@loaders.gl/graphs 5.0.0-alpha.9 → 5.0.0-beta.1

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 (78) hide show
  1. package/LICENSE +24 -0
  2. package/README.md +121 -4
  3. package/THIRD-PARTY-NOTICES.md +17 -0
  4. package/dist/dot-loader-types.d.ts +25 -8
  5. package/dist/dot-loader-types.d.ts.map +1 -1
  6. package/dist/dot-loader-types.js +2 -2
  7. package/dist/dot-loader-types.js.map +1 -1
  8. package/dist/dot-loader.cjs +638 -47
  9. package/dist/dot-loader.cjs.map +3 -3
  10. package/dist/dot-loader.d.ts +10 -6
  11. package/dist/dot-loader.d.ts.map +1 -1
  12. package/dist/dot-loader.js +3 -3
  13. package/dist/dot-loader.js.map +1 -1
  14. package/dist/dot-types.d.ts +6 -1
  15. package/dist/dot-types.d.ts.map +1 -1
  16. package/dist/gexf-loader-types.d.ts +53 -0
  17. package/dist/gexf-loader-types.d.ts.map +1 -0
  18. package/dist/gexf-loader-types.js +26 -0
  19. package/dist/gexf-loader-types.js.map +1 -0
  20. package/dist/gexf-loader.cjs +1304 -0
  21. package/dist/gexf-loader.cjs.map +6 -0
  22. package/dist/gexf-loader.d.ts +24 -0
  23. package/dist/gexf-loader.d.ts.map +1 -0
  24. package/dist/gexf-loader.js +13 -0
  25. package/dist/gexf-loader.js.map +1 -0
  26. package/dist/gexf-types.d.ts +16 -0
  27. package/dist/gexf-types.d.ts.map +1 -0
  28. package/dist/gexf-types.js +5 -0
  29. package/dist/gexf-types.js.map +1 -0
  30. package/dist/graph-types.d.ts +28 -0
  31. package/dist/graph-types.d.ts.map +1 -1
  32. package/dist/graphml-loader-types.d.ts +24 -8
  33. package/dist/graphml-loader-types.d.ts.map +1 -1
  34. package/dist/graphml-loader-types.js +2 -2
  35. package/dist/graphml-loader-types.js.map +1 -1
  36. package/dist/graphml-loader.cjs +670 -60
  37. package/dist/graphml-loader.cjs.map +3 -3
  38. package/dist/graphml-loader.d.ts +10 -6
  39. package/dist/graphml-loader.d.ts.map +1 -1
  40. package/dist/graphml-loader.js +3 -3
  41. package/dist/graphml-loader.js.map +1 -1
  42. package/dist/index.cjs +927 -59
  43. package/dist/index.cjs.map +3 -3
  44. package/dist/index.d.ts +7 -0
  45. package/dist/index.d.ts.map +1 -1
  46. package/dist/index.js +1 -0
  47. package/dist/index.js.map +1 -1
  48. package/dist/lib/build-graph-tables.d.ts +4 -0
  49. package/dist/lib/build-graph-tables.d.ts.map +1 -0
  50. package/dist/lib/build-graph-tables.js +143 -0
  51. package/dist/lib/build-graph-tables.js.map +1 -0
  52. package/dist/lib/parse-dot.d.ts +3 -1
  53. package/dist/lib/parse-dot.d.ts.map +1 -1
  54. package/dist/lib/parse-dot.js +38 -21
  55. package/dist/lib/parse-dot.js.map +1 -1
  56. package/dist/lib/parse-gexf.d.ts +5 -0
  57. package/dist/lib/parse-gexf.d.ts.map +1 -0
  58. package/dist/lib/parse-gexf.js +380 -0
  59. package/dist/lib/parse-gexf.js.map +1 -0
  60. package/dist/lib/parse-graphml.d.ts +2 -1
  61. package/dist/lib/parse-graphml.d.ts.map +1 -1
  62. package/dist/lib/parse-graphml.js +50 -27
  63. package/dist/lib/parse-graphml.js.map +1 -1
  64. package/package.json +13 -5
  65. package/src/dot-loader-types.ts +15 -5
  66. package/src/dot-loader.ts +7 -4
  67. package/src/dot-types.ts +7 -1
  68. package/src/gexf-loader-types.ts +41 -0
  69. package/src/gexf-loader.ts +17 -0
  70. package/src/gexf-types.ts +21 -0
  71. package/src/graph-types.ts +35 -0
  72. package/src/graphml-loader-types.ts +14 -5
  73. package/src/graphml-loader.ts +7 -4
  74. package/src/index.ts +9 -0
  75. package/src/lib/build-graph-tables.ts +208 -0
  76. package/src/lib/parse-dot.ts +44 -21
  77. package/src/lib/parse-gexf.ts +388 -0
  78. package/src/lib/parse-graphml.ts +86 -34
@@ -1,7 +1,9 @@
1
1
  "use strict";
2
+ var __create = Object.create;
2
3
  var __defProp = Object.defineProperty;
3
4
  var __getOwnPropDesc = Object.getOwnPropertyDescriptor;
4
5
  var __getOwnPropNames = Object.getOwnPropertyNames;
6
+ var __getProtoOf = Object.getPrototypeOf;
5
7
  var __hasOwnProp = Object.prototype.hasOwnProperty;
6
8
  var __esm = (fn, res, err) => function __init() {
7
9
  if (err) throw err[0];
@@ -23,6 +25,14 @@ var __copyProps = (to, from, except, desc) => {
23
25
  }
24
26
  return to;
25
27
  };
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
+ ));
26
36
  var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: true }), mod);
27
37
 
28
38
  // src/graphml-loader-types.ts
@@ -46,14 +56,427 @@ var init_graphml_loader_types = __esm({
46
56
  worker: false,
47
57
  dataType: null,
48
58
  batchType: null,
49
- options: {},
59
+ options: { graphml: { shape: "arrow-table" } },
50
60
  preload
51
61
  };
52
62
  }
53
63
  });
54
64
 
65
+ // ../core/src/schema-utils/lib/schema/convert-arrow-schema.ts
66
+ function convertArrowToSchema(arrowSchema) {
67
+ return serializeArrowSchema(arrowSchema);
68
+ }
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";
89
+ }
90
+ if (isArrowViewType(arrowType, "Utf8View")) {
91
+ return "utf8-view";
92
+ }
93
+ if (isOptionalArrowType(arrowType, "LargeList")) {
94
+ const largeListType = arrowType;
95
+ return {
96
+ type: "large-list",
97
+ children: [serializeArrowField(largeListType.valueField)]
98
+ };
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
109
+ };
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
158
+ };
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";
179
+ }
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";
201
+ }
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";
219
+ }
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: ${arrowType.constructor.name}`);
271
+ }
272
+ }
273
+ function getOptionalArrowConstructor(typeName) {
274
+ return arrow[typeName];
275
+ }
276
+ function isArrowViewType(arrowType, viewTypeName) {
277
+ const constructor = getOptionalArrowConstructor(viewTypeName);
278
+ if (constructor && arrowType instanceof constructor) {
279
+ return true;
280
+ }
281
+ const guards = arrow.DataType;
282
+ const guard = viewTypeName === "BinaryView" ? guards.isBinaryView : guards.isUtf8View;
283
+ return Boolean(guard == null ? void 0 : guard(arrowType));
284
+ }
285
+ function isOptionalArrowType(arrowType, typeName) {
286
+ const constructor = getOptionalArrowConstructor(typeName);
287
+ return Boolean(constructor && arrowType instanceof constructor);
288
+ }
289
+ function serializeArrowDictionaryKeyType(arrowType) {
290
+ if (arrowType instanceof arrow.Int) {
291
+ const prefix = arrowType.isSigned ? "int" : "uint";
292
+ const keyType = `${prefix}${arrowType.bitWidth}`;
293
+ if (isSchemaDictionaryKeyType(keyType)) {
294
+ return keyType;
295
+ }
296
+ }
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}`);
312
+ }
313
+ }
314
+ function isSchemaDictionaryKeyType(keyType) {
315
+ return ["int8", "int16", "int32", "uint8", "uint16", "uint32"].includes(keyType);
316
+ }
317
+ var arrow;
318
+ var init_convert_arrow_schema = __esm({
319
+ "../core/src/schema-utils/lib/schema/convert-arrow-schema.ts"() {
320
+ "use strict";
321
+ arrow = __toESM(require("apache-arrow"), 1);
322
+ }
323
+ });
324
+
325
+ // ../core/src/schema-utils/index.ts
326
+ var init_schema_utils = __esm({
327
+ "../core/src/schema-utils/index.ts"() {
328
+ "use strict";
329
+ init_convert_arrow_schema();
330
+ }
331
+ });
332
+
333
+ // src/lib/build-graph-tables.ts
334
+ function buildGraphTables(nodes, edges, shape, declarations) {
335
+ if (shape === "plain-graph-data") {
336
+ return { shape, nodes: Array.from(nodes), edges: Array.from(edges) };
337
+ }
338
+ if (shape !== "arrow-table" && shape !== "object-row-table") {
339
+ throw new Error(`Unsupported graph shape: ${shape}`);
340
+ }
341
+ const nodeTable = buildGraphTable(nodes, false, shape, declarations == null ? void 0 : declarations.nodes);
342
+ const edgeTable = buildGraphTable(edges, true, shape, declarations == null ? void 0 : declarations.edges);
343
+ return {
344
+ shape: "tables",
345
+ tables: [
346
+ { name: "nodes", table: nodeTable },
347
+ { name: "edges", table: edgeTable }
348
+ ]
349
+ };
350
+ }
351
+ function buildGraphTable(records, isEdge, shape, declarations) {
352
+ const columnNames = isEdge ? ["id", "sourceId", "targetId", "directed", "label", "attributes"] : ["id", "label", "attributes"];
353
+ const columns = new Map(columnNames.map((name) => [name, []]));
354
+ for (const record of records) {
355
+ for (const name of columnNames) {
356
+ const value = record[name];
357
+ columns.get(name).push(
358
+ name === "id" || name === "sourceId" || name === "targetId" ? String(value) : value ?? null
359
+ );
360
+ }
361
+ }
362
+ if (shape === "object-row-table") {
363
+ return {
364
+ shape,
365
+ data: columns.get("id").map(
366
+ (_, index) => Object.fromEntries(columnNames.map((name) => [name, columns.get(name)[index]]))
367
+ )
368
+ };
369
+ }
370
+ const vectors = /* @__PURE__ */ Object.create(null);
371
+ for (const [name, values] of columns) {
372
+ const type = name === "attributes" ? inferAttributeType(values, declarations) : name === "directed" ? new import_apache_arrow.Bool() : new import_apache_arrow.Utf8();
373
+ vectors[name] = (0, import_apache_arrow.vectorFromArray)(
374
+ values.map((value) => normalizeValue(value, type)),
375
+ type
376
+ );
377
+ }
378
+ const data = new import_apache_arrow.Table(vectors);
379
+ return { shape, schema: convertArrowToSchema(data.schema), data };
380
+ }
381
+ function inferAttributeType(values, declarations) {
382
+ const names = new Set(declarations == null ? void 0 : declarations.keys());
383
+ for (const value of values) {
384
+ if (value && typeof value === "object") {
385
+ for (const name of Object.keys(value)) {
386
+ names.add(name);
387
+ }
388
+ }
389
+ }
390
+ return new import_apache_arrow.Struct(
391
+ Array.from(names, (name) => {
392
+ const fieldValues = values.map(
393
+ (value) => (value == null ? void 0 : value[name]) ?? null
394
+ );
395
+ return new import_apache_arrow.Field(name, inferValueType(fieldValues, declarations == null ? void 0 : declarations.get(name)), true);
396
+ })
397
+ );
398
+ }
399
+ function inferValueType(values, declaration) {
400
+ const present = values.filter((value) => value !== null && value !== void 0);
401
+ if (declaration && typeof declaration === "object") {
402
+ if (!present.every((value) => Array.isArray(value))) return new import_apache_arrow.Utf8();
403
+ return new import_apache_arrow.List(
404
+ new import_apache_arrow.Field(
405
+ "item",
406
+ inferValueType(
407
+ present.flatMap((value) => value),
408
+ declaration.list
409
+ ),
410
+ true
411
+ )
412
+ );
413
+ }
414
+ if (declaration) {
415
+ const declaredTypes = {
416
+ boolean: new import_apache_arrow.Bool(),
417
+ int: new import_apache_arrow.Int32(),
418
+ long: new import_apache_arrow.Int64(),
419
+ float: new import_apache_arrow.Float32(),
420
+ double: new import_apache_arrow.Float64(),
421
+ string: new import_apache_arrow.Utf8()
422
+ };
423
+ const valid = present.every((value) => {
424
+ switch (declaration) {
425
+ case "boolean":
426
+ return typeof value === "boolean";
427
+ case "int":
428
+ return typeof value === "number" && Number.isInteger(value) && value >= -2147483648 && value <= 2147483647;
429
+ case "long":
430
+ return typeof value === "bigint" && value >= -(2n ** 63n) && value < 2n ** 63n;
431
+ case "float":
432
+ case "double":
433
+ return typeof value === "number";
434
+ case "string":
435
+ return typeof value === "string";
436
+ }
437
+ });
438
+ return valid ? declaredTypes[declaration] : new import_apache_arrow.Utf8();
439
+ }
440
+ if (present.length && present.every((value) => typeof value === "boolean")) return new import_apache_arrow.Bool();
441
+ if (present.length && present.every((value) => typeof value === "number")) return new import_apache_arrow.Float64();
442
+ if (present.length && present.every((value) => Array.isArray(value))) {
443
+ return new import_apache_arrow.List(
444
+ new import_apache_arrow.Field("item", inferValueType(present.flatMap((value) => value)), true)
445
+ );
446
+ }
447
+ if (present.length && present.every((value) => typeof value === "object" && !Array.isArray(value))) {
448
+ return inferAttributeType(present);
449
+ }
450
+ return new import_apache_arrow.Utf8();
451
+ }
452
+ function normalizeValue(value, type) {
453
+ if (value === null || value === void 0) return null;
454
+ if (type instanceof import_apache_arrow.Struct) {
455
+ return Object.fromEntries(
456
+ type.children.map((field) => [
457
+ field.name,
458
+ normalizeValue(value[field.name], field.type)
459
+ ])
460
+ );
461
+ }
462
+ if (type instanceof import_apache_arrow.List)
463
+ return value.map((item) => normalizeValue(item, type.valueType));
464
+ if (type instanceof import_apache_arrow.Utf8 && typeof value !== "string") {
465
+ return typeof value === "object" ? JSON.stringify(value) : String(value);
466
+ }
467
+ return value;
468
+ }
469
+ var import_apache_arrow;
470
+ var init_build_graph_tables = __esm({
471
+ "src/lib/build-graph-tables.ts"() {
472
+ "use strict";
473
+ import_apache_arrow = require("apache-arrow");
474
+ init_schema_utils();
475
+ }
476
+ });
477
+
55
478
  // src/lib/parse-graphml.ts
56
- function parseGraphML(graphml) {
479
+ function parseGraphML(graphml, shape = "plain-graph-data") {
57
480
  const xmlText = decodeGraphML(graphml);
58
481
  const document = graphmlParser.parse(xmlText);
59
482
  const graphmlRoot = getGraphMLRoot(document);
@@ -64,19 +487,38 @@ function parseGraphML(graphml) {
64
487
  if (!graphElement) {
65
488
  throw new Error("GraphML document does not contain a <graph> element.");
66
489
  }
67
- const keyDefinitions = collectKeyDefinitions(graphmlRoot, graphElement);
490
+ const keyDefinitions = collectKeyDefinitions(
491
+ graphmlRoot,
492
+ graphElement,
493
+ shape !== "plain-graph-data"
494
+ );
68
495
  const defaultDirected = parseEdgeDefault(graphElement[`${XML_ATTRIBUTE_PREFIX}edgedefault`]);
69
- const nodes = normalizeArray(graphElement.node).map((node) => parseNode(node, keyDefinitions));
70
- const edges = normalizeArray(graphElement.edge).map(
71
- (edge, index) => parseEdge(edge, index, keyDefinitions, defaultDirected)
496
+ const declarations = { nodes: /* @__PURE__ */ new Map(), edges: /* @__PURE__ */ new Map() };
497
+ for (const definition of keyDefinitions.values()) {
498
+ if (definition.domain === "all" || definition.domain === "node")
499
+ declarations.nodes.set(definition.name, definition.type);
500
+ if (definition.domain === "all" || definition.domain === "edge")
501
+ declarations.edges.set(definition.name, definition.type);
502
+ }
503
+ return buildGraphTables(
504
+ iterateNodes(graphElement.node, keyDefinitions),
505
+ iterateEdges(graphElement.edge, keyDefinitions, defaultDirected),
506
+ shape,
507
+ declarations
72
508
  );
73
- const filteredNodes = nodes.filter((node) => Boolean(node));
74
- const filteredEdges = edges.filter((edge) => Boolean(edge));
75
- return {
76
- shape: "plain-graph-data",
77
- nodes: filteredNodes,
78
- edges: filteredEdges
79
- };
509
+ }
510
+ function* iterateNodes(nodes, definitions) {
511
+ for (const node of normalizeArray(nodes)) {
512
+ const record = parseNode(node, definitions);
513
+ if (record) yield record;
514
+ }
515
+ }
516
+ function* iterateEdges(edges, definitions, directed) {
517
+ let index = 0;
518
+ for (const edge of normalizeArray(edges)) {
519
+ const record = parseEdge(edge, index++, definitions, directed);
520
+ if (record) yield record;
521
+ }
80
522
  }
81
523
  function decodeGraphML(graphml) {
82
524
  if (typeof graphml === "string") {
@@ -99,16 +541,21 @@ function getGraphMLRoot(document) {
99
541
  }
100
542
  function getGraphElement(graphmlRoot) {
101
543
  const graph = graphmlRoot.graph;
544
+ if (typeof graph === "string" && !graph.trim()) {
545
+ return {};
546
+ }
102
547
  if (!graph) {
103
548
  return null;
104
549
  }
105
550
  if (Array.isArray(graph)) {
106
- const firstGraph = graph.find((entry) => isObject(entry));
107
- return isObject(firstGraph) ? firstGraph : null;
551
+ const firstGraph = graph.find(
552
+ (entry) => isObject(entry) || typeof entry === "string" && !entry.trim()
553
+ );
554
+ return isObject(firstGraph) ? firstGraph : typeof firstGraph === "string" ? {} : null;
108
555
  }
109
556
  return isObject(graph) ? graph : null;
110
557
  }
111
- function collectKeyDefinitions(graphmlRoot, graphElement) {
558
+ function collectKeyDefinitions(graphmlRoot, graphElement, lossless) {
112
559
  const keys = /* @__PURE__ */ new Map();
113
560
  for (const candidate of [
114
561
  ...normalizeArray(graphmlRoot.key),
@@ -121,8 +568,8 @@ function collectKeyDefinitions(graphmlRoot, graphElement) {
121
568
  const name = String(candidate[`${XML_ATTRIBUTE_PREFIX}attr.name`] ?? id).trim();
122
569
  const type = normalizeType(candidate[`${XML_ATTRIBUTE_PREFIX}attr.type`]);
123
570
  const defaultNode = candidate.default ?? null;
124
- const defaultValue = defaultNode !== null ? castDataValue(defaultNode, type) : void 0;
125
- keys.set(id, { id, domain, name, type, defaultValue });
571
+ const defaultValue = defaultNode !== null ? castDataValue(defaultNode, type, lossless) : void 0;
572
+ keys.set(id, { id, domain, name, type, defaultValue, lossless });
126
573
  }
127
574
  }
128
575
  }
@@ -200,12 +647,12 @@ function assignAttributeFromDataEntry(entry, keyDefinitions, attributes) {
200
647
  }
201
648
  const definition = keyDefinitions.get(keyId);
202
649
  const attributeName = (definition == null ? void 0 : definition.name) ?? keyId;
203
- const value = castDataValue(entry, (definition == null ? void 0 : definition.type) ?? "string");
650
+ const value = (definition == null ? void 0 : definition.type) === "boolean" && extractTextContent(entry) === void 0 ? false : castDataValue(entry, (definition == null ? void 0 : definition.type) ?? "string", definition == null ? void 0 : definition.lossless);
204
651
  if (value !== void 0) {
205
652
  attributes[attributeName] = value;
206
653
  }
207
654
  }
208
- function castDataValue(value, type) {
655
+ function castDataValue(value, type, lossless = false) {
209
656
  if (value === null || typeof value === "undefined") {
210
657
  return void 0;
211
658
  }
@@ -216,6 +663,9 @@ function castDataValue(value, type) {
216
663
  if (type === "boolean") {
217
664
  return parseBoolean(text);
218
665
  }
666
+ if (type === "long" && lossless) {
667
+ return /^[+-]?\d+$/.test(text.trim()) ? BigInt(text.trim()) : text;
668
+ }
219
669
  if (type === "int" || type === "long") {
220
670
  const parsed = Number.parseInt(text, 10);
221
671
  return Number.isNaN(parsed) ? text : parsed;
@@ -238,14 +688,14 @@ function extractTextContent(value) {
238
688
  }
239
689
  if (isObject(value)) {
240
690
  const text = value[XML_TEXT_KEY];
241
- if (typeof text === "string") {
242
- return text;
243
- }
244
691
  const nonAttributeEntries = Object.entries(value).filter(
245
- ([key]) => !key.startsWith(XML_ATTRIBUTE_PREFIX)
692
+ ([key]) => !key.startsWith(XML_ATTRIBUTE_PREFIX) && key !== XML_TEXT_KEY
246
693
  );
247
694
  if (nonAttributeEntries.length === 0) {
248
- return void 0;
695
+ return typeof text === "string" ? text : void 0;
696
+ }
697
+ if (typeof text === "string" && text.trim()) {
698
+ nonAttributeEntries.push([XML_TEXT_KEY, text]);
249
699
  }
250
700
  return JSON.stringify(Object.fromEntries(nonAttributeEntries));
251
701
  }
@@ -306,20 +756,21 @@ function parseBoolean(value) {
306
756
  if (normalized === "false" || normalized === "0" || normalized === "no" || normalized === "n") {
307
757
  return false;
308
758
  }
309
- return Boolean(value);
759
+ return Boolean(normalized);
310
760
  }
311
761
  var import_fast_xml_parser, XML_ATTRIBUTE_PREFIX, XML_TEXT_KEY, graphmlParser;
312
762
  var init_parse_graphml = __esm({
313
763
  "src/lib/parse-graphml.ts"() {
314
764
  "use strict";
315
765
  import_fast_xml_parser = require("fast-xml-parser");
766
+ init_build_graph_tables();
316
767
  XML_ATTRIBUTE_PREFIX = "@_";
317
768
  XML_TEXT_KEY = "#text";
318
769
  graphmlParser = new import_fast_xml_parser.XMLParser({
319
770
  ignoreAttributes: false,
320
771
  attributeNamePrefix: XML_ATTRIBUTE_PREFIX,
321
772
  textNodeName: XML_TEXT_KEY,
322
- trimValues: true,
773
+ trimValues: false,
323
774
  parseAttributeValue: false,
324
775
  parseTagValue: false,
325
776
  removeNSPrefix: true
@@ -340,9 +791,18 @@ var init_graphml_loader = __esm({
340
791
  init_parse_graphml();
341
792
  GraphMLLoaderWithParser = {
342
793
  ...GraphMLLoader,
343
- parse: async (data) => parseGraphML(data),
344
- parseSync: parseGraphML,
345
- parseTextSync: parseGraphML
794
+ parse: async (data, options) => {
795
+ var _a;
796
+ return parseGraphML(data, ((_a = options == null ? void 0 : options.graphml) == null ? void 0 : _a.shape) ?? "arrow-table");
797
+ },
798
+ parseSync: (data, options) => {
799
+ var _a;
800
+ return parseGraphML(data, ((_a = options == null ? void 0 : options.graphml) == null ? void 0 : _a.shape) ?? "arrow-table");
801
+ },
802
+ parseTextSync: (data, options) => {
803
+ var _a;
804
+ return parseGraphML(data, ((_a = options == null ? void 0 : options.graphml) == null ? void 0 : _a.shape) ?? "arrow-table");
805
+ }
346
806
  };
347
807
  }
348
808
  });
@@ -355,7 +815,7 @@ __export(graphml_loader_exports2, {
355
815
  module.exports = __toCommonJS(graphml_loader_exports2);
356
816
 
357
817
  // dist/graphml-loader-types.js
358
- var VERSION2 = true ? "5.0.0-alpha.9" : "latest";
818
+ var VERSION2 = true ? "5.0.0-beta.1" : "latest";
359
819
  async function preload2() {
360
820
  const { GraphMLLoaderWithParser: GraphMLLoaderWithParser3 } = await Promise.resolve().then(() => (init_graphml_loader(), graphml_loader_exports));
361
821
  return GraphMLLoaderWithParser3;
@@ -371,24 +831,145 @@ var GraphMLLoader2 = {
371
831
  worker: false,
372
832
  dataType: null,
373
833
  batchType: null,
374
- options: {},
834
+ options: { graphml: { shape: "arrow-table" } },
375
835
  preload: preload2
376
836
  };
377
837
 
378
838
  // dist/lib/parse-graphml.js
379
839
  var import_fast_xml_parser2 = require("fast-xml-parser");
840
+
841
+ // dist/lib/build-graph-tables.js
842
+ var import_apache_arrow2 = require("apache-arrow");
843
+ init_schema_utils();
844
+ function buildGraphTables2(nodes, edges, shape, declarations) {
845
+ if (shape === "plain-graph-data") {
846
+ return { shape, nodes: Array.from(nodes), edges: Array.from(edges) };
847
+ }
848
+ if (shape !== "arrow-table" && shape !== "object-row-table") {
849
+ throw new Error(`Unsupported graph shape: ${shape}`);
850
+ }
851
+ const nodeTable = buildGraphTable2(nodes, false, shape, declarations == null ? void 0 : declarations.nodes);
852
+ const edgeTable = buildGraphTable2(edges, true, shape, declarations == null ? void 0 : declarations.edges);
853
+ return {
854
+ shape: "tables",
855
+ tables: [
856
+ { name: "nodes", table: nodeTable },
857
+ { name: "edges", table: edgeTable }
858
+ ]
859
+ };
860
+ }
861
+ function buildGraphTable2(records, isEdge, shape, declarations) {
862
+ const columnNames = isEdge ? ["id", "sourceId", "targetId", "directed", "label", "attributes"] : ["id", "label", "attributes"];
863
+ const columns = new Map(columnNames.map((name) => [name, []]));
864
+ for (const record of records) {
865
+ for (const name of columnNames) {
866
+ const value = record[name];
867
+ columns.get(name).push(name === "id" || name === "sourceId" || name === "targetId" ? String(value) : value ?? null);
868
+ }
869
+ }
870
+ if (shape === "object-row-table") {
871
+ return {
872
+ shape,
873
+ data: columns.get("id").map((_, index) => Object.fromEntries(columnNames.map((name) => [name, columns.get(name)[index]])))
874
+ };
875
+ }
876
+ const vectors = /* @__PURE__ */ Object.create(null);
877
+ for (const [name, values] of columns) {
878
+ const type = name === "attributes" ? inferAttributeType2(values, declarations) : name === "directed" ? new import_apache_arrow2.Bool() : new import_apache_arrow2.Utf8();
879
+ vectors[name] = (0, import_apache_arrow2.vectorFromArray)(values.map((value) => normalizeValue2(value, type)), type);
880
+ }
881
+ const data = new import_apache_arrow2.Table(vectors);
882
+ return { shape, schema: convertArrowToSchema(data.schema), data };
883
+ }
884
+ function inferAttributeType2(values, declarations) {
885
+ const names = new Set(declarations == null ? void 0 : declarations.keys());
886
+ for (const value of values) {
887
+ if (value && typeof value === "object") {
888
+ for (const name of Object.keys(value)) {
889
+ names.add(name);
890
+ }
891
+ }
892
+ }
893
+ return new import_apache_arrow2.Struct(Array.from(names, (name) => {
894
+ const fieldValues = values.map((value) => (value == null ? void 0 : value[name]) ?? null);
895
+ return new import_apache_arrow2.Field(name, inferValueType2(fieldValues, declarations == null ? void 0 : declarations.get(name)), true);
896
+ }));
897
+ }
898
+ function inferValueType2(values, declaration) {
899
+ const present = values.filter((value) => value !== null && value !== void 0);
900
+ if (declaration && typeof declaration === "object") {
901
+ if (!present.every((value) => Array.isArray(value)))
902
+ return new import_apache_arrow2.Utf8();
903
+ return new import_apache_arrow2.List(new import_apache_arrow2.Field("item", inferValueType2(present.flatMap((value) => value), declaration.list), true));
904
+ }
905
+ if (declaration) {
906
+ const declaredTypes = {
907
+ boolean: new import_apache_arrow2.Bool(),
908
+ int: new import_apache_arrow2.Int32(),
909
+ long: new import_apache_arrow2.Int64(),
910
+ float: new import_apache_arrow2.Float32(),
911
+ double: new import_apache_arrow2.Float64(),
912
+ string: new import_apache_arrow2.Utf8()
913
+ };
914
+ const valid = present.every((value) => {
915
+ switch (declaration) {
916
+ case "boolean":
917
+ return typeof value === "boolean";
918
+ case "int":
919
+ return typeof value === "number" && Number.isInteger(value) && value >= -2147483648 && value <= 2147483647;
920
+ case "long":
921
+ return typeof value === "bigint" && value >= -(2n ** 63n) && value < 2n ** 63n;
922
+ case "float":
923
+ case "double":
924
+ return typeof value === "number";
925
+ case "string":
926
+ return typeof value === "string";
927
+ }
928
+ });
929
+ return valid ? declaredTypes[declaration] : new import_apache_arrow2.Utf8();
930
+ }
931
+ if (present.length && present.every((value) => typeof value === "boolean"))
932
+ return new import_apache_arrow2.Bool();
933
+ if (present.length && present.every((value) => typeof value === "number"))
934
+ return new import_apache_arrow2.Float64();
935
+ if (present.length && present.every((value) => Array.isArray(value))) {
936
+ return new import_apache_arrow2.List(new import_apache_arrow2.Field("item", inferValueType2(present.flatMap((value) => value)), true));
937
+ }
938
+ if (present.length && present.every((value) => typeof value === "object" && !Array.isArray(value))) {
939
+ return inferAttributeType2(present);
940
+ }
941
+ return new import_apache_arrow2.Utf8();
942
+ }
943
+ function normalizeValue2(value, type) {
944
+ if (value === null || value === void 0)
945
+ return null;
946
+ if (type instanceof import_apache_arrow2.Struct) {
947
+ return Object.fromEntries(type.children.map((field) => [
948
+ field.name,
949
+ normalizeValue2(value[field.name], field.type)
950
+ ]));
951
+ }
952
+ if (type instanceof import_apache_arrow2.List)
953
+ return value.map((item) => normalizeValue2(item, type.valueType));
954
+ if (type instanceof import_apache_arrow2.Utf8 && typeof value !== "string") {
955
+ return typeof value === "object" ? JSON.stringify(value) : String(value);
956
+ }
957
+ return value;
958
+ }
959
+
960
+ // dist/lib/parse-graphml.js
380
961
  var XML_ATTRIBUTE_PREFIX2 = "@_";
381
962
  var XML_TEXT_KEY2 = "#text";
382
963
  var graphmlParser2 = new import_fast_xml_parser2.XMLParser({
383
964
  ignoreAttributes: false,
384
965
  attributeNamePrefix: XML_ATTRIBUTE_PREFIX2,
385
966
  textNodeName: XML_TEXT_KEY2,
386
- trimValues: true,
967
+ trimValues: false,
387
968
  parseAttributeValue: false,
388
969
  parseTagValue: false,
389
970
  removeNSPrefix: true
390
971
  });
391
- function parseGraphML2(graphml) {
972
+ function parseGraphML2(graphml, shape = "plain-graph-data") {
392
973
  const xmlText = decodeGraphML2(graphml);
393
974
  const document = graphmlParser2.parse(xmlText);
394
975
  const graphmlRoot = getGraphMLRoot2(document);
@@ -399,17 +980,31 @@ function parseGraphML2(graphml) {
399
980
  if (!graphElement) {
400
981
  throw new Error("GraphML document does not contain a <graph> element.");
401
982
  }
402
- const keyDefinitions = collectKeyDefinitions2(graphmlRoot, graphElement);
983
+ const keyDefinitions = collectKeyDefinitions2(graphmlRoot, graphElement, shape !== "plain-graph-data");
403
984
  const defaultDirected = parseEdgeDefault2(graphElement[`${XML_ATTRIBUTE_PREFIX2}edgedefault`]);
404
- const nodes = normalizeArray2(graphElement.node).map((node) => parseNode2(node, keyDefinitions));
405
- const edges = normalizeArray2(graphElement.edge).map((edge, index) => parseEdge2(edge, index, keyDefinitions, defaultDirected));
406
- const filteredNodes = nodes.filter((node) => Boolean(node));
407
- const filteredEdges = edges.filter((edge) => Boolean(edge));
408
- return {
409
- shape: "plain-graph-data",
410
- nodes: filteredNodes,
411
- edges: filteredEdges
412
- };
985
+ const declarations = { nodes: /* @__PURE__ */ new Map(), edges: /* @__PURE__ */ new Map() };
986
+ for (const definition of keyDefinitions.values()) {
987
+ if (definition.domain === "all" || definition.domain === "node")
988
+ declarations.nodes.set(definition.name, definition.type);
989
+ if (definition.domain === "all" || definition.domain === "edge")
990
+ declarations.edges.set(definition.name, definition.type);
991
+ }
992
+ return buildGraphTables2(iterateNodes2(graphElement.node, keyDefinitions), iterateEdges2(graphElement.edge, keyDefinitions, defaultDirected), shape, declarations);
993
+ }
994
+ function* iterateNodes2(nodes, definitions) {
995
+ for (const node of normalizeArray2(nodes)) {
996
+ const record = parseNode2(node, definitions);
997
+ if (record)
998
+ yield record;
999
+ }
1000
+ }
1001
+ function* iterateEdges2(edges, definitions, directed) {
1002
+ let index = 0;
1003
+ for (const edge of normalizeArray2(edges)) {
1004
+ const record = parseEdge2(edge, index++, definitions, directed);
1005
+ if (record)
1006
+ yield record;
1007
+ }
413
1008
  }
414
1009
  function decodeGraphML2(graphml) {
415
1010
  if (typeof graphml === "string") {
@@ -432,16 +1027,19 @@ function getGraphMLRoot2(document) {
432
1027
  }
433
1028
  function getGraphElement2(graphmlRoot) {
434
1029
  const graph = graphmlRoot.graph;
1030
+ if (typeof graph === "string" && !graph.trim()) {
1031
+ return {};
1032
+ }
435
1033
  if (!graph) {
436
1034
  return null;
437
1035
  }
438
1036
  if (Array.isArray(graph)) {
439
- const firstGraph = graph.find((entry) => isObject2(entry));
440
- return isObject2(firstGraph) ? firstGraph : null;
1037
+ const firstGraph = graph.find((entry) => isObject2(entry) || typeof entry === "string" && !entry.trim());
1038
+ return isObject2(firstGraph) ? firstGraph : typeof firstGraph === "string" ? {} : null;
441
1039
  }
442
1040
  return isObject2(graph) ? graph : null;
443
1041
  }
444
- function collectKeyDefinitions2(graphmlRoot, graphElement) {
1042
+ function collectKeyDefinitions2(graphmlRoot, graphElement, lossless) {
445
1043
  const keys = /* @__PURE__ */ new Map();
446
1044
  for (const candidate of [
447
1045
  ...normalizeArray2(graphmlRoot.key),
@@ -454,8 +1052,8 @@ function collectKeyDefinitions2(graphmlRoot, graphElement) {
454
1052
  const name = String(candidate[`${XML_ATTRIBUTE_PREFIX2}attr.name`] ?? id).trim();
455
1053
  const type = normalizeType2(candidate[`${XML_ATTRIBUTE_PREFIX2}attr.type`]);
456
1054
  const defaultNode = candidate.default ?? null;
457
- const defaultValue = defaultNode !== null ? castDataValue2(defaultNode, type) : void 0;
458
- keys.set(id, { id, domain, name, type, defaultValue });
1055
+ const defaultValue = defaultNode !== null ? castDataValue2(defaultNode, type, lossless) : void 0;
1056
+ keys.set(id, { id, domain, name, type, defaultValue, lossless });
459
1057
  }
460
1058
  }
461
1059
  }
@@ -533,12 +1131,12 @@ function assignAttributeFromDataEntry2(entry, keyDefinitions, attributes) {
533
1131
  }
534
1132
  const definition = keyDefinitions.get(keyId);
535
1133
  const attributeName = (definition == null ? void 0 : definition.name) ?? keyId;
536
- const value = castDataValue2(entry, (definition == null ? void 0 : definition.type) ?? "string");
1134
+ const value = (definition == null ? void 0 : definition.type) === "boolean" && extractTextContent2(entry) === void 0 ? false : castDataValue2(entry, (definition == null ? void 0 : definition.type) ?? "string", definition == null ? void 0 : definition.lossless);
537
1135
  if (value !== void 0) {
538
1136
  attributes[attributeName] = value;
539
1137
  }
540
1138
  }
541
- function castDataValue2(value, type) {
1139
+ function castDataValue2(value, type, lossless = false) {
542
1140
  if (value === null || typeof value === "undefined") {
543
1141
  return void 0;
544
1142
  }
@@ -549,6 +1147,9 @@ function castDataValue2(value, type) {
549
1147
  if (type === "boolean") {
550
1148
  return parseBoolean2(text);
551
1149
  }
1150
+ if (type === "long" && lossless) {
1151
+ return /^[+-]?\d+$/.test(text.trim()) ? BigInt(text.trim()) : text;
1152
+ }
552
1153
  if (type === "int" || type === "long") {
553
1154
  const parsed = Number.parseInt(text, 10);
554
1155
  return Number.isNaN(parsed) ? text : parsed;
@@ -571,12 +1172,12 @@ function extractTextContent2(value) {
571
1172
  }
572
1173
  if (isObject2(value)) {
573
1174
  const text = value[XML_TEXT_KEY2];
574
- if (typeof text === "string") {
575
- return text;
576
- }
577
- const nonAttributeEntries = Object.entries(value).filter(([key]) => !key.startsWith(XML_ATTRIBUTE_PREFIX2));
1175
+ const nonAttributeEntries = Object.entries(value).filter(([key]) => !key.startsWith(XML_ATTRIBUTE_PREFIX2) && key !== XML_TEXT_KEY2);
578
1176
  if (nonAttributeEntries.length === 0) {
579
- return void 0;
1177
+ return typeof text === "string" ? text : void 0;
1178
+ }
1179
+ if (typeof text === "string" && text.trim()) {
1180
+ nonAttributeEntries.push([XML_TEXT_KEY2, text]);
580
1181
  }
581
1182
  return JSON.stringify(Object.fromEntries(nonAttributeEntries));
582
1183
  }
@@ -637,14 +1238,23 @@ function parseBoolean2(value) {
637
1238
  if (normalized === "false" || normalized === "0" || normalized === "no" || normalized === "n") {
638
1239
  return false;
639
1240
  }
640
- return Boolean(value);
1241
+ return Boolean(normalized);
641
1242
  }
642
1243
 
643
1244
  // dist/graphml-loader.js
644
1245
  var GraphMLLoaderWithParser2 = {
645
1246
  ...GraphMLLoader2,
646
- parse: async (data) => parseGraphML2(data),
647
- parseSync: parseGraphML2,
648
- parseTextSync: parseGraphML2
1247
+ parse: async (data, options) => {
1248
+ var _a;
1249
+ return parseGraphML2(data, ((_a = options == null ? void 0 : options.graphml) == null ? void 0 : _a.shape) ?? "arrow-table");
1250
+ },
1251
+ parseSync: (data, options) => {
1252
+ var _a;
1253
+ return parseGraphML2(data, ((_a = options == null ? void 0 : options.graphml) == null ? void 0 : _a.shape) ?? "arrow-table");
1254
+ },
1255
+ parseTextSync: (data, options) => {
1256
+ var _a;
1257
+ return parseGraphML2(data, ((_a = options == null ? void 0 : options.graphml) == null ? void 0 : _a.shape) ?? "arrow-table");
1258
+ }
649
1259
  };
650
1260
  //# sourceMappingURL=graphml-loader.cjs.map