing
|
|
265
|
-
exports.getFirstChild = function (node) {
|
|
266
|
-
return node.children[0];
|
|
267
|
-
};
|
|
268
|
-
|
|
269
|
-
exports.getChildNodes = function (node) {
|
|
270
|
-
return node.children;
|
|
271
|
-
};
|
|
272
|
-
|
|
273
|
-
exports.getParentNode = function (node) {
|
|
274
|
-
return node.parent;
|
|
275
|
-
};
|
|
276
|
-
|
|
277
|
-
exports.getAttrList = function (element) {
|
|
278
|
-
var attrList = [];
|
|
279
|
-
|
|
280
|
-
for (var name in element.attribs) {
|
|
281
|
-
attrList.push({
|
|
282
|
-
name: name,
|
|
283
|
-
value: element.attribs[name],
|
|
284
|
-
namespace: element['x-attribsNamespace'][name],
|
|
285
|
-
prefix: element['x-attribsPrefix'][name]
|
|
286
|
-
});
|
|
287
|
-
}
|
|
288
|
-
|
|
289
|
-
return attrList;
|
|
290
|
-
};
|
|
291
|
-
|
|
292
|
-
|
|
293
|
-
//Node data
|
|
294
|
-
exports.getTagName = function (element) {
|
|
295
|
-
return element.name;
|
|
296
|
-
};
|
|
297
|
-
|
|
298
|
-
exports.getNamespaceURI = function (element) {
|
|
299
|
-
return element.namespace;
|
|
300
|
-
};
|
|
301
|
-
|
|
302
|
-
exports.getTextNodeContent = function (textNode) {
|
|
303
|
-
return textNode.data;
|
|
304
|
-
};
|
|
305
|
-
|
|
306
|
-
exports.getCommentNodeContent = function (commentNode) {
|
|
307
|
-
return commentNode.data;
|
|
308
|
-
};
|
|
309
|
-
|
|
310
|
-
exports.getDocumentTypeNodeName = function (doctypeNode) {
|
|
311
|
-
return doctypeNode['x-name'];
|
|
312
|
-
};
|
|
313
|
-
|
|
314
|
-
exports.getDocumentTypeNodePublicId = function (doctypeNode) {
|
|
315
|
-
return doctypeNode['x-publicId'];
|
|
316
|
-
};
|
|
317
|
-
|
|
318
|
-
exports.getDocumentTypeNodeSystemId = function (doctypeNode) {
|
|
319
|
-
return doctypeNode['x-systemId'];
|
|
320
|
-
};
|
|
321
|
-
|
|
322
|
-
|
|
323
|
-
//Node types
|
|
324
|
-
exports.isTextNode = function (node) {
|
|
325
|
-
return node.type === 'text';
|
|
326
|
-
};
|
|
327
|
-
|
|
328
|
-
exports.isCommentNode = function (node) {
|
|
329
|
-
return node.type === 'comment';
|
|
330
|
-
};
|
|
331
|
-
|
|
332
|
-
exports.isDocumentTypeNode = function (node) {
|
|
333
|
-
return node.type === 'directive' && node.name === '!doctype';
|
|
334
|
-
};
|
|
335
|
-
|
|
336
|
-
exports.isElementNode = function (node) {
|
|
337
|
-
return !!node.attribs;
|
|
338
|
-
};
|