xml-tokenizer 0.0.32 → 0.0.33
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/dist/cjs/config.js +1 -43
- package/dist/cjs/get-q-name.js +1 -8
- package/dist/cjs/index.js +1 -80
- package/dist/cjs/processor/process.js +1 -0
- package/dist/cjs/processor/processors/path-tracker.js +1 -0
- package/dist/cjs/processor/validate-processor-order.js +5 -0
- package/dist/cjs/selector/TokenSelectState.js +1 -107
- package/dist/cjs/selector/TokenSelectStateMachine.js +1 -151
- package/dist/cjs/selector/TokenSelector.js +1 -140
- package/dist/cjs/selector/match-string.js +1 -22
- package/dist/cjs/selector/select.js +1 -20
- package/dist/cjs/token-to-xml.js +1 -45
- package/dist/cjs/tokenizer/XmlError.js +1 -97
- package/dist/cjs/tokenizer/XmlStream.js +1 -479
- package/dist/cjs/tokenizer/ascii-constants.js +1 -70
- package/dist/cjs/tokenizer/tokenize.js +1 -420
- package/dist/cjs/tokenizer/utils.js +1 -48
- package/dist/cjs/tokens-to-xml.js +1 -14
- package/dist/cjs/xml-to-object.js +1 -70
- package/dist/cjs/xml-to-simplified-object.js +1 -87
- package/dist/esm/config.js +1 -39
- package/dist/esm/get-q-name.js +1 -6
- package/dist/esm/index.js +1 -16
- package/dist/esm/processor/process.js +1 -0
- package/dist/esm/processor/processors/path-tracker.js +1 -0
- package/dist/esm/processor/validate-processor-order.js +5 -0
- package/dist/esm/selector/TokenSelectState.js +1 -103
- package/dist/esm/selector/TokenSelectStateMachine.js +1 -149
- package/dist/esm/selector/TokenSelector.js +1 -138
- package/dist/esm/selector/match-string.js +1 -20
- package/dist/esm/selector/select.js +1 -18
- package/dist/esm/token-to-xml.js +1 -43
- package/dist/esm/tokenizer/XmlError.js +1 -95
- package/dist/esm/tokenizer/XmlStream.js +1 -477
- package/dist/esm/tokenizer/ascii-constants.js +1 -36
- package/dist/esm/tokenizer/tokenize.js +1 -417
- package/dist/esm/tokenizer/utils.js +1 -41
- package/dist/esm/tokens-to-xml.js +1 -12
- package/dist/esm/xml-to-object.js +1 -67
- package/dist/esm/xml-to-simplified-object.js +1 -84
- package/dist/types/config.d.ts +0 -1
- package/dist/types/get-q-name.d.ts +0 -1
- package/dist/types/index.d.ts +1 -1
- package/dist/types/processor/index.d.ts +4 -0
- package/dist/types/processor/process.d.ts +3 -0
- package/dist/types/processor/processors/index.d.ts +1 -0
- package/dist/types/processor/processors/path-tracker.d.ts +5 -0
- package/dist/types/processor/types.d.ts +10 -0
- package/dist/types/processor/validate-processor-order.d.ts +2 -0
- package/dist/types/selector/TokenSelectState.d.ts +0 -1
- package/dist/types/selector/TokenSelectStateMachine.d.ts +0 -1
- package/dist/types/selector/TokenSelector.d.ts +0 -1
- package/dist/types/selector/index.d.ts +0 -1
- package/dist/types/selector/match-string.d.ts +0 -1
- package/dist/types/selector/select.d.ts +0 -1
- package/dist/types/selector/types.d.ts +0 -1
- package/dist/types/token-to-xml.d.ts +0 -1
- package/dist/types/tokenizer/XmlError.d.ts +0 -1
- package/dist/types/tokenizer/XmlStream.d.ts +7 -1
- package/dist/types/tokenizer/ascii-constants.d.ts +0 -1
- package/dist/types/tokenizer/index.d.ts +0 -1
- package/dist/types/tokenizer/tokenize.d.ts +0 -1
- package/dist/types/tokenizer/types.d.ts +0 -1
- package/dist/types/tokenizer/utils.d.ts +0 -1
- package/dist/types/tokens-to-xml.d.ts +0 -1
- package/dist/types/xml-to-object.d.ts +0 -1
- package/dist/types/xml-to-simplified-object.d.ts +0 -1
- package/package.json +6 -6
- package/dist/cjs/config.js.map +0 -1
- package/dist/cjs/get-q-name.js.map +0 -1
- package/dist/cjs/index.js.map +0 -1
- package/dist/cjs/selector/TokenSelectState.js.map +0 -1
- package/dist/cjs/selector/TokenSelectStateMachine.js.map +0 -1
- package/dist/cjs/selector/TokenSelector.js.map +0 -1
- package/dist/cjs/selector/match-string.js.map +0 -1
- package/dist/cjs/selector/select.js.map +0 -1
- package/dist/cjs/token-to-xml.js.map +0 -1
- package/dist/cjs/tokenizer/XmlError.js.map +0 -1
- package/dist/cjs/tokenizer/XmlStream.js.map +0 -1
- package/dist/cjs/tokenizer/ascii-constants.js.map +0 -1
- package/dist/cjs/tokenizer/tokenize.js.map +0 -1
- package/dist/cjs/tokenizer/utils.js.map +0 -1
- package/dist/cjs/tokens-to-xml.js.map +0 -1
- package/dist/cjs/xml-to-object.js.map +0 -1
- package/dist/cjs/xml-to-simplified-object.js.map +0 -1
- package/dist/esm/config.js.map +0 -1
- package/dist/esm/get-q-name.js.map +0 -1
- package/dist/esm/index.js.map +0 -1
- package/dist/esm/selector/TokenSelectState.js.map +0 -1
- package/dist/esm/selector/TokenSelectStateMachine.js.map +0 -1
- package/dist/esm/selector/TokenSelector.js.map +0 -1
- package/dist/esm/selector/match-string.js.map +0 -1
- package/dist/esm/selector/select.js.map +0 -1
- package/dist/esm/token-to-xml.js.map +0 -1
- package/dist/esm/tokenizer/XmlError.js.map +0 -1
- package/dist/esm/tokenizer/XmlStream.js.map +0 -1
- package/dist/esm/tokenizer/ascii-constants.js.map +0 -1
- package/dist/esm/tokenizer/tokenize.js.map +0 -1
- package/dist/esm/tokenizer/utils.js.map +0 -1
- package/dist/esm/tokens-to-xml.js.map +0 -1
- package/dist/esm/xml-to-object.js.map +0 -1
- package/dist/esm/xml-to-simplified-object.js.map +0 -1
- package/dist/types/config.d.ts.map +0 -1
- package/dist/types/get-q-name.d.ts.map +0 -1
- package/dist/types/index.d.ts.map +0 -1
- package/dist/types/selector/TokenSelectState.d.ts.map +0 -1
- package/dist/types/selector/TokenSelectStateMachine.d.ts.map +0 -1
- package/dist/types/selector/TokenSelector.d.ts.map +0 -1
- package/dist/types/selector/index.d.ts.map +0 -1
- package/dist/types/selector/match-string.d.ts.map +0 -1
- package/dist/types/selector/select.d.ts.map +0 -1
- package/dist/types/selector/types.d.ts.map +0 -1
- package/dist/types/token-to-xml.d.ts.map +0 -1
- package/dist/types/tokenizer/XmlError.d.ts.map +0 -1
- package/dist/types/tokenizer/XmlStream.d.ts.map +0 -1
- package/dist/types/tokenizer/ascii-constants.d.ts.map +0 -1
- package/dist/types/tokenizer/index.d.ts.map +0 -1
- package/dist/types/tokenizer/tokenize.d.ts.map +0 -1
- package/dist/types/tokenizer/types.d.ts.map +0 -1
- package/dist/types/tokenizer/utils.d.ts.map +0 -1
- package/dist/types/tokens-to-xml.d.ts.map +0 -1
- package/dist/types/xml-to-object.d.ts.map +0 -1
- package/dist/types/xml-to-simplified-object.d.ts.map +0 -1
|
@@ -1,420 +1 @@
|
|
|
1
|
-
|
|
2
|
-
|
|
3
|
-
var getQName = require('../get-q-name.js');
|
|
4
|
-
var asciiConstants = require('./ascii-constants.js');
|
|
5
|
-
var utils = require('./utils.js');
|
|
6
|
-
var XmlError = require('./XmlError.js');
|
|
7
|
-
var XmlStream = require('./XmlStream.js');
|
|
8
|
-
|
|
9
|
-
const BOM = "\uFEFF";
|
|
10
|
-
const XML_DECLARATION_START = "<?xml ";
|
|
11
|
-
const XML_DECLARATION_END = "?>";
|
|
12
|
-
const ENTITY_DECLARATION_START = "<!ENTITY";
|
|
13
|
-
const ELEMENT_DECLARATION_START = "<!ELEMENT";
|
|
14
|
-
const ATTLIST_DECLARATION_START = "<!ATTLIST";
|
|
15
|
-
const NOTATION_DECLARATION_START = "<!NOTATION";
|
|
16
|
-
const DOCTYPE_START = "<!DOCTYPE";
|
|
17
|
-
const PI_START = "<?";
|
|
18
|
-
const PI_END = "?>";
|
|
19
|
-
const COMMENT_START = "<!--";
|
|
20
|
-
const COMMENT_END = "-->";
|
|
21
|
-
const CDATA_START = "<![CDATA[";
|
|
22
|
-
const CDATA_END = "]]>";
|
|
23
|
-
const NDATA = "NDATA";
|
|
24
|
-
const PUBLIC = "PUBLIC";
|
|
25
|
-
const SYSTEM = "SYSTEM";
|
|
26
|
-
const VERSION = "version";
|
|
27
|
-
const ENCODING = "encoding";
|
|
28
|
-
const STANDALONE = "standalone";
|
|
29
|
-
function tokenize(xmlString, tokenCallback, options = {}) {
|
|
30
|
-
tokenizeXmlStream(new XmlStream.XmlStream(xmlString, options), tokenCallback);
|
|
31
|
-
}
|
|
32
|
-
function tokenizeXmlStream(s, tokenCallback) {
|
|
33
|
-
if (s.startsWith(BOM)) {
|
|
34
|
-
s.advance(1);
|
|
35
|
-
}
|
|
36
|
-
if (s.startsWith(XML_DECLARATION_START)) {
|
|
37
|
-
parseDeclaration(s);
|
|
38
|
-
}
|
|
39
|
-
parseMisc(s, tokenCallback);
|
|
40
|
-
s.skipSpaces();
|
|
41
|
-
if (s.startsWith(DOCTYPE_START) || s.startsWith(DOCTYPE_START.toLowerCase())) {
|
|
42
|
-
if (!s.config.allowDtd) {
|
|
43
|
-
throw new XmlError.XmlError({ type: "DtdDetected" });
|
|
44
|
-
}
|
|
45
|
-
parseDoctype(s, tokenCallback);
|
|
46
|
-
parseMisc(s, tokenCallback);
|
|
47
|
-
}
|
|
48
|
-
s.skipSpaces();
|
|
49
|
-
if (s.config.strictDocument) {
|
|
50
|
-
if (!s.atEnd() && s.currCodeUnit() === asciiConstants.LESS_THAN) {
|
|
51
|
-
parseElement(s, tokenCallback);
|
|
52
|
-
}
|
|
53
|
-
parseMisc(s, tokenCallback);
|
|
54
|
-
if (!s.atEnd()) {
|
|
55
|
-
throw new XmlError.XmlError({ type: "UnknownToken", message: "Not at end" }, s.genTextPos());
|
|
56
|
-
}
|
|
57
|
-
} else {
|
|
58
|
-
while (!s.atEnd()) {
|
|
59
|
-
if (s.currCodeUnit() === asciiConstants.LESS_THAN) {
|
|
60
|
-
parseElement(s, tokenCallback);
|
|
61
|
-
} else {
|
|
62
|
-
parseText(s, tokenCallback);
|
|
63
|
-
}
|
|
64
|
-
parseMisc(s, tokenCallback);
|
|
65
|
-
}
|
|
66
|
-
}
|
|
67
|
-
}
|
|
68
|
-
function parseMisc(s, tokenCallback) {
|
|
69
|
-
while (!s.atEnd()) {
|
|
70
|
-
s.skipSpaces();
|
|
71
|
-
if (s.startsWith(COMMENT_START)) {
|
|
72
|
-
parseComment(s, tokenCallback);
|
|
73
|
-
} else if (s.startsWith(PI_START)) {
|
|
74
|
-
parsePi(s, tokenCallback);
|
|
75
|
-
} else {
|
|
76
|
-
break;
|
|
77
|
-
}
|
|
78
|
-
}
|
|
79
|
-
}
|
|
80
|
-
function parseDeclaration(s) {
|
|
81
|
-
function consumeSpaces(_s) {
|
|
82
|
-
if (_s.startsWithSpace()) {
|
|
83
|
-
_s.skipSpaces();
|
|
84
|
-
} else if (!_s.startsWith(XML_DECLARATION_END) && !_s.atEnd()) {
|
|
85
|
-
throw new XmlError.XmlError(
|
|
86
|
-
{ type: "InvalidChar", expected: "a whitespace", actual: _s.currCodeUnitUnchecked() },
|
|
87
|
-
_s.genTextPos()
|
|
88
|
-
);
|
|
89
|
-
}
|
|
90
|
-
}
|
|
91
|
-
s.advance(5);
|
|
92
|
-
consumeSpaces(s);
|
|
93
|
-
if (!s.startsWith(VERSION)) {
|
|
94
|
-
throw new XmlError.XmlError({ type: "InvalidString", expected: "version" }, s.genTextPos());
|
|
95
|
-
}
|
|
96
|
-
parseAttribute(s);
|
|
97
|
-
consumeSpaces(s);
|
|
98
|
-
if (s.startsWith(ENCODING)) {
|
|
99
|
-
parseAttribute(s);
|
|
100
|
-
consumeSpaces(s);
|
|
101
|
-
}
|
|
102
|
-
if (s.startsWith(STANDALONE)) {
|
|
103
|
-
parseAttribute(s);
|
|
104
|
-
}
|
|
105
|
-
s.skipSpaces();
|
|
106
|
-
s.skipString(XML_DECLARATION_END);
|
|
107
|
-
}
|
|
108
|
-
function parseComment(s, tokenCallback) {
|
|
109
|
-
const start = s.getPos();
|
|
110
|
-
s.advance(4);
|
|
111
|
-
const text = s.consumeCodeUnitsWhile((c, _s) => !(c === asciiConstants.HYPHEN && _s.startsWith(COMMENT_END)));
|
|
112
|
-
s.skipString(COMMENT_END);
|
|
113
|
-
if (text.includes("--") || text.endsWith("-")) {
|
|
114
|
-
throw new XmlError.XmlError({ type: "InvalidComment" }, s.genTextPosFrom(start));
|
|
115
|
-
}
|
|
116
|
-
tokenCallback({ type: "Comment", text, range: s.rangeFrom(start) }, s);
|
|
117
|
-
}
|
|
118
|
-
function parsePi(s, tokenCallback) {
|
|
119
|
-
if (s.startsWith(XML_DECLARATION_START)) {
|
|
120
|
-
throw new XmlError.XmlError({ type: "UnexpectedDeclaration" }, s.genTextPos());
|
|
121
|
-
}
|
|
122
|
-
const start = s.getPos();
|
|
123
|
-
s.advance(2);
|
|
124
|
-
const target = s.consumeName();
|
|
125
|
-
s.skipSpaces();
|
|
126
|
-
const content = s.consumeCodeUnitsWhile(
|
|
127
|
-
(c, _s) => !(c === asciiConstants.QUESTION_MARK && _s.startsWith(PI_END))
|
|
128
|
-
);
|
|
129
|
-
s.skipString(PI_END);
|
|
130
|
-
tokenCallback(
|
|
131
|
-
{
|
|
132
|
-
type: "ProcessingInstruction",
|
|
133
|
-
target,
|
|
134
|
-
content: content.length === 0 ? void 0 : content,
|
|
135
|
-
range: s.rangeFrom(start)
|
|
136
|
-
},
|
|
137
|
-
s
|
|
138
|
-
);
|
|
139
|
-
}
|
|
140
|
-
function parseDoctype(s, tokenCallback) {
|
|
141
|
-
const start = s.getPos();
|
|
142
|
-
parseDoctypeStart(s);
|
|
143
|
-
s.skipSpaces();
|
|
144
|
-
if (s.currCodeUnit() === asciiConstants.GREATER_THAN) {
|
|
145
|
-
s.advance(1);
|
|
146
|
-
return;
|
|
147
|
-
}
|
|
148
|
-
s.advance(1);
|
|
149
|
-
while (!s.atEnd()) {
|
|
150
|
-
s.skipSpaces();
|
|
151
|
-
if (s.startsWith(ENTITY_DECLARATION_START)) {
|
|
152
|
-
parseEntityDecl(s, tokenCallback);
|
|
153
|
-
} else if (s.startsWith(COMMENT_START)) {
|
|
154
|
-
parseComment(s, tokenCallback);
|
|
155
|
-
} else if (s.startsWith(PI_START)) {
|
|
156
|
-
parsePi(s, tokenCallback);
|
|
157
|
-
} else if (s.startsWith("]")) {
|
|
158
|
-
s.advance(1);
|
|
159
|
-
s.skipSpaces();
|
|
160
|
-
if (s.currCodeUnit() === asciiConstants.GREATER_THAN) {
|
|
161
|
-
s.advance(1);
|
|
162
|
-
break;
|
|
163
|
-
} else {
|
|
164
|
-
throw new XmlError.XmlError(
|
|
165
|
-
{ type: "InvalidChar", expected: "'>'", actual: s.currCodeUnitUnchecked() },
|
|
166
|
-
s.genTextPos()
|
|
167
|
-
);
|
|
168
|
-
}
|
|
169
|
-
} else if (s.startsWith(ELEMENT_DECLARATION_START) || s.startsWith(ATTLIST_DECLARATION_START) || s.startsWith(NOTATION_DECLARATION_START)) {
|
|
170
|
-
try {
|
|
171
|
-
consumeDecl(s);
|
|
172
|
-
} catch (e) {
|
|
173
|
-
throw new XmlError.XmlError(
|
|
174
|
-
{ type: "UnknownToken", message: "Failed to consume declaration" },
|
|
175
|
-
s.genTextPosFrom(start)
|
|
176
|
-
);
|
|
177
|
-
}
|
|
178
|
-
} else {
|
|
179
|
-
throw new XmlError.XmlError(
|
|
180
|
-
{ type: "UnknownToken", message: "Failed to parse doctype" },
|
|
181
|
-
s.genTextPos()
|
|
182
|
-
);
|
|
183
|
-
}
|
|
184
|
-
}
|
|
185
|
-
}
|
|
186
|
-
function parseDoctypeStart(s) {
|
|
187
|
-
s.advance(9);
|
|
188
|
-
s.consumeSpaces();
|
|
189
|
-
s.skipName();
|
|
190
|
-
s.skipSpaces();
|
|
191
|
-
parseExternalId(s);
|
|
192
|
-
s.skipSpaces();
|
|
193
|
-
const currCodeUnit = s.currCodeUnit();
|
|
194
|
-
if (currCodeUnit !== asciiConstants.OPEN_SQUARE_BRACKET && currCodeUnit !== asciiConstants.GREATER_THAN) {
|
|
195
|
-
throw new XmlError.XmlError(
|
|
196
|
-
{ type: "InvalidChar", expected: "'[' or '>'", actual: currCodeUnit },
|
|
197
|
-
s.genTextPos()
|
|
198
|
-
);
|
|
199
|
-
}
|
|
200
|
-
}
|
|
201
|
-
function parseExternalId(s) {
|
|
202
|
-
if (s.startsWith(SYSTEM) || s.startsWith(PUBLIC)) {
|
|
203
|
-
const start = s.getPos();
|
|
204
|
-
s.advance(6);
|
|
205
|
-
const id = s.sliceBack(start);
|
|
206
|
-
s.consumeSpaces();
|
|
207
|
-
const quote = s.consumeQuote();
|
|
208
|
-
s.consumeCodeUnitsWhile((c) => c !== quote);
|
|
209
|
-
s.consumeCodeUnit(quote);
|
|
210
|
-
if (id === PUBLIC) {
|
|
211
|
-
s.consumeSpaces();
|
|
212
|
-
const _quote = s.consumeQuote();
|
|
213
|
-
s.consumeCodeUnitsWhile((c) => c !== _quote);
|
|
214
|
-
s.consumeCodeUnit(_quote);
|
|
215
|
-
}
|
|
216
|
-
return true;
|
|
217
|
-
}
|
|
218
|
-
return false;
|
|
219
|
-
}
|
|
220
|
-
function parseEntityDecl(s, tokenCallback) {
|
|
221
|
-
s.advance(8);
|
|
222
|
-
s.consumeSpaces();
|
|
223
|
-
const isGe = !s.tryConsumeCodeUnit(asciiConstants.PERCENT);
|
|
224
|
-
if (!isGe) {
|
|
225
|
-
s.consumeSpaces();
|
|
226
|
-
}
|
|
227
|
-
const name = s.consumeName();
|
|
228
|
-
s.consumeSpaces();
|
|
229
|
-
const definition = parseEntityDef(s, isGe);
|
|
230
|
-
if (definition !== null) {
|
|
231
|
-
tokenCallback({ type: "EntityDeclaration", name, definition }, s);
|
|
232
|
-
}
|
|
233
|
-
s.skipSpaces();
|
|
234
|
-
s.consumeCodeUnit(asciiConstants.GREATER_THAN);
|
|
235
|
-
}
|
|
236
|
-
function parseEntityDef(s, isGe) {
|
|
237
|
-
const currCodeUnit = s.currCodeUnit();
|
|
238
|
-
if (currCodeUnit === asciiConstants.DOUBLE_QUOTE || currCodeUnit === asciiConstants.SINGLE_QUOTE) {
|
|
239
|
-
const quote = s.consumeQuote();
|
|
240
|
-
const start = s.getPos();
|
|
241
|
-
s.skipCodeUnitsWhile((c) => c !== quote);
|
|
242
|
-
const value = s.sliceBack(start);
|
|
243
|
-
s.consumeCodeUnit(quote);
|
|
244
|
-
return value;
|
|
245
|
-
} else if (currCodeUnit === asciiConstants.UPPERCASE_S || currCodeUnit === asciiConstants.UPPERCASE_P) {
|
|
246
|
-
if (parseExternalId(s)) {
|
|
247
|
-
if (isGe) {
|
|
248
|
-
s.skipSpaces();
|
|
249
|
-
if (s.startsWith(NDATA)) {
|
|
250
|
-
s.advance(5);
|
|
251
|
-
s.consumeSpaces();
|
|
252
|
-
s.skipName();
|
|
253
|
-
}
|
|
254
|
-
}
|
|
255
|
-
return null;
|
|
256
|
-
}
|
|
257
|
-
throw new XmlError.XmlError({ type: "InvalidExternalID" }, s.genTextPos());
|
|
258
|
-
} else {
|
|
259
|
-
throw new XmlError.XmlError(
|
|
260
|
-
{ type: "InvalidChar", expected: "a quote, SYSTEM or PUBLIC", actual: currCodeUnit },
|
|
261
|
-
s.genTextPos()
|
|
262
|
-
);
|
|
263
|
-
}
|
|
264
|
-
}
|
|
265
|
-
function consumeDecl(s) {
|
|
266
|
-
s.skipCodeUnitsWhile((c) => c !== asciiConstants.GREATER_THAN);
|
|
267
|
-
s.consumeCodeUnit(asciiConstants.GREATER_THAN);
|
|
268
|
-
}
|
|
269
|
-
function parseElement(s, tokenCallback) {
|
|
270
|
-
const start = s.getPos();
|
|
271
|
-
s.advance(1);
|
|
272
|
-
const [prefix, local] = s.consumeQName();
|
|
273
|
-
tokenCallback({ type: "ElementStart", prefix, local, start }, s);
|
|
274
|
-
let open = false;
|
|
275
|
-
while (!s.atEnd()) {
|
|
276
|
-
const hasSpace = s.startsWithSpace();
|
|
277
|
-
s.skipSpaces();
|
|
278
|
-
const _start = s.getPos();
|
|
279
|
-
const currCodeUnit = s.currCodeUnit();
|
|
280
|
-
if (currCodeUnit === asciiConstants.SLASH) {
|
|
281
|
-
s.advance(1);
|
|
282
|
-
s.consumeCodeUnit(asciiConstants.GREATER_THAN);
|
|
283
|
-
const range = s.rangeFrom(_start);
|
|
284
|
-
tokenCallback({ type: "ElementEnd", end: { type: "Empty" }, range }, s);
|
|
285
|
-
break;
|
|
286
|
-
} else if (currCodeUnit === asciiConstants.GREATER_THAN) {
|
|
287
|
-
s.advance(1);
|
|
288
|
-
const range = s.rangeFrom(_start);
|
|
289
|
-
if (s.config.implicitSelfClosingElements != null && s.config.implicitSelfClosingElements.includes(getQName.getQName(local, prefix))) {
|
|
290
|
-
tokenCallback({ type: "ElementEnd", end: { type: "Empty" }, range }, s);
|
|
291
|
-
break;
|
|
292
|
-
} else {
|
|
293
|
-
tokenCallback({ type: "ElementEnd", end: { type: "Open" }, range }, s);
|
|
294
|
-
open = true;
|
|
295
|
-
break;
|
|
296
|
-
}
|
|
297
|
-
} else {
|
|
298
|
-
if (!hasSpace) {
|
|
299
|
-
throw new XmlError.XmlError(
|
|
300
|
-
{ type: "InvalidChar", expected: "a whitespace", actual: s.currCodeUnitUnchecked() },
|
|
301
|
-
s.genTextPos()
|
|
302
|
-
);
|
|
303
|
-
}
|
|
304
|
-
const [_prefix, _local, value] = parseAttribute(s);
|
|
305
|
-
const end = s.getPos();
|
|
306
|
-
tokenCallback(
|
|
307
|
-
{
|
|
308
|
-
type: "Attribute",
|
|
309
|
-
range: { start: _start, end },
|
|
310
|
-
prefix: _prefix,
|
|
311
|
-
local: _local,
|
|
312
|
-
value
|
|
313
|
-
},
|
|
314
|
-
s
|
|
315
|
-
);
|
|
316
|
-
}
|
|
317
|
-
}
|
|
318
|
-
if (open) {
|
|
319
|
-
if (s.config.rawTextElements != null) {
|
|
320
|
-
const qName = getQName.getQName(local, prefix);
|
|
321
|
-
if (s.config.rawTextElements.includes(qName)) {
|
|
322
|
-
parseContent(s, tokenCallback, `</${qName}>`);
|
|
323
|
-
return;
|
|
324
|
-
}
|
|
325
|
-
}
|
|
326
|
-
parseContent(s, tokenCallback);
|
|
327
|
-
}
|
|
328
|
-
}
|
|
329
|
-
function parseAttribute(s) {
|
|
330
|
-
const [prefix, local] = s.consumeQName();
|
|
331
|
-
let value;
|
|
332
|
-
const start = s.getPos();
|
|
333
|
-
s.skipSpaces();
|
|
334
|
-
if (s.tryConsumeCodeUnit(asciiConstants.EQUALS)) {
|
|
335
|
-
s.skipSpaces();
|
|
336
|
-
const currCodeUnit = s.currCodeUnit();
|
|
337
|
-
if (currCodeUnit === asciiConstants.SINGLE_QUOTE || currCodeUnit === asciiConstants.DOUBLE_QUOTE) {
|
|
338
|
-
const quote = s.consumeQuote();
|
|
339
|
-
value = s.consumeCodeUnitsWhile((c) => c !== quote && c !== asciiConstants.LESS_THAN);
|
|
340
|
-
s.consumeCodeUnit(quote);
|
|
341
|
-
} else if (!s.config.strictDocument) {
|
|
342
|
-
value = s.consumeCodeUnitsWhile(
|
|
343
|
-
(c) => !utils.isXmlSpaceByte(c) && c !== asciiConstants.GREATER_THAN && c !== asciiConstants.SLASH && c !== asciiConstants.LESS_THAN
|
|
344
|
-
);
|
|
345
|
-
} else {
|
|
346
|
-
throw new XmlError.XmlError(
|
|
347
|
-
{ type: "InvalidChar", expected: "a quote", actual: currCodeUnit },
|
|
348
|
-
s.genTextPos()
|
|
349
|
-
);
|
|
350
|
-
}
|
|
351
|
-
} else if (!s.config.strictDocument) {
|
|
352
|
-
s.goTo(start);
|
|
353
|
-
value = "true";
|
|
354
|
-
} else {
|
|
355
|
-
throw new XmlError.XmlError(
|
|
356
|
-
{ type: "InvalidChar", expected: asciiConstants.EQUALS, actual: s.currCodeUnit() },
|
|
357
|
-
s.genTextPos()
|
|
358
|
-
);
|
|
359
|
-
}
|
|
360
|
-
return [prefix, local, value];
|
|
361
|
-
}
|
|
362
|
-
function parseContent(s, tokenCallback, parseContentUntil = asciiConstants.LESS_THAN) {
|
|
363
|
-
while (!s.atEnd()) {
|
|
364
|
-
if (typeof parseContentUntil === "number" ? s.currCodeUnit() === parseContentUntil : s.startsWith(parseContentUntil)) {
|
|
365
|
-
const nextCodeUnit = s.nextCodeUnit();
|
|
366
|
-
if (nextCodeUnit === asciiConstants.EXCLAMATION_MARK) {
|
|
367
|
-
if (s.startsWith(COMMENT_START)) {
|
|
368
|
-
parseComment(s, tokenCallback);
|
|
369
|
-
} else if (s.startsWith(CDATA_START)) {
|
|
370
|
-
parseCdata(s, tokenCallback);
|
|
371
|
-
} else {
|
|
372
|
-
throw new XmlError.XmlError(
|
|
373
|
-
{ type: "UnknownToken", message: "Failed to parse content" },
|
|
374
|
-
s.genTextPos()
|
|
375
|
-
);
|
|
376
|
-
}
|
|
377
|
-
} else if (nextCodeUnit === asciiConstants.QUESTION_MARK) {
|
|
378
|
-
parsePi(s, tokenCallback);
|
|
379
|
-
} else if (nextCodeUnit === asciiConstants.SLASH) {
|
|
380
|
-
parseCloseElement(s, tokenCallback);
|
|
381
|
-
break;
|
|
382
|
-
} else {
|
|
383
|
-
parseElement(s, tokenCallback);
|
|
384
|
-
}
|
|
385
|
-
} else {
|
|
386
|
-
parseText(s, tokenCallback, parseContentUntil);
|
|
387
|
-
}
|
|
388
|
-
}
|
|
389
|
-
}
|
|
390
|
-
function parseCdata(s, tokenCallback) {
|
|
391
|
-
const start = s.getPos();
|
|
392
|
-
s.advance(9);
|
|
393
|
-
const text = s.consumeCodeUnitsWhile(
|
|
394
|
-
(c, _s) => !(c === asciiConstants.CLOSE_SQUARE_BRACKET && _s.startsWith(CDATA_END))
|
|
395
|
-
);
|
|
396
|
-
s.skipString(CDATA_END);
|
|
397
|
-
const range = s.rangeFrom(start);
|
|
398
|
-
tokenCallback({ type: "Cdata", text, range }, s);
|
|
399
|
-
}
|
|
400
|
-
function parseCloseElement(s, tokenCallback) {
|
|
401
|
-
const start = s.getPos();
|
|
402
|
-
s.advance(2);
|
|
403
|
-
const [prefix, local] = s.consumeQName();
|
|
404
|
-
s.skipSpaces();
|
|
405
|
-
s.consumeCodeUnit(asciiConstants.GREATER_THAN);
|
|
406
|
-
const range = s.rangeFrom(start);
|
|
407
|
-
tokenCallback({ type: "ElementEnd", end: { type: "Close", prefix, local }, range }, s);
|
|
408
|
-
}
|
|
409
|
-
function parseText(s, tokenCallback, parseTextUntil = asciiConstants.LESS_THAN) {
|
|
410
|
-
const start = s.getPos();
|
|
411
|
-
const text = typeof parseTextUntil === "number" ? s.consumeCodeUnitsWhile((c) => c !== parseTextUntil) : s.consumeCodeUnitsWhile((_, _s) => !_s.startsWith(parseTextUntil));
|
|
412
|
-
if (text.includes(CDATA_END)) {
|
|
413
|
-
throw new XmlError.XmlError({ type: "InvalidCharacterData" }, s.genTextPos());
|
|
414
|
-
}
|
|
415
|
-
tokenCallback({ type: "Text", text, range: s.rangeFrom(start) }, s);
|
|
416
|
-
}
|
|
417
|
-
|
|
418
|
-
exports.tokenize = tokenize;
|
|
419
|
-
exports.tokenizeXmlStream = tokenizeXmlStream;
|
|
420
|
-
//# sourceMappingURL=tokenize.js.map
|
|
1
|
+
"use strict";var U=require("../get-q-name.js"),s=require("./ascii-constants.js"),L=require("./utils.js"),c=require("./XmlError.js"),H=require("./XmlStream.js");const Q="\uFEFF",A="<?xml ",k="?>",D="<!ENTITY",F="<!ELEMENT",O="<!ATTLIST",G="<!NOTATION",p="<!DOCTYPE",w="<?",W="?>",E="<!--",y="-->",b="<![CDATA[",f="]]>",B="NDATA",N="PUBLIC",M="SYSTEM",q="version",K="encoding",Y="standalone";function z(t,n,e={}){P(new H.XmlStream(t,e),n)}function P(t,n){if(t.startsWith(Q)&&t.advance(1),t.startsWith(A)&&$(t),m(t,n),t.config.strictDocument){if(t.skipSpaces(),t.startsWithIgnoreCase(p)&&(S(t,n),m(t,n)),t.skipSpaces(),!t.atEnd()&&t.currCodeUnit()===s.LESS_THAN&&T(t,n),m(t,n),!t.atEnd())throw new c.XmlError({type:"UnknownToken",message:"Not at end"},t.genTextPos())}else for(;!t.atEnd();)t.currCodeUnit()===s.LESS_THAN?T(t,n):R(t,n),m(t,n)}function m(t,n){for(;!t.atEnd();)if(t.skipSpaces(),t.startsWith(E))g(t,n);else if(!t.config.strictDocument&&t.startsWithIgnoreCase(p))S(t,n);else if(t.startsWith(w))h(t,n);else break}function $(t){function n(e){if(e.startsWithSpace())e.skipSpaces();else if(!e.startsWith(k)&&!e.atEnd())throw new c.XmlError({type:"InvalidChar",expected:"a whitespace",actual:e.currCodeUnitUnchecked()},e.genTextPos())}if(t.advance(5),n(t),!t.startsWith(q))throw new c.XmlError({type:"InvalidString",expected:"version"},t.genTextPos());d(t),n(t),t.startsWith(K)&&(d(t),n(t)),t.startsWith(Y)&&d(t),t.skipSpaces(),t.skipString(k)}function g(t,n){const e=t.getPos();t.advance(4);const o=t.consumeCodeUnitsWhile((i,r)=>!(i===s.HYPHEN&&r.startsWith(y)));if(t.skipString(y),o.includes("--")||o.endsWith("-"))throw new c.XmlError({type:"InvalidComment"},t.genTextPosFrom(e));n({type:"Comment",text:o,range:t.rangeFrom(e)},t)}function h(t,n){if(t.startsWith(A))throw new c.XmlError({type:"UnexpectedDeclaration"},t.genTextPos());const e=t.getPos();t.advance(2);const o=t.consumeName();t.skipSpaces();const i=t.consumeCodeUnitsWhile((r,a)=>!(r===s.QUESTION_MARK&&a.startsWith(W)));t.skipString(W),n({type:"ProcessingInstruction",target:o,content:i.length===0?void 0:i,range:t.rangeFrom(e)},t)}function S(t,n){if(!t.config.allowDtd)throw new c.XmlError({type:"DtdDetected"});const e=t.getPos();if(j(t),t.skipSpaces(),t.currCodeUnit()===s.GREATER_THAN){t.advance(1);return}for(t.advance(1);!t.atEnd();)if(t.skipSpaces(),t.startsWith(D))J(t,n);else if(t.startsWith(E))g(t,n);else if(t.startsWith(w))h(t,n);else if(t.startsWith("]"))if(t.advance(1),t.skipSpaces(),t.currCodeUnit()===s.GREATER_THAN){t.advance(1);break}else throw new c.XmlError({type:"InvalidChar",expected:"'>'",actual:t.currCodeUnitUnchecked()},t.genTextPos());else if(t.startsWith(F)||t.startsWith(O)||t.startsWith(G))try{Z(t)}catch(o){throw new c.XmlError({type:"UnknownToken",message:"Failed to consume declaration"},t.genTextPosFrom(e))}else throw new c.XmlError({type:"UnknownToken",message:"Failed to parse doctype"},t.genTextPos())}function j(t){t.advance(9),t.consumeSpaces(),t.skipName(),t.skipSpaces(),x(t),t.skipSpaces();const n=t.currCodeUnit();if(n!==s.OPEN_SQUARE_BRACKET&&n!==s.GREATER_THAN)throw new c.XmlError({type:"InvalidChar",expected:"'[' or '>'",actual:n},t.genTextPos())}function x(t){if(t.startsWith(M)||t.startsWith(N)){const n=t.getPos();t.advance(6);const e=t.sliceBack(n);t.consumeSpaces();const o=t.consumeQuote();if(t.consumeCodeUnitsWhile(i=>i!==o),t.consumeCodeUnit(o),e===N){t.consumeSpaces();const i=t.consumeQuote();t.consumeCodeUnitsWhile(r=>r!==i),t.consumeCodeUnit(i)}return!0}return!1}function J(t,n){t.advance(8),t.consumeSpaces();const e=!t.tryConsumeCodeUnit(s.PERCENT);e||t.consumeSpaces();const o=t.consumeName();t.consumeSpaces();const i=V(t,e);i!==null&&n({type:"EntityDeclaration",name:o,definition:i},t),t.skipSpaces(),t.consumeCodeUnit(s.GREATER_THAN)}function V(t,n){const e=t.currCodeUnit();if(e===s.DOUBLE_QUOTE||e===s.SINGLE_QUOTE){const o=t.consumeQuote(),i=t.getPos();t.skipCodeUnitsWhile(a=>a!==o);const r=t.sliceBack(i);return t.consumeCodeUnit(o),r}else if(e===s.UPPERCASE_S||e===s.UPPERCASE_P){if(x(t))return n&&(t.skipSpaces(),t.startsWith(B)&&(t.advance(5),t.consumeSpaces(),t.skipName())),null;throw new c.XmlError({type:"InvalidExternalID"},t.genTextPos())}else throw new c.XmlError({type:"InvalidChar",expected:"a quote, SYSTEM or PUBLIC",actual:e},t.genTextPos())}function Z(t){t.skipCodeUnitsWhile(n=>n!==s.GREATER_THAN),t.consumeCodeUnit(s.GREATER_THAN)}function T(t,n){const e=t.getPos();t.advance(1);const[o,i]=t.consumeQName();n({type:"ElementStart",prefix:o,local:i,start:e},t);let r=!1;for(;!t.atEnd();){const a=t.startsWithSpace();t.skipSpaces();const l=t.getPos(),C=t.currCodeUnit();if(C===s.SLASH){t.advance(1),t.consumeCodeUnit(s.GREATER_THAN);const u=t.rangeFrom(l);n({type:"ElementEnd",end:{type:"Empty"},range:u},t);break}else if(C===s.GREATER_THAN){t.advance(1);const u=t.rangeFrom(l);if(t.config.implicitSelfClosingElements!=null&&t.config.implicitSelfClosingElements.includes(U.getQName(i,o))){n({type:"ElementEnd",end:{type:"Empty"},range:u},t);break}else{n({type:"ElementEnd",end:{type:"Open"},range:u},t),r=!0;break}}else{if(!a)throw new c.XmlError({type:"InvalidChar",expected:"a whitespace",actual:t.currCodeUnitUnchecked()},t.genTextPos());const[u,_,I]=d(t),X=t.getPos();n({type:"Attribute",range:{start:l,end:X},prefix:u,local:_,value:I},t)}}if(r){if(t.config.rawTextElements!=null){const a=U.getQName(i,o);if(t.config.rawTextElements.includes(a)){v(t,n,`</${a}>`);return}}v(t,n)}}function d(t){const[n,e]=t.consumeQName();let o;const i=t.getPos();if(t.skipSpaces(),t.tryConsumeCodeUnit(s.EQUALS)){t.skipSpaces();const r=t.currCodeUnit();if(r===s.SINGLE_QUOTE||r===s.DOUBLE_QUOTE){const a=t.consumeQuote();o=t.consumeCodeUnitsWhile(l=>l!==a&&l!==s.LESS_THAN),t.consumeCodeUnit(a)}else if(!t.config.strictDocument)o=t.consumeCodeUnitsWhile(a=>!L.isXmlSpaceByte(a)&&a!==s.GREATER_THAN&&a!==s.SLASH&&a!==s.LESS_THAN);else throw new c.XmlError({type:"InvalidChar",expected:"a quote",actual:r},t.genTextPos())}else if(!t.config.strictDocument)t.goTo(i),o="true";else throw new c.XmlError({type:"InvalidChar",expected:s.EQUALS,actual:t.currCodeUnit()},t.genTextPos());return[n,e,o]}function v(t,n,e=s.LESS_THAN){for(;!t.atEnd();)if(typeof e=="number"?t.currCodeUnit()===e:t.startsWith(e)){const o=t.nextCodeUnit();if(o===s.EXCLAMATION_MARK)if(t.startsWith(E))g(t,n);else if(t.startsWith(b))tt(t,n);else if(!t.config.strictDocument&&t.startsWithIgnoreCase(p))S(t,n);else throw new c.XmlError({type:"UnknownToken",message:"Failed to parse content"},t.genTextPos());else if(o===s.QUESTION_MARK)h(t,n);else if(o===s.SLASH){nt(t,n);break}else T(t,n)}else R(t,n,e)}function tt(t,n){const e=t.getPos();t.advance(9);const o=t.consumeCodeUnitsWhile((r,a)=>!(r===s.CLOSE_SQUARE_BRACKET&&a.startsWith(f)));t.skipString(f);const i=t.rangeFrom(e);n({type:"Cdata",text:o,range:i},t)}function nt(t,n){const e=t.getPos();t.advance(2);const[o,i]=t.consumeQName();t.skipSpaces(),t.consumeCodeUnit(s.GREATER_THAN);const r=t.rangeFrom(e);n({type:"ElementEnd",end:{type:"Close",prefix:o,local:i},range:r},t)}function R(t,n,e=s.LESS_THAN){const o=t.getPos(),i=typeof e=="number"?t.consumeCodeUnitsWhile(r=>r!==e):t.consumeCodeUnitsWhile((r,a)=>!a.startsWith(e));if(t.config.strictDocument&&i.includes(f))throw new c.XmlError({type:"InvalidCharacterData"},t.genTextPos());n({type:"Text",text:i,range:t.rangeFrom(o)},t)}exports.tokenize=z,exports.tokenizeXmlStream=P;
|
|
@@ -1,48 +1 @@
|
|
|
1
|
-
|
|
2
|
-
|
|
3
|
-
var asciiConstants = require('./ascii-constants.js');
|
|
4
|
-
|
|
5
|
-
function isAsciiDigit(byte) {
|
|
6
|
-
return byte >= asciiConstants.ZERO && byte <= asciiConstants.NINE;
|
|
7
|
-
}
|
|
8
|
-
function isXmlNameStart(codePoint) {
|
|
9
|
-
if (codePoint == null) {
|
|
10
|
-
return false;
|
|
11
|
-
}
|
|
12
|
-
if (codePoint <= 128) {
|
|
13
|
-
return codePoint >= asciiConstants.UPPERCASE_A && codePoint <= asciiConstants.UPPERCASE_Z || codePoint >= asciiConstants.LOWERCASE_A && codePoint <= asciiConstants.LOWERCASE_Z || codePoint === asciiConstants.COLON || codePoint === asciiConstants.UNDERSCORE;
|
|
14
|
-
}
|
|
15
|
-
return codePoint >= 192 && codePoint <= 214 || codePoint >= 216 && codePoint <= 246 || codePoint >= 248 && codePoint <= 767 || codePoint >= 880 && codePoint <= 893 || codePoint >= 895 && codePoint <= 8191 || codePoint >= 8204 && codePoint <= 8205 || codePoint >= 8304 && codePoint <= 8591 || codePoint >= 11264 && codePoint <= 12271 || codePoint >= 12289 && codePoint <= 55295 || codePoint >= 63744 && codePoint <= 64975 || codePoint >= 65008 && codePoint <= 65533 || codePoint >= 65536 && codePoint <= 983039;
|
|
16
|
-
}
|
|
17
|
-
function isXmlName(codePoint) {
|
|
18
|
-
if (codePoint == null) {
|
|
19
|
-
return false;
|
|
20
|
-
}
|
|
21
|
-
if (codePoint <= 128) {
|
|
22
|
-
return isXmlNameByte(codePoint);
|
|
23
|
-
}
|
|
24
|
-
return codePoint === 183 || codePoint >= 192 && codePoint <= 214 || codePoint >= 216 && codePoint <= 246 || codePoint >= 248 && codePoint <= 767 || codePoint >= 768 && codePoint <= 879 || codePoint >= 880 && codePoint <= 893 || codePoint >= 895 && codePoint <= 8191 || codePoint >= 8204 && codePoint <= 8205 || codePoint >= 8255 && codePoint <= 8256 || codePoint >= 8304 && codePoint <= 8591 || codePoint >= 11264 && codePoint <= 12271 || codePoint >= 12289 && codePoint <= 55295 || codePoint >= 63744 && codePoint <= 64975 || codePoint >= 65008 && codePoint <= 65533 || codePoint >= 65536 && codePoint <= 983039;
|
|
25
|
-
}
|
|
26
|
-
function isXmlChar(codePoint) {
|
|
27
|
-
if (codePoint == null) {
|
|
28
|
-
return false;
|
|
29
|
-
}
|
|
30
|
-
if (codePoint < 32) {
|
|
31
|
-
return isXmlSpaceByte(codePoint);
|
|
32
|
-
}
|
|
33
|
-
return codePoint !== 65535 && codePoint !== 65534;
|
|
34
|
-
}
|
|
35
|
-
function isXmlSpaceByte(byte) {
|
|
36
|
-
return byte === asciiConstants.SPACE || byte === asciiConstants.HORIZONTAL_TAB || byte === asciiConstants.LINE_FEED || byte === asciiConstants.CARRIAGE_RETURN;
|
|
37
|
-
}
|
|
38
|
-
function isXmlNameByte(byte) {
|
|
39
|
-
return byte >= asciiConstants.ZERO && byte <= asciiConstants.NINE || byte >= asciiConstants.UPPERCASE_A && byte <= asciiConstants.UPPERCASE_Z || byte >= asciiConstants.LOWERCASE_A && byte <= asciiConstants.LOWERCASE_Z || byte === asciiConstants.COLON || byte === asciiConstants.UNDERSCORE || byte === asciiConstants.HYPHEN || byte === asciiConstants.PERIOD;
|
|
40
|
-
}
|
|
41
|
-
|
|
42
|
-
exports.isAsciiDigit = isAsciiDigit;
|
|
43
|
-
exports.isXmlChar = isXmlChar;
|
|
44
|
-
exports.isXmlName = isXmlName;
|
|
45
|
-
exports.isXmlNameByte = isXmlNameByte;
|
|
46
|
-
exports.isXmlNameStart = isXmlNameStart;
|
|
47
|
-
exports.isXmlSpaceByte = isXmlSpaceByte;
|
|
48
|
-
//# sourceMappingURL=utils.js.map
|
|
1
|
+
"use strict";var i=require("./ascii-constants.js");function n(E){return E>=i.ZERO&&E<=i.NINE}function s(E){return E==null?!1:E<=128?E>=i.UPPERCASE_A&&E<=i.UPPERCASE_Z||E>=i.LOWERCASE_A&&E<=i.LOWERCASE_Z||E===i.COLON||E===i.UNDERSCORE:E>=192&&E<=214||E>=216&&E<=246||E>=248&&E<=767||E>=880&&E<=893||E>=895&&E<=8191||E>=8204&&E<=8205||E>=8304&&E<=8591||E>=11264&&E<=12271||E>=12289&&E<=55295||E>=63744&&E<=64975||E>=65008&&E<=65533||E>=65536&&E<=983039}function r(E){return E==null?!1:E<=128?e(E):E===183||E>=192&&E<=214||E>=216&&E<=246||E>=248&&E<=767||E>=768&&E<=879||E>=880&&E<=893||E>=895&&E<=8191||E>=8204&&E<=8205||E>=8255&&E<=8256||E>=8304&&E<=8591||E>=11264&&E<=12271||E>=12289&&E<=55295||E>=63744&&E<=64975||E>=65008&&E<=65533||E>=65536&&E<=983039}function R(E){return E==null?!1:E<32?t(E):E!==65535&&E!==65534}function t(E){return E===i.SPACE||E===i.HORIZONTAL_TAB||E===i.LINE_FEED||E===i.CARRIAGE_RETURN}function e(E){return E>=i.ZERO&&E<=i.NINE||E>=i.UPPERCASE_A&&E<=i.UPPERCASE_Z||E>=i.LOWERCASE_A&&E<=i.LOWERCASE_Z||E===i.COLON||E===i.UNDERSCORE||E===i.HYPHEN||E===i.PERIOD}exports.isAsciiDigit=n,exports.isXmlChar=R,exports.isXmlName=r,exports.isXmlNameByte=e,exports.isXmlNameStart=s,exports.isXmlSpaceByte=t;
|
|
@@ -1,14 +1 @@
|
|
|
1
|
-
|
|
2
|
-
|
|
3
|
-
var tokenToXml = require('./token-to-xml.js');
|
|
4
|
-
|
|
5
|
-
function tokensToXml(tokens) {
|
|
6
|
-
let xmlString = "";
|
|
7
|
-
tokens.forEach((token) => {
|
|
8
|
-
xmlString += tokenToXml.tokenToXml(token);
|
|
9
|
-
});
|
|
10
|
-
return xmlString;
|
|
11
|
-
}
|
|
12
|
-
|
|
13
|
-
exports.tokensToXml = tokensToXml;
|
|
14
|
-
//# sourceMappingURL=tokens-to-xml.js.map
|
|
1
|
+
"use strict";var n=require("./token-to-xml.js");function r(e){let o="";return e.forEach(t=>{o+=n.tokenToXml(t)}),o}exports.tokensToXml=r;
|
|
@@ -1,70 +1 @@
|
|
|
1
|
-
|
|
2
|
-
|
|
3
|
-
var tokenize = require('./tokenizer/tokenize.js');
|
|
4
|
-
|
|
5
|
-
function xmlToObject(xmlString, options = {}) {
|
|
6
|
-
const root = {
|
|
7
|
-
local: "root",
|
|
8
|
-
attributes: [],
|
|
9
|
-
content: []
|
|
10
|
-
};
|
|
11
|
-
const stack = [root];
|
|
12
|
-
tokenize.tokenize(
|
|
13
|
-
xmlString,
|
|
14
|
-
(token) => {
|
|
15
|
-
processTokenForObject(token, stack);
|
|
16
|
-
},
|
|
17
|
-
options
|
|
18
|
-
);
|
|
19
|
-
return root.content[0];
|
|
20
|
-
}
|
|
21
|
-
function processTokenForObject(token, stack) {
|
|
22
|
-
switch (token.type) {
|
|
23
|
-
case "ElementStart": {
|
|
24
|
-
const newNode = {
|
|
25
|
-
local: token.local,
|
|
26
|
-
prefix: token.prefix.length > 0 ? token.prefix : void 0,
|
|
27
|
-
attributes: [],
|
|
28
|
-
content: []
|
|
29
|
-
};
|
|
30
|
-
const currentNode = stack[stack.length - 1];
|
|
31
|
-
if (currentNode != null) {
|
|
32
|
-
currentNode.content.push(newNode);
|
|
33
|
-
}
|
|
34
|
-
stack.push(newNode);
|
|
35
|
-
break;
|
|
36
|
-
}
|
|
37
|
-
case "ElementEnd": {
|
|
38
|
-
if (token.end.type === "Close" || token.end.type === "Empty") {
|
|
39
|
-
stack.pop();
|
|
40
|
-
}
|
|
41
|
-
break;
|
|
42
|
-
}
|
|
43
|
-
case "Attribute": {
|
|
44
|
-
const currentNode = stack[stack.length - 1];
|
|
45
|
-
if (currentNode != null) {
|
|
46
|
-
currentNode.attributes.push({
|
|
47
|
-
local: token.local,
|
|
48
|
-
prefix: token.prefix.length > 0 ? token.prefix : void 0,
|
|
49
|
-
value: token.value
|
|
50
|
-
});
|
|
51
|
-
}
|
|
52
|
-
break;
|
|
53
|
-
}
|
|
54
|
-
case "Text":
|
|
55
|
-
case "Cdata": {
|
|
56
|
-
const currentNode = stack[stack.length - 1];
|
|
57
|
-
if (currentNode != null) {
|
|
58
|
-
const trimmedText = token.text.trim();
|
|
59
|
-
if (trimmedText.length > 0) {
|
|
60
|
-
currentNode.content.push(token.text);
|
|
61
|
-
}
|
|
62
|
-
}
|
|
63
|
-
break;
|
|
64
|
-
}
|
|
65
|
-
}
|
|
66
|
-
}
|
|
67
|
-
|
|
68
|
-
exports.processTokenForObject = processTokenForObject;
|
|
69
|
-
exports.xmlToObject = xmlToObject;
|
|
70
|
-
//# sourceMappingURL=xml-to-object.js.map
|
|
1
|
+
"use strict";var r=require("./tokenizer/tokenize.js");function s(t,o={}){const e={local:"root",attributes:[],content:[]},n=[e];return r.tokenize(t,l=>{c(l,n)},o),e.content[0]}function c(t,o){switch(t.type){case"ElementStart":{const e={local:t.local,prefix:t.prefix.length>0?t.prefix:void 0,attributes:[],content:[]},n=o[o.length-1];n!=null&&n.content.push(e),o.push(e);break}case"ElementEnd":{(t.end.type==="Close"||t.end.type==="Empty")&&o.pop();break}case"Attribute":{const e=o[o.length-1];e!=null&&e.attributes.push({local:t.local,prefix:t.prefix.length>0?t.prefix:void 0,value:t.value});break}case"Text":case"Cdata":{const e=o[o.length-1];e!=null&&t.text.trim().length>0&&e.content.push(t.text);break}}}exports.processTokenForObject=c,exports.xmlToObject=s;
|
|
@@ -1,87 +1 @@
|
|
|
1
|
-
|
|
2
|
-
|
|
3
|
-
var getQName = require('./get-q-name.js');
|
|
4
|
-
var tokenize = require('./tokenizer/tokenize.js');
|
|
5
|
-
|
|
6
|
-
function xmlToSimplifiedObject(xmlString, options = {}) {
|
|
7
|
-
const result = {};
|
|
8
|
-
const stack = [result];
|
|
9
|
-
tokenize.tokenize(
|
|
10
|
-
xmlString,
|
|
11
|
-
(token) => {
|
|
12
|
-
processTokenForSimplifiedObject(token, stack);
|
|
13
|
-
},
|
|
14
|
-
options
|
|
15
|
-
);
|
|
16
|
-
return result;
|
|
17
|
-
}
|
|
18
|
-
function processTokenForSimplifiedObject(token, stack) {
|
|
19
|
-
switch (token.type) {
|
|
20
|
-
case "ElementStart": {
|
|
21
|
-
const tagName = getQName.getQName(token.local, token.prefix);
|
|
22
|
-
const tagNameWithPrefix = `_${tagName}`;
|
|
23
|
-
const newNode = {};
|
|
24
|
-
const currentNode = stack[stack.length - 1];
|
|
25
|
-
if (currentNode != null) {
|
|
26
|
-
if (currentNode.content != null) {
|
|
27
|
-
newNode.tag = tagName;
|
|
28
|
-
currentNode.content.push(newNode);
|
|
29
|
-
} else if (currentNode.text != null) {
|
|
30
|
-
newNode.tag = tagName;
|
|
31
|
-
currentNode.content = [currentNode.text, newNode];
|
|
32
|
-
delete currentNode.text;
|
|
33
|
-
} else if (currentNode[tagNameWithPrefix] == null) {
|
|
34
|
-
currentNode[tagNameWithPrefix] = [newNode];
|
|
35
|
-
} else if (Array.isArray(currentNode[tagNameWithPrefix])) {
|
|
36
|
-
currentNode[tagNameWithPrefix].push(newNode);
|
|
37
|
-
} else {
|
|
38
|
-
currentNode[tagNameWithPrefix] = [currentNode[tagNameWithPrefix], newNode];
|
|
39
|
-
}
|
|
40
|
-
}
|
|
41
|
-
stack.push(newNode);
|
|
42
|
-
break;
|
|
43
|
-
}
|
|
44
|
-
case "ElementEnd": {
|
|
45
|
-
if (token.end.type === "Close" || token.end.type === "Empty") {
|
|
46
|
-
stack.pop();
|
|
47
|
-
}
|
|
48
|
-
break;
|
|
49
|
-
}
|
|
50
|
-
case "Attribute": {
|
|
51
|
-
const currentNode = stack[stack.length - 1];
|
|
52
|
-
if (currentNode != null) {
|
|
53
|
-
const attrName = getQName.getQName(token.local, token.prefix);
|
|
54
|
-
if (currentNode.attributes != null) {
|
|
55
|
-
currentNode.attributes[attrName] = token.value;
|
|
56
|
-
} else {
|
|
57
|
-
currentNode.attributes = {
|
|
58
|
-
[attrName]: token.value
|
|
59
|
-
};
|
|
60
|
-
}
|
|
61
|
-
}
|
|
62
|
-
break;
|
|
63
|
-
}
|
|
64
|
-
case "Text":
|
|
65
|
-
case "Cdata": {
|
|
66
|
-
const currentNode = stack[stack.length - 1];
|
|
67
|
-
if (currentNode != null) {
|
|
68
|
-
const trimmedText = token.text.trim();
|
|
69
|
-
if (trimmedText.length > 0) {
|
|
70
|
-
if (currentNode.content != null) {
|
|
71
|
-
currentNode.content.push(trimmedText);
|
|
72
|
-
} else if (currentNode.text != null) {
|
|
73
|
-
currentNode.content = [currentNode.text, trimmedText];
|
|
74
|
-
delete currentNode.text;
|
|
75
|
-
} else {
|
|
76
|
-
currentNode.text = trimmedText;
|
|
77
|
-
}
|
|
78
|
-
}
|
|
79
|
-
}
|
|
80
|
-
break;
|
|
81
|
-
}
|
|
82
|
-
}
|
|
83
|
-
}
|
|
84
|
-
|
|
85
|
-
exports.processTokenForSimplifiedObject = processTokenForSimplifiedObject;
|
|
86
|
-
exports.xmlToSimplifiedObject = xmlToSimplifiedObject;
|
|
87
|
-
//# sourceMappingURL=xml-to-simplified-object.js.map
|
|
1
|
+
"use strict";var c=require("./get-q-name.js"),o=require("./tokenizer/tokenize.js");function s(i,l={}){const e={},t=[e];return o.tokenize(i,r=>{a(r,t)},l),e}function a(i,l){switch(i.type){case"ElementStart":{const e=c.getQName(i.local,i.prefix),t=`_${e}`,r={},n=l[l.length-1];n!=null&&(n.content!=null?(r.tag=e,n.content.push(r)):n.text!=null?(r.tag=e,n.content=[n.text,r],delete n.text):n[t]==null?n[t]=[r]:Array.isArray(n[t])?n[t].push(r):n[t]=[n[t],r]),l.push(r);break}case"ElementEnd":{(i.end.type==="Close"||i.end.type==="Empty")&&l.pop();break}case"Attribute":{const e=l[l.length-1];if(e!=null){const t=c.getQName(i.local,i.prefix);e.attributes!=null?e.attributes[t]=i.value:e.attributes={[t]:i.value}}break}case"Text":case"Cdata":{const e=l[l.length-1];if(e!=null){const t=i.text.trim();t.length>0&&(e.content!=null?e.content.push(t):e.text!=null?(e.content=[e.text,t],delete e.text):e.text=t)}break}}}exports.processTokenForSimplifiedObject=a,exports.xmlToSimplifiedObject=s;
|