@loaders.gl/graphs 5.0.0-beta.1 → 5.0.0-beta.10

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.
Files changed (61) hide show
  1. package/README.md +4 -2
  2. package/THIRD-PARTY-NOTICES.md +1 -1
  3. package/dist/dot-loader-types.cjs +60 -0
  4. package/dist/dot-loader-types.cjs.map +7 -0
  5. package/dist/dot-loader-types.d.ts +4 -2
  6. package/dist/dot-loader-types.d.ts.map +1 -1
  7. package/dist/dot-loader-types.js +3 -2
  8. package/dist/dot-loader-types.js.map +1 -1
  9. package/dist/dot-loader.cjs +15 -295
  10. package/dist/dot-loader.cjs.map +2 -2
  11. package/dist/dot-loader.d.ts +2 -1
  12. package/dist/dot-loader.d.ts.map +1 -1
  13. package/dist/dot-loader.js +3 -3
  14. package/dist/dot-loader.js.map +1 -1
  15. package/dist/gexf-loader-types.cjs +60 -0
  16. package/dist/gexf-loader-types.cjs.map +7 -0
  17. package/dist/gexf-loader-types.d.ts +4 -2
  18. package/dist/gexf-loader-types.d.ts.map +1 -1
  19. package/dist/gexf-loader-types.js +3 -2
  20. package/dist/gexf-loader-types.js.map +1 -1
  21. package/dist/gexf-loader.cjs +1652 -287
  22. package/dist/gexf-loader.cjs.map +3 -3
  23. package/dist/gexf-loader.d.ts +2 -1
  24. package/dist/gexf-loader.d.ts.map +1 -1
  25. package/dist/gexf-loader.js +3 -3
  26. package/dist/gexf-loader.js.map +1 -1
  27. package/dist/graph-types.d.ts +3 -2
  28. package/dist/graph-types.d.ts.map +1 -1
  29. package/dist/graphml-loader-types.cjs +60 -0
  30. package/dist/graphml-loader-types.cjs.map +7 -0
  31. package/dist/graphml-loader-types.d.ts +4 -2
  32. package/dist/graphml-loader-types.d.ts.map +1 -1
  33. package/dist/graphml-loader-types.js +3 -2
  34. package/dist/graphml-loader-types.js.map +1 -1
  35. package/dist/graphml-loader.cjs +1633 -280
  36. package/dist/graphml-loader.cjs.map +3 -3
  37. package/dist/graphml-loader.d.ts +2 -1
  38. package/dist/graphml-loader.d.ts.map +1 -1
  39. package/dist/graphml-loader.js +3 -3
  40. package/dist/graphml-loader.js.map +1 -1
  41. package/dist/index.cjs +36 -1930
  42. package/dist/index.cjs.map +4 -3
  43. package/dist/lib/build-graph-tables.d.ts.map +1 -1
  44. package/dist/lib/build-graph-tables.js +2 -4
  45. package/dist/lib/build-graph-tables.js.map +1 -1
  46. package/dist/lib/parse-gexf.d.ts.map +1 -1
  47. package/dist/lib/parse-gexf.js +10 -8
  48. package/dist/lib/parse-gexf.js.map +1 -1
  49. package/dist/lib/parse-graphml.js +4 -4
  50. package/dist/lib/parse-graphml.js.map +1 -1
  51. package/package.json +9 -7
  52. package/src/dot-loader-types.ts +2 -1
  53. package/src/dot-loader.ts +3 -3
  54. package/src/gexf-loader-types.ts +2 -1
  55. package/src/gexf-loader.ts +3 -3
  56. package/src/graph-types.ts +3 -2
  57. package/src/graphml-loader-types.ts +2 -1
  58. package/src/graphml-loader.ts +3 -3
  59. package/src/lib/build-graph-tables.ts +5 -7
  60. package/src/lib/parse-gexf.ts +10 -8
  61. package/src/lib/parse-graphml.ts +5 -5
@@ -1,9 +1,7 @@
1
1
  "use strict";
2
- var __create = Object.create;
3
2
  var __defProp = Object.defineProperty;
4
3
  var __getOwnPropDesc = Object.getOwnPropertyDescriptor;
5
4
  var __getOwnPropNames = Object.getOwnPropertyNames;
6
- var __getProtoOf = Object.getPrototypeOf;
7
5
  var __hasOwnProp = Object.prototype.hasOwnProperty;
8
6
  var __esm = (fn, res, err) => function __init() {
9
7
  if (err) throw err[0];
@@ -25,14 +23,6 @@ var __copyProps = (to, from, except, desc) => {
25
23
  }
26
24
  return to;
27
25
  };
28
- var __toESM = (mod, isNodeMode, target) => (target = mod != null ? __create(__getProtoOf(mod)) : {}, __copyProps(
29
- // If the importer is in node compatibility mode or this is not an ESM
30
- // file that has been converted to a CommonJS file using a Babel-
31
- // compatible transform (i.e. "__esModule" has not been set), then set
32
- // "default" to the CommonJS "module.exports" for node compatibility.
33
- isNodeMode || !mod || !mod.__esModule ? __defProp(target, "default", { value: mod, enumerable: true }) : target,
34
- mod
35
- ));
36
26
  var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: true }), mod);
37
27
 
38
28
  // src/gexf-loader-types.ts
@@ -54,279 +44,1653 @@ var init_gexf_loader_types = __esm({
54
44
  mimeTypes: ["application/gexf+xml"],
55
45
  text: true,
56
46
  worker: false,
47
+ supportsNativeArrow: true,
57
48
  dataType: null,
58
49
  batchType: null,
59
- options: { gexf: { shape: "arrow-table" } },
50
+ options: { gexf: { shape: "arrow" } },
60
51
  preload
61
52
  };
62
53
  }
63
54
  });
64
55
 
65
- // ../core/src/schema-utils/lib/schema/convert-arrow-schema.ts
66
- function convertArrowToSchema(arrowSchema) {
67
- return serializeArrowSchema(arrowSchema);
56
+ // ../documents/src/lib/xml-utils/uncapitalize.ts
57
+ function uncapitalize(str) {
58
+ return typeof str === "string" ? str.charAt(0).toLowerCase() + str.slice(1) : str;
68
59
  }
69
- function serializeArrowSchema(arrowSchema) {
70
- return {
71
- fields: arrowSchema.fields.map((arrowField) => serializeArrowField(arrowField)),
72
- metadata: serializeArrowMetadata(arrowSchema.metadata)
73
- };
74
- }
75
- function serializeArrowMetadata(arrowMetadata) {
76
- return Object.fromEntries(arrowMetadata);
77
- }
78
- function serializeArrowField(field) {
79
- return {
80
- name: field.name,
81
- type: serializeArrowType(field.type),
82
- nullable: field.nullable,
83
- metadata: serializeArrowMetadata(field.metadata)
84
- };
85
- }
86
- function serializeArrowType(arrowType) {
87
- if (isArrowViewType(arrowType, "BinaryView")) {
88
- return "binary-view";
60
+ function uncapitalizeKeys(object) {
61
+ if (Array.isArray(object)) {
62
+ return object.map((element) => uncapitalizeKeys(element));
63
+ }
64
+ if (object && typeof object === "object") {
65
+ const newObject = {};
66
+ for (const [key, value] of Object.entries(object)) {
67
+ Object.defineProperty(newObject, uncapitalize(key), {
68
+ value: uncapitalizeKeys(value),
69
+ enumerable: true,
70
+ configurable: true,
71
+ writable: true
72
+ });
73
+ }
74
+ return newObject;
89
75
  }
90
- if (isArrowViewType(arrowType, "Utf8View")) {
91
- return "utf8-view";
76
+ return object;
77
+ }
78
+ var init_uncapitalize = __esm({
79
+ "../documents/src/lib/xml-utils/uncapitalize.ts"() {
80
+ "use strict";
92
81
  }
93
- if (isOptionalArrowType(arrowType, "LargeList")) {
94
- const largeListType = arrowType;
95
- return {
96
- type: "large-list",
97
- children: [serializeArrowField(largeListType.valueField)]
82
+ });
83
+
84
+ // ../documents/src/sax-ts/sax.ts
85
+ var DEFAULT_SAX_EVENTS, DEFAULT_SAX_PARSER_OPTIONS, EVENTS, BUFFERS, nameStart, nameBody, entityStart, entityBody, ENTITIES, SAX, SAXParser;
86
+ var init_sax = __esm({
87
+ "../documents/src/sax-ts/sax.ts"() {
88
+ "use strict";
89
+ DEFAULT_SAX_EVENTS = {
90
+ ontext: () => {
91
+ },
92
+ onprocessinginstruction: () => {
93
+ },
94
+ onsgmldeclaration: () => {
95
+ },
96
+ ondoctype: () => {
97
+ },
98
+ oncomment: () => {
99
+ },
100
+ onopentagstart: () => {
101
+ },
102
+ onattribute: () => {
103
+ },
104
+ onopentag: () => {
105
+ },
106
+ onclosetag: () => {
107
+ },
108
+ onopencdata: () => {
109
+ },
110
+ oncdata: () => {
111
+ },
112
+ onclosecdata: () => {
113
+ },
114
+ onerror: () => {
115
+ },
116
+ onend: () => {
117
+ },
118
+ onready: () => {
119
+ },
120
+ onscript: () => {
121
+ },
122
+ onopennamespace: () => {
123
+ },
124
+ onclosenamespace: () => {
125
+ }
98
126
  };
99
- }
100
- switch (arrowType.constructor) {
101
- case arrow.Null:
102
- return "null";
103
- case arrow.Binary:
104
- return "binary";
105
- case arrow.FixedSizeBinary:
106
- return {
107
- type: "fixed-size-binary",
108
- byteWidth: arrowType.byteWidth
127
+ DEFAULT_SAX_PARSER_OPTIONS = {
128
+ ...DEFAULT_SAX_EVENTS,
129
+ strict: false,
130
+ MAX_BUFFER_LENGTH: 64 * 1024,
131
+ lowercase: false,
132
+ lowercasetags: false,
133
+ noscript: false,
134
+ strictEntities: false,
135
+ preserveCase: false,
136
+ caseInsensitiveTags: false,
137
+ processEntities: true,
138
+ xmlns: void 0,
139
+ position: void 0,
140
+ trim: void 0,
141
+ normalize: void 0
142
+ };
143
+ EVENTS = [
144
+ "text",
145
+ "processinginstruction",
146
+ "sgmldeclaration",
147
+ "doctype",
148
+ "comment",
149
+ "opentagstart",
150
+ "attribute",
151
+ "opentag",
152
+ "closetag",
153
+ "opencdata",
154
+ "cdata",
155
+ "closecdata",
156
+ "error",
157
+ "end",
158
+ "ready",
159
+ "script",
160
+ "opennamespace",
161
+ "closenamespace"
162
+ ];
163
+ BUFFERS = [
164
+ "comment",
165
+ "sgmlDecl",
166
+ "textNode",
167
+ "tagName",
168
+ "doctype",
169
+ "procInstName",
170
+ "procInstBody",
171
+ "entity",
172
+ "attribName",
173
+ "attribValue",
174
+ "cdata",
175
+ "script"
176
+ ];
177
+ 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]/;
178
+ 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-]/;
179
+ 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]/;
180
+ 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-]/;
181
+ ENTITIES = {
182
+ amp: "&",
183
+ gt: ">",
184
+ lt: "<",
185
+ quot: '"',
186
+ apos: "'",
187
+ AElig: 198,
188
+ Aacute: 193,
189
+ Acirc: 194,
190
+ Agrave: 192,
191
+ Aring: 197,
192
+ Atilde: 195,
193
+ Auml: 196,
194
+ Ccedil: 199,
195
+ ETH: 208,
196
+ Eacute: 201,
197
+ Ecirc: 202,
198
+ Egrave: 200,
199
+ Euml: 203,
200
+ Iacute: 205,
201
+ Icirc: 206,
202
+ Igrave: 204,
203
+ Iuml: 207,
204
+ Ntilde: 209,
205
+ Oacute: 211,
206
+ Ocirc: 212,
207
+ Ograve: 210,
208
+ Oslash: 216,
209
+ Otilde: 213,
210
+ Ouml: 214,
211
+ THORN: 222,
212
+ Uacute: 218,
213
+ Ucirc: 219,
214
+ Ugrave: 217,
215
+ Uuml: 220,
216
+ Yacute: 221,
217
+ aacute: 225,
218
+ acirc: 226,
219
+ aelig: 230,
220
+ agrave: 224,
221
+ aring: 229,
222
+ atilde: 227,
223
+ auml: 228,
224
+ ccedil: 231,
225
+ eacute: 233,
226
+ ecirc: 234,
227
+ egrave: 232,
228
+ eth: 240,
229
+ euml: 235,
230
+ iacute: 237,
231
+ icirc: 238,
232
+ igrave: 236,
233
+ iuml: 239,
234
+ ntilde: 241,
235
+ oacute: 243,
236
+ ocirc: 244,
237
+ ograve: 242,
238
+ oslash: 248,
239
+ otilde: 245,
240
+ ouml: 246,
241
+ szlig: 223,
242
+ thorn: 254,
243
+ uacute: 250,
244
+ ucirc: 251,
245
+ ugrave: 249,
246
+ uuml: 252,
247
+ yacute: 253,
248
+ yuml: 255,
249
+ copy: 169,
250
+ reg: 174,
251
+ nbsp: 160,
252
+ iexcl: 161,
253
+ cent: 162,
254
+ pound: 163,
255
+ curren: 164,
256
+ yen: 165,
257
+ brvbar: 166,
258
+ sect: 167,
259
+ uml: 168,
260
+ ordf: 170,
261
+ laquo: 171,
262
+ not: 172,
263
+ shy: 173,
264
+ macr: 175,
265
+ deg: 176,
266
+ plusmn: 177,
267
+ sup1: 185,
268
+ sup2: 178,
269
+ sup3: 179,
270
+ acute: 180,
271
+ micro: 181,
272
+ para: 182,
273
+ middot: 183,
274
+ cedil: 184,
275
+ ordm: 186,
276
+ raquo: 187,
277
+ frac14: 188,
278
+ frac12: 189,
279
+ frac34: 190,
280
+ iquest: 191,
281
+ times: 215,
282
+ divide: 247,
283
+ OElig: 338,
284
+ oelig: 339,
285
+ Scaron: 352,
286
+ scaron: 353,
287
+ Yuml: 376,
288
+ fnof: 402,
289
+ circ: 710,
290
+ tilde: 732,
291
+ Alpha: 913,
292
+ Beta: 914,
293
+ Gamma: 915,
294
+ Delta: 916,
295
+ Epsilon: 917,
296
+ Zeta: 918,
297
+ Eta: 919,
298
+ Theta: 920,
299
+ Iota: 921,
300
+ Kappa: 922,
301
+ Lambda: 923,
302
+ Mu: 924,
303
+ Nu: 925,
304
+ Xi: 926,
305
+ Omicron: 927,
306
+ Pi: 928,
307
+ Rho: 929,
308
+ Sigma: 931,
309
+ Tau: 932,
310
+ Upsilon: 933,
311
+ Phi: 934,
312
+ Chi: 935,
313
+ Psi: 936,
314
+ Omega: 937,
315
+ alpha: 945,
316
+ beta: 946,
317
+ gamma: 947,
318
+ delta: 948,
319
+ epsilon: 949,
320
+ zeta: 950,
321
+ eta: 951,
322
+ theta: 952,
323
+ iota: 953,
324
+ kappa: 954,
325
+ lambda: 955,
326
+ mu: 956,
327
+ nu: 957,
328
+ xi: 958,
329
+ omicron: 959,
330
+ pi: 960,
331
+ rho: 961,
332
+ sigmaf: 962,
333
+ sigma: 963,
334
+ tau: 964,
335
+ upsilon: 965,
336
+ phi: 966,
337
+ chi: 967,
338
+ psi: 968,
339
+ omega: 969,
340
+ thetasym: 977,
341
+ upsih: 978,
342
+ piv: 982,
343
+ ensp: 8194,
344
+ emsp: 8195,
345
+ thinsp: 8201,
346
+ zwnj: 8204,
347
+ zwj: 8205,
348
+ lrm: 8206,
349
+ rlm: 8207,
350
+ ndash: 8211,
351
+ mdash: 8212,
352
+ lsquo: 8216,
353
+ rsquo: 8217,
354
+ sbquo: 8218,
355
+ ldquo: 8220,
356
+ rdquo: 8221,
357
+ bdquo: 8222,
358
+ dagger: 8224,
359
+ Dagger: 8225,
360
+ bull: 8226,
361
+ hellip: 8230,
362
+ permil: 8240,
363
+ prime: 8242,
364
+ Prime: 8243,
365
+ lsaquo: 8249,
366
+ rsaquo: 8250,
367
+ oline: 8254,
368
+ frasl: 8260,
369
+ euro: 8364,
370
+ image: 8465,
371
+ weierp: 8472,
372
+ real: 8476,
373
+ trade: 8482,
374
+ alefsym: 8501,
375
+ larr: 8592,
376
+ uarr: 8593,
377
+ rarr: 8594,
378
+ darr: 8595,
379
+ harr: 8596,
380
+ crarr: 8629,
381
+ lArr: 8656,
382
+ uArr: 8657,
383
+ rArr: 8658,
384
+ dArr: 8659,
385
+ hArr: 8660,
386
+ forall: 8704,
387
+ part: 8706,
388
+ exist: 8707,
389
+ empty: 8709,
390
+ nabla: 8711,
391
+ isin: 8712,
392
+ notin: 8713,
393
+ ni: 8715,
394
+ prod: 8719,
395
+ sum: 8721,
396
+ minus: 8722,
397
+ lowast: 8727,
398
+ radic: 8730,
399
+ prop: 8733,
400
+ infin: 8734,
401
+ ang: 8736,
402
+ and: 8743,
403
+ or: 8744,
404
+ cap: 8745,
405
+ cup: 8746,
406
+ int: 8747,
407
+ there4: 8756,
408
+ sim: 8764,
409
+ cong: 8773,
410
+ asymp: 8776,
411
+ ne: 8800,
412
+ equiv: 8801,
413
+ le: 8804,
414
+ ge: 8805,
415
+ sub: 8834,
416
+ sup: 8835,
417
+ nsub: 8836,
418
+ sube: 8838,
419
+ supe: 8839,
420
+ oplus: 8853,
421
+ otimes: 8855,
422
+ perp: 8869,
423
+ sdot: 8901,
424
+ lceil: 8968,
425
+ rceil: 8969,
426
+ lfloor: 8970,
427
+ rfloor: 8971,
428
+ lang: 9001,
429
+ rang: 9002,
430
+ loz: 9674,
431
+ spades: 9824,
432
+ clubs: 9827,
433
+ hearts: 9829,
434
+ diams: 9830
435
+ };
436
+ Object.keys(ENTITIES).forEach((key) => {
437
+ const e = ENTITIES[key];
438
+ ENTITIES[key] = typeof e === "number" ? String.fromCharCode(e) : e;
439
+ });
440
+ SAX = class _SAX {
441
+ EVENTS = EVENTS;
442
+ ENTITIES = {
443
+ // TODO: make it readonly, needed for entity-mega test
444
+ // amp, gt, lt, quot and apos are resolved to strings instead of numerical
445
+ // codes, IDK why
446
+ ...ENTITIES
109
447
  };
110
- case arrow.Bool:
111
- return "bool";
112
- case arrow.Int:
113
- const intType = arrowType;
114
- return `${intType.isSigned ? "" : "u"}int${intType.bitWidth}`;
115
- case arrow.Int8:
116
- return "int8";
117
- case arrow.Int16:
118
- return "int16";
119
- case arrow.Int32:
120
- return "int32";
121
- case arrow.Int64:
122
- return "int64";
123
- case arrow.Uint8:
124
- return "uint8";
125
- case arrow.Uint16:
126
- return "uint16";
127
- case arrow.Uint32:
128
- return "uint32";
129
- case arrow.Uint64:
130
- return "uint64";
131
- case arrow.Float:
132
- const precision = arrowType.precision;
133
- switch (precision) {
134
- case arrow.Precision.HALF:
135
- return "float16";
136
- case arrow.Precision.SINGLE:
137
- return "float32";
138
- case arrow.Precision.DOUBLE:
139
- return "float64";
140
- default:
141
- return "float16";
142
- }
143
- case arrow.Float16:
144
- return "float16";
145
- case arrow.Float32:
146
- return "float32";
147
- case arrow.Float64:
148
- return "float64";
149
- case arrow.Utf8:
150
- return "utf8";
151
- case arrow.Decimal:
152
- const decimal = arrowType;
153
- return {
154
- type: "decimal",
155
- bitWidth: decimal.bitWidth,
156
- precision: decimal.precision,
157
- scale: decimal.scale
448
+ XML_ENTITIES = {
449
+ amp: "&",
450
+ gt: ">",
451
+ lt: "<",
452
+ quot: '"',
453
+ apos: "'"
158
454
  };
159
- case arrow.Date_:
160
- const dateUnit = arrowType.unit;
161
- return dateUnit === arrow.DateUnit.DAY ? "date-day" : "date-millisecond";
162
- case arrow.DateDay:
163
- return "date-day";
164
- case arrow.DateMillisecond:
165
- return "date-millisecond";
166
- case arrow.Time:
167
- const timeUnit = arrowType.unit;
168
- switch (timeUnit) {
169
- case arrow.TimeUnit.SECOND:
170
- return "time-second";
171
- case arrow.TimeUnit.MILLISECOND:
172
- return "time-millisecond";
173
- case arrow.TimeUnit.MICROSECOND:
174
- return "time-microsecond";
175
- case arrow.TimeUnit.NANOSECOND:
176
- return "time-nanosecond";
177
- default:
178
- return "time-second";
455
+ S = 0;
456
+ opt;
457
+ trackPosition = false;
458
+ column = 0;
459
+ line = 0;
460
+ c = "";
461
+ error;
462
+ q = "";
463
+ bufferCheckPosition;
464
+ closed = false;
465
+ tags = [];
466
+ looseCase = "";
467
+ closedRoot = false;
468
+ sawRoot = false;
469
+ strict = false;
470
+ tag;
471
+ strictEntities;
472
+ state;
473
+ noscript = false;
474
+ attribList = [];
475
+ ns;
476
+ position = 0;
477
+ STATE = {
478
+ BEGIN: this.S++,
479
+ // leading byte order mark or whitespace
480
+ BEGIN_WHITESPACE: this.S++,
481
+ // leading whitespace
482
+ TEXT: this.S++,
483
+ // general stuff
484
+ TEXT_ENTITY: this.S++,
485
+ // &amp and such.
486
+ OPEN_WAKA: this.S++,
487
+ // <
488
+ SGML_DECL: this.S++,
489
+ // <!BLARG
490
+ SGML_DECL_QUOTED: this.S++,
491
+ // <!BLARG foo "bar
492
+ DOCTYPE: this.S++,
493
+ // <!DOCTYPE
494
+ DOCTYPE_QUOTED: this.S++,
495
+ // <!DOCTYPE "//blah
496
+ DOCTYPE_DTD: this.S++,
497
+ // <!DOCTYPE "//blah" [ ...
498
+ DOCTYPE_DTD_QUOTED: this.S++,
499
+ // <!DOCTYPE "//blah" [ "foo
500
+ COMMENT_STARTING: this.S++,
501
+ // <!-
502
+ COMMENT: this.S++,
503
+ // <!--
504
+ COMMENT_ENDING: this.S++,
505
+ // <!-- blah -
506
+ COMMENT_ENDED: this.S++,
507
+ // <!-- blah --
508
+ CDATA: this.S++,
509
+ // <![CDATA[ something
510
+ CDATA_ENDING: this.S++,
511
+ // ]
512
+ CDATA_ENDING_2: this.S++,
513
+ // ]]
514
+ PROC_INST: this.S++,
515
+ // <?hi
516
+ PROC_INST_BODY: this.S++,
517
+ // <?hi there
518
+ PROC_INST_ENDING: this.S++,
519
+ // <?hi "there" ?
520
+ OPEN_TAG: this.S++,
521
+ // <strong
522
+ OPEN_TAG_SLASH: this.S++,
523
+ // <strong /
524
+ ATTRIB: this.S++,
525
+ // <a
526
+ ATTRIB_NAME: this.S++,
527
+ // <a foo
528
+ ATTRIB_NAME_SAW_WHITE: this.S++,
529
+ // <a foo _
530
+ ATTRIB_VALUE: this.S++,
531
+ // <a foo=
532
+ ATTRIB_VALUE_QUOTED: this.S++,
533
+ // <a foo="bar
534
+ ATTRIB_VALUE_CLOSED: this.S++,
535
+ // <a foo="bar"
536
+ ATTRIB_VALUE_UNQUOTED: this.S++,
537
+ // <a foo=bar
538
+ ATTRIB_VALUE_ENTITY_Q: this.S++,
539
+ // <foo bar="&quot;"
540
+ ATTRIB_VALUE_ENTITY_U: this.S++,
541
+ // <foo bar=&quot
542
+ CLOSE_TAG: this.S++,
543
+ // </a
544
+ CLOSE_TAG_SAW_WHITE: this.S++,
545
+ // </a >
546
+ SCRIPT: this.S++,
547
+ // <script> ...
548
+ SCRIPT_ENDING: this.S++
549
+ // <script> ... <
550
+ };
551
+ BUFFERS = BUFFERS;
552
+ // private parser: (strict: boolean, opt: any) => SAXParser;
553
+ CDATA = "[CDATA[";
554
+ DOCTYPE = "DOCTYPE";
555
+ XML_NAMESPACE = "http://www.w3.org/XML/1998/namespace";
556
+ XMLNS_NAMESPACE = "http://www.w3.org/2000/xmlns/";
557
+ rootNS = {
558
+ xml: this.XML_NAMESPACE,
559
+ xmlns: this.XMLNS_NAMESPACE
560
+ };
561
+ comment;
562
+ sgmlDecl;
563
+ textNode = "";
564
+ tagName;
565
+ doctype;
566
+ procInstName;
567
+ procInstBody;
568
+ entity = "";
569
+ attribName;
570
+ attribValue;
571
+ cdata = "";
572
+ script = "";
573
+ startTagPosition = 0;
574
+ constructor() {
575
+ this.S = 0;
576
+ for (const s in this.STATE) {
577
+ if (this.STATE.hasOwnProperty(s)) {
578
+ this.STATE[this.STATE[s]] = s;
579
+ }
580
+ }
581
+ this.S = this.STATE;
179
582
  }
180
- case arrow.TimeMillisecond:
181
- return "time-millisecond";
182
- case arrow.TimeSecond:
183
- return "time-second";
184
- case arrow.TimeMicrosecond:
185
- return "time-microsecond";
186
- case arrow.TimeNanosecond:
187
- return "time-nanosecond";
188
- case arrow.Timestamp:
189
- const timeStampUnit = arrowType.unit;
190
- switch (timeStampUnit) {
191
- case arrow.TimeUnit.SECOND:
192
- return "timestamp-second";
193
- case arrow.TimeUnit.MILLISECOND:
194
- return "timestamp-millisecond";
195
- case arrow.TimeUnit.MICROSECOND:
196
- return "timestamp-microsecond";
197
- case arrow.TimeUnit.NANOSECOND:
198
- return "timestamp-nanosecond";
199
- default:
200
- return "timestamp-second";
583
+ static charAt(chunk, i) {
584
+ let result = "";
585
+ if (i < chunk.length) {
586
+ result = chunk.charAt(i);
587
+ }
588
+ return result;
201
589
  }
202
- case arrow.TimestampSecond:
203
- return "timestamp-second";
204
- case arrow.TimestampMillisecond:
205
- return "timestamp-millisecond";
206
- case arrow.TimestampMicrosecond:
207
- return "timestamp-microsecond";
208
- case arrow.TimestampNanosecond:
209
- return "timestamp-nanosecond";
210
- case arrow.Interval:
211
- const intervalUnit = arrowType.unit;
212
- switch (intervalUnit) {
213
- case arrow.IntervalUnit.DAY_TIME:
214
- return "interval-daytime";
215
- case arrow.IntervalUnit.YEAR_MONTH:
216
- return "interval-yearmonth";
217
- default:
218
- return "interval-daytime";
590
+ static isWhitespace(c) {
591
+ return c === " " || c === "\n" || c === "\r" || c === " ";
219
592
  }
220
- case arrow.IntervalDayTime:
221
- return "interval-daytime";
222
- case arrow.IntervalYearMonth:
223
- return "interval-yearmonth";
224
- case arrow.Map_:
225
- const mapType = arrowType;
226
- return {
227
- type: "map",
228
- keysSorted: mapType.keysSorted,
229
- children: mapType.children.map((arrowField) => serializeArrowField(arrowField))
230
- };
231
- case arrow.List:
232
- const listType = arrowType;
233
- const listField = listType.valueField;
234
- return {
235
- type: "list",
236
- children: [serializeArrowField(listField)]
237
- };
238
- case arrow.FixedSizeList:
239
- const fixedSizeList = arrowType;
240
- return {
241
- type: "fixed-size-list",
242
- listSize: fixedSizeList.listSize,
243
- children: [serializeArrowField(fixedSizeList.children[0])]
244
- };
245
- case arrow.Struct:
246
- const structType = arrowType;
247
- return {
248
- type: "struct",
249
- children: structType.children.map((arrowField) => serializeArrowField(arrowField))
250
- };
251
- case arrow.SparseUnion:
252
- case arrow.DenseUnion:
253
- const unionType = arrowType;
254
- return {
255
- type: unionType instanceof arrow.DenseUnion ? "dense-union" : "sparse-union",
256
- typeIds: unionType.typeIds,
257
- children: unionType.children.map((arrowField) => serializeArrowField(arrowField)),
258
- typeIdToChildIndex: { ...unionType.typeIdToChildIndex }
259
- };
260
- case arrow.Dictionary:
261
- const dictionaryType = arrowType;
262
- return {
263
- type: "dictionary",
264
- id: dictionaryType.id,
265
- indices: serializeArrowDictionaryKeyType(dictionaryType.indices),
266
- dictionary: serializeArrowType(dictionaryType.dictionary),
267
- isOrdered: dictionaryType.isOrdered
268
- };
269
- default:
270
- throw new Error(`arrow type not supported: ${arrowType.constructor.name}`);
593
+ static isQuote(c) {
594
+ return c === '"' || c === "'";
595
+ }
596
+ static isAttribEnd(c) {
597
+ return c === ">" || _SAX.isWhitespace(c);
598
+ }
599
+ static isMatch(regex, c) {
600
+ return regex.test(c);
601
+ }
602
+ static notMatch(regex, c) {
603
+ return !_SAX.isMatch(regex, c);
604
+ }
605
+ static qname(name, attribute) {
606
+ const i = name.indexOf(":");
607
+ const qualName = i < 0 ? ["", name] : name.split(":");
608
+ let prefix = qualName[0];
609
+ let local = qualName[1];
610
+ if (attribute && name === "xmlns") {
611
+ prefix = "xmlns";
612
+ local = "";
613
+ }
614
+ return { prefix, local };
615
+ }
616
+ write(chunk) {
617
+ if (this.error) {
618
+ throw this.error;
619
+ }
620
+ if (this.closed) {
621
+ return this.errorFunction("Cannot write after close. Assign an onready handler.");
622
+ }
623
+ if (chunk === null) {
624
+ return this.end();
625
+ }
626
+ if (typeof chunk === "object") {
627
+ chunk = chunk.toString();
628
+ }
629
+ let i = 0;
630
+ let c;
631
+ while (true) {
632
+ c = _SAX.charAt(chunk, i++);
633
+ this.c = c;
634
+ if (!c) {
635
+ break;
636
+ }
637
+ if (this.trackPosition) {
638
+ this.position++;
639
+ if (c === "\n") {
640
+ this.line++;
641
+ this.column = 0;
642
+ } else {
643
+ this.column++;
644
+ }
645
+ }
646
+ switch (this.state) {
647
+ case this.S.BEGIN:
648
+ this.state = this.S.BEGIN_WHITESPACE;
649
+ if (c === "\uFEFF") {
650
+ continue;
651
+ }
652
+ this.beginWhiteSpace(c);
653
+ continue;
654
+ case this.S.BEGIN_WHITESPACE:
655
+ this.beginWhiteSpace(c);
656
+ continue;
657
+ case this.S.TEXT:
658
+ if (this.sawRoot && !this.closedRoot) {
659
+ const starti = i - 1;
660
+ while (c && c !== "<" && c !== "&") {
661
+ c = _SAX.charAt(chunk, i++);
662
+ if (c && this.trackPosition) {
663
+ this.position++;
664
+ if (c === "\n") {
665
+ this.line++;
666
+ this.column = 0;
667
+ } else {
668
+ this.column++;
669
+ }
670
+ }
671
+ }
672
+ this.textNode += chunk.substring(starti, i - 1);
673
+ }
674
+ if (c === "<" && !(this.sawRoot && this.closedRoot && !this.strict)) {
675
+ this.state = this.S.OPEN_WAKA;
676
+ this.startTagPosition = this.position;
677
+ } else {
678
+ if (!_SAX.isWhitespace(c) && (!this.sawRoot || this.closedRoot)) {
679
+ this.strictFail("Text data outside of root node.");
680
+ }
681
+ if (c === "&") {
682
+ this.state = this.S.TEXT_ENTITY;
683
+ } else {
684
+ this.textNode += c;
685
+ }
686
+ }
687
+ continue;
688
+ case this.S.SCRIPT:
689
+ if (c === "<") {
690
+ this.state = this.S.SCRIPT_ENDING;
691
+ } else {
692
+ this.script += c;
693
+ }
694
+ continue;
695
+ case this.S.SCRIPT_ENDING:
696
+ if (c === "/") {
697
+ this.state = this.S.CLOSE_TAG;
698
+ } else {
699
+ this.script += `<${c}`;
700
+ this.state = this.S.SCRIPT;
701
+ }
702
+ continue;
703
+ case this.S.OPEN_WAKA:
704
+ if (c === "!") {
705
+ this.state = this.S.SGML_DECL;
706
+ this.sgmlDecl = "";
707
+ } else if (_SAX.isWhitespace(c)) {
708
+ } else if (_SAX.isMatch(nameStart, c)) {
709
+ this.state = this.S.OPEN_TAG;
710
+ this.tagName = c;
711
+ } else if (c === "/") {
712
+ this.state = this.S.CLOSE_TAG;
713
+ this.tagName = "";
714
+ } else if (c === "?") {
715
+ this.state = this.S.PROC_INST;
716
+ this.procInstName = this.procInstBody = "";
717
+ } else {
718
+ this.strictFail("Unencoded <");
719
+ if (this.startTagPosition + 1 < this.position) {
720
+ const pad = this.position - this.startTagPosition;
721
+ c = new Array(pad).join(" ") + c;
722
+ }
723
+ this.textNode += `<${c}`;
724
+ this.state = this.S.TEXT;
725
+ }
726
+ continue;
727
+ case this.S.SGML_DECL:
728
+ if ((this.sgmlDecl + c).toUpperCase() === this.CDATA) {
729
+ this.emitNode("onopencdata");
730
+ this.state = this.S.CDATA;
731
+ this.sgmlDecl = "";
732
+ this.cdata = "";
733
+ } else if (this.sgmlDecl + c === "--") {
734
+ this.state = this.S.COMMENT;
735
+ this.comment = "";
736
+ this.sgmlDecl = "";
737
+ } else if ((this.sgmlDecl + c).toUpperCase() === this.DOCTYPE) {
738
+ this.state = this.S.DOCTYPE;
739
+ if (this.doctype || this.sawRoot) {
740
+ this.strictFail("Inappropriately located doctype declaration");
741
+ }
742
+ this.doctype = "";
743
+ this.sgmlDecl = "";
744
+ } else if (c === ">") {
745
+ this.emitNode("onsgmldeclaration", this.sgmlDecl);
746
+ this.sgmlDecl = "";
747
+ this.state = this.S.TEXT;
748
+ } else if (_SAX.isQuote(c)) {
749
+ this.state = this.S.SGML_DECL_QUOTED;
750
+ this.sgmlDecl += c;
751
+ } else {
752
+ this.sgmlDecl += c;
753
+ }
754
+ continue;
755
+ case this.S.SGML_DECL_QUOTED:
756
+ if (c === this.q) {
757
+ this.state = this.S.SGML_DECL;
758
+ this.q = "";
759
+ }
760
+ this.sgmlDecl += c;
761
+ continue;
762
+ case this.S.DOCTYPE:
763
+ if (c === ">") {
764
+ this.state = this.S.TEXT;
765
+ this.emitNode("ondoctype", this.doctype);
766
+ this.doctype = true;
767
+ } else {
768
+ this.doctype += c;
769
+ if (c === "[") {
770
+ this.state = this.S.DOCTYPE_DTD;
771
+ } else if (_SAX.isQuote(c)) {
772
+ this.state = this.S.DOCTYPE_QUOTED;
773
+ this.q = c;
774
+ }
775
+ }
776
+ continue;
777
+ case this.S.DOCTYPE_QUOTED:
778
+ this.doctype += c;
779
+ if (c === this.q) {
780
+ this.q = "";
781
+ this.state = this.S.DOCTYPE;
782
+ }
783
+ continue;
784
+ case this.S.DOCTYPE_DTD:
785
+ this.doctype += c;
786
+ if (c === "]") {
787
+ this.state = this.S.DOCTYPE;
788
+ } else if (_SAX.isQuote(c)) {
789
+ this.state = this.S.DOCTYPE_DTD_QUOTED;
790
+ this.q = c;
791
+ }
792
+ continue;
793
+ case this.S.DOCTYPE_DTD_QUOTED:
794
+ this.doctype += c;
795
+ if (c === this.q) {
796
+ this.state = this.S.DOCTYPE_DTD;
797
+ this.q = "";
798
+ }
799
+ continue;
800
+ case this.S.COMMENT:
801
+ if (c === "-") {
802
+ this.state = this.S.COMMENT_ENDING;
803
+ } else {
804
+ this.comment += c;
805
+ }
806
+ continue;
807
+ case this.S.COMMENT_ENDING:
808
+ if (c === "-") {
809
+ this.state = this.S.COMMENT_ENDED;
810
+ this.comment = this.textApplyOptions(this.comment);
811
+ if (this.comment) {
812
+ this.emitNode("oncomment", this.comment);
813
+ }
814
+ this.comment = "";
815
+ } else {
816
+ this.comment += `-${c}`;
817
+ this.state = this.S.COMMENT;
818
+ }
819
+ continue;
820
+ case this.S.COMMENT_ENDED:
821
+ if (c !== ">") {
822
+ this.strictFail("Malformed comment");
823
+ this.comment += `--${c}`;
824
+ this.state = this.S.COMMENT;
825
+ } else {
826
+ this.state = this.S.TEXT;
827
+ }
828
+ continue;
829
+ case this.S.CDATA:
830
+ if (c === "]") {
831
+ this.state = this.S.CDATA_ENDING;
832
+ } else {
833
+ this.cdata += c;
834
+ }
835
+ continue;
836
+ case this.S.CDATA_ENDING:
837
+ if (c === "]") {
838
+ this.state = this.S.CDATA_ENDING_2;
839
+ } else {
840
+ this.cdata += `]${c}`;
841
+ this.state = this.S.CDATA;
842
+ }
843
+ continue;
844
+ case this.S.CDATA_ENDING_2:
845
+ if (c === ">") {
846
+ if (this.cdata) {
847
+ this.emitNode("oncdata", this.cdata);
848
+ }
849
+ this.emitNode("onclosecdata");
850
+ this.cdata = "";
851
+ this.state = this.S.TEXT;
852
+ } else if (c === "]") {
853
+ this.cdata += "]";
854
+ } else {
855
+ this.cdata += `]]${c}`;
856
+ this.state = this.S.CDATA;
857
+ }
858
+ continue;
859
+ case this.S.PROC_INST:
860
+ if (c === "?") {
861
+ this.state = this.S.PROC_INST_ENDING;
862
+ } else if (_SAX.isWhitespace(c)) {
863
+ this.state = this.S.PROC_INST_BODY;
864
+ } else {
865
+ this.procInstName += c;
866
+ }
867
+ continue;
868
+ case this.S.PROC_INST_BODY:
869
+ if (!this.procInstBody && _SAX.isWhitespace(c)) {
870
+ continue;
871
+ } else if (c === "?") {
872
+ this.state = this.S.PROC_INST_ENDING;
873
+ } else {
874
+ this.procInstBody += c;
875
+ }
876
+ continue;
877
+ case this.S.PROC_INST_ENDING:
878
+ if (c === ">") {
879
+ this.emitNode("onprocessinginstruction", {
880
+ name: this.procInstName,
881
+ body: this.procInstBody
882
+ });
883
+ this.procInstName = this.procInstBody = "";
884
+ this.state = this.S.TEXT;
885
+ } else {
886
+ this.procInstBody += `?${c}`;
887
+ this.state = this.S.PROC_INST_BODY;
888
+ }
889
+ continue;
890
+ case this.S.OPEN_TAG:
891
+ if (_SAX.isMatch(nameBody, c)) {
892
+ this.tagName += c;
893
+ } else {
894
+ this.newTag();
895
+ if (c === ">") {
896
+ this.openTag();
897
+ } else if (c === "/") {
898
+ this.state = this.S.OPEN_TAG_SLASH;
899
+ } else {
900
+ if (!_SAX.isWhitespace(c)) {
901
+ this.strictFail("Invalid character in tag name");
902
+ }
903
+ this.state = this.S.ATTRIB;
904
+ }
905
+ }
906
+ continue;
907
+ case this.S.OPEN_TAG_SLASH:
908
+ if (c === ">") {
909
+ this.openTag(true);
910
+ this.closeTag();
911
+ } else {
912
+ this.strictFail("Forward-slash in opening tag not followed by >");
913
+ this.state = this.S.ATTRIB;
914
+ }
915
+ continue;
916
+ case this.S.ATTRIB:
917
+ if (_SAX.isWhitespace(c)) {
918
+ continue;
919
+ } else if (c === ">") {
920
+ this.openTag();
921
+ } else if (c === "/") {
922
+ this.state = this.S.OPEN_TAG_SLASH;
923
+ } else if (_SAX.isMatch(nameStart, c)) {
924
+ this.attribName = c;
925
+ this.attribValue = "";
926
+ this.state = this.S.ATTRIB_NAME;
927
+ } else {
928
+ this.strictFail("Invalid attribute name");
929
+ }
930
+ continue;
931
+ case this.S.ATTRIB_NAME:
932
+ if (c === "=") {
933
+ this.state = this.S.ATTRIB_VALUE;
934
+ } else if (c === ">") {
935
+ this.strictFail("Attribute without value");
936
+ this.attribValue = this.attribName;
937
+ this.attrib();
938
+ this.openTag();
939
+ } else if (_SAX.isWhitespace(c)) {
940
+ this.state = this.S.ATTRIB_NAME_SAW_WHITE;
941
+ } else if (_SAX.isMatch(nameBody, c)) {
942
+ this.attribName += c;
943
+ } else {
944
+ this.strictFail("Invalid attribute name");
945
+ }
946
+ continue;
947
+ case this.S.ATTRIB_NAME_SAW_WHITE:
948
+ if (c === "=") {
949
+ this.state = this.S.ATTRIB_VALUE;
950
+ } else if (_SAX.isWhitespace(c)) {
951
+ continue;
952
+ } else {
953
+ this.strictFail("Attribute without value");
954
+ Object.defineProperty(this.tag.attributes, this.attribName, {
955
+ value: "",
956
+ enumerable: true,
957
+ configurable: true,
958
+ writable: true
959
+ });
960
+ this.attribValue = "";
961
+ this.emitNode("onattribute", {
962
+ name: this.attribName,
963
+ value: ""
964
+ });
965
+ this.attribName = "";
966
+ if (c === ">") {
967
+ this.openTag();
968
+ } else if (_SAX.isMatch(nameStart, c)) {
969
+ this.attribName = c;
970
+ this.state = this.S.ATTRIB_NAME;
971
+ } else {
972
+ this.strictFail("Invalid attribute name");
973
+ this.state = this.S.ATTRIB;
974
+ }
975
+ }
976
+ continue;
977
+ case this.S.ATTRIB_VALUE:
978
+ if (_SAX.isWhitespace(c)) {
979
+ continue;
980
+ } else if (_SAX.isQuote(c)) {
981
+ this.q = c;
982
+ this.state = this.S.ATTRIB_VALUE_QUOTED;
983
+ } else {
984
+ this.strictFail("Unquoted attribute value");
985
+ this.state = this.S.ATTRIB_VALUE_UNQUOTED;
986
+ this.attribValue = c;
987
+ }
988
+ continue;
989
+ case this.S.ATTRIB_VALUE_QUOTED:
990
+ if (c !== this.q) {
991
+ if (c === "&") {
992
+ this.state = this.S.ATTRIB_VALUE_ENTITY_Q;
993
+ } else {
994
+ this.attribValue += c;
995
+ }
996
+ continue;
997
+ }
998
+ this.attrib();
999
+ this.q = "";
1000
+ this.state = this.S.ATTRIB_VALUE_CLOSED;
1001
+ continue;
1002
+ case this.S.ATTRIB_VALUE_CLOSED:
1003
+ if (_SAX.isWhitespace(c)) {
1004
+ this.state = this.S.ATTRIB;
1005
+ } else if (c === ">") {
1006
+ this.openTag();
1007
+ } else if (c === "/") {
1008
+ this.state = this.S.OPEN_TAG_SLASH;
1009
+ } else if (_SAX.isMatch(nameStart, c)) {
1010
+ this.strictFail("No whitespace between attributes");
1011
+ this.attribName = c;
1012
+ this.attribValue = "";
1013
+ this.state = this.S.ATTRIB_NAME;
1014
+ } else {
1015
+ this.strictFail("Invalid attribute name");
1016
+ }
1017
+ continue;
1018
+ case this.S.ATTRIB_VALUE_UNQUOTED:
1019
+ if (!_SAX.isAttribEnd(c)) {
1020
+ if (c === "&") {
1021
+ this.state = this.S.ATTRIB_VALUE_ENTITY_U;
1022
+ } else {
1023
+ this.attribValue += c;
1024
+ }
1025
+ continue;
1026
+ }
1027
+ this.attrib();
1028
+ if (c === ">") {
1029
+ this.openTag();
1030
+ } else {
1031
+ this.state = this.S.ATTRIB;
1032
+ }
1033
+ continue;
1034
+ case this.S.CLOSE_TAG:
1035
+ if (!this.tagName) {
1036
+ if (_SAX.isWhitespace(c)) {
1037
+ continue;
1038
+ } else if (_SAX.notMatch(nameStart, c)) {
1039
+ if (this.script) {
1040
+ this.script += `</${c}`;
1041
+ this.state = this.S.SCRIPT;
1042
+ } else {
1043
+ this.strictFail("Invalid tagname in closing tag.");
1044
+ }
1045
+ } else {
1046
+ this.tagName = c;
1047
+ }
1048
+ } else if (c === ">") {
1049
+ this.closeTag();
1050
+ } else if (_SAX.isMatch(nameBody, c)) {
1051
+ this.tagName += c;
1052
+ } else if (this.script) {
1053
+ this.script += `</${this.tagName}`;
1054
+ this.tagName = "";
1055
+ this.state = this.S.SCRIPT;
1056
+ } else {
1057
+ if (!_SAX.isWhitespace(c)) {
1058
+ this.strictFail("Invalid tagname in closing tag");
1059
+ }
1060
+ this.state = this.S.CLOSE_TAG_SAW_WHITE;
1061
+ }
1062
+ continue;
1063
+ case this.S.CLOSE_TAG_SAW_WHITE:
1064
+ if (_SAX.isWhitespace(c)) {
1065
+ continue;
1066
+ }
1067
+ if (c === ">") {
1068
+ this.closeTag();
1069
+ } else {
1070
+ this.strictFail("Invalid characters in closing tag");
1071
+ }
1072
+ continue;
1073
+ case this.S.TEXT_ENTITY:
1074
+ case this.S.ATTRIB_VALUE_ENTITY_Q:
1075
+ case this.S.ATTRIB_VALUE_ENTITY_U:
1076
+ let returnState;
1077
+ let buffer;
1078
+ switch (this.state) {
1079
+ case this.S.TEXT_ENTITY:
1080
+ returnState = this.S.TEXT;
1081
+ buffer = "textNode";
1082
+ break;
1083
+ case this.S.ATTRIB_VALUE_ENTITY_Q:
1084
+ returnState = this.S.ATTRIB_VALUE_QUOTED;
1085
+ buffer = "attribValue";
1086
+ break;
1087
+ case this.S.ATTRIB_VALUE_ENTITY_U:
1088
+ returnState = this.S.ATTRIB_VALUE_UNQUOTED;
1089
+ buffer = "attribValue";
1090
+ break;
1091
+ default:
1092
+ throw new Error(`Unknown state: ${this.state}`);
1093
+ }
1094
+ if (c === ";") {
1095
+ this[buffer] += this.parseEntity();
1096
+ this.entity = "";
1097
+ this.state = returnState;
1098
+ } else if (_SAX.isMatch(this.entity.length ? entityBody : entityStart, c)) {
1099
+ this.entity += c;
1100
+ } else {
1101
+ this.strictFail("Invalid character in entity name");
1102
+ this[buffer] += `&${this.entity}${c}`;
1103
+ this.entity = "";
1104
+ this.state = returnState;
1105
+ }
1106
+ continue;
1107
+ default:
1108
+ throw new Error(`Unknown state: ${this.state}`);
1109
+ }
1110
+ }
1111
+ if (this.position >= this.bufferCheckPosition) {
1112
+ this.checkBufferLength();
1113
+ }
1114
+ return this;
1115
+ }
1116
+ emit(event, data) {
1117
+ if (this.events.hasOwnProperty(event)) {
1118
+ const eventName = event.replace(/^on/, "");
1119
+ this.events[event](data, eventName, this);
1120
+ }
1121
+ }
1122
+ clearBuffers() {
1123
+ for (let i = 0, l = this.BUFFERS.length; i < l; i++) {
1124
+ this[this[i]] = "";
1125
+ }
1126
+ }
1127
+ flushBuffers() {
1128
+ this.closeText();
1129
+ if (this.cdata !== "") {
1130
+ this.emitNode("oncdata", this.cdata);
1131
+ this.cdata = "";
1132
+ }
1133
+ if (this.script !== "") {
1134
+ this.emitNode("onscript", this.script);
1135
+ this.script = "";
1136
+ }
1137
+ }
1138
+ end() {
1139
+ if (this.sawRoot && !this.closedRoot) this.strictFail("Unclosed root tag");
1140
+ if (this.strict && this.state !== this.S.BEGIN && this.state !== this.S.BEGIN_WHITESPACE && this.state !== this.S.TEXT) {
1141
+ this.errorFunction("Unexpected end");
1142
+ }
1143
+ this.closeText();
1144
+ this.c = "";
1145
+ this.closed = true;
1146
+ this.emit("onend");
1147
+ return new SAXParser(this.opt);
1148
+ }
1149
+ errorFunction(er) {
1150
+ this.closeText();
1151
+ if (this.trackPosition) {
1152
+ er += `
1153
+ Line: ${this.line}
1154
+ Column: ${this.column}
1155
+ Char: ${this.c}`;
1156
+ }
1157
+ const error = new Error(er);
1158
+ this.error = error;
1159
+ this.emit("onerror", error);
1160
+ return this;
1161
+ }
1162
+ attrib() {
1163
+ if (!this.strict && !this.opt.preserveCase) {
1164
+ this.attribName = this.attribName[this.looseCase]();
1165
+ }
1166
+ if (this.attribList.indexOf(this.attribName) !== -1 || Object.hasOwn(this.tag.attributes, this.attribName)) {
1167
+ this.strictFail("Duplicate attribute");
1168
+ this.attribName = this.attribValue = "";
1169
+ return;
1170
+ }
1171
+ if (this.opt.xmlns) {
1172
+ const qn = _SAX.qname(this.attribName, true);
1173
+ const prefix = qn.prefix;
1174
+ const local = qn.local;
1175
+ if (prefix === "xmlns") {
1176
+ if (local === "xml" && this.attribValue !== this.XML_NAMESPACE) {
1177
+ this.strictFail(
1178
+ `xml: prefix must be bound to ${this.XML_NAMESPACE}
1179
+ Actual: ${this.attribValue}`
1180
+ );
1181
+ } else if (local === "xmlns" && this.attribValue !== this.XMLNS_NAMESPACE) {
1182
+ this.strictFail(
1183
+ `xmlns: prefix must be bound to ${this.XMLNS_NAMESPACE}
1184
+ Actual: ${this.attribValue}`
1185
+ );
1186
+ } else {
1187
+ const tag = this.tag;
1188
+ const parent = this.tags[this.tags.length - 1] || this;
1189
+ if (tag.ns === parent.ns) {
1190
+ tag.ns = Object.create(parent.ns);
1191
+ }
1192
+ tag.ns[local] = this.attribValue;
1193
+ }
1194
+ }
1195
+ this.attribList.push([this.attribName, this.attribValue]);
1196
+ } else {
1197
+ Object.defineProperty(this.tag.attributes, this.attribName, {
1198
+ value: this.attribValue,
1199
+ enumerable: true,
1200
+ configurable: true,
1201
+ writable: true
1202
+ });
1203
+ this.emitNode("onattribute", {
1204
+ name: this.attribName,
1205
+ value: this.attribValue
1206
+ });
1207
+ }
1208
+ this.attribName = this.attribValue = "";
1209
+ }
1210
+ newTag() {
1211
+ if (!this.strict && !this.opt.preserveCase) this.tagName = this.tagName[this.looseCase]();
1212
+ const parent = this.tags[this.tags.length - 1] || this;
1213
+ const tag = { name: this.tagName, attributes: {} };
1214
+ this.tag = tag;
1215
+ if (this.opt.xmlns) {
1216
+ tag.ns = parent.ns;
1217
+ }
1218
+ this.attribList.length = 0;
1219
+ this.emitNode("onopentagstart", tag);
1220
+ }
1221
+ /** Validates an entity before optionally preserving its authored spelling. */
1222
+ parseEntity() {
1223
+ const decoded = this.decodeEntity();
1224
+ return this.opt.processEntities === false ? `&${this.entity};` : decoded;
1225
+ }
1226
+ /** Decodes predefined, configured, or numeric entities without resolving document types. */
1227
+ decodeEntity() {
1228
+ let entity = this.entity;
1229
+ const entityLC = entity.toLowerCase();
1230
+ let num = NaN;
1231
+ let numStr = "";
1232
+ if (typeof this.ENTITIES[entity] === "string" || typeof this.ENTITIES[entity] === "number") {
1233
+ return this.ENTITIES[entity];
1234
+ }
1235
+ if (typeof this.ENTITIES[entityLC] === "string" || typeof this.ENTITIES[entityLC] === "number") {
1236
+ return this.ENTITIES[entityLC];
1237
+ }
1238
+ entity = entityLC;
1239
+ if (entity.charAt(0) === "#") {
1240
+ if (entity.charAt(1) === "x") {
1241
+ entity = entity.slice(2);
1242
+ num = parseInt(entity, 16);
1243
+ numStr = num.toString(16);
1244
+ } else {
1245
+ entity = entity.slice(1);
1246
+ num = parseInt(entity, 10);
1247
+ numStr = num.toString(10);
1248
+ }
1249
+ }
1250
+ entity = entity.replace(/^0+/, "");
1251
+ if (Number.isNaN(num) || numStr.toLowerCase() !== entity) {
1252
+ this.strictFail("Invalid character entity");
1253
+ return `&${this.entity};`;
1254
+ }
1255
+ if (num < 0 || num > 1114111 || num >= 55296 && num <= 57343 || num < 32 && ![9, 10, 13].includes(num) || num === 65534 || num === 65535) {
1256
+ this.strictFail("Invalid character entity");
1257
+ return `&${this.entity};`;
1258
+ }
1259
+ return String.fromCodePoint(num);
1260
+ }
1261
+ beginWhiteSpace(c) {
1262
+ if (c === "<") {
1263
+ this.state = this.S.OPEN_WAKA;
1264
+ this.startTagPosition = this.position;
1265
+ } else if (!_SAX.isWhitespace(c)) {
1266
+ this.strictFail("Non-whitespace before first tag.");
1267
+ this.textNode = c;
1268
+ this.state = this.S.TEXT;
1269
+ } else {
1270
+ }
1271
+ }
1272
+ strictFail(message) {
1273
+ if (typeof this !== "object" || !(this instanceof SAXParser)) {
1274
+ throw new Error("bad call to strictFail");
1275
+ }
1276
+ if (this.strict) {
1277
+ this.errorFunction(message);
1278
+ }
1279
+ }
1280
+ textApplyOptions(text) {
1281
+ if (this.opt.trim) text = text.trim();
1282
+ if (this.opt.normalize) text = text.replace(/\s+/g, " ");
1283
+ return text;
1284
+ }
1285
+ emitNode(nodeType, data) {
1286
+ if (this.textNode) this.closeText();
1287
+ this.emit(nodeType, data);
1288
+ }
1289
+ closeText() {
1290
+ this.textNode = this.textApplyOptions(this.textNode);
1291
+ if (this.textNode !== void 0 && this.textNode !== "" && this.textNode !== "undefined") {
1292
+ this.emit("ontext", this.textNode);
1293
+ }
1294
+ this.textNode = "";
1295
+ }
1296
+ checkBufferLength() {
1297
+ var _a;
1298
+ const maxAllowed = Math.max(this.opt.MAX_BUFFER_LENGTH, 10);
1299
+ let maxActual = 0;
1300
+ for (let i = 0, l = this.BUFFERS.length; i < l; i++) {
1301
+ const len = ((_a = this[this.BUFFERS[i]]) == null ? void 0 : _a.length) || 0;
1302
+ if (len > maxAllowed) {
1303
+ switch (this.BUFFERS[i]) {
1304
+ case "textNode":
1305
+ this.closeText();
1306
+ break;
1307
+ case "cdata":
1308
+ this.emitNode("oncdata", this.cdata);
1309
+ this.cdata = "";
1310
+ break;
1311
+ case "script":
1312
+ this.emitNode("onscript", this.script);
1313
+ this.script = "";
1314
+ break;
1315
+ default:
1316
+ this.errorFunction(`Max buffer length exceeded: ${this.BUFFERS[i]}`);
1317
+ }
1318
+ }
1319
+ maxActual = Math.max(maxActual, len);
1320
+ }
1321
+ const m = this.opt.MAX_BUFFER_LENGTH - maxActual;
1322
+ this.bufferCheckPosition = m + this.position;
1323
+ }
1324
+ openTag(selfClosing) {
1325
+ if (this.opt.xmlns) {
1326
+ const tag = this.tag;
1327
+ const qn = _SAX.qname(this.tagName);
1328
+ tag.prefix = qn.prefix;
1329
+ tag.local = qn.local;
1330
+ tag.uri = tag.ns[qn.prefix] || "";
1331
+ if (tag.prefix && !tag.uri) {
1332
+ this.strictFail(`Unbound namespace prefix: ${JSON.stringify(this.tagName)}`);
1333
+ tag.uri = qn.prefix;
1334
+ }
1335
+ const parent = this.tags[this.tags.length - 1] || this;
1336
+ if (tag.ns && parent.ns !== tag.ns) {
1337
+ const that = this;
1338
+ Object.keys(tag.ns).forEach((p) => {
1339
+ that.emitNode("onopennamespace", {
1340
+ prefix: p,
1341
+ uri: tag.ns[p]
1342
+ });
1343
+ });
1344
+ }
1345
+ for (let i = 0, l = this.attribList.length; i < l; i++) {
1346
+ const nv = this.attribList[i];
1347
+ const name = nv[0];
1348
+ const value = nv[1];
1349
+ const qualName = _SAX.qname(name, true);
1350
+ const prefix = qualName.prefix;
1351
+ const local = qualName.local;
1352
+ const uri = prefix === "" ? "" : tag.ns[prefix] || "";
1353
+ const a = {
1354
+ name,
1355
+ value,
1356
+ prefix,
1357
+ local,
1358
+ uri
1359
+ };
1360
+ if (prefix && prefix !== "xmlns" && !uri) {
1361
+ this.strictFail(`Unbound namespace prefix: ${JSON.stringify(prefix)}`);
1362
+ a.uri = prefix;
1363
+ }
1364
+ Object.defineProperty(this.tag.attributes, name, {
1365
+ value: a,
1366
+ enumerable: true,
1367
+ configurable: true,
1368
+ writable: true
1369
+ });
1370
+ this.emitNode("onattribute", a);
1371
+ }
1372
+ this.attribList.length = 0;
1373
+ }
1374
+ this.tag.isSelfClosing = Boolean(selfClosing);
1375
+ this.sawRoot = true;
1376
+ this.tags.push(this.tag);
1377
+ this.emitNode("onopentag", this.tag);
1378
+ if (!selfClosing) {
1379
+ if (!this.noscript && this.tagName.toLowerCase() === "script") {
1380
+ this.state = this.S.SCRIPT;
1381
+ } else {
1382
+ this.state = this.S.TEXT;
1383
+ }
1384
+ this.tag = null;
1385
+ this.tagName = "";
1386
+ }
1387
+ this.attribName = this.attribValue = "";
1388
+ this.attribList.length = 0;
1389
+ }
1390
+ closeTag() {
1391
+ if (!this.tagName) {
1392
+ this.strictFail("Weird empty close tag.");
1393
+ this.textNode += "</>";
1394
+ this.state = this.S.TEXT;
1395
+ return;
1396
+ }
1397
+ const caseInsensitiveTags = !this.strict && (this.opt.caseInsensitiveTags || !this.opt.preserveCase);
1398
+ if (this.script) {
1399
+ const normalizedScriptCloseTag = caseInsensitiveTags ? this.tagName.toLowerCase() : this.tagName;
1400
+ if (normalizedScriptCloseTag !== "script") {
1401
+ this.script += `</${this.tagName}>`;
1402
+ this.tagName = "";
1403
+ this.state = this.S.SCRIPT;
1404
+ return;
1405
+ }
1406
+ this.emitNode("onscript", this.script);
1407
+ this.script = "";
1408
+ }
1409
+ let t = this.tags.length;
1410
+ let tagName = this.tagName;
1411
+ const normalizedTagName = caseInsensitiveTags ? tagName.toLowerCase() : tagName;
1412
+ while (t--) {
1413
+ const close = this.tags[t];
1414
+ const normalizedOpenTagName = caseInsensitiveTags ? close.name.toLowerCase() : close.name;
1415
+ if (normalizedOpenTagName !== normalizedTagName) {
1416
+ this.strictFail("Unexpected close tag");
1417
+ } else {
1418
+ break;
1419
+ }
1420
+ }
1421
+ if (t < 0) {
1422
+ this.strictFail(`Unmatched closing tag: ${this.tagName}`);
1423
+ this.textNode += `</${this.tagName}>`;
1424
+ this.state = this.S.TEXT;
1425
+ return;
1426
+ }
1427
+ this.tagName = tagName;
1428
+ let s = this.tags.length;
1429
+ while (s-- > t) {
1430
+ const tag = this.tags.pop();
1431
+ this.tag = tag;
1432
+ this.tagName = this.tag.name;
1433
+ this.emitNode("onclosetag", this.tagName);
1434
+ const x = {};
1435
+ for (const i in tag.ns) {
1436
+ if (tag.ns.hasOwnProperty(i)) {
1437
+ x[i] = tag.ns[i];
1438
+ }
1439
+ }
1440
+ const parent = this.tags[this.tags.length - 1] || this;
1441
+ if (this.opt.xmlns && tag.ns !== parent.ns) {
1442
+ const that = this;
1443
+ Object.keys(tag.ns).forEach((p) => {
1444
+ const n = tag.ns[p];
1445
+ that.emitNode("onclosenamespace", { prefix: p, uri: n });
1446
+ });
1447
+ }
1448
+ }
1449
+ if (t === 0) this.closedRoot = true;
1450
+ this.tagName = this.attribValue = this.attribName = "";
1451
+ this.attribList.length = 0;
1452
+ this.state = this.S.TEXT;
1453
+ }
1454
+ };
1455
+ SAXParser = class extends SAX {
1456
+ static ENTITIES = ENTITIES;
1457
+ opt = DEFAULT_SAX_PARSER_OPTIONS;
1458
+ events = DEFAULT_SAX_EVENTS;
1459
+ constructor(opt) {
1460
+ super();
1461
+ this.clearBuffers();
1462
+ this.opt = opt = { ...this.opt, ...opt };
1463
+ this.events = { ...this.events, ...opt };
1464
+ this.q = this.c = "";
1465
+ this.opt.lowercase = this.opt.lowercase || this.opt.lowercasetags;
1466
+ this.bufferCheckPosition = this.opt.MAX_BUFFER_LENGTH;
1467
+ this.looseCase = this.opt.lowercase ? "toLowerCase" : "toUpperCase";
1468
+ this.tags = [];
1469
+ this.closed = this.closedRoot = this.sawRoot = false;
1470
+ this.tag = this.error = null;
1471
+ this.strict = Boolean(this.opt.strict);
1472
+ this.noscript = Boolean(this.opt.strict || this.opt.noscript);
1473
+ this.state = this.S.BEGIN;
1474
+ this.strictEntities = this.opt.strictEntities;
1475
+ this.ENTITIES = this.strictEntities ? Object.create(this.XML_ENTITIES) : Object.create(this.ENTITIES);
1476
+ this.attribList = [];
1477
+ if (this.opt.xmlns) {
1478
+ this.ns = Object.create(this.rootNS);
1479
+ }
1480
+ this.trackPosition = this.opt.position !== false;
1481
+ if (this.trackPosition) {
1482
+ this.position = this.line = this.column = 0;
1483
+ }
1484
+ this.emit("onready");
1485
+ }
1486
+ resume() {
1487
+ this.error = null;
1488
+ return this;
1489
+ }
1490
+ close() {
1491
+ return this.write(null);
1492
+ }
1493
+ flush() {
1494
+ this.flushBuffers();
1495
+ }
1496
+ };
271
1497
  }
1498
+ });
1499
+
1500
+ // ../documents/src/lib/parsers/parse-xml-internal.ts
1501
+ function setXMLField(object, name, value) {
1502
+ Object.defineProperty(object, name, {
1503
+ value,
1504
+ enumerable: true,
1505
+ configurable: true,
1506
+ writable: true
1507
+ });
272
1508
  }
273
- function getOptionalArrowConstructor(typeName) {
274
- return arrow[typeName];
275
- }
276
- function isArrowViewType(arrowType, viewTypeName) {
277
- const constructor = getOptionalArrowConstructor(viewTypeName);
278
- if (constructor && arrowType instanceof constructor) {
279
- return true;
1509
+ function parseXMLInternal(text, options = {}) {
1510
+ if (options.strict !== false && /[^\u0009\u000a\u000d\u0020-\ud7ff\ue000-\ufffd\u{10000}-\u{10ffff}]/u.test(text)) {
1511
+ throw new Error("XML: invalid XML character");
1512
+ }
1513
+ const maximumDepth = options.maxDepth ?? 256;
1514
+ if (!Number.isSafeInteger(maximumDepth) || maximumDepth < 1 || maximumDepth > 4096) {
1515
+ throw new Error("XML: maxDepth must be an integer between 1 and 4096");
1516
+ }
1517
+ const state = new InternalXMLParserState(options, maximumDepth);
1518
+ const parser = new SAXParser({
1519
+ strict: options.strict !== false,
1520
+ strictEntities: !options.htmlEntities,
1521
+ preserveCase: true,
1522
+ caseInsensitiveTags: options.strict === false,
1523
+ processEntities: options.processEntities !== false,
1524
+ onopentag: (tag) => state.openElement(tag.name, tag.attributes || {}),
1525
+ ontext: (value) => state.addText(value),
1526
+ oncdata: (value) => state.addText(value),
1527
+ onscript: (value) => state.addText(value),
1528
+ onclosetag: () => state.closeElement(),
1529
+ onerror: (error) => {
1530
+ throw error;
1531
+ }
1532
+ });
1533
+ parser.write(text).close();
1534
+ if (options.strict === false) {
1535
+ while (state.stack.length) state.closeElement();
280
1536
  }
281
- const guards = arrow.DataType;
282
- const guard = viewTypeName === "BinaryView" ? guards.isBinaryView : guards.isUtf8View;
283
- return Boolean(guard == null ? void 0 : guard(arrowType));
284
- }
285
- function isOptionalArrowType(arrowType, typeName) {
286
- const constructor = getOptionalArrowConstructor(typeName);
287
- return Boolean(constructor && arrowType instanceof constructor);
1537
+ if (state.result === void 0) throw new Error("XML: document requires a root element");
1538
+ return state.result;
288
1539
  }
289
- function serializeArrowDictionaryKeyType(arrowType) {
290
- if (arrowType instanceof arrow.Int) {
291
- const prefix = arrowType.isSigned ? "int" : "uint";
292
- const keyType = `${prefix}${arrowType.bitWidth}`;
293
- if (isSchemaDictionaryKeyType(keyType)) {
294
- return keyType;
295
- }
1540
+ var InternalXMLParserState;
1541
+ var init_parse_xml_internal = __esm({
1542
+ "../documents/src/lib/parsers/parse-xml-internal.ts"() {
1543
+ "use strict";
1544
+ init_sax();
1545
+ InternalXMLParserState = class {
1546
+ /** Initializes conversion options and the element nesting budget. */
1547
+ constructor(options, maximumDepth) {
1548
+ this.options = options;
1549
+ this.maximumDepth = maximumDepth;
1550
+ }
1551
+ options;
1552
+ maximumDepth;
1553
+ /** Open element stack. */
1554
+ stack = [];
1555
+ /** Parsed root document. */
1556
+ result;
1557
+ /** Starts an element, rejecting multiple roots and excessive nesting. */
1558
+ openElement(name, attributes) {
1559
+ if (this.stack.length >= this.maximumDepth)
1560
+ throw new Error("XML: maximum nesting depth exceeded");
1561
+ if (!this.stack.length && this.result !== void 0 && this.options.strict !== false) {
1562
+ throw new Error("XML: document requires a single root element");
1563
+ }
1564
+ const elementName = this.normalizeName(name);
1565
+ const parent = this.stack[this.stack.length - 1];
1566
+ this.stack.push({
1567
+ name: elementName,
1568
+ path: parent ? `${parent.path}.${elementName}` : elementName,
1569
+ attributes: this.normalizeAttributes(attributes),
1570
+ children: {},
1571
+ orderedChildren: [],
1572
+ text: ""
1573
+ });
1574
+ }
1575
+ /** Accumulates text without losing whitespace between text and CDATA chunks. */
1576
+ addText(text) {
1577
+ const current = this.stack[this.stack.length - 1];
1578
+ if (current) current.text += text;
1579
+ }
1580
+ /** Finishes an element and attaches it to its parent or the document root. */
1581
+ closeElement() {
1582
+ const node = this.stack.pop();
1583
+ if (!node) return;
1584
+ const value = this.finalizeElement(node);
1585
+ const parent = this.stack[this.stack.length - 1];
1586
+ if (this.options.preserveOrder) {
1587
+ if (parent) parent.orderedChildren.push(value);
1588
+ else this.result = value;
1589
+ return;
1590
+ }
1591
+ if (parent) this.addChild(parent.children, node.name, node.path, value);
1592
+ else {
1593
+ if (this.result === void 0) this.result = {};
1594
+ this.addChild(this.result, node.name, node.path, value);
1595
+ }
1596
+ }
1597
+ /** Converts a completed element, keeping grouping and coercion independent. */
1598
+ finalizeElement(node) {
1599
+ const text = this.options.trimValues === false ? node.text : node.text.trim();
1600
+ if (this.options.preserveOrder) {
1601
+ return { name: node.name, attributes: node.attributes, children: node.orderedChildren, text };
1602
+ }
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
+ };
296
1659
  }
297
- switch (arrowType.constructor) {
298
- case arrow.Int8:
299
- return "int8";
300
- case arrow.Int16:
301
- return "int16";
302
- case arrow.Int32:
303
- return "int32";
304
- case arrow.Uint8:
305
- return "uint8";
306
- case arrow.Uint16:
307
- return "uint16";
308
- case arrow.Uint32:
309
- return "uint32";
310
- default:
311
- throw new Error(`arrow dictionary index type not supported: ${arrowType.constructor.name}`);
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");
312
1669
  }
1670
+ const parsed = parseXMLInternal(text, options);
1671
+ return options.uncapitalizeKeys ? uncapitalizeKeys(parsed) : parsed;
313
1672
  }
314
- function isSchemaDictionaryKeyType(keyType) {
315
- return ["int8", "int16", "int32", "uint8", "uint16", "uint32"].includes(keyType);
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
+ }
316
1680
  }
317
- var arrow;
318
- var init_convert_arrow_schema = __esm({
319
- "../core/src/schema-utils/lib/schema/convert-arrow-schema.ts"() {
1681
+ var init_parse_xml = __esm({
1682
+ "../documents/src/lib/parsers/parse-xml.ts"() {
320
1683
  "use strict";
321
- arrow = __toESM(require("apache-arrow"), 1);
1684
+ init_uncapitalize();
1685
+ init_parse_xml_internal();
322
1686
  }
323
1687
  });
324
1688
 
325
- // ../core/src/schema-utils/index.ts
326
- var init_schema_utils = __esm({
327
- "../core/src/schema-utils/index.ts"() {
1689
+ // ../documents/src/xml-parser.ts
1690
+ var init_xml_parser = __esm({
1691
+ "../documents/src/xml-parser.ts"() {
328
1692
  "use strict";
329
- init_convert_arrow_schema();
1693
+ init_parse_xml();
330
1694
  }
331
1695
  });
332
1696
 
@@ -335,7 +1699,7 @@ function buildGraphTables(nodes, edges, shape, declarations) {
335
1699
  if (shape === "plain-graph-data") {
336
1700
  return { shape, nodes: Array.from(nodes), edges: Array.from(edges) };
337
1701
  }
338
- if (shape !== "arrow-table" && shape !== "object-row-table") {
1702
+ if (shape !== "arrow" && shape !== "object-row-table") {
339
1703
  throw new Error(`Unsupported graph shape: ${shape}`);
340
1704
  }
341
1705
  const nodeTable = buildGraphTable(nodes, false, shape, declarations == null ? void 0 : declarations.nodes);
@@ -375,8 +1739,7 @@ function buildGraphTable(records, isEdge, shape, declarations) {
375
1739
  type
376
1740
  );
377
1741
  }
378
- const data = new import_apache_arrow.Table(vectors);
379
- return { shape, schema: convertArrowToSchema(data.schema), data };
1742
+ return new import_apache_arrow.Table(vectors);
380
1743
  }
381
1744
  function inferAttributeType(values, declarations) {
382
1745
  const names = new Set(declarations == null ? void 0 : declarations.keys());
@@ -471,17 +1834,16 @@ var init_build_graph_tables = __esm({
471
1834
  "src/lib/build-graph-tables.ts"() {
472
1835
  "use strict";
473
1836
  import_apache_arrow = require("apache-arrow");
474
- init_schema_utils();
475
1837
  }
476
1838
  });
477
1839
 
478
1840
  // src/lib/parse-gexf.ts
479
- function parseGEXF(input, shape = "arrow-table") {
1841
+ function parseGEXF(input, shape = "arrow") {
480
1842
  const text = typeof input === "string" ? input : new TextDecoder().decode(input);
481
- const validation = import_fast_xml_parser.XMLValidator.validate(text);
482
- if (validation !== true) throw new Error(`Invalid GEXF XML: ${validation.err.msg}`);
1843
+ const validation = validateXML(text);
1844
+ if (validation !== true) throw new Error(`Invalid GEXF XML: ${validation.message}`);
483
1845
  if (/<!DOCTYPE\s/i.test(text)) throw new Error("GEXF document types are unsupported.");
484
- const document = normalizeNamespaces(parser.parse(text), {});
1846
+ const document = normalizeNamespaces(parseXMLSync(text, XML_OPTIONS), {});
485
1847
  const root = getElement(document.gexf, "gexf");
486
1848
  const version = root["@_version"];
487
1849
  if (version && !["1.2", "1.2draft", "1.3"].includes(version)) {
@@ -755,18 +2117,20 @@ function requireValue(element) {
755
2117
  if (typeof value !== "string") throw new Error("GEXF attvalue requires a value attribute.");
756
2118
  return value;
757
2119
  }
758
- var import_fast_xml_parser, parser, GEXF_NAMESPACE, TEMPORAL_ATTRIBUTES;
2120
+ var XML_OPTIONS, GEXF_NAMESPACE, TEMPORAL_ATTRIBUTES;
759
2121
  var init_parse_gexf = __esm({
760
2122
  "src/lib/parse-gexf.ts"() {
761
2123
  "use strict";
762
- import_fast_xml_parser = require("fast-xml-parser");
2124
+ init_xml_parser();
763
2125
  init_build_graph_tables();
764
- parser = new import_fast_xml_parser.XMLParser({
2126
+ XML_OPTIONS = {
765
2127
  ignoreAttributes: false,
766
2128
  parseAttributeValue: false,
767
2129
  parseTagValue: false,
768
- trimValues: false
769
- });
2130
+ trimValues: false,
2131
+ attributeNamePrefix: "@_",
2132
+ textNodeName: "#text"
2133
+ };
770
2134
  GEXF_NAMESPACE = /^https?:\/\/(?:www\.)?gexf.net\/1\.(?:2(?:draft)?|3)(\/viz)?$/;
771
2135
  TEMPORAL_ATTRIBUTES = [
772
2136
  "start",
@@ -795,15 +2159,15 @@ var init_gexf_loader = __esm({
795
2159
  ...GEXFLoader,
796
2160
  parse: async (data, options) => {
797
2161
  var _a;
798
- return parseGEXF(data, ((_a = options == null ? void 0 : options.gexf) == null ? void 0 : _a.shape) ?? "arrow-table");
2162
+ return parseGEXF(data, ((_a = options == null ? void 0 : options.gexf) == null ? void 0 : _a.shape) ?? "arrow");
799
2163
  },
800
2164
  parseSync: (data, options) => {
801
2165
  var _a;
802
- return parseGEXF(data, ((_a = options == null ? void 0 : options.gexf) == null ? void 0 : _a.shape) ?? "arrow-table");
2166
+ return parseGEXF(data, ((_a = options == null ? void 0 : options.gexf) == null ? void 0 : _a.shape) ?? "arrow");
803
2167
  },
804
2168
  parseTextSync: (data, options) => {
805
2169
  var _a;
806
- return parseGEXF(data, ((_a = options == null ? void 0 : options.gexf) == null ? void 0 : _a.shape) ?? "arrow-table");
2170
+ return parseGEXF(data, ((_a = options == null ? void 0 : options.gexf) == null ? void 0 : _a.shape) ?? "arrow");
807
2171
  }
808
2172
  };
809
2173
  }
@@ -817,7 +2181,7 @@ __export(gexf_loader_exports2, {
817
2181
  module.exports = __toCommonJS(gexf_loader_exports2);
818
2182
 
819
2183
  // dist/gexf-loader-types.js
820
- var VERSION2 = true ? "5.0.0-beta.1" : "latest";
2184
+ var VERSION2 = true ? "5.0.0-beta.10" : "latest";
821
2185
  async function preload2() {
822
2186
  const { GEXFLoaderWithParser: GEXFLoaderWithParser3 } = await Promise.resolve().then(() => (init_gexf_loader(), gexf_loader_exports));
823
2187
  return GEXFLoaderWithParser3;
@@ -831,23 +2195,23 @@ var GEXFLoader2 = {
831
2195
  mimeTypes: ["application/gexf+xml"],
832
2196
  text: true,
833
2197
  worker: false,
2198
+ supportsNativeArrow: true,
834
2199
  dataType: null,
835
2200
  batchType: null,
836
- options: { gexf: { shape: "arrow-table" } },
2201
+ options: { gexf: { shape: "arrow" } },
837
2202
  preload: preload2
838
2203
  };
839
2204
 
840
2205
  // dist/lib/parse-gexf.js
841
- var import_fast_xml_parser2 = require("fast-xml-parser");
2206
+ init_xml_parser();
842
2207
 
843
2208
  // dist/lib/build-graph-tables.js
844
2209
  var import_apache_arrow2 = require("apache-arrow");
845
- init_schema_utils();
846
2210
  function buildGraphTables2(nodes, edges, shape, declarations) {
847
2211
  if (shape === "plain-graph-data") {
848
2212
  return { shape, nodes: Array.from(nodes), edges: Array.from(edges) };
849
2213
  }
850
- if (shape !== "arrow-table" && shape !== "object-row-table") {
2214
+ if (shape !== "arrow" && shape !== "object-row-table") {
851
2215
  throw new Error(`Unsupported graph shape: ${shape}`);
852
2216
  }
853
2217
  const nodeTable = buildGraphTable2(nodes, false, shape, declarations == null ? void 0 : declarations.nodes);
@@ -880,8 +2244,7 @@ function buildGraphTable2(records, isEdge, shape, declarations) {
880
2244
  const type = name === "attributes" ? inferAttributeType2(values, declarations) : name === "directed" ? new import_apache_arrow2.Bool() : new import_apache_arrow2.Utf8();
881
2245
  vectors[name] = (0, import_apache_arrow2.vectorFromArray)(values.map((value) => normalizeValue2(value, type)), type);
882
2246
  }
883
- const data = new import_apache_arrow2.Table(vectors);
884
- return { shape, schema: convertArrowToSchema(data.schema), data };
2247
+ return new import_apache_arrow2.Table(vectors);
885
2248
  }
886
2249
  function inferAttributeType2(values, declarations) {
887
2250
  const names = new Set(declarations == null ? void 0 : declarations.keys());
@@ -960,12 +2323,14 @@ function normalizeValue2(value, type) {
960
2323
  }
961
2324
 
962
2325
  // dist/lib/parse-gexf.js
963
- var parser2 = new import_fast_xml_parser2.XMLParser({
2326
+ var XML_OPTIONS2 = {
964
2327
  ignoreAttributes: false,
965
2328
  parseAttributeValue: false,
966
2329
  parseTagValue: false,
967
- trimValues: false
968
- });
2330
+ trimValues: false,
2331
+ attributeNamePrefix: "@_",
2332
+ textNodeName: "#text"
2333
+ };
969
2334
  var GEXF_NAMESPACE2 = /^https?:\/\/(?:www\.)?gexf.net\/1\.(?:2(?:draft)?|3)(\/viz)?$/;
970
2335
  var TEMPORAL_ATTRIBUTES2 = [
971
2336
  "start",
@@ -976,14 +2341,14 @@ var TEMPORAL_ATTRIBUTES2 = [
976
2341
  "timestamps",
977
2342
  "intervals"
978
2343
  ];
979
- function parseGEXF2(input, shape = "arrow-table") {
2344
+ function parseGEXF2(input, shape = "arrow") {
980
2345
  const text = typeof input === "string" ? input : new TextDecoder().decode(input);
981
- const validation = import_fast_xml_parser2.XMLValidator.validate(text);
2346
+ const validation = validateXML(text);
982
2347
  if (validation !== true)
983
- throw new Error(`Invalid GEXF XML: ${validation.err.msg}`);
2348
+ throw new Error(`Invalid GEXF XML: ${validation.message}`);
984
2349
  if (/<!DOCTYPE\s/i.test(text))
985
2350
  throw new Error("GEXF document types are unsupported.");
986
- const document = normalizeNamespaces2(parser2.parse(text), {});
2351
+ const document = normalizeNamespaces2(parseXMLSync(text, XML_OPTIONS2), {});
987
2352
  const root = getElement2(document.gexf, "gexf");
988
2353
  const version = root["@_version"];
989
2354
  if (version && !["1.2", "1.2draft", "1.3"].includes(version)) {
@@ -1290,15 +2655,15 @@ var GEXFLoaderWithParser2 = {
1290
2655
  ...GEXFLoader2,
1291
2656
  parse: async (data, options) => {
1292
2657
  var _a;
1293
- return parseGEXF2(data, ((_a = options == null ? void 0 : options.gexf) == null ? void 0 : _a.shape) ?? "arrow-table");
2658
+ return parseGEXF2(data, ((_a = options == null ? void 0 : options.gexf) == null ? void 0 : _a.shape) ?? "arrow");
1294
2659
  },
1295
2660
  parseSync: (data, options) => {
1296
2661
  var _a;
1297
- return parseGEXF2(data, ((_a = options == null ? void 0 : options.gexf) == null ? void 0 : _a.shape) ?? "arrow-table");
2662
+ return parseGEXF2(data, ((_a = options == null ? void 0 : options.gexf) == null ? void 0 : _a.shape) ?? "arrow");
1298
2663
  },
1299
2664
  parseTextSync: (data, options) => {
1300
2665
  var _a;
1301
- return parseGEXF2(data, ((_a = options == null ? void 0 : options.gexf) == null ? void 0 : _a.shape) ?? "arrow-table");
2666
+ return parseGEXF2(data, ((_a = options == null ? void 0 : options.gexf) == null ? void 0 : _a.shape) ?? "arrow");
1302
2667
  }
1303
2668
  };
1304
2669
  //# sourceMappingURL=gexf-loader.cjs.map