@loaders.gl/graphs 5.0.0-beta.1 → 5.0.0-beta.2
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 +3 -1
- package/THIRD-PARTY-NOTICES.md +1 -1
- package/dist/dot-loader-types.cjs +59 -0
- package/dist/dot-loader-types.cjs.map +7 -0
- package/dist/dot-loader-types.js +1 -1
- package/dist/dot-loader.cjs +1 -1
- package/dist/gexf-loader-types.cjs +59 -0
- package/dist/gexf-loader-types.cjs.map +7 -0
- package/dist/gexf-loader-types.js +1 -1
- package/dist/gexf-loader.cjs +1652 -16
- package/dist/gexf-loader.cjs.map +3 -3
- package/dist/graphml-loader-types.cjs +59 -0
- package/dist/graphml-loader-types.cjs.map +7 -0
- package/dist/graphml-loader-types.js +1 -1
- package/dist/graphml-loader.cjs +1634 -10
- package/dist/graphml-loader.cjs.map +3 -3
- package/dist/index.cjs +30 -1927
- package/dist/index.cjs.map +4 -3
- package/dist/lib/parse-gexf.d.ts.map +1 -1
- package/dist/lib/parse-gexf.js +9 -7
- package/dist/lib/parse-gexf.js.map +1 -1
- package/dist/lib/parse-graphml.js +4 -4
- package/dist/lib/parse-graphml.js.map +1 -1
- package/package.json +9 -7
- package/src/lib/parse-gexf.ts +9 -7
- package/src/lib/parse-graphml.ts +5 -5
package/dist/gexf-loader.cjs
CHANGED
|
@@ -62,6 +62,1638 @@ var init_gexf_loader_types = __esm({
|
|
|
62
62
|
}
|
|
63
63
|
});
|
|
64
64
|
|
|
65
|
+
// ../documents/src/lib/xml-utils/uncapitalize.ts
|
|
66
|
+
function uncapitalize(str) {
|
|
67
|
+
return typeof str === "string" ? str.charAt(0).toLowerCase() + str.slice(1) : str;
|
|
68
|
+
}
|
|
69
|
+
function uncapitalizeKeys(object) {
|
|
70
|
+
if (Array.isArray(object)) {
|
|
71
|
+
return object.map((element) => uncapitalizeKeys(element));
|
|
72
|
+
}
|
|
73
|
+
if (object && typeof object === "object") {
|
|
74
|
+
const newObject = {};
|
|
75
|
+
for (const [key, value] of Object.entries(object)) {
|
|
76
|
+
Object.defineProperty(newObject, uncapitalize(key), {
|
|
77
|
+
value: uncapitalizeKeys(value),
|
|
78
|
+
enumerable: true,
|
|
79
|
+
configurable: true,
|
|
80
|
+
writable: true
|
|
81
|
+
});
|
|
82
|
+
}
|
|
83
|
+
return newObject;
|
|
84
|
+
}
|
|
85
|
+
return object;
|
|
86
|
+
}
|
|
87
|
+
var init_uncapitalize = __esm({
|
|
88
|
+
"../documents/src/lib/xml-utils/uncapitalize.ts"() {
|
|
89
|
+
"use strict";
|
|
90
|
+
}
|
|
91
|
+
});
|
|
92
|
+
|
|
93
|
+
// ../documents/src/sax-ts/sax.ts
|
|
94
|
+
var DEFAULT_SAX_EVENTS, DEFAULT_SAX_PARSER_OPTIONS, EVENTS, BUFFERS, nameStart, nameBody, entityStart, entityBody, ENTITIES, SAX, SAXParser;
|
|
95
|
+
var init_sax = __esm({
|
|
96
|
+
"../documents/src/sax-ts/sax.ts"() {
|
|
97
|
+
"use strict";
|
|
98
|
+
DEFAULT_SAX_EVENTS = {
|
|
99
|
+
ontext: () => {
|
|
100
|
+
},
|
|
101
|
+
onprocessinginstruction: () => {
|
|
102
|
+
},
|
|
103
|
+
onsgmldeclaration: () => {
|
|
104
|
+
},
|
|
105
|
+
ondoctype: () => {
|
|
106
|
+
},
|
|
107
|
+
oncomment: () => {
|
|
108
|
+
},
|
|
109
|
+
onopentagstart: () => {
|
|
110
|
+
},
|
|
111
|
+
onattribute: () => {
|
|
112
|
+
},
|
|
113
|
+
onopentag: () => {
|
|
114
|
+
},
|
|
115
|
+
onclosetag: () => {
|
|
116
|
+
},
|
|
117
|
+
onopencdata: () => {
|
|
118
|
+
},
|
|
119
|
+
oncdata: () => {
|
|
120
|
+
},
|
|
121
|
+
onclosecdata: () => {
|
|
122
|
+
},
|
|
123
|
+
onerror: () => {
|
|
124
|
+
},
|
|
125
|
+
onend: () => {
|
|
126
|
+
},
|
|
127
|
+
onready: () => {
|
|
128
|
+
},
|
|
129
|
+
onscript: () => {
|
|
130
|
+
},
|
|
131
|
+
onopennamespace: () => {
|
|
132
|
+
},
|
|
133
|
+
onclosenamespace: () => {
|
|
134
|
+
}
|
|
135
|
+
};
|
|
136
|
+
DEFAULT_SAX_PARSER_OPTIONS = {
|
|
137
|
+
...DEFAULT_SAX_EVENTS,
|
|
138
|
+
strict: false,
|
|
139
|
+
MAX_BUFFER_LENGTH: 64 * 1024,
|
|
140
|
+
lowercase: false,
|
|
141
|
+
lowercasetags: false,
|
|
142
|
+
noscript: false,
|
|
143
|
+
strictEntities: false,
|
|
144
|
+
preserveCase: false,
|
|
145
|
+
caseInsensitiveTags: false,
|
|
146
|
+
processEntities: true,
|
|
147
|
+
xmlns: void 0,
|
|
148
|
+
position: void 0,
|
|
149
|
+
trim: void 0,
|
|
150
|
+
normalize: void 0
|
|
151
|
+
};
|
|
152
|
+
EVENTS = [
|
|
153
|
+
"text",
|
|
154
|
+
"processinginstruction",
|
|
155
|
+
"sgmldeclaration",
|
|
156
|
+
"doctype",
|
|
157
|
+
"comment",
|
|
158
|
+
"opentagstart",
|
|
159
|
+
"attribute",
|
|
160
|
+
"opentag",
|
|
161
|
+
"closetag",
|
|
162
|
+
"opencdata",
|
|
163
|
+
"cdata",
|
|
164
|
+
"closecdata",
|
|
165
|
+
"error",
|
|
166
|
+
"end",
|
|
167
|
+
"ready",
|
|
168
|
+
"script",
|
|
169
|
+
"opennamespace",
|
|
170
|
+
"closenamespace"
|
|
171
|
+
];
|
|
172
|
+
BUFFERS = [
|
|
173
|
+
"comment",
|
|
174
|
+
"sgmlDecl",
|
|
175
|
+
"textNode",
|
|
176
|
+
"tagName",
|
|
177
|
+
"doctype",
|
|
178
|
+
"procInstName",
|
|
179
|
+
"procInstBody",
|
|
180
|
+
"entity",
|
|
181
|
+
"attribName",
|
|
182
|
+
"attribValue",
|
|
183
|
+
"cdata",
|
|
184
|
+
"script"
|
|
185
|
+
];
|
|
186
|
+
nameStart = /[:_A-Za-z\u00C0-\u00D6\u00D8-\u00F6\u00F8-\u02FF\u0370-\u037D\u037F-\u1FFF\u200C-\u200D\u2070-\u218F\u2C00-\u2FEF\u3001-\uD7FF\uF900-\uFDCF\uFDF0-\uFFFD]/;
|
|
187
|
+
nameBody = /[:_A-Za-z\u00C0-\u00D6\u00D8-\u00F6\u00F8-\u02FF\u0370-\u037D\u037F-\u1FFF\u200C-\u200D\u2070-\u218F\u2C00-\u2FEF\u3001-\uD7FF\uF900-\uFDCF\uFDF0-\uFFFD\u00B7\u0300-\u036F\u203F-\u2040.\d-]/;
|
|
188
|
+
entityStart = /[#:_A-Za-z\u00C0-\u00D6\u00D8-\u00F6\u00F8-\u02FF\u0370-\u037D\u037F-\u1FFF\u200C-\u200D\u2070-\u218F\u2C00-\u2FEF\u3001-\uD7FF\uF900-\uFDCF\uFDF0-\uFFFD]/;
|
|
189
|
+
entityBody = /[#:_A-Za-z\u00C0-\u00D6\u00D8-\u00F6\u00F8-\u02FF\u0370-\u037D\u037F-\u1FFF\u200C-\u200D\u2070-\u218F\u2C00-\u2FEF\u3001-\uD7FF\uF900-\uFDCF\uFDF0-\uFFFD\u00B7\u0300-\u036F\u203F-\u2040.\d-]/;
|
|
190
|
+
ENTITIES = {
|
|
191
|
+
amp: "&",
|
|
192
|
+
gt: ">",
|
|
193
|
+
lt: "<",
|
|
194
|
+
quot: '"',
|
|
195
|
+
apos: "'",
|
|
196
|
+
AElig: 198,
|
|
197
|
+
Aacute: 193,
|
|
198
|
+
Acirc: 194,
|
|
199
|
+
Agrave: 192,
|
|
200
|
+
Aring: 197,
|
|
201
|
+
Atilde: 195,
|
|
202
|
+
Auml: 196,
|
|
203
|
+
Ccedil: 199,
|
|
204
|
+
ETH: 208,
|
|
205
|
+
Eacute: 201,
|
|
206
|
+
Ecirc: 202,
|
|
207
|
+
Egrave: 200,
|
|
208
|
+
Euml: 203,
|
|
209
|
+
Iacute: 205,
|
|
210
|
+
Icirc: 206,
|
|
211
|
+
Igrave: 204,
|
|
212
|
+
Iuml: 207,
|
|
213
|
+
Ntilde: 209,
|
|
214
|
+
Oacute: 211,
|
|
215
|
+
Ocirc: 212,
|
|
216
|
+
Ograve: 210,
|
|
217
|
+
Oslash: 216,
|
|
218
|
+
Otilde: 213,
|
|
219
|
+
Ouml: 214,
|
|
220
|
+
THORN: 222,
|
|
221
|
+
Uacute: 218,
|
|
222
|
+
Ucirc: 219,
|
|
223
|
+
Ugrave: 217,
|
|
224
|
+
Uuml: 220,
|
|
225
|
+
Yacute: 221,
|
|
226
|
+
aacute: 225,
|
|
227
|
+
acirc: 226,
|
|
228
|
+
aelig: 230,
|
|
229
|
+
agrave: 224,
|
|
230
|
+
aring: 229,
|
|
231
|
+
atilde: 227,
|
|
232
|
+
auml: 228,
|
|
233
|
+
ccedil: 231,
|
|
234
|
+
eacute: 233,
|
|
235
|
+
ecirc: 234,
|
|
236
|
+
egrave: 232,
|
|
237
|
+
eth: 240,
|
|
238
|
+
euml: 235,
|
|
239
|
+
iacute: 237,
|
|
240
|
+
icirc: 238,
|
|
241
|
+
igrave: 236,
|
|
242
|
+
iuml: 239,
|
|
243
|
+
ntilde: 241,
|
|
244
|
+
oacute: 243,
|
|
245
|
+
ocirc: 244,
|
|
246
|
+
ograve: 242,
|
|
247
|
+
oslash: 248,
|
|
248
|
+
otilde: 245,
|
|
249
|
+
ouml: 246,
|
|
250
|
+
szlig: 223,
|
|
251
|
+
thorn: 254,
|
|
252
|
+
uacute: 250,
|
|
253
|
+
ucirc: 251,
|
|
254
|
+
ugrave: 249,
|
|
255
|
+
uuml: 252,
|
|
256
|
+
yacute: 253,
|
|
257
|
+
yuml: 255,
|
|
258
|
+
copy: 169,
|
|
259
|
+
reg: 174,
|
|
260
|
+
nbsp: 160,
|
|
261
|
+
iexcl: 161,
|
|
262
|
+
cent: 162,
|
|
263
|
+
pound: 163,
|
|
264
|
+
curren: 164,
|
|
265
|
+
yen: 165,
|
|
266
|
+
brvbar: 166,
|
|
267
|
+
sect: 167,
|
|
268
|
+
uml: 168,
|
|
269
|
+
ordf: 170,
|
|
270
|
+
laquo: 171,
|
|
271
|
+
not: 172,
|
|
272
|
+
shy: 173,
|
|
273
|
+
macr: 175,
|
|
274
|
+
deg: 176,
|
|
275
|
+
plusmn: 177,
|
|
276
|
+
sup1: 185,
|
|
277
|
+
sup2: 178,
|
|
278
|
+
sup3: 179,
|
|
279
|
+
acute: 180,
|
|
280
|
+
micro: 181,
|
|
281
|
+
para: 182,
|
|
282
|
+
middot: 183,
|
|
283
|
+
cedil: 184,
|
|
284
|
+
ordm: 186,
|
|
285
|
+
raquo: 187,
|
|
286
|
+
frac14: 188,
|
|
287
|
+
frac12: 189,
|
|
288
|
+
frac34: 190,
|
|
289
|
+
iquest: 191,
|
|
290
|
+
times: 215,
|
|
291
|
+
divide: 247,
|
|
292
|
+
OElig: 338,
|
|
293
|
+
oelig: 339,
|
|
294
|
+
Scaron: 352,
|
|
295
|
+
scaron: 353,
|
|
296
|
+
Yuml: 376,
|
|
297
|
+
fnof: 402,
|
|
298
|
+
circ: 710,
|
|
299
|
+
tilde: 732,
|
|
300
|
+
Alpha: 913,
|
|
301
|
+
Beta: 914,
|
|
302
|
+
Gamma: 915,
|
|
303
|
+
Delta: 916,
|
|
304
|
+
Epsilon: 917,
|
|
305
|
+
Zeta: 918,
|
|
306
|
+
Eta: 919,
|
|
307
|
+
Theta: 920,
|
|
308
|
+
Iota: 921,
|
|
309
|
+
Kappa: 922,
|
|
310
|
+
Lambda: 923,
|
|
311
|
+
Mu: 924,
|
|
312
|
+
Nu: 925,
|
|
313
|
+
Xi: 926,
|
|
314
|
+
Omicron: 927,
|
|
315
|
+
Pi: 928,
|
|
316
|
+
Rho: 929,
|
|
317
|
+
Sigma: 931,
|
|
318
|
+
Tau: 932,
|
|
319
|
+
Upsilon: 933,
|
|
320
|
+
Phi: 934,
|
|
321
|
+
Chi: 935,
|
|
322
|
+
Psi: 936,
|
|
323
|
+
Omega: 937,
|
|
324
|
+
alpha: 945,
|
|
325
|
+
beta: 946,
|
|
326
|
+
gamma: 947,
|
|
327
|
+
delta: 948,
|
|
328
|
+
epsilon: 949,
|
|
329
|
+
zeta: 950,
|
|
330
|
+
eta: 951,
|
|
331
|
+
theta: 952,
|
|
332
|
+
iota: 953,
|
|
333
|
+
kappa: 954,
|
|
334
|
+
lambda: 955,
|
|
335
|
+
mu: 956,
|
|
336
|
+
nu: 957,
|
|
337
|
+
xi: 958,
|
|
338
|
+
omicron: 959,
|
|
339
|
+
pi: 960,
|
|
340
|
+
rho: 961,
|
|
341
|
+
sigmaf: 962,
|
|
342
|
+
sigma: 963,
|
|
343
|
+
tau: 964,
|
|
344
|
+
upsilon: 965,
|
|
345
|
+
phi: 966,
|
|
346
|
+
chi: 967,
|
|
347
|
+
psi: 968,
|
|
348
|
+
omega: 969,
|
|
349
|
+
thetasym: 977,
|
|
350
|
+
upsih: 978,
|
|
351
|
+
piv: 982,
|
|
352
|
+
ensp: 8194,
|
|
353
|
+
emsp: 8195,
|
|
354
|
+
thinsp: 8201,
|
|
355
|
+
zwnj: 8204,
|
|
356
|
+
zwj: 8205,
|
|
357
|
+
lrm: 8206,
|
|
358
|
+
rlm: 8207,
|
|
359
|
+
ndash: 8211,
|
|
360
|
+
mdash: 8212,
|
|
361
|
+
lsquo: 8216,
|
|
362
|
+
rsquo: 8217,
|
|
363
|
+
sbquo: 8218,
|
|
364
|
+
ldquo: 8220,
|
|
365
|
+
rdquo: 8221,
|
|
366
|
+
bdquo: 8222,
|
|
367
|
+
dagger: 8224,
|
|
368
|
+
Dagger: 8225,
|
|
369
|
+
bull: 8226,
|
|
370
|
+
hellip: 8230,
|
|
371
|
+
permil: 8240,
|
|
372
|
+
prime: 8242,
|
|
373
|
+
Prime: 8243,
|
|
374
|
+
lsaquo: 8249,
|
|
375
|
+
rsaquo: 8250,
|
|
376
|
+
oline: 8254,
|
|
377
|
+
frasl: 8260,
|
|
378
|
+
euro: 8364,
|
|
379
|
+
image: 8465,
|
|
380
|
+
weierp: 8472,
|
|
381
|
+
real: 8476,
|
|
382
|
+
trade: 8482,
|
|
383
|
+
alefsym: 8501,
|
|
384
|
+
larr: 8592,
|
|
385
|
+
uarr: 8593,
|
|
386
|
+
rarr: 8594,
|
|
387
|
+
darr: 8595,
|
|
388
|
+
harr: 8596,
|
|
389
|
+
crarr: 8629,
|
|
390
|
+
lArr: 8656,
|
|
391
|
+
uArr: 8657,
|
|
392
|
+
rArr: 8658,
|
|
393
|
+
dArr: 8659,
|
|
394
|
+
hArr: 8660,
|
|
395
|
+
forall: 8704,
|
|
396
|
+
part: 8706,
|
|
397
|
+
exist: 8707,
|
|
398
|
+
empty: 8709,
|
|
399
|
+
nabla: 8711,
|
|
400
|
+
isin: 8712,
|
|
401
|
+
notin: 8713,
|
|
402
|
+
ni: 8715,
|
|
403
|
+
prod: 8719,
|
|
404
|
+
sum: 8721,
|
|
405
|
+
minus: 8722,
|
|
406
|
+
lowast: 8727,
|
|
407
|
+
radic: 8730,
|
|
408
|
+
prop: 8733,
|
|
409
|
+
infin: 8734,
|
|
410
|
+
ang: 8736,
|
|
411
|
+
and: 8743,
|
|
412
|
+
or: 8744,
|
|
413
|
+
cap: 8745,
|
|
414
|
+
cup: 8746,
|
|
415
|
+
int: 8747,
|
|
416
|
+
there4: 8756,
|
|
417
|
+
sim: 8764,
|
|
418
|
+
cong: 8773,
|
|
419
|
+
asymp: 8776,
|
|
420
|
+
ne: 8800,
|
|
421
|
+
equiv: 8801,
|
|
422
|
+
le: 8804,
|
|
423
|
+
ge: 8805,
|
|
424
|
+
sub: 8834,
|
|
425
|
+
sup: 8835,
|
|
426
|
+
nsub: 8836,
|
|
427
|
+
sube: 8838,
|
|
428
|
+
supe: 8839,
|
|
429
|
+
oplus: 8853,
|
|
430
|
+
otimes: 8855,
|
|
431
|
+
perp: 8869,
|
|
432
|
+
sdot: 8901,
|
|
433
|
+
lceil: 8968,
|
|
434
|
+
rceil: 8969,
|
|
435
|
+
lfloor: 8970,
|
|
436
|
+
rfloor: 8971,
|
|
437
|
+
lang: 9001,
|
|
438
|
+
rang: 9002,
|
|
439
|
+
loz: 9674,
|
|
440
|
+
spades: 9824,
|
|
441
|
+
clubs: 9827,
|
|
442
|
+
hearts: 9829,
|
|
443
|
+
diams: 9830
|
|
444
|
+
};
|
|
445
|
+
Object.keys(ENTITIES).forEach((key) => {
|
|
446
|
+
const e = ENTITIES[key];
|
|
447
|
+
ENTITIES[key] = typeof e === "number" ? String.fromCharCode(e) : e;
|
|
448
|
+
});
|
|
449
|
+
SAX = class _SAX {
|
|
450
|
+
EVENTS = EVENTS;
|
|
451
|
+
ENTITIES = {
|
|
452
|
+
// TODO: make it readonly, needed for entity-mega test
|
|
453
|
+
// amp, gt, lt, quot and apos are resolved to strings instead of numerical
|
|
454
|
+
// codes, IDK why
|
|
455
|
+
...ENTITIES
|
|
456
|
+
};
|
|
457
|
+
XML_ENTITIES = {
|
|
458
|
+
amp: "&",
|
|
459
|
+
gt: ">",
|
|
460
|
+
lt: "<",
|
|
461
|
+
quot: '"',
|
|
462
|
+
apos: "'"
|
|
463
|
+
};
|
|
464
|
+
S = 0;
|
|
465
|
+
opt;
|
|
466
|
+
trackPosition = false;
|
|
467
|
+
column = 0;
|
|
468
|
+
line = 0;
|
|
469
|
+
c = "";
|
|
470
|
+
error;
|
|
471
|
+
q = "";
|
|
472
|
+
bufferCheckPosition;
|
|
473
|
+
closed = false;
|
|
474
|
+
tags = [];
|
|
475
|
+
looseCase = "";
|
|
476
|
+
closedRoot = false;
|
|
477
|
+
sawRoot = false;
|
|
478
|
+
strict = false;
|
|
479
|
+
tag;
|
|
480
|
+
strictEntities;
|
|
481
|
+
state;
|
|
482
|
+
noscript = false;
|
|
483
|
+
attribList = [];
|
|
484
|
+
ns;
|
|
485
|
+
position = 0;
|
|
486
|
+
STATE = {
|
|
487
|
+
BEGIN: this.S++,
|
|
488
|
+
// leading byte order mark or whitespace
|
|
489
|
+
BEGIN_WHITESPACE: this.S++,
|
|
490
|
+
// leading whitespace
|
|
491
|
+
TEXT: this.S++,
|
|
492
|
+
// general stuff
|
|
493
|
+
TEXT_ENTITY: this.S++,
|
|
494
|
+
// & and such.
|
|
495
|
+
OPEN_WAKA: this.S++,
|
|
496
|
+
// <
|
|
497
|
+
SGML_DECL: this.S++,
|
|
498
|
+
// <!BLARG
|
|
499
|
+
SGML_DECL_QUOTED: this.S++,
|
|
500
|
+
// <!BLARG foo "bar
|
|
501
|
+
DOCTYPE: this.S++,
|
|
502
|
+
// <!DOCTYPE
|
|
503
|
+
DOCTYPE_QUOTED: this.S++,
|
|
504
|
+
// <!DOCTYPE "//blah
|
|
505
|
+
DOCTYPE_DTD: this.S++,
|
|
506
|
+
// <!DOCTYPE "//blah" [ ...
|
|
507
|
+
DOCTYPE_DTD_QUOTED: this.S++,
|
|
508
|
+
// <!DOCTYPE "//blah" [ "foo
|
|
509
|
+
COMMENT_STARTING: this.S++,
|
|
510
|
+
// <!-
|
|
511
|
+
COMMENT: this.S++,
|
|
512
|
+
// <!--
|
|
513
|
+
COMMENT_ENDING: this.S++,
|
|
514
|
+
// <!-- blah -
|
|
515
|
+
COMMENT_ENDED: this.S++,
|
|
516
|
+
// <!-- blah --
|
|
517
|
+
CDATA: this.S++,
|
|
518
|
+
// <![CDATA[ something
|
|
519
|
+
CDATA_ENDING: this.S++,
|
|
520
|
+
// ]
|
|
521
|
+
CDATA_ENDING_2: this.S++,
|
|
522
|
+
// ]]
|
|
523
|
+
PROC_INST: this.S++,
|
|
524
|
+
// <?hi
|
|
525
|
+
PROC_INST_BODY: this.S++,
|
|
526
|
+
// <?hi there
|
|
527
|
+
PROC_INST_ENDING: this.S++,
|
|
528
|
+
// <?hi "there" ?
|
|
529
|
+
OPEN_TAG: this.S++,
|
|
530
|
+
// <strong
|
|
531
|
+
OPEN_TAG_SLASH: this.S++,
|
|
532
|
+
// <strong /
|
|
533
|
+
ATTRIB: this.S++,
|
|
534
|
+
// <a
|
|
535
|
+
ATTRIB_NAME: this.S++,
|
|
536
|
+
// <a foo
|
|
537
|
+
ATTRIB_NAME_SAW_WHITE: this.S++,
|
|
538
|
+
// <a foo _
|
|
539
|
+
ATTRIB_VALUE: this.S++,
|
|
540
|
+
// <a foo=
|
|
541
|
+
ATTRIB_VALUE_QUOTED: this.S++,
|
|
542
|
+
// <a foo="bar
|
|
543
|
+
ATTRIB_VALUE_CLOSED: this.S++,
|
|
544
|
+
// <a foo="bar"
|
|
545
|
+
ATTRIB_VALUE_UNQUOTED: this.S++,
|
|
546
|
+
// <a foo=bar
|
|
547
|
+
ATTRIB_VALUE_ENTITY_Q: this.S++,
|
|
548
|
+
// <foo bar="""
|
|
549
|
+
ATTRIB_VALUE_ENTITY_U: this.S++,
|
|
550
|
+
// <foo bar="
|
|
551
|
+
CLOSE_TAG: this.S++,
|
|
552
|
+
// </a
|
|
553
|
+
CLOSE_TAG_SAW_WHITE: this.S++,
|
|
554
|
+
// </a >
|
|
555
|
+
SCRIPT: this.S++,
|
|
556
|
+
// <script> ...
|
|
557
|
+
SCRIPT_ENDING: this.S++
|
|
558
|
+
// <script> ... <
|
|
559
|
+
};
|
|
560
|
+
BUFFERS = BUFFERS;
|
|
561
|
+
// private parser: (strict: boolean, opt: any) => SAXParser;
|
|
562
|
+
CDATA = "[CDATA[";
|
|
563
|
+
DOCTYPE = "DOCTYPE";
|
|
564
|
+
XML_NAMESPACE = "http://www.w3.org/XML/1998/namespace";
|
|
565
|
+
XMLNS_NAMESPACE = "http://www.w3.org/2000/xmlns/";
|
|
566
|
+
rootNS = {
|
|
567
|
+
xml: this.XML_NAMESPACE,
|
|
568
|
+
xmlns: this.XMLNS_NAMESPACE
|
|
569
|
+
};
|
|
570
|
+
comment;
|
|
571
|
+
sgmlDecl;
|
|
572
|
+
textNode = "";
|
|
573
|
+
tagName;
|
|
574
|
+
doctype;
|
|
575
|
+
procInstName;
|
|
576
|
+
procInstBody;
|
|
577
|
+
entity = "";
|
|
578
|
+
attribName;
|
|
579
|
+
attribValue;
|
|
580
|
+
cdata = "";
|
|
581
|
+
script = "";
|
|
582
|
+
startTagPosition = 0;
|
|
583
|
+
constructor() {
|
|
584
|
+
this.S = 0;
|
|
585
|
+
for (const s in this.STATE) {
|
|
586
|
+
if (this.STATE.hasOwnProperty(s)) {
|
|
587
|
+
this.STATE[this.STATE[s]] = s;
|
|
588
|
+
}
|
|
589
|
+
}
|
|
590
|
+
this.S = this.STATE;
|
|
591
|
+
}
|
|
592
|
+
static charAt(chunk, i) {
|
|
593
|
+
let result = "";
|
|
594
|
+
if (i < chunk.length) {
|
|
595
|
+
result = chunk.charAt(i);
|
|
596
|
+
}
|
|
597
|
+
return result;
|
|
598
|
+
}
|
|
599
|
+
static isWhitespace(c) {
|
|
600
|
+
return c === " " || c === "\n" || c === "\r" || c === " ";
|
|
601
|
+
}
|
|
602
|
+
static isQuote(c) {
|
|
603
|
+
return c === '"' || c === "'";
|
|
604
|
+
}
|
|
605
|
+
static isAttribEnd(c) {
|
|
606
|
+
return c === ">" || _SAX.isWhitespace(c);
|
|
607
|
+
}
|
|
608
|
+
static isMatch(regex, c) {
|
|
609
|
+
return regex.test(c);
|
|
610
|
+
}
|
|
611
|
+
static notMatch(regex, c) {
|
|
612
|
+
return !_SAX.isMatch(regex, c);
|
|
613
|
+
}
|
|
614
|
+
static qname(name, attribute) {
|
|
615
|
+
const i = name.indexOf(":");
|
|
616
|
+
const qualName = i < 0 ? ["", name] : name.split(":");
|
|
617
|
+
let prefix = qualName[0];
|
|
618
|
+
let local = qualName[1];
|
|
619
|
+
if (attribute && name === "xmlns") {
|
|
620
|
+
prefix = "xmlns";
|
|
621
|
+
local = "";
|
|
622
|
+
}
|
|
623
|
+
return { prefix, local };
|
|
624
|
+
}
|
|
625
|
+
write(chunk) {
|
|
626
|
+
if (this.error) {
|
|
627
|
+
throw this.error;
|
|
628
|
+
}
|
|
629
|
+
if (this.closed) {
|
|
630
|
+
return this.errorFunction("Cannot write after close. Assign an onready handler.");
|
|
631
|
+
}
|
|
632
|
+
if (chunk === null) {
|
|
633
|
+
return this.end();
|
|
634
|
+
}
|
|
635
|
+
if (typeof chunk === "object") {
|
|
636
|
+
chunk = chunk.toString();
|
|
637
|
+
}
|
|
638
|
+
let i = 0;
|
|
639
|
+
let c;
|
|
640
|
+
while (true) {
|
|
641
|
+
c = _SAX.charAt(chunk, i++);
|
|
642
|
+
this.c = c;
|
|
643
|
+
if (!c) {
|
|
644
|
+
break;
|
|
645
|
+
}
|
|
646
|
+
if (this.trackPosition) {
|
|
647
|
+
this.position++;
|
|
648
|
+
if (c === "\n") {
|
|
649
|
+
this.line++;
|
|
650
|
+
this.column = 0;
|
|
651
|
+
} else {
|
|
652
|
+
this.column++;
|
|
653
|
+
}
|
|
654
|
+
}
|
|
655
|
+
switch (this.state) {
|
|
656
|
+
case this.S.BEGIN:
|
|
657
|
+
this.state = this.S.BEGIN_WHITESPACE;
|
|
658
|
+
if (c === "\uFEFF") {
|
|
659
|
+
continue;
|
|
660
|
+
}
|
|
661
|
+
this.beginWhiteSpace(c);
|
|
662
|
+
continue;
|
|
663
|
+
case this.S.BEGIN_WHITESPACE:
|
|
664
|
+
this.beginWhiteSpace(c);
|
|
665
|
+
continue;
|
|
666
|
+
case this.S.TEXT:
|
|
667
|
+
if (this.sawRoot && !this.closedRoot) {
|
|
668
|
+
const starti = i - 1;
|
|
669
|
+
while (c && c !== "<" && c !== "&") {
|
|
670
|
+
c = _SAX.charAt(chunk, i++);
|
|
671
|
+
if (c && this.trackPosition) {
|
|
672
|
+
this.position++;
|
|
673
|
+
if (c === "\n") {
|
|
674
|
+
this.line++;
|
|
675
|
+
this.column = 0;
|
|
676
|
+
} else {
|
|
677
|
+
this.column++;
|
|
678
|
+
}
|
|
679
|
+
}
|
|
680
|
+
}
|
|
681
|
+
this.textNode += chunk.substring(starti, i - 1);
|
|
682
|
+
}
|
|
683
|
+
if (c === "<" && !(this.sawRoot && this.closedRoot && !this.strict)) {
|
|
684
|
+
this.state = this.S.OPEN_WAKA;
|
|
685
|
+
this.startTagPosition = this.position;
|
|
686
|
+
} else {
|
|
687
|
+
if (!_SAX.isWhitespace(c) && (!this.sawRoot || this.closedRoot)) {
|
|
688
|
+
this.strictFail("Text data outside of root node.");
|
|
689
|
+
}
|
|
690
|
+
if (c === "&") {
|
|
691
|
+
this.state = this.S.TEXT_ENTITY;
|
|
692
|
+
} else {
|
|
693
|
+
this.textNode += c;
|
|
694
|
+
}
|
|
695
|
+
}
|
|
696
|
+
continue;
|
|
697
|
+
case this.S.SCRIPT:
|
|
698
|
+
if (c === "<") {
|
|
699
|
+
this.state = this.S.SCRIPT_ENDING;
|
|
700
|
+
} else {
|
|
701
|
+
this.script += c;
|
|
702
|
+
}
|
|
703
|
+
continue;
|
|
704
|
+
case this.S.SCRIPT_ENDING:
|
|
705
|
+
if (c === "/") {
|
|
706
|
+
this.state = this.S.CLOSE_TAG;
|
|
707
|
+
} else {
|
|
708
|
+
this.script += `<${c}`;
|
|
709
|
+
this.state = this.S.SCRIPT;
|
|
710
|
+
}
|
|
711
|
+
continue;
|
|
712
|
+
case this.S.OPEN_WAKA:
|
|
713
|
+
if (c === "!") {
|
|
714
|
+
this.state = this.S.SGML_DECL;
|
|
715
|
+
this.sgmlDecl = "";
|
|
716
|
+
} else if (_SAX.isWhitespace(c)) {
|
|
717
|
+
} else if (_SAX.isMatch(nameStart, c)) {
|
|
718
|
+
this.state = this.S.OPEN_TAG;
|
|
719
|
+
this.tagName = c;
|
|
720
|
+
} else if (c === "/") {
|
|
721
|
+
this.state = this.S.CLOSE_TAG;
|
|
722
|
+
this.tagName = "";
|
|
723
|
+
} else if (c === "?") {
|
|
724
|
+
this.state = this.S.PROC_INST;
|
|
725
|
+
this.procInstName = this.procInstBody = "";
|
|
726
|
+
} else {
|
|
727
|
+
this.strictFail("Unencoded <");
|
|
728
|
+
if (this.startTagPosition + 1 < this.position) {
|
|
729
|
+
const pad = this.position - this.startTagPosition;
|
|
730
|
+
c = new Array(pad).join(" ") + c;
|
|
731
|
+
}
|
|
732
|
+
this.textNode += `<${c}`;
|
|
733
|
+
this.state = this.S.TEXT;
|
|
734
|
+
}
|
|
735
|
+
continue;
|
|
736
|
+
case this.S.SGML_DECL:
|
|
737
|
+
if ((this.sgmlDecl + c).toUpperCase() === this.CDATA) {
|
|
738
|
+
this.emitNode("onopencdata");
|
|
739
|
+
this.state = this.S.CDATA;
|
|
740
|
+
this.sgmlDecl = "";
|
|
741
|
+
this.cdata = "";
|
|
742
|
+
} else if (this.sgmlDecl + c === "--") {
|
|
743
|
+
this.state = this.S.COMMENT;
|
|
744
|
+
this.comment = "";
|
|
745
|
+
this.sgmlDecl = "";
|
|
746
|
+
} else if ((this.sgmlDecl + c).toUpperCase() === this.DOCTYPE) {
|
|
747
|
+
this.state = this.S.DOCTYPE;
|
|
748
|
+
if (this.doctype || this.sawRoot) {
|
|
749
|
+
this.strictFail("Inappropriately located doctype declaration");
|
|
750
|
+
}
|
|
751
|
+
this.doctype = "";
|
|
752
|
+
this.sgmlDecl = "";
|
|
753
|
+
} else if (c === ">") {
|
|
754
|
+
this.emitNode("onsgmldeclaration", this.sgmlDecl);
|
|
755
|
+
this.sgmlDecl = "";
|
|
756
|
+
this.state = this.S.TEXT;
|
|
757
|
+
} else if (_SAX.isQuote(c)) {
|
|
758
|
+
this.state = this.S.SGML_DECL_QUOTED;
|
|
759
|
+
this.sgmlDecl += c;
|
|
760
|
+
} else {
|
|
761
|
+
this.sgmlDecl += c;
|
|
762
|
+
}
|
|
763
|
+
continue;
|
|
764
|
+
case this.S.SGML_DECL_QUOTED:
|
|
765
|
+
if (c === this.q) {
|
|
766
|
+
this.state = this.S.SGML_DECL;
|
|
767
|
+
this.q = "";
|
|
768
|
+
}
|
|
769
|
+
this.sgmlDecl += c;
|
|
770
|
+
continue;
|
|
771
|
+
case this.S.DOCTYPE:
|
|
772
|
+
if (c === ">") {
|
|
773
|
+
this.state = this.S.TEXT;
|
|
774
|
+
this.emitNode("ondoctype", this.doctype);
|
|
775
|
+
this.doctype = true;
|
|
776
|
+
} else {
|
|
777
|
+
this.doctype += c;
|
|
778
|
+
if (c === "[") {
|
|
779
|
+
this.state = this.S.DOCTYPE_DTD;
|
|
780
|
+
} else if (_SAX.isQuote(c)) {
|
|
781
|
+
this.state = this.S.DOCTYPE_QUOTED;
|
|
782
|
+
this.q = c;
|
|
783
|
+
}
|
|
784
|
+
}
|
|
785
|
+
continue;
|
|
786
|
+
case this.S.DOCTYPE_QUOTED:
|
|
787
|
+
this.doctype += c;
|
|
788
|
+
if (c === this.q) {
|
|
789
|
+
this.q = "";
|
|
790
|
+
this.state = this.S.DOCTYPE;
|
|
791
|
+
}
|
|
792
|
+
continue;
|
|
793
|
+
case this.S.DOCTYPE_DTD:
|
|
794
|
+
this.doctype += c;
|
|
795
|
+
if (c === "]") {
|
|
796
|
+
this.state = this.S.DOCTYPE;
|
|
797
|
+
} else if (_SAX.isQuote(c)) {
|
|
798
|
+
this.state = this.S.DOCTYPE_DTD_QUOTED;
|
|
799
|
+
this.q = c;
|
|
800
|
+
}
|
|
801
|
+
continue;
|
|
802
|
+
case this.S.DOCTYPE_DTD_QUOTED:
|
|
803
|
+
this.doctype += c;
|
|
804
|
+
if (c === this.q) {
|
|
805
|
+
this.state = this.S.DOCTYPE_DTD;
|
|
806
|
+
this.q = "";
|
|
807
|
+
}
|
|
808
|
+
continue;
|
|
809
|
+
case this.S.COMMENT:
|
|
810
|
+
if (c === "-") {
|
|
811
|
+
this.state = this.S.COMMENT_ENDING;
|
|
812
|
+
} else {
|
|
813
|
+
this.comment += c;
|
|
814
|
+
}
|
|
815
|
+
continue;
|
|
816
|
+
case this.S.COMMENT_ENDING:
|
|
817
|
+
if (c === "-") {
|
|
818
|
+
this.state = this.S.COMMENT_ENDED;
|
|
819
|
+
this.comment = this.textApplyOptions(this.comment);
|
|
820
|
+
if (this.comment) {
|
|
821
|
+
this.emitNode("oncomment", this.comment);
|
|
822
|
+
}
|
|
823
|
+
this.comment = "";
|
|
824
|
+
} else {
|
|
825
|
+
this.comment += `-${c}`;
|
|
826
|
+
this.state = this.S.COMMENT;
|
|
827
|
+
}
|
|
828
|
+
continue;
|
|
829
|
+
case this.S.COMMENT_ENDED:
|
|
830
|
+
if (c !== ">") {
|
|
831
|
+
this.strictFail("Malformed comment");
|
|
832
|
+
this.comment += `--${c}`;
|
|
833
|
+
this.state = this.S.COMMENT;
|
|
834
|
+
} else {
|
|
835
|
+
this.state = this.S.TEXT;
|
|
836
|
+
}
|
|
837
|
+
continue;
|
|
838
|
+
case this.S.CDATA:
|
|
839
|
+
if (c === "]") {
|
|
840
|
+
this.state = this.S.CDATA_ENDING;
|
|
841
|
+
} else {
|
|
842
|
+
this.cdata += c;
|
|
843
|
+
}
|
|
844
|
+
continue;
|
|
845
|
+
case this.S.CDATA_ENDING:
|
|
846
|
+
if (c === "]") {
|
|
847
|
+
this.state = this.S.CDATA_ENDING_2;
|
|
848
|
+
} else {
|
|
849
|
+
this.cdata += `]${c}`;
|
|
850
|
+
this.state = this.S.CDATA;
|
|
851
|
+
}
|
|
852
|
+
continue;
|
|
853
|
+
case this.S.CDATA_ENDING_2:
|
|
854
|
+
if (c === ">") {
|
|
855
|
+
if (this.cdata) {
|
|
856
|
+
this.emitNode("oncdata", this.cdata);
|
|
857
|
+
}
|
|
858
|
+
this.emitNode("onclosecdata");
|
|
859
|
+
this.cdata = "";
|
|
860
|
+
this.state = this.S.TEXT;
|
|
861
|
+
} else if (c === "]") {
|
|
862
|
+
this.cdata += "]";
|
|
863
|
+
} else {
|
|
864
|
+
this.cdata += `]]${c}`;
|
|
865
|
+
this.state = this.S.CDATA;
|
|
866
|
+
}
|
|
867
|
+
continue;
|
|
868
|
+
case this.S.PROC_INST:
|
|
869
|
+
if (c === "?") {
|
|
870
|
+
this.state = this.S.PROC_INST_ENDING;
|
|
871
|
+
} else if (_SAX.isWhitespace(c)) {
|
|
872
|
+
this.state = this.S.PROC_INST_BODY;
|
|
873
|
+
} else {
|
|
874
|
+
this.procInstName += c;
|
|
875
|
+
}
|
|
876
|
+
continue;
|
|
877
|
+
case this.S.PROC_INST_BODY:
|
|
878
|
+
if (!this.procInstBody && _SAX.isWhitespace(c)) {
|
|
879
|
+
continue;
|
|
880
|
+
} else if (c === "?") {
|
|
881
|
+
this.state = this.S.PROC_INST_ENDING;
|
|
882
|
+
} else {
|
|
883
|
+
this.procInstBody += c;
|
|
884
|
+
}
|
|
885
|
+
continue;
|
|
886
|
+
case this.S.PROC_INST_ENDING:
|
|
887
|
+
if (c === ">") {
|
|
888
|
+
this.emitNode("onprocessinginstruction", {
|
|
889
|
+
name: this.procInstName,
|
|
890
|
+
body: this.procInstBody
|
|
891
|
+
});
|
|
892
|
+
this.procInstName = this.procInstBody = "";
|
|
893
|
+
this.state = this.S.TEXT;
|
|
894
|
+
} else {
|
|
895
|
+
this.procInstBody += `?${c}`;
|
|
896
|
+
this.state = this.S.PROC_INST_BODY;
|
|
897
|
+
}
|
|
898
|
+
continue;
|
|
899
|
+
case this.S.OPEN_TAG:
|
|
900
|
+
if (_SAX.isMatch(nameBody, c)) {
|
|
901
|
+
this.tagName += c;
|
|
902
|
+
} else {
|
|
903
|
+
this.newTag();
|
|
904
|
+
if (c === ">") {
|
|
905
|
+
this.openTag();
|
|
906
|
+
} else if (c === "/") {
|
|
907
|
+
this.state = this.S.OPEN_TAG_SLASH;
|
|
908
|
+
} else {
|
|
909
|
+
if (!_SAX.isWhitespace(c)) {
|
|
910
|
+
this.strictFail("Invalid character in tag name");
|
|
911
|
+
}
|
|
912
|
+
this.state = this.S.ATTRIB;
|
|
913
|
+
}
|
|
914
|
+
}
|
|
915
|
+
continue;
|
|
916
|
+
case this.S.OPEN_TAG_SLASH:
|
|
917
|
+
if (c === ">") {
|
|
918
|
+
this.openTag(true);
|
|
919
|
+
this.closeTag();
|
|
920
|
+
} else {
|
|
921
|
+
this.strictFail("Forward-slash in opening tag not followed by >");
|
|
922
|
+
this.state = this.S.ATTRIB;
|
|
923
|
+
}
|
|
924
|
+
continue;
|
|
925
|
+
case this.S.ATTRIB:
|
|
926
|
+
if (_SAX.isWhitespace(c)) {
|
|
927
|
+
continue;
|
|
928
|
+
} else if (c === ">") {
|
|
929
|
+
this.openTag();
|
|
930
|
+
} else if (c === "/") {
|
|
931
|
+
this.state = this.S.OPEN_TAG_SLASH;
|
|
932
|
+
} else if (_SAX.isMatch(nameStart, c)) {
|
|
933
|
+
this.attribName = c;
|
|
934
|
+
this.attribValue = "";
|
|
935
|
+
this.state = this.S.ATTRIB_NAME;
|
|
936
|
+
} else {
|
|
937
|
+
this.strictFail("Invalid attribute name");
|
|
938
|
+
}
|
|
939
|
+
continue;
|
|
940
|
+
case this.S.ATTRIB_NAME:
|
|
941
|
+
if (c === "=") {
|
|
942
|
+
this.state = this.S.ATTRIB_VALUE;
|
|
943
|
+
} else if (c === ">") {
|
|
944
|
+
this.strictFail("Attribute without value");
|
|
945
|
+
this.attribValue = this.attribName;
|
|
946
|
+
this.attrib();
|
|
947
|
+
this.openTag();
|
|
948
|
+
} else if (_SAX.isWhitespace(c)) {
|
|
949
|
+
this.state = this.S.ATTRIB_NAME_SAW_WHITE;
|
|
950
|
+
} else if (_SAX.isMatch(nameBody, c)) {
|
|
951
|
+
this.attribName += c;
|
|
952
|
+
} else {
|
|
953
|
+
this.strictFail("Invalid attribute name");
|
|
954
|
+
}
|
|
955
|
+
continue;
|
|
956
|
+
case this.S.ATTRIB_NAME_SAW_WHITE:
|
|
957
|
+
if (c === "=") {
|
|
958
|
+
this.state = this.S.ATTRIB_VALUE;
|
|
959
|
+
} else if (_SAX.isWhitespace(c)) {
|
|
960
|
+
continue;
|
|
961
|
+
} else {
|
|
962
|
+
this.strictFail("Attribute without value");
|
|
963
|
+
Object.defineProperty(this.tag.attributes, this.attribName, {
|
|
964
|
+
value: "",
|
|
965
|
+
enumerable: true,
|
|
966
|
+
configurable: true,
|
|
967
|
+
writable: true
|
|
968
|
+
});
|
|
969
|
+
this.attribValue = "";
|
|
970
|
+
this.emitNode("onattribute", {
|
|
971
|
+
name: this.attribName,
|
|
972
|
+
value: ""
|
|
973
|
+
});
|
|
974
|
+
this.attribName = "";
|
|
975
|
+
if (c === ">") {
|
|
976
|
+
this.openTag();
|
|
977
|
+
} else if (_SAX.isMatch(nameStart, c)) {
|
|
978
|
+
this.attribName = c;
|
|
979
|
+
this.state = this.S.ATTRIB_NAME;
|
|
980
|
+
} else {
|
|
981
|
+
this.strictFail("Invalid attribute name");
|
|
982
|
+
this.state = this.S.ATTRIB;
|
|
983
|
+
}
|
|
984
|
+
}
|
|
985
|
+
continue;
|
|
986
|
+
case this.S.ATTRIB_VALUE:
|
|
987
|
+
if (_SAX.isWhitespace(c)) {
|
|
988
|
+
continue;
|
|
989
|
+
} else if (_SAX.isQuote(c)) {
|
|
990
|
+
this.q = c;
|
|
991
|
+
this.state = this.S.ATTRIB_VALUE_QUOTED;
|
|
992
|
+
} else {
|
|
993
|
+
this.strictFail("Unquoted attribute value");
|
|
994
|
+
this.state = this.S.ATTRIB_VALUE_UNQUOTED;
|
|
995
|
+
this.attribValue = c;
|
|
996
|
+
}
|
|
997
|
+
continue;
|
|
998
|
+
case this.S.ATTRIB_VALUE_QUOTED:
|
|
999
|
+
if (c !== this.q) {
|
|
1000
|
+
if (c === "&") {
|
|
1001
|
+
this.state = this.S.ATTRIB_VALUE_ENTITY_Q;
|
|
1002
|
+
} else {
|
|
1003
|
+
this.attribValue += c;
|
|
1004
|
+
}
|
|
1005
|
+
continue;
|
|
1006
|
+
}
|
|
1007
|
+
this.attrib();
|
|
1008
|
+
this.q = "";
|
|
1009
|
+
this.state = this.S.ATTRIB_VALUE_CLOSED;
|
|
1010
|
+
continue;
|
|
1011
|
+
case this.S.ATTRIB_VALUE_CLOSED:
|
|
1012
|
+
if (_SAX.isWhitespace(c)) {
|
|
1013
|
+
this.state = this.S.ATTRIB;
|
|
1014
|
+
} else if (c === ">") {
|
|
1015
|
+
this.openTag();
|
|
1016
|
+
} else if (c === "/") {
|
|
1017
|
+
this.state = this.S.OPEN_TAG_SLASH;
|
|
1018
|
+
} else if (_SAX.isMatch(nameStart, c)) {
|
|
1019
|
+
this.strictFail("No whitespace between attributes");
|
|
1020
|
+
this.attribName = c;
|
|
1021
|
+
this.attribValue = "";
|
|
1022
|
+
this.state = this.S.ATTRIB_NAME;
|
|
1023
|
+
} else {
|
|
1024
|
+
this.strictFail("Invalid attribute name");
|
|
1025
|
+
}
|
|
1026
|
+
continue;
|
|
1027
|
+
case this.S.ATTRIB_VALUE_UNQUOTED:
|
|
1028
|
+
if (!_SAX.isAttribEnd(c)) {
|
|
1029
|
+
if (c === "&") {
|
|
1030
|
+
this.state = this.S.ATTRIB_VALUE_ENTITY_U;
|
|
1031
|
+
} else {
|
|
1032
|
+
this.attribValue += c;
|
|
1033
|
+
}
|
|
1034
|
+
continue;
|
|
1035
|
+
}
|
|
1036
|
+
this.attrib();
|
|
1037
|
+
if (c === ">") {
|
|
1038
|
+
this.openTag();
|
|
1039
|
+
} else {
|
|
1040
|
+
this.state = this.S.ATTRIB;
|
|
1041
|
+
}
|
|
1042
|
+
continue;
|
|
1043
|
+
case this.S.CLOSE_TAG:
|
|
1044
|
+
if (!this.tagName) {
|
|
1045
|
+
if (_SAX.isWhitespace(c)) {
|
|
1046
|
+
continue;
|
|
1047
|
+
} else if (_SAX.notMatch(nameStart, c)) {
|
|
1048
|
+
if (this.script) {
|
|
1049
|
+
this.script += `</${c}`;
|
|
1050
|
+
this.state = this.S.SCRIPT;
|
|
1051
|
+
} else {
|
|
1052
|
+
this.strictFail("Invalid tagname in closing tag.");
|
|
1053
|
+
}
|
|
1054
|
+
} else {
|
|
1055
|
+
this.tagName = c;
|
|
1056
|
+
}
|
|
1057
|
+
} else if (c === ">") {
|
|
1058
|
+
this.closeTag();
|
|
1059
|
+
} else if (_SAX.isMatch(nameBody, c)) {
|
|
1060
|
+
this.tagName += c;
|
|
1061
|
+
} else if (this.script) {
|
|
1062
|
+
this.script += `</${this.tagName}`;
|
|
1063
|
+
this.tagName = "";
|
|
1064
|
+
this.state = this.S.SCRIPT;
|
|
1065
|
+
} else {
|
|
1066
|
+
if (!_SAX.isWhitespace(c)) {
|
|
1067
|
+
this.strictFail("Invalid tagname in closing tag");
|
|
1068
|
+
}
|
|
1069
|
+
this.state = this.S.CLOSE_TAG_SAW_WHITE;
|
|
1070
|
+
}
|
|
1071
|
+
continue;
|
|
1072
|
+
case this.S.CLOSE_TAG_SAW_WHITE:
|
|
1073
|
+
if (_SAX.isWhitespace(c)) {
|
|
1074
|
+
continue;
|
|
1075
|
+
}
|
|
1076
|
+
if (c === ">") {
|
|
1077
|
+
this.closeTag();
|
|
1078
|
+
} else {
|
|
1079
|
+
this.strictFail("Invalid characters in closing tag");
|
|
1080
|
+
}
|
|
1081
|
+
continue;
|
|
1082
|
+
case this.S.TEXT_ENTITY:
|
|
1083
|
+
case this.S.ATTRIB_VALUE_ENTITY_Q:
|
|
1084
|
+
case this.S.ATTRIB_VALUE_ENTITY_U:
|
|
1085
|
+
let returnState;
|
|
1086
|
+
let buffer;
|
|
1087
|
+
switch (this.state) {
|
|
1088
|
+
case this.S.TEXT_ENTITY:
|
|
1089
|
+
returnState = this.S.TEXT;
|
|
1090
|
+
buffer = "textNode";
|
|
1091
|
+
break;
|
|
1092
|
+
case this.S.ATTRIB_VALUE_ENTITY_Q:
|
|
1093
|
+
returnState = this.S.ATTRIB_VALUE_QUOTED;
|
|
1094
|
+
buffer = "attribValue";
|
|
1095
|
+
break;
|
|
1096
|
+
case this.S.ATTRIB_VALUE_ENTITY_U:
|
|
1097
|
+
returnState = this.S.ATTRIB_VALUE_UNQUOTED;
|
|
1098
|
+
buffer = "attribValue";
|
|
1099
|
+
break;
|
|
1100
|
+
default:
|
|
1101
|
+
throw new Error(`Unknown state: ${this.state}`);
|
|
1102
|
+
}
|
|
1103
|
+
if (c === ";") {
|
|
1104
|
+
this[buffer] += this.parseEntity();
|
|
1105
|
+
this.entity = "";
|
|
1106
|
+
this.state = returnState;
|
|
1107
|
+
} else if (_SAX.isMatch(this.entity.length ? entityBody : entityStart, c)) {
|
|
1108
|
+
this.entity += c;
|
|
1109
|
+
} else {
|
|
1110
|
+
this.strictFail("Invalid character in entity name");
|
|
1111
|
+
this[buffer] += `&${this.entity}${c}`;
|
|
1112
|
+
this.entity = "";
|
|
1113
|
+
this.state = returnState;
|
|
1114
|
+
}
|
|
1115
|
+
continue;
|
|
1116
|
+
default:
|
|
1117
|
+
throw new Error(`Unknown state: ${this.state}`);
|
|
1118
|
+
}
|
|
1119
|
+
}
|
|
1120
|
+
if (this.position >= this.bufferCheckPosition) {
|
|
1121
|
+
this.checkBufferLength();
|
|
1122
|
+
}
|
|
1123
|
+
return this;
|
|
1124
|
+
}
|
|
1125
|
+
emit(event, data) {
|
|
1126
|
+
if (this.events.hasOwnProperty(event)) {
|
|
1127
|
+
const eventName = event.replace(/^on/, "");
|
|
1128
|
+
this.events[event](data, eventName, this);
|
|
1129
|
+
}
|
|
1130
|
+
}
|
|
1131
|
+
clearBuffers() {
|
|
1132
|
+
for (let i = 0, l = this.BUFFERS.length; i < l; i++) {
|
|
1133
|
+
this[this[i]] = "";
|
|
1134
|
+
}
|
|
1135
|
+
}
|
|
1136
|
+
flushBuffers() {
|
|
1137
|
+
this.closeText();
|
|
1138
|
+
if (this.cdata !== "") {
|
|
1139
|
+
this.emitNode("oncdata", this.cdata);
|
|
1140
|
+
this.cdata = "";
|
|
1141
|
+
}
|
|
1142
|
+
if (this.script !== "") {
|
|
1143
|
+
this.emitNode("onscript", this.script);
|
|
1144
|
+
this.script = "";
|
|
1145
|
+
}
|
|
1146
|
+
}
|
|
1147
|
+
end() {
|
|
1148
|
+
if (this.sawRoot && !this.closedRoot) this.strictFail("Unclosed root tag");
|
|
1149
|
+
if (this.strict && this.state !== this.S.BEGIN && this.state !== this.S.BEGIN_WHITESPACE && this.state !== this.S.TEXT) {
|
|
1150
|
+
this.errorFunction("Unexpected end");
|
|
1151
|
+
}
|
|
1152
|
+
this.closeText();
|
|
1153
|
+
this.c = "";
|
|
1154
|
+
this.closed = true;
|
|
1155
|
+
this.emit("onend");
|
|
1156
|
+
return new SAXParser(this.opt);
|
|
1157
|
+
}
|
|
1158
|
+
errorFunction(er) {
|
|
1159
|
+
this.closeText();
|
|
1160
|
+
if (this.trackPosition) {
|
|
1161
|
+
er += `
|
|
1162
|
+
Line: ${this.line}
|
|
1163
|
+
Column: ${this.column}
|
|
1164
|
+
Char: ${this.c}`;
|
|
1165
|
+
}
|
|
1166
|
+
const error = new Error(er);
|
|
1167
|
+
this.error = error;
|
|
1168
|
+
this.emit("onerror", error);
|
|
1169
|
+
return this;
|
|
1170
|
+
}
|
|
1171
|
+
attrib() {
|
|
1172
|
+
if (!this.strict && !this.opt.preserveCase) {
|
|
1173
|
+
this.attribName = this.attribName[this.looseCase]();
|
|
1174
|
+
}
|
|
1175
|
+
if (this.attribList.indexOf(this.attribName) !== -1 || Object.hasOwn(this.tag.attributes, this.attribName)) {
|
|
1176
|
+
this.strictFail("Duplicate attribute");
|
|
1177
|
+
this.attribName = this.attribValue = "";
|
|
1178
|
+
return;
|
|
1179
|
+
}
|
|
1180
|
+
if (this.opt.xmlns) {
|
|
1181
|
+
const qn = _SAX.qname(this.attribName, true);
|
|
1182
|
+
const prefix = qn.prefix;
|
|
1183
|
+
const local = qn.local;
|
|
1184
|
+
if (prefix === "xmlns") {
|
|
1185
|
+
if (local === "xml" && this.attribValue !== this.XML_NAMESPACE) {
|
|
1186
|
+
this.strictFail(
|
|
1187
|
+
`xml: prefix must be bound to ${this.XML_NAMESPACE}
|
|
1188
|
+
Actual: ${this.attribValue}`
|
|
1189
|
+
);
|
|
1190
|
+
} else if (local === "xmlns" && this.attribValue !== this.XMLNS_NAMESPACE) {
|
|
1191
|
+
this.strictFail(
|
|
1192
|
+
`xmlns: prefix must be bound to ${this.XMLNS_NAMESPACE}
|
|
1193
|
+
Actual: ${this.attribValue}`
|
|
1194
|
+
);
|
|
1195
|
+
} else {
|
|
1196
|
+
const tag = this.tag;
|
|
1197
|
+
const parent = this.tags[this.tags.length - 1] || this;
|
|
1198
|
+
if (tag.ns === parent.ns) {
|
|
1199
|
+
tag.ns = Object.create(parent.ns);
|
|
1200
|
+
}
|
|
1201
|
+
tag.ns[local] = this.attribValue;
|
|
1202
|
+
}
|
|
1203
|
+
}
|
|
1204
|
+
this.attribList.push([this.attribName, this.attribValue]);
|
|
1205
|
+
} else {
|
|
1206
|
+
Object.defineProperty(this.tag.attributes, this.attribName, {
|
|
1207
|
+
value: this.attribValue,
|
|
1208
|
+
enumerable: true,
|
|
1209
|
+
configurable: true,
|
|
1210
|
+
writable: true
|
|
1211
|
+
});
|
|
1212
|
+
this.emitNode("onattribute", {
|
|
1213
|
+
name: this.attribName,
|
|
1214
|
+
value: this.attribValue
|
|
1215
|
+
});
|
|
1216
|
+
}
|
|
1217
|
+
this.attribName = this.attribValue = "";
|
|
1218
|
+
}
|
|
1219
|
+
newTag() {
|
|
1220
|
+
if (!this.strict && !this.opt.preserveCase) this.tagName = this.tagName[this.looseCase]();
|
|
1221
|
+
const parent = this.tags[this.tags.length - 1] || this;
|
|
1222
|
+
const tag = { name: this.tagName, attributes: {} };
|
|
1223
|
+
this.tag = tag;
|
|
1224
|
+
if (this.opt.xmlns) {
|
|
1225
|
+
tag.ns = parent.ns;
|
|
1226
|
+
}
|
|
1227
|
+
this.attribList.length = 0;
|
|
1228
|
+
this.emitNode("onopentagstart", tag);
|
|
1229
|
+
}
|
|
1230
|
+
/** Validates an entity before optionally preserving its authored spelling. */
|
|
1231
|
+
parseEntity() {
|
|
1232
|
+
const decoded = this.decodeEntity();
|
|
1233
|
+
return this.opt.processEntities === false ? `&${this.entity};` : decoded;
|
|
1234
|
+
}
|
|
1235
|
+
/** Decodes predefined, configured, or numeric entities without resolving document types. */
|
|
1236
|
+
decodeEntity() {
|
|
1237
|
+
let entity = this.entity;
|
|
1238
|
+
const entityLC = entity.toLowerCase();
|
|
1239
|
+
let num = NaN;
|
|
1240
|
+
let numStr = "";
|
|
1241
|
+
if (typeof this.ENTITIES[entity] === "string" || typeof this.ENTITIES[entity] === "number") {
|
|
1242
|
+
return this.ENTITIES[entity];
|
|
1243
|
+
}
|
|
1244
|
+
if (typeof this.ENTITIES[entityLC] === "string" || typeof this.ENTITIES[entityLC] === "number") {
|
|
1245
|
+
return this.ENTITIES[entityLC];
|
|
1246
|
+
}
|
|
1247
|
+
entity = entityLC;
|
|
1248
|
+
if (entity.charAt(0) === "#") {
|
|
1249
|
+
if (entity.charAt(1) === "x") {
|
|
1250
|
+
entity = entity.slice(2);
|
|
1251
|
+
num = parseInt(entity, 16);
|
|
1252
|
+
numStr = num.toString(16);
|
|
1253
|
+
} else {
|
|
1254
|
+
entity = entity.slice(1);
|
|
1255
|
+
num = parseInt(entity, 10);
|
|
1256
|
+
numStr = num.toString(10);
|
|
1257
|
+
}
|
|
1258
|
+
}
|
|
1259
|
+
entity = entity.replace(/^0+/, "");
|
|
1260
|
+
if (Number.isNaN(num) || numStr.toLowerCase() !== entity) {
|
|
1261
|
+
this.strictFail("Invalid character entity");
|
|
1262
|
+
return `&${this.entity};`;
|
|
1263
|
+
}
|
|
1264
|
+
if (num < 0 || num > 1114111 || num >= 55296 && num <= 57343 || num < 32 && ![9, 10, 13].includes(num) || num === 65534 || num === 65535) {
|
|
1265
|
+
this.strictFail("Invalid character entity");
|
|
1266
|
+
return `&${this.entity};`;
|
|
1267
|
+
}
|
|
1268
|
+
return String.fromCodePoint(num);
|
|
1269
|
+
}
|
|
1270
|
+
beginWhiteSpace(c) {
|
|
1271
|
+
if (c === "<") {
|
|
1272
|
+
this.state = this.S.OPEN_WAKA;
|
|
1273
|
+
this.startTagPosition = this.position;
|
|
1274
|
+
} else if (!_SAX.isWhitespace(c)) {
|
|
1275
|
+
this.strictFail("Non-whitespace before first tag.");
|
|
1276
|
+
this.textNode = c;
|
|
1277
|
+
this.state = this.S.TEXT;
|
|
1278
|
+
} else {
|
|
1279
|
+
}
|
|
1280
|
+
}
|
|
1281
|
+
strictFail(message) {
|
|
1282
|
+
if (typeof this !== "object" || !(this instanceof SAXParser)) {
|
|
1283
|
+
throw new Error("bad call to strictFail");
|
|
1284
|
+
}
|
|
1285
|
+
if (this.strict) {
|
|
1286
|
+
this.errorFunction(message);
|
|
1287
|
+
}
|
|
1288
|
+
}
|
|
1289
|
+
textApplyOptions(text) {
|
|
1290
|
+
if (this.opt.trim) text = text.trim();
|
|
1291
|
+
if (this.opt.normalize) text = text.replace(/\s+/g, " ");
|
|
1292
|
+
return text;
|
|
1293
|
+
}
|
|
1294
|
+
emitNode(nodeType, data) {
|
|
1295
|
+
if (this.textNode) this.closeText();
|
|
1296
|
+
this.emit(nodeType, data);
|
|
1297
|
+
}
|
|
1298
|
+
closeText() {
|
|
1299
|
+
this.textNode = this.textApplyOptions(this.textNode);
|
|
1300
|
+
if (this.textNode !== void 0 && this.textNode !== "" && this.textNode !== "undefined") {
|
|
1301
|
+
this.emit("ontext", this.textNode);
|
|
1302
|
+
}
|
|
1303
|
+
this.textNode = "";
|
|
1304
|
+
}
|
|
1305
|
+
checkBufferLength() {
|
|
1306
|
+
var _a;
|
|
1307
|
+
const maxAllowed = Math.max(this.opt.MAX_BUFFER_LENGTH, 10);
|
|
1308
|
+
let maxActual = 0;
|
|
1309
|
+
for (let i = 0, l = this.BUFFERS.length; i < l; i++) {
|
|
1310
|
+
const len = ((_a = this[this.BUFFERS[i]]) == null ? void 0 : _a.length) || 0;
|
|
1311
|
+
if (len > maxAllowed) {
|
|
1312
|
+
switch (this.BUFFERS[i]) {
|
|
1313
|
+
case "textNode":
|
|
1314
|
+
this.closeText();
|
|
1315
|
+
break;
|
|
1316
|
+
case "cdata":
|
|
1317
|
+
this.emitNode("oncdata", this.cdata);
|
|
1318
|
+
this.cdata = "";
|
|
1319
|
+
break;
|
|
1320
|
+
case "script":
|
|
1321
|
+
this.emitNode("onscript", this.script);
|
|
1322
|
+
this.script = "";
|
|
1323
|
+
break;
|
|
1324
|
+
default:
|
|
1325
|
+
this.errorFunction(`Max buffer length exceeded: ${this.BUFFERS[i]}`);
|
|
1326
|
+
}
|
|
1327
|
+
}
|
|
1328
|
+
maxActual = Math.max(maxActual, len);
|
|
1329
|
+
}
|
|
1330
|
+
const m = this.opt.MAX_BUFFER_LENGTH - maxActual;
|
|
1331
|
+
this.bufferCheckPosition = m + this.position;
|
|
1332
|
+
}
|
|
1333
|
+
openTag(selfClosing) {
|
|
1334
|
+
if (this.opt.xmlns) {
|
|
1335
|
+
const tag = this.tag;
|
|
1336
|
+
const qn = _SAX.qname(this.tagName);
|
|
1337
|
+
tag.prefix = qn.prefix;
|
|
1338
|
+
tag.local = qn.local;
|
|
1339
|
+
tag.uri = tag.ns[qn.prefix] || "";
|
|
1340
|
+
if (tag.prefix && !tag.uri) {
|
|
1341
|
+
this.strictFail(`Unbound namespace prefix: ${JSON.stringify(this.tagName)}`);
|
|
1342
|
+
tag.uri = qn.prefix;
|
|
1343
|
+
}
|
|
1344
|
+
const parent = this.tags[this.tags.length - 1] || this;
|
|
1345
|
+
if (tag.ns && parent.ns !== tag.ns) {
|
|
1346
|
+
const that = this;
|
|
1347
|
+
Object.keys(tag.ns).forEach((p) => {
|
|
1348
|
+
that.emitNode("onopennamespace", {
|
|
1349
|
+
prefix: p,
|
|
1350
|
+
uri: tag.ns[p]
|
|
1351
|
+
});
|
|
1352
|
+
});
|
|
1353
|
+
}
|
|
1354
|
+
for (let i = 0, l = this.attribList.length; i < l; i++) {
|
|
1355
|
+
const nv = this.attribList[i];
|
|
1356
|
+
const name = nv[0];
|
|
1357
|
+
const value = nv[1];
|
|
1358
|
+
const qualName = _SAX.qname(name, true);
|
|
1359
|
+
const prefix = qualName.prefix;
|
|
1360
|
+
const local = qualName.local;
|
|
1361
|
+
const uri = prefix === "" ? "" : tag.ns[prefix] || "";
|
|
1362
|
+
const a = {
|
|
1363
|
+
name,
|
|
1364
|
+
value,
|
|
1365
|
+
prefix,
|
|
1366
|
+
local,
|
|
1367
|
+
uri
|
|
1368
|
+
};
|
|
1369
|
+
if (prefix && prefix !== "xmlns" && !uri) {
|
|
1370
|
+
this.strictFail(`Unbound namespace prefix: ${JSON.stringify(prefix)}`);
|
|
1371
|
+
a.uri = prefix;
|
|
1372
|
+
}
|
|
1373
|
+
Object.defineProperty(this.tag.attributes, name, {
|
|
1374
|
+
value: a,
|
|
1375
|
+
enumerable: true,
|
|
1376
|
+
configurable: true,
|
|
1377
|
+
writable: true
|
|
1378
|
+
});
|
|
1379
|
+
this.emitNode("onattribute", a);
|
|
1380
|
+
}
|
|
1381
|
+
this.attribList.length = 0;
|
|
1382
|
+
}
|
|
1383
|
+
this.tag.isSelfClosing = Boolean(selfClosing);
|
|
1384
|
+
this.sawRoot = true;
|
|
1385
|
+
this.tags.push(this.tag);
|
|
1386
|
+
this.emitNode("onopentag", this.tag);
|
|
1387
|
+
if (!selfClosing) {
|
|
1388
|
+
if (!this.noscript && this.tagName.toLowerCase() === "script") {
|
|
1389
|
+
this.state = this.S.SCRIPT;
|
|
1390
|
+
} else {
|
|
1391
|
+
this.state = this.S.TEXT;
|
|
1392
|
+
}
|
|
1393
|
+
this.tag = null;
|
|
1394
|
+
this.tagName = "";
|
|
1395
|
+
}
|
|
1396
|
+
this.attribName = this.attribValue = "";
|
|
1397
|
+
this.attribList.length = 0;
|
|
1398
|
+
}
|
|
1399
|
+
closeTag() {
|
|
1400
|
+
if (!this.tagName) {
|
|
1401
|
+
this.strictFail("Weird empty close tag.");
|
|
1402
|
+
this.textNode += "</>";
|
|
1403
|
+
this.state = this.S.TEXT;
|
|
1404
|
+
return;
|
|
1405
|
+
}
|
|
1406
|
+
const caseInsensitiveTags = !this.strict && (this.opt.caseInsensitiveTags || !this.opt.preserveCase);
|
|
1407
|
+
if (this.script) {
|
|
1408
|
+
const normalizedScriptCloseTag = caseInsensitiveTags ? this.tagName.toLowerCase() : this.tagName;
|
|
1409
|
+
if (normalizedScriptCloseTag !== "script") {
|
|
1410
|
+
this.script += `</${this.tagName}>`;
|
|
1411
|
+
this.tagName = "";
|
|
1412
|
+
this.state = this.S.SCRIPT;
|
|
1413
|
+
return;
|
|
1414
|
+
}
|
|
1415
|
+
this.emitNode("onscript", this.script);
|
|
1416
|
+
this.script = "";
|
|
1417
|
+
}
|
|
1418
|
+
let t = this.tags.length;
|
|
1419
|
+
let tagName = this.tagName;
|
|
1420
|
+
const normalizedTagName = caseInsensitiveTags ? tagName.toLowerCase() : tagName;
|
|
1421
|
+
while (t--) {
|
|
1422
|
+
const close = this.tags[t];
|
|
1423
|
+
const normalizedOpenTagName = caseInsensitiveTags ? close.name.toLowerCase() : close.name;
|
|
1424
|
+
if (normalizedOpenTagName !== normalizedTagName) {
|
|
1425
|
+
this.strictFail("Unexpected close tag");
|
|
1426
|
+
} else {
|
|
1427
|
+
break;
|
|
1428
|
+
}
|
|
1429
|
+
}
|
|
1430
|
+
if (t < 0) {
|
|
1431
|
+
this.strictFail(`Unmatched closing tag: ${this.tagName}`);
|
|
1432
|
+
this.textNode += `</${this.tagName}>`;
|
|
1433
|
+
this.state = this.S.TEXT;
|
|
1434
|
+
return;
|
|
1435
|
+
}
|
|
1436
|
+
this.tagName = tagName;
|
|
1437
|
+
let s = this.tags.length;
|
|
1438
|
+
while (s-- > t) {
|
|
1439
|
+
const tag = this.tags.pop();
|
|
1440
|
+
this.tag = tag;
|
|
1441
|
+
this.tagName = this.tag.name;
|
|
1442
|
+
this.emitNode("onclosetag", this.tagName);
|
|
1443
|
+
const x = {};
|
|
1444
|
+
for (const i in tag.ns) {
|
|
1445
|
+
if (tag.ns.hasOwnProperty(i)) {
|
|
1446
|
+
x[i] = tag.ns[i];
|
|
1447
|
+
}
|
|
1448
|
+
}
|
|
1449
|
+
const parent = this.tags[this.tags.length - 1] || this;
|
|
1450
|
+
if (this.opt.xmlns && tag.ns !== parent.ns) {
|
|
1451
|
+
const that = this;
|
|
1452
|
+
Object.keys(tag.ns).forEach((p) => {
|
|
1453
|
+
const n = tag.ns[p];
|
|
1454
|
+
that.emitNode("onclosenamespace", { prefix: p, uri: n });
|
|
1455
|
+
});
|
|
1456
|
+
}
|
|
1457
|
+
}
|
|
1458
|
+
if (t === 0) this.closedRoot = true;
|
|
1459
|
+
this.tagName = this.attribValue = this.attribName = "";
|
|
1460
|
+
this.attribList.length = 0;
|
|
1461
|
+
this.state = this.S.TEXT;
|
|
1462
|
+
}
|
|
1463
|
+
};
|
|
1464
|
+
SAXParser = class extends SAX {
|
|
1465
|
+
static ENTITIES = ENTITIES;
|
|
1466
|
+
opt = DEFAULT_SAX_PARSER_OPTIONS;
|
|
1467
|
+
events = DEFAULT_SAX_EVENTS;
|
|
1468
|
+
constructor(opt) {
|
|
1469
|
+
super();
|
|
1470
|
+
this.clearBuffers();
|
|
1471
|
+
this.opt = opt = { ...this.opt, ...opt };
|
|
1472
|
+
this.events = { ...this.events, ...opt };
|
|
1473
|
+
this.q = this.c = "";
|
|
1474
|
+
this.opt.lowercase = this.opt.lowercase || this.opt.lowercasetags;
|
|
1475
|
+
this.bufferCheckPosition = this.opt.MAX_BUFFER_LENGTH;
|
|
1476
|
+
this.looseCase = this.opt.lowercase ? "toLowerCase" : "toUpperCase";
|
|
1477
|
+
this.tags = [];
|
|
1478
|
+
this.closed = this.closedRoot = this.sawRoot = false;
|
|
1479
|
+
this.tag = this.error = null;
|
|
1480
|
+
this.strict = Boolean(this.opt.strict);
|
|
1481
|
+
this.noscript = Boolean(this.opt.strict || this.opt.noscript);
|
|
1482
|
+
this.state = this.S.BEGIN;
|
|
1483
|
+
this.strictEntities = this.opt.strictEntities;
|
|
1484
|
+
this.ENTITIES = this.strictEntities ? Object.create(this.XML_ENTITIES) : Object.create(this.ENTITIES);
|
|
1485
|
+
this.attribList = [];
|
|
1486
|
+
if (this.opt.xmlns) {
|
|
1487
|
+
this.ns = Object.create(this.rootNS);
|
|
1488
|
+
}
|
|
1489
|
+
this.trackPosition = this.opt.position !== false;
|
|
1490
|
+
if (this.trackPosition) {
|
|
1491
|
+
this.position = this.line = this.column = 0;
|
|
1492
|
+
}
|
|
1493
|
+
this.emit("onready");
|
|
1494
|
+
}
|
|
1495
|
+
resume() {
|
|
1496
|
+
this.error = null;
|
|
1497
|
+
return this;
|
|
1498
|
+
}
|
|
1499
|
+
close() {
|
|
1500
|
+
return this.write(null);
|
|
1501
|
+
}
|
|
1502
|
+
flush() {
|
|
1503
|
+
this.flushBuffers();
|
|
1504
|
+
}
|
|
1505
|
+
};
|
|
1506
|
+
}
|
|
1507
|
+
});
|
|
1508
|
+
|
|
1509
|
+
// ../documents/src/lib/parsers/parse-xml-internal.ts
|
|
1510
|
+
function setXMLField(object, name, value) {
|
|
1511
|
+
Object.defineProperty(object, name, {
|
|
1512
|
+
value,
|
|
1513
|
+
enumerable: true,
|
|
1514
|
+
configurable: true,
|
|
1515
|
+
writable: true
|
|
1516
|
+
});
|
|
1517
|
+
}
|
|
1518
|
+
function parseXMLInternal(text, options = {}) {
|
|
1519
|
+
if (options.strict !== false && /[^\u0009\u000a\u000d\u0020-\ud7ff\ue000-\ufffd\u{10000}-\u{10ffff}]/u.test(text)) {
|
|
1520
|
+
throw new Error("XML: invalid XML character");
|
|
1521
|
+
}
|
|
1522
|
+
const maximumDepth = options.maxDepth ?? 256;
|
|
1523
|
+
if (!Number.isSafeInteger(maximumDepth) || maximumDepth < 1 || maximumDepth > 4096) {
|
|
1524
|
+
throw new Error("XML: maxDepth must be an integer between 1 and 4096");
|
|
1525
|
+
}
|
|
1526
|
+
const state = new InternalXMLParserState(options, maximumDepth);
|
|
1527
|
+
const parser = new SAXParser({
|
|
1528
|
+
strict: options.strict !== false,
|
|
1529
|
+
strictEntities: !options.htmlEntities,
|
|
1530
|
+
preserveCase: true,
|
|
1531
|
+
caseInsensitiveTags: options.strict === false,
|
|
1532
|
+
processEntities: options.processEntities !== false,
|
|
1533
|
+
onopentag: (tag) => state.openElement(tag.name, tag.attributes || {}),
|
|
1534
|
+
ontext: (value) => state.addText(value),
|
|
1535
|
+
oncdata: (value) => state.addText(value),
|
|
1536
|
+
onscript: (value) => state.addText(value),
|
|
1537
|
+
onclosetag: () => state.closeElement(),
|
|
1538
|
+
onerror: (error) => {
|
|
1539
|
+
throw error;
|
|
1540
|
+
}
|
|
1541
|
+
});
|
|
1542
|
+
parser.write(text).close();
|
|
1543
|
+
if (options.strict === false) {
|
|
1544
|
+
while (state.stack.length) state.closeElement();
|
|
1545
|
+
}
|
|
1546
|
+
if (state.result === void 0) throw new Error("XML: document requires a root element");
|
|
1547
|
+
return state.result;
|
|
1548
|
+
}
|
|
1549
|
+
var InternalXMLParserState;
|
|
1550
|
+
var init_parse_xml_internal = __esm({
|
|
1551
|
+
"../documents/src/lib/parsers/parse-xml-internal.ts"() {
|
|
1552
|
+
"use strict";
|
|
1553
|
+
init_sax();
|
|
1554
|
+
InternalXMLParserState = class {
|
|
1555
|
+
/** Initializes conversion options and the element nesting budget. */
|
|
1556
|
+
constructor(options, maximumDepth) {
|
|
1557
|
+
this.options = options;
|
|
1558
|
+
this.maximumDepth = maximumDepth;
|
|
1559
|
+
}
|
|
1560
|
+
options;
|
|
1561
|
+
maximumDepth;
|
|
1562
|
+
/** Open element stack. */
|
|
1563
|
+
stack = [];
|
|
1564
|
+
/** Parsed root document. */
|
|
1565
|
+
result;
|
|
1566
|
+
/** Starts an element, rejecting multiple roots and excessive nesting. */
|
|
1567
|
+
openElement(name, attributes) {
|
|
1568
|
+
if (this.stack.length >= this.maximumDepth)
|
|
1569
|
+
throw new Error("XML: maximum nesting depth exceeded");
|
|
1570
|
+
if (!this.stack.length && this.result !== void 0 && this.options.strict !== false) {
|
|
1571
|
+
throw new Error("XML: document requires a single root element");
|
|
1572
|
+
}
|
|
1573
|
+
const elementName = this.normalizeName(name);
|
|
1574
|
+
const parent = this.stack[this.stack.length - 1];
|
|
1575
|
+
this.stack.push({
|
|
1576
|
+
name: elementName,
|
|
1577
|
+
path: parent ? `${parent.path}.${elementName}` : elementName,
|
|
1578
|
+
attributes: this.normalizeAttributes(attributes),
|
|
1579
|
+
children: {},
|
|
1580
|
+
text: ""
|
|
1581
|
+
});
|
|
1582
|
+
}
|
|
1583
|
+
/** Accumulates text without losing whitespace between text and CDATA chunks. */
|
|
1584
|
+
addText(text) {
|
|
1585
|
+
const current = this.stack[this.stack.length - 1];
|
|
1586
|
+
if (current) current.text += text;
|
|
1587
|
+
}
|
|
1588
|
+
/** Finishes an element and attaches it to its parent or the document root. */
|
|
1589
|
+
closeElement() {
|
|
1590
|
+
const node = this.stack.pop();
|
|
1591
|
+
if (!node) return;
|
|
1592
|
+
const value = this.finalizeElement(node);
|
|
1593
|
+
const parent = this.stack[this.stack.length - 1];
|
|
1594
|
+
if (parent) this.addChild(parent.children, node.name, node.path, value);
|
|
1595
|
+
else {
|
|
1596
|
+
if (this.result === void 0) this.result = {};
|
|
1597
|
+
this.addChild(this.result, node.name, node.path, value);
|
|
1598
|
+
}
|
|
1599
|
+
}
|
|
1600
|
+
/** Converts a completed element, keeping grouping and coercion independent. */
|
|
1601
|
+
finalizeElement(node) {
|
|
1602
|
+
const text = this.options.trimValues === false ? node.text : node.text.trim();
|
|
1603
|
+
const hasAttributes = Object.keys(node.attributes).length > 0;
|
|
1604
|
+
if (!hasAttributes && !Object.keys(node.children).length) {
|
|
1605
|
+
return this.parseValue(text, this.options.parseTagValue !== false);
|
|
1606
|
+
}
|
|
1607
|
+
const object = node.children;
|
|
1608
|
+
if (text)
|
|
1609
|
+
setXMLField(
|
|
1610
|
+
object,
|
|
1611
|
+
this.options.textNodeName ?? "value",
|
|
1612
|
+
this.parseValue(text, this.options.parseTagValue !== false)
|
|
1613
|
+
);
|
|
1614
|
+
if (hasAttributes && this.options.attributesGroupName) {
|
|
1615
|
+
setXMLField(object, this.options.attributesGroupName, node.attributes);
|
|
1616
|
+
} else {
|
|
1617
|
+
for (const [name, value] of Object.entries(node.attributes)) setXMLField(object, name, value);
|
|
1618
|
+
}
|
|
1619
|
+
return object;
|
|
1620
|
+
}
|
|
1621
|
+
/** Appends an element while preserving repeated and explicitly array-valued fields. */
|
|
1622
|
+
addChild(object, name, path, value) {
|
|
1623
|
+
var _a;
|
|
1624
|
+
if (!Object.hasOwn(object, name)) {
|
|
1625
|
+
setXMLField(object, name, ((_a = this.options.arrayPaths) == null ? void 0 : _a.includes(path)) ? [value] : value);
|
|
1626
|
+
} else {
|
|
1627
|
+
const existing = object[name];
|
|
1628
|
+
if (Array.isArray(existing)) existing.push(value);
|
|
1629
|
+
else setXMLField(object, name, [existing, value]);
|
|
1630
|
+
}
|
|
1631
|
+
}
|
|
1632
|
+
/** Applies namespace, prefix, grouping and scalar options to attributes. */
|
|
1633
|
+
normalizeAttributes(attributes) {
|
|
1634
|
+
const normalized = {};
|
|
1635
|
+
if (this.options.ignoreAttributes) return normalized;
|
|
1636
|
+
for (const [name, value] of Object.entries(attributes)) {
|
|
1637
|
+
if (this.options.removeNSPrefix && (name === "xmlns" || name.startsWith("xmlns:"))) continue;
|
|
1638
|
+
const text = this.options.trimValues === false ? value : value.trim();
|
|
1639
|
+
setXMLField(
|
|
1640
|
+
normalized,
|
|
1641
|
+
(this.options.attributeNamePrefix ?? "") + this.normalizeName(name),
|
|
1642
|
+
this.parseValue(text, this.options.parseAttributeValue === true)
|
|
1643
|
+
);
|
|
1644
|
+
}
|
|
1645
|
+
return normalized;
|
|
1646
|
+
}
|
|
1647
|
+
/** Removes a namespace prefix only when requested. */
|
|
1648
|
+
normalizeName(name) {
|
|
1649
|
+
return this.options.removeNSPrefix ? name.slice(name.indexOf(":") + 1) : name;
|
|
1650
|
+
}
|
|
1651
|
+
/** Coerces only finite numeric values and lowercase XML boolean literals. */
|
|
1652
|
+
parseValue(value, convert) {
|
|
1653
|
+
if (!convert) return value;
|
|
1654
|
+
if (value === "true" || value === "false") return value === "true";
|
|
1655
|
+
if (value !== "" && value.trim() !== "" && Number.isFinite(Number(value))) return Number(value);
|
|
1656
|
+
return value;
|
|
1657
|
+
}
|
|
1658
|
+
};
|
|
1659
|
+
}
|
|
1660
|
+
});
|
|
1661
|
+
|
|
1662
|
+
// ../documents/src/lib/parsers/parse-xml.ts
|
|
1663
|
+
function parseXMLSync(text, options = {}) {
|
|
1664
|
+
if (Object.hasOwn(options, "_fastXML") || Object.hasOwn(options, "_sax")) {
|
|
1665
|
+
throw new Error("XML: backend-specific options were removed; use native XML options");
|
|
1666
|
+
}
|
|
1667
|
+
if (options._parser && !["internal", "sax"].includes(options._parser)) {
|
|
1668
|
+
throw new Error("XML: unsupported parser backend; use the native documents parser");
|
|
1669
|
+
}
|
|
1670
|
+
const parsed = parseXMLInternal(text, options);
|
|
1671
|
+
return options.uncapitalizeKeys ? uncapitalizeKeys(parsed) : parsed;
|
|
1672
|
+
}
|
|
1673
|
+
function validateXML(text) {
|
|
1674
|
+
try {
|
|
1675
|
+
parseXMLInternal(text, {});
|
|
1676
|
+
return true;
|
|
1677
|
+
} catch (error) {
|
|
1678
|
+
return error instanceof Error ? error : new Error(String(error));
|
|
1679
|
+
}
|
|
1680
|
+
}
|
|
1681
|
+
var init_parse_xml = __esm({
|
|
1682
|
+
"../documents/src/lib/parsers/parse-xml.ts"() {
|
|
1683
|
+
"use strict";
|
|
1684
|
+
init_uncapitalize();
|
|
1685
|
+
init_parse_xml_internal();
|
|
1686
|
+
}
|
|
1687
|
+
});
|
|
1688
|
+
|
|
1689
|
+
// ../documents/src/xml-parser.ts
|
|
1690
|
+
var init_xml_parser = __esm({
|
|
1691
|
+
"../documents/src/xml-parser.ts"() {
|
|
1692
|
+
"use strict";
|
|
1693
|
+
init_parse_xml();
|
|
1694
|
+
}
|
|
1695
|
+
});
|
|
1696
|
+
|
|
65
1697
|
// ../core/src/schema-utils/lib/schema/convert-arrow-schema.ts
|
|
66
1698
|
function convertArrowToSchema(arrowSchema) {
|
|
67
1699
|
return serializeArrowSchema(arrowSchema);
|
|
@@ -478,10 +2110,10 @@ var init_build_graph_tables = __esm({
|
|
|
478
2110
|
// src/lib/parse-gexf.ts
|
|
479
2111
|
function parseGEXF(input, shape = "arrow-table") {
|
|
480
2112
|
const text = typeof input === "string" ? input : new TextDecoder().decode(input);
|
|
481
|
-
const validation =
|
|
482
|
-
if (validation !== true) throw new Error(`Invalid GEXF XML: ${validation.
|
|
2113
|
+
const validation = validateXML(text);
|
|
2114
|
+
if (validation !== true) throw new Error(`Invalid GEXF XML: ${validation.message}`);
|
|
483
2115
|
if (/<!DOCTYPE\s/i.test(text)) throw new Error("GEXF document types are unsupported.");
|
|
484
|
-
const document = normalizeNamespaces(
|
|
2116
|
+
const document = normalizeNamespaces(parseXMLSync(text, XML_OPTIONS), {});
|
|
485
2117
|
const root = getElement(document.gexf, "gexf");
|
|
486
2118
|
const version = root["@_version"];
|
|
487
2119
|
if (version && !["1.2", "1.2draft", "1.3"].includes(version)) {
|
|
@@ -755,18 +2387,20 @@ function requireValue(element) {
|
|
|
755
2387
|
if (typeof value !== "string") throw new Error("GEXF attvalue requires a value attribute.");
|
|
756
2388
|
return value;
|
|
757
2389
|
}
|
|
758
|
-
var
|
|
2390
|
+
var XML_OPTIONS, GEXF_NAMESPACE, TEMPORAL_ATTRIBUTES;
|
|
759
2391
|
var init_parse_gexf = __esm({
|
|
760
2392
|
"src/lib/parse-gexf.ts"() {
|
|
761
2393
|
"use strict";
|
|
762
|
-
|
|
2394
|
+
init_xml_parser();
|
|
763
2395
|
init_build_graph_tables();
|
|
764
|
-
|
|
2396
|
+
XML_OPTIONS = {
|
|
765
2397
|
ignoreAttributes: false,
|
|
766
2398
|
parseAttributeValue: false,
|
|
767
2399
|
parseTagValue: false,
|
|
768
|
-
trimValues: false
|
|
769
|
-
|
|
2400
|
+
trimValues: false,
|
|
2401
|
+
attributeNamePrefix: "@_",
|
|
2402
|
+
textNodeName: "#text"
|
|
2403
|
+
};
|
|
770
2404
|
GEXF_NAMESPACE = /^https?:\/\/(?:www\.)?gexf.net\/1\.(?:2(?:draft)?|3)(\/viz)?$/;
|
|
771
2405
|
TEMPORAL_ATTRIBUTES = [
|
|
772
2406
|
"start",
|
|
@@ -817,7 +2451,7 @@ __export(gexf_loader_exports2, {
|
|
|
817
2451
|
module.exports = __toCommonJS(gexf_loader_exports2);
|
|
818
2452
|
|
|
819
2453
|
// dist/gexf-loader-types.js
|
|
820
|
-
var VERSION2 = true ? "5.0.0-beta.
|
|
2454
|
+
var VERSION2 = true ? "5.0.0-beta.2" : "latest";
|
|
821
2455
|
async function preload2() {
|
|
822
2456
|
const { GEXFLoaderWithParser: GEXFLoaderWithParser3 } = await Promise.resolve().then(() => (init_gexf_loader(), gexf_loader_exports));
|
|
823
2457
|
return GEXFLoaderWithParser3;
|
|
@@ -838,7 +2472,7 @@ var GEXFLoader2 = {
|
|
|
838
2472
|
};
|
|
839
2473
|
|
|
840
2474
|
// dist/lib/parse-gexf.js
|
|
841
|
-
|
|
2475
|
+
init_xml_parser();
|
|
842
2476
|
|
|
843
2477
|
// dist/lib/build-graph-tables.js
|
|
844
2478
|
var import_apache_arrow2 = require("apache-arrow");
|
|
@@ -960,12 +2594,14 @@ function normalizeValue2(value, type) {
|
|
|
960
2594
|
}
|
|
961
2595
|
|
|
962
2596
|
// dist/lib/parse-gexf.js
|
|
963
|
-
var
|
|
2597
|
+
var XML_OPTIONS2 = {
|
|
964
2598
|
ignoreAttributes: false,
|
|
965
2599
|
parseAttributeValue: false,
|
|
966
2600
|
parseTagValue: false,
|
|
967
|
-
trimValues: false
|
|
968
|
-
|
|
2601
|
+
trimValues: false,
|
|
2602
|
+
attributeNamePrefix: "@_",
|
|
2603
|
+
textNodeName: "#text"
|
|
2604
|
+
};
|
|
969
2605
|
var GEXF_NAMESPACE2 = /^https?:\/\/(?:www\.)?gexf.net\/1\.(?:2(?:draft)?|3)(\/viz)?$/;
|
|
970
2606
|
var TEMPORAL_ATTRIBUTES2 = [
|
|
971
2607
|
"start",
|
|
@@ -978,12 +2614,12 @@ var TEMPORAL_ATTRIBUTES2 = [
|
|
|
978
2614
|
];
|
|
979
2615
|
function parseGEXF2(input, shape = "arrow-table") {
|
|
980
2616
|
const text = typeof input === "string" ? input : new TextDecoder().decode(input);
|
|
981
|
-
const validation =
|
|
2617
|
+
const validation = validateXML(text);
|
|
982
2618
|
if (validation !== true)
|
|
983
|
-
throw new Error(`Invalid GEXF XML: ${validation.
|
|
2619
|
+
throw new Error(`Invalid GEXF XML: ${validation.message}`);
|
|
984
2620
|
if (/<!DOCTYPE\s/i.test(text))
|
|
985
2621
|
throw new Error("GEXF document types are unsupported.");
|
|
986
|
-
const document = normalizeNamespaces2(
|
|
2622
|
+
const document = normalizeNamespaces2(parseXMLSync(text, XML_OPTIONS2), {});
|
|
987
2623
|
const root = getElement2(document.gexf, "gexf");
|
|
988
2624
|
const version = root["@_version"];
|
|
989
2625
|
if (version && !["1.2", "1.2draft", "1.3"].includes(version)) {
|