docling-core 0.0.1__py3-none-any.whl
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.
Potentially problematic release.
This version of docling-core might be problematic. Click here for more details.
- docling_core/__init__.py +6 -0
- docling_core/py.typed +0 -0
- docling_core/resources/schemas/doc/ANN.json +171 -0
- docling_core/resources/schemas/doc/DOC.json +300 -0
- docling_core/resources/schemas/doc/OCR-output.json +166 -0
- docling_core/resources/schemas/doc/RAW.json +158 -0
- docling_core/resources/schemas/generated/ccs_document_schema.json +1071 -0
- docling_core/resources/schemas/generated/minimal_document_schema_flat.json +1129 -0
- docling_core/resources/schemas/search/search_doc_mapping.json +104 -0
- docling_core/resources/schemas/search/search_doc_mapping_v2.json +256 -0
- docling_core/search/__init__.py +6 -0
- docling_core/search/json_schema_to_search_mapper.py +406 -0
- docling_core/search/mapping.py +29 -0
- docling_core/search/meta.py +93 -0
- docling_core/search/package.py +56 -0
- docling_core/types/__init__.py +25 -0
- docling_core/types/base.py +248 -0
- docling_core/types/doc/__init__.py +6 -0
- docling_core/types/doc/base.py +199 -0
- docling_core/types/doc/doc_ann.py +76 -0
- docling_core/types/doc/doc_ocr.py +83 -0
- docling_core/types/doc/doc_raw.py +187 -0
- docling_core/types/doc/document.py +393 -0
- docling_core/types/gen/__init__.py +6 -0
- docling_core/types/gen/generic.py +33 -0
- docling_core/types/nlp/__init__.py +6 -0
- docling_core/types/nlp/qa.py +74 -0
- docling_core/types/nlp/qa_labels.py +118 -0
- docling_core/types/rec/__init__.py +6 -0
- docling_core/types/rec/attribute.py +55 -0
- docling_core/types/rec/base.py +90 -0
- docling_core/types/rec/predicate.py +133 -0
- docling_core/types/rec/record.py +95 -0
- docling_core/types/rec/statement.py +41 -0
- docling_core/types/rec/subject.py +77 -0
- docling_core/utils/__init__.py +6 -0
- docling_core/utils/alias.py +27 -0
- docling_core/utils/ds_generate_docs.py +144 -0
- docling_core/utils/ds_generate_jsonschema.py +62 -0
- docling_core/utils/validate.py +86 -0
- docling_core/utils/validators.py +100 -0
- docling_core-0.0.1.dist-info/LICENSE +21 -0
- docling_core-0.0.1.dist-info/METADATA +133 -0
- docling_core-0.0.1.dist-info/RECORD +46 -0
- docling_core-0.0.1.dist-info/WHEEL +4 -0
- docling_core-0.0.1.dist-info/entry_points.txt +5 -0
|
@@ -0,0 +1,1071 @@
|
|
|
1
|
+
{
|
|
2
|
+
"title": "ExportedCCSDocument",
|
|
3
|
+
"type": "object",
|
|
4
|
+
"properties": {
|
|
5
|
+
"name": {
|
|
6
|
+
"title": "Name",
|
|
7
|
+
"type": "string"
|
|
8
|
+
},
|
|
9
|
+
"type": {
|
|
10
|
+
"title": "Type",
|
|
11
|
+
"default": "pdf-document",
|
|
12
|
+
"x-es": {
|
|
13
|
+
"type": "keyword",
|
|
14
|
+
"ignore_above": 8191
|
|
15
|
+
},
|
|
16
|
+
"type": "string"
|
|
17
|
+
},
|
|
18
|
+
"description": {
|
|
19
|
+
"$ref": "#/definitions/CCSDocumentDescription"
|
|
20
|
+
},
|
|
21
|
+
"file-info": {
|
|
22
|
+
"$ref": "#/definitions/CCSFileInfoObject"
|
|
23
|
+
},
|
|
24
|
+
"main-text": {
|
|
25
|
+
"title": "Main-Text",
|
|
26
|
+
"type": "array",
|
|
27
|
+
"items": {
|
|
28
|
+
"anyOf": [
|
|
29
|
+
{
|
|
30
|
+
"$ref": "#/definitions/Ref"
|
|
31
|
+
},
|
|
32
|
+
{
|
|
33
|
+
"$ref": "#/definitions/RecursiveList"
|
|
34
|
+
},
|
|
35
|
+
{
|
|
36
|
+
"$ref": "#/definitions/BaseText"
|
|
37
|
+
}
|
|
38
|
+
]
|
|
39
|
+
}
|
|
40
|
+
},
|
|
41
|
+
"figures": {
|
|
42
|
+
"title": "Figures",
|
|
43
|
+
"type": "array",
|
|
44
|
+
"items": {
|
|
45
|
+
"$ref": "#/definitions/BaseCell"
|
|
46
|
+
}
|
|
47
|
+
},
|
|
48
|
+
"tables": {
|
|
49
|
+
"title": "Tables",
|
|
50
|
+
"type": "array",
|
|
51
|
+
"items": {
|
|
52
|
+
"$ref": "#/definitions/Table"
|
|
53
|
+
}
|
|
54
|
+
},
|
|
55
|
+
"bitmaps": {
|
|
56
|
+
"title": "Bitmaps",
|
|
57
|
+
"type": "array",
|
|
58
|
+
"items": {
|
|
59
|
+
"$ref": "#/definitions/BitmapObject"
|
|
60
|
+
}
|
|
61
|
+
},
|
|
62
|
+
"equations": {
|
|
63
|
+
"title": "Equations",
|
|
64
|
+
"type": "array",
|
|
65
|
+
"items": {
|
|
66
|
+
"$ref": "#/definitions/BaseCell"
|
|
67
|
+
}
|
|
68
|
+
},
|
|
69
|
+
"footnotes": {
|
|
70
|
+
"title": "Footnotes",
|
|
71
|
+
"type": "array",
|
|
72
|
+
"items": {
|
|
73
|
+
"$ref": "#/definitions/BaseText"
|
|
74
|
+
}
|
|
75
|
+
},
|
|
76
|
+
"page-dimensions": {
|
|
77
|
+
"title": "Page-Dimensions",
|
|
78
|
+
"type": "array",
|
|
79
|
+
"items": {
|
|
80
|
+
"$ref": "#/definitions/PageDimensions"
|
|
81
|
+
}
|
|
82
|
+
},
|
|
83
|
+
"page-footers": {
|
|
84
|
+
"title": "Page-Footers",
|
|
85
|
+
"type": "array",
|
|
86
|
+
"items": {
|
|
87
|
+
"$ref": "#/definitions/BaseText"
|
|
88
|
+
}
|
|
89
|
+
},
|
|
90
|
+
"page-headers": {
|
|
91
|
+
"title": "Page-Headers",
|
|
92
|
+
"type": "array",
|
|
93
|
+
"items": {
|
|
94
|
+
"$ref": "#/definitions/BaseText"
|
|
95
|
+
}
|
|
96
|
+
},
|
|
97
|
+
"_s3_data": {
|
|
98
|
+
"$ref": "#/definitions/S3Data"
|
|
99
|
+
}
|
|
100
|
+
},
|
|
101
|
+
"required": [
|
|
102
|
+
"name",
|
|
103
|
+
"description",
|
|
104
|
+
"file-info",
|
|
105
|
+
"main-text"
|
|
106
|
+
],
|
|
107
|
+
"definitions": {
|
|
108
|
+
"Affiliation": {
|
|
109
|
+
"title": "Affiliation",
|
|
110
|
+
"type": "object",
|
|
111
|
+
"properties": {
|
|
112
|
+
"name": {
|
|
113
|
+
"title": "Name",
|
|
114
|
+
"x-es": {
|
|
115
|
+
"fields": {
|
|
116
|
+
"lower": {
|
|
117
|
+
"normalizer": "lowercase_asciifolding",
|
|
118
|
+
"type": "keyword"
|
|
119
|
+
},
|
|
120
|
+
"keyword": {
|
|
121
|
+
"type": "keyword"
|
|
122
|
+
}
|
|
123
|
+
},
|
|
124
|
+
"type": "keyword"
|
|
125
|
+
},
|
|
126
|
+
"type": "string"
|
|
127
|
+
},
|
|
128
|
+
"id": {
|
|
129
|
+
"title": "Id",
|
|
130
|
+
"x-es": {
|
|
131
|
+
"type": "keyword",
|
|
132
|
+
"ignore_above": 8191
|
|
133
|
+
},
|
|
134
|
+
"type": "string"
|
|
135
|
+
},
|
|
136
|
+
"source": {
|
|
137
|
+
"title": "Source",
|
|
138
|
+
"x-es": {
|
|
139
|
+
"type": "keyword",
|
|
140
|
+
"ignore_above": 8191
|
|
141
|
+
},
|
|
142
|
+
"type": "string"
|
|
143
|
+
}
|
|
144
|
+
},
|
|
145
|
+
"required": [
|
|
146
|
+
"name",
|
|
147
|
+
"id",
|
|
148
|
+
"source"
|
|
149
|
+
]
|
|
150
|
+
},
|
|
151
|
+
"Author": {
|
|
152
|
+
"title": "Author",
|
|
153
|
+
"type": "object",
|
|
154
|
+
"properties": {
|
|
155
|
+
"name": {
|
|
156
|
+
"title": "Name",
|
|
157
|
+
"x-es": {
|
|
158
|
+
"fields": {
|
|
159
|
+
"lower": {
|
|
160
|
+
"normalizer": "lowercase_asciifolding",
|
|
161
|
+
"type": "keyword"
|
|
162
|
+
},
|
|
163
|
+
"keyword": {
|
|
164
|
+
"type": "keyword"
|
|
165
|
+
}
|
|
166
|
+
},
|
|
167
|
+
"type": "keyword"
|
|
168
|
+
},
|
|
169
|
+
"type": "string"
|
|
170
|
+
},
|
|
171
|
+
"id": {
|
|
172
|
+
"title": "Id",
|
|
173
|
+
"x-es": {
|
|
174
|
+
"type": "keyword",
|
|
175
|
+
"ignore_above": 8191
|
|
176
|
+
},
|
|
177
|
+
"type": "string"
|
|
178
|
+
},
|
|
179
|
+
"source": {
|
|
180
|
+
"title": "Source",
|
|
181
|
+
"x-es": {
|
|
182
|
+
"type": "keyword",
|
|
183
|
+
"ignore_above": 8191
|
|
184
|
+
},
|
|
185
|
+
"type": "string"
|
|
186
|
+
},
|
|
187
|
+
"affiliations": {
|
|
188
|
+
"title": "Affiliations",
|
|
189
|
+
"type": "array",
|
|
190
|
+
"items": {
|
|
191
|
+
"$ref": "#/definitions/Affiliation"
|
|
192
|
+
}
|
|
193
|
+
}
|
|
194
|
+
},
|
|
195
|
+
"required": [
|
|
196
|
+
"name",
|
|
197
|
+
"id",
|
|
198
|
+
"source",
|
|
199
|
+
"affiliations"
|
|
200
|
+
]
|
|
201
|
+
},
|
|
202
|
+
"Identifier": {
|
|
203
|
+
"title": "Identifier",
|
|
204
|
+
"type": "object",
|
|
205
|
+
"properties": {
|
|
206
|
+
"type": {
|
|
207
|
+
"title": "Type",
|
|
208
|
+
"x-es": {
|
|
209
|
+
"type": "keyword",
|
|
210
|
+
"ignore_above": 8191
|
|
211
|
+
},
|
|
212
|
+
"type": "string"
|
|
213
|
+
},
|
|
214
|
+
"value": {
|
|
215
|
+
"title": "Value",
|
|
216
|
+
"x-es": {
|
|
217
|
+
"type": "keyword",
|
|
218
|
+
"ignore_above": 8191
|
|
219
|
+
},
|
|
220
|
+
"type": "string"
|
|
221
|
+
},
|
|
222
|
+
"_name": {
|
|
223
|
+
"title": " Name",
|
|
224
|
+
"x-es": {
|
|
225
|
+
"type": "keyword",
|
|
226
|
+
"ignore_above": 8191
|
|
227
|
+
},
|
|
228
|
+
"type": "string"
|
|
229
|
+
}
|
|
230
|
+
},
|
|
231
|
+
"required": [
|
|
232
|
+
"type",
|
|
233
|
+
"value",
|
|
234
|
+
"_name"
|
|
235
|
+
]
|
|
236
|
+
},
|
|
237
|
+
"Log": {
|
|
238
|
+
"title": "Log",
|
|
239
|
+
"type": "object",
|
|
240
|
+
"properties": {
|
|
241
|
+
"agent": {
|
|
242
|
+
"title": "Agent",
|
|
243
|
+
"x-es": {
|
|
244
|
+
"type": "keyword",
|
|
245
|
+
"ignore_above": 8191
|
|
246
|
+
},
|
|
247
|
+
"type": "string"
|
|
248
|
+
},
|
|
249
|
+
"type": {
|
|
250
|
+
"title": "Type",
|
|
251
|
+
"x-es": {
|
|
252
|
+
"type": "keyword",
|
|
253
|
+
"ignore_above": 8191
|
|
254
|
+
},
|
|
255
|
+
"type": "string"
|
|
256
|
+
},
|
|
257
|
+
"comment": {
|
|
258
|
+
"title": "Comment",
|
|
259
|
+
"type": "string"
|
|
260
|
+
},
|
|
261
|
+
"date": {
|
|
262
|
+
"title": "Date",
|
|
263
|
+
"x-es": {
|
|
264
|
+
"type": "date"
|
|
265
|
+
},
|
|
266
|
+
"type": "string",
|
|
267
|
+
"format": "date-time"
|
|
268
|
+
}
|
|
269
|
+
},
|
|
270
|
+
"required": [
|
|
271
|
+
"agent",
|
|
272
|
+
"type",
|
|
273
|
+
"comment",
|
|
274
|
+
"date"
|
|
275
|
+
]
|
|
276
|
+
},
|
|
277
|
+
"CCSDocumentDescription": {
|
|
278
|
+
"title": "CCSDocumentDescription",
|
|
279
|
+
"type": "object",
|
|
280
|
+
"properties": {
|
|
281
|
+
"title": {
|
|
282
|
+
"title": "Title",
|
|
283
|
+
"type": "string"
|
|
284
|
+
},
|
|
285
|
+
"abstract": {
|
|
286
|
+
"title": "Abstract",
|
|
287
|
+
"type": "array",
|
|
288
|
+
"items": {
|
|
289
|
+
"type": "string"
|
|
290
|
+
}
|
|
291
|
+
},
|
|
292
|
+
"authors": {
|
|
293
|
+
"title": "Authors",
|
|
294
|
+
"type": "array",
|
|
295
|
+
"items": {
|
|
296
|
+
"$ref": "#/definitions/Author"
|
|
297
|
+
}
|
|
298
|
+
},
|
|
299
|
+
"affiliations": {
|
|
300
|
+
"title": "Affiliations",
|
|
301
|
+
"type": "array",
|
|
302
|
+
"items": {
|
|
303
|
+
"$ref": "#/definitions/Affiliation"
|
|
304
|
+
}
|
|
305
|
+
},
|
|
306
|
+
"subjects": {
|
|
307
|
+
"title": "Subjects",
|
|
308
|
+
"x-es": {
|
|
309
|
+
"fields": {
|
|
310
|
+
"keyword": {
|
|
311
|
+
"ignore_above": 8191,
|
|
312
|
+
"type": "keyword"
|
|
313
|
+
}
|
|
314
|
+
}
|
|
315
|
+
},
|
|
316
|
+
"type": "array",
|
|
317
|
+
"items": {
|
|
318
|
+
"type": "string"
|
|
319
|
+
}
|
|
320
|
+
},
|
|
321
|
+
"keywords": {
|
|
322
|
+
"title": "Keywords",
|
|
323
|
+
"x-es": {
|
|
324
|
+
"type": "keyword",
|
|
325
|
+
"ignore_above": 8191
|
|
326
|
+
},
|
|
327
|
+
"type": "array",
|
|
328
|
+
"items": {
|
|
329
|
+
"type": "string"
|
|
330
|
+
}
|
|
331
|
+
},
|
|
332
|
+
"publication_date": {
|
|
333
|
+
"title": "Publication Date",
|
|
334
|
+
"x-es": {
|
|
335
|
+
"type": "date"
|
|
336
|
+
},
|
|
337
|
+
"type": "string",
|
|
338
|
+
"format": "date-time"
|
|
339
|
+
},
|
|
340
|
+
"languages": {
|
|
341
|
+
"title": "Languages",
|
|
342
|
+
"x-es": {
|
|
343
|
+
"type": "keyword",
|
|
344
|
+
"ignore_above": 8191
|
|
345
|
+
},
|
|
346
|
+
"type": "array",
|
|
347
|
+
"items": {
|
|
348
|
+
"type": "string"
|
|
349
|
+
}
|
|
350
|
+
},
|
|
351
|
+
"publishers": {
|
|
352
|
+
"title": "Publishers",
|
|
353
|
+
"x-es": {
|
|
354
|
+
"type": "keyword",
|
|
355
|
+
"ignore_above": 8191
|
|
356
|
+
},
|
|
357
|
+
"type": "array",
|
|
358
|
+
"items": {
|
|
359
|
+
"type": "string"
|
|
360
|
+
}
|
|
361
|
+
},
|
|
362
|
+
"url_refs": {
|
|
363
|
+
"title": "Url Refs",
|
|
364
|
+
"x-es": {
|
|
365
|
+
"type": "keyword",
|
|
366
|
+
"ignore_above": 8191
|
|
367
|
+
},
|
|
368
|
+
"type": "array",
|
|
369
|
+
"items": {
|
|
370
|
+
"type": "string"
|
|
371
|
+
}
|
|
372
|
+
},
|
|
373
|
+
"references": {
|
|
374
|
+
"title": "References",
|
|
375
|
+
"type": "array",
|
|
376
|
+
"items": {
|
|
377
|
+
"$ref": "#/definitions/Identifier"
|
|
378
|
+
}
|
|
379
|
+
},
|
|
380
|
+
"advanced": {
|
|
381
|
+
"title": "Advanced",
|
|
382
|
+
"additionalProperties": {},
|
|
383
|
+
"properties": {},
|
|
384
|
+
"type": "object"
|
|
385
|
+
},
|
|
386
|
+
"analytics": {
|
|
387
|
+
"title": "Analytics",
|
|
388
|
+
"additionalProperties": {},
|
|
389
|
+
"properties": {},
|
|
390
|
+
"type": "object"
|
|
391
|
+
},
|
|
392
|
+
"logs": {
|
|
393
|
+
"title": "Logs",
|
|
394
|
+
"type": "array",
|
|
395
|
+
"items": {
|
|
396
|
+
"$ref": "#/definitions/Log"
|
|
397
|
+
}
|
|
398
|
+
}
|
|
399
|
+
},
|
|
400
|
+
"required": [
|
|
401
|
+
"title",
|
|
402
|
+
"abstract",
|
|
403
|
+
"authors",
|
|
404
|
+
"affiliations",
|
|
405
|
+
"subjects",
|
|
406
|
+
"keywords",
|
|
407
|
+
"publication_date",
|
|
408
|
+
"languages",
|
|
409
|
+
"publishers",
|
|
410
|
+
"url_refs",
|
|
411
|
+
"references",
|
|
412
|
+
"advanced",
|
|
413
|
+
"analytics",
|
|
414
|
+
"logs"
|
|
415
|
+
]
|
|
416
|
+
},
|
|
417
|
+
"CCSFileInfoDescription": {
|
|
418
|
+
"title": "CCSFileInfoDescription",
|
|
419
|
+
"type": "object",
|
|
420
|
+
"properties": {
|
|
421
|
+
"author": {
|
|
422
|
+
"title": "Author",
|
|
423
|
+
"type": "array",
|
|
424
|
+
"items": {
|
|
425
|
+
"type": "string"
|
|
426
|
+
}
|
|
427
|
+
},
|
|
428
|
+
"keywords": {
|
|
429
|
+
"title": "Keywords",
|
|
430
|
+
"type": "string"
|
|
431
|
+
},
|
|
432
|
+
"subject": {
|
|
433
|
+
"title": "Subject",
|
|
434
|
+
"type": "string"
|
|
435
|
+
},
|
|
436
|
+
"title": {
|
|
437
|
+
"title": "Title",
|
|
438
|
+
"type": "string"
|
|
439
|
+
}
|
|
440
|
+
}
|
|
441
|
+
},
|
|
442
|
+
"PageReference": {
|
|
443
|
+
"title": "PageReference",
|
|
444
|
+
"type": "object",
|
|
445
|
+
"properties": {
|
|
446
|
+
"hash": {
|
|
447
|
+
"title": "Hash",
|
|
448
|
+
"type": "string"
|
|
449
|
+
},
|
|
450
|
+
"model": {
|
|
451
|
+
"title": "Model",
|
|
452
|
+
"type": "string"
|
|
453
|
+
},
|
|
454
|
+
"page": {
|
|
455
|
+
"title": "Page",
|
|
456
|
+
"type": "integer"
|
|
457
|
+
}
|
|
458
|
+
},
|
|
459
|
+
"required": [
|
|
460
|
+
"hash",
|
|
461
|
+
"model",
|
|
462
|
+
"page"
|
|
463
|
+
]
|
|
464
|
+
},
|
|
465
|
+
"CCSFileInfoObject": {
|
|
466
|
+
"title": "CCSFileInfoObject",
|
|
467
|
+
"type": "object",
|
|
468
|
+
"properties": {
|
|
469
|
+
"filename": {
|
|
470
|
+
"title": "Filename",
|
|
471
|
+
"x-es": {
|
|
472
|
+
"type": "keyword",
|
|
473
|
+
"ignore_above": 8191
|
|
474
|
+
},
|
|
475
|
+
"type": "string"
|
|
476
|
+
},
|
|
477
|
+
"#-pages": {
|
|
478
|
+
"title": "#-Pages",
|
|
479
|
+
"type": "integer"
|
|
480
|
+
},
|
|
481
|
+
"document-hash": {
|
|
482
|
+
"title": "Document-Hash",
|
|
483
|
+
"type": "string"
|
|
484
|
+
},
|
|
485
|
+
"collection-name": {
|
|
486
|
+
"title": "Collection-Name",
|
|
487
|
+
"type": "string"
|
|
488
|
+
},
|
|
489
|
+
"description": {
|
|
490
|
+
"$ref": "#/definitions/CCSFileInfoDescription"
|
|
491
|
+
},
|
|
492
|
+
"page-hashes": {
|
|
493
|
+
"title": "Page-Hashes",
|
|
494
|
+
"type": "array",
|
|
495
|
+
"items": {
|
|
496
|
+
"$ref": "#/definitions/PageReference"
|
|
497
|
+
}
|
|
498
|
+
},
|
|
499
|
+
"filename-prov": {
|
|
500
|
+
"title": "Filename-Prov",
|
|
501
|
+
"x-es": {
|
|
502
|
+
"type": "keyword",
|
|
503
|
+
"ignore_above": 8191
|
|
504
|
+
},
|
|
505
|
+
"type": "string"
|
|
506
|
+
}
|
|
507
|
+
},
|
|
508
|
+
"required": [
|
|
509
|
+
"filename",
|
|
510
|
+
"#-pages",
|
|
511
|
+
"document-hash",
|
|
512
|
+
"collection-name",
|
|
513
|
+
"description",
|
|
514
|
+
"page-hashes",
|
|
515
|
+
"filename-prov"
|
|
516
|
+
]
|
|
517
|
+
},
|
|
518
|
+
"Ref": {
|
|
519
|
+
"title": "Ref",
|
|
520
|
+
"type": "object",
|
|
521
|
+
"properties": {
|
|
522
|
+
"name": {
|
|
523
|
+
"title": "Name",
|
|
524
|
+
"type": "string"
|
|
525
|
+
},
|
|
526
|
+
"type": {
|
|
527
|
+
"title": "Type",
|
|
528
|
+
"type": "string"
|
|
529
|
+
},
|
|
530
|
+
"__ref": {
|
|
531
|
+
"title": " Ref",
|
|
532
|
+
"type": "string"
|
|
533
|
+
}
|
|
534
|
+
},
|
|
535
|
+
"required": [
|
|
536
|
+
"name",
|
|
537
|
+
"type",
|
|
538
|
+
"__ref"
|
|
539
|
+
]
|
|
540
|
+
},
|
|
541
|
+
"Prov": {
|
|
542
|
+
"title": "Prov",
|
|
543
|
+
"type": "object",
|
|
544
|
+
"properties": {
|
|
545
|
+
"bbox": {
|
|
546
|
+
"title": "Bbox",
|
|
547
|
+
"type": "array",
|
|
548
|
+
"items": [
|
|
549
|
+
{
|
|
550
|
+
"type": "number"
|
|
551
|
+
},
|
|
552
|
+
{
|
|
553
|
+
"type": "number"
|
|
554
|
+
},
|
|
555
|
+
{
|
|
556
|
+
"type": "number"
|
|
557
|
+
},
|
|
558
|
+
{
|
|
559
|
+
"type": "number"
|
|
560
|
+
}
|
|
561
|
+
]
|
|
562
|
+
},
|
|
563
|
+
"page": {
|
|
564
|
+
"title": "Page",
|
|
565
|
+
"type": "integer"
|
|
566
|
+
},
|
|
567
|
+
"span": {
|
|
568
|
+
"title": "Span",
|
|
569
|
+
"type": "array",
|
|
570
|
+
"items": [
|
|
571
|
+
{
|
|
572
|
+
"type": "integer"
|
|
573
|
+
},
|
|
574
|
+
{
|
|
575
|
+
"type": "integer"
|
|
576
|
+
}
|
|
577
|
+
]
|
|
578
|
+
}
|
|
579
|
+
},
|
|
580
|
+
"required": [
|
|
581
|
+
"bbox",
|
|
582
|
+
"page",
|
|
583
|
+
"span"
|
|
584
|
+
]
|
|
585
|
+
},
|
|
586
|
+
"ListItem": {
|
|
587
|
+
"title": "ListItem",
|
|
588
|
+
"type": "object",
|
|
589
|
+
"properties": {
|
|
590
|
+
"text": {
|
|
591
|
+
"title": "Text",
|
|
592
|
+
"type": "string"
|
|
593
|
+
},
|
|
594
|
+
"type": {
|
|
595
|
+
"title": "Type",
|
|
596
|
+
"x-es": {
|
|
597
|
+
"type": "keyword",
|
|
598
|
+
"ignore_above": 8191
|
|
599
|
+
},
|
|
600
|
+
"type": "string"
|
|
601
|
+
},
|
|
602
|
+
"name": {
|
|
603
|
+
"title": "Name",
|
|
604
|
+
"x-es": {
|
|
605
|
+
"type": "keyword",
|
|
606
|
+
"ignore_above": 8191
|
|
607
|
+
},
|
|
608
|
+
"type": "string"
|
|
609
|
+
},
|
|
610
|
+
"font": {
|
|
611
|
+
"title": "Font",
|
|
612
|
+
"type": "string"
|
|
613
|
+
},
|
|
614
|
+
"prov": {
|
|
615
|
+
"title": "Prov",
|
|
616
|
+
"type": "array",
|
|
617
|
+
"items": {
|
|
618
|
+
"$ref": "#/definitions/Prov"
|
|
619
|
+
}
|
|
620
|
+
},
|
|
621
|
+
"identifier": {
|
|
622
|
+
"title": "Identifier",
|
|
623
|
+
"type": "string"
|
|
624
|
+
}
|
|
625
|
+
},
|
|
626
|
+
"required": [
|
|
627
|
+
"text",
|
|
628
|
+
"type",
|
|
629
|
+
"identifier"
|
|
630
|
+
]
|
|
631
|
+
},
|
|
632
|
+
"RecursiveList": {
|
|
633
|
+
"title": "RecursiveList",
|
|
634
|
+
"type": "object",
|
|
635
|
+
"properties": {
|
|
636
|
+
"data": {
|
|
637
|
+
"title": "Data",
|
|
638
|
+
"type": "array",
|
|
639
|
+
"items": {
|
|
640
|
+
"anyOf": [
|
|
641
|
+
{
|
|
642
|
+
"$ref": "#/definitions/ListItem"
|
|
643
|
+
},
|
|
644
|
+
{
|
|
645
|
+
"$ref": "#/definitions/RecursiveList"
|
|
646
|
+
}
|
|
647
|
+
]
|
|
648
|
+
}
|
|
649
|
+
},
|
|
650
|
+
"name": {
|
|
651
|
+
"title": "Name",
|
|
652
|
+
"type": "string"
|
|
653
|
+
},
|
|
654
|
+
"prov": {
|
|
655
|
+
"title": "Prov",
|
|
656
|
+
"type": "array",
|
|
657
|
+
"items": {
|
|
658
|
+
"$ref": "#/definitions/Prov"
|
|
659
|
+
}
|
|
660
|
+
},
|
|
661
|
+
"type": {
|
|
662
|
+
"title": "Type",
|
|
663
|
+
"type": "string"
|
|
664
|
+
}
|
|
665
|
+
},
|
|
666
|
+
"required": [
|
|
667
|
+
"data",
|
|
668
|
+
"type"
|
|
669
|
+
]
|
|
670
|
+
},
|
|
671
|
+
"BaseText": {
|
|
672
|
+
"title": "BaseText",
|
|
673
|
+
"type": "object",
|
|
674
|
+
"properties": {
|
|
675
|
+
"text": {
|
|
676
|
+
"title": "Text",
|
|
677
|
+
"type": "string"
|
|
678
|
+
},
|
|
679
|
+
"type": {
|
|
680
|
+
"title": "Type",
|
|
681
|
+
"x-es": {
|
|
682
|
+
"type": "keyword",
|
|
683
|
+
"ignore_above": 8191
|
|
684
|
+
},
|
|
685
|
+
"type": "string"
|
|
686
|
+
},
|
|
687
|
+
"name": {
|
|
688
|
+
"title": "Name",
|
|
689
|
+
"x-es": {
|
|
690
|
+
"type": "keyword",
|
|
691
|
+
"ignore_above": 8191
|
|
692
|
+
},
|
|
693
|
+
"type": "string"
|
|
694
|
+
},
|
|
695
|
+
"font": {
|
|
696
|
+
"title": "Font",
|
|
697
|
+
"type": "string"
|
|
698
|
+
},
|
|
699
|
+
"prov": {
|
|
700
|
+
"title": "Prov",
|
|
701
|
+
"type": "array",
|
|
702
|
+
"items": {
|
|
703
|
+
"$ref": "#/definitions/Prov"
|
|
704
|
+
}
|
|
705
|
+
}
|
|
706
|
+
},
|
|
707
|
+
"required": [
|
|
708
|
+
"text",
|
|
709
|
+
"type"
|
|
710
|
+
]
|
|
711
|
+
},
|
|
712
|
+
"BoundingBoxContainer": {
|
|
713
|
+
"title": "BoundingBoxContainer",
|
|
714
|
+
"type": "object",
|
|
715
|
+
"properties": {
|
|
716
|
+
"min": {
|
|
717
|
+
"title": "Min",
|
|
718
|
+
"type": "array",
|
|
719
|
+
"items": [
|
|
720
|
+
{
|
|
721
|
+
"type": "number"
|
|
722
|
+
},
|
|
723
|
+
{
|
|
724
|
+
"type": "number"
|
|
725
|
+
},
|
|
726
|
+
{
|
|
727
|
+
"type": "number"
|
|
728
|
+
},
|
|
729
|
+
{
|
|
730
|
+
"type": "number"
|
|
731
|
+
}
|
|
732
|
+
]
|
|
733
|
+
},
|
|
734
|
+
"max": {
|
|
735
|
+
"title": "Max",
|
|
736
|
+
"type": "array",
|
|
737
|
+
"items": [
|
|
738
|
+
{
|
|
739
|
+
"type": "number"
|
|
740
|
+
},
|
|
741
|
+
{
|
|
742
|
+
"type": "number"
|
|
743
|
+
},
|
|
744
|
+
{
|
|
745
|
+
"type": "number"
|
|
746
|
+
},
|
|
747
|
+
{
|
|
748
|
+
"type": "number"
|
|
749
|
+
}
|
|
750
|
+
]
|
|
751
|
+
}
|
|
752
|
+
},
|
|
753
|
+
"required": [
|
|
754
|
+
"min",
|
|
755
|
+
"max"
|
|
756
|
+
]
|
|
757
|
+
},
|
|
758
|
+
"CellsContainer": {
|
|
759
|
+
"title": "CellsContainer",
|
|
760
|
+
"type": "object",
|
|
761
|
+
"properties": {
|
|
762
|
+
"data": {
|
|
763
|
+
"title": "Data",
|
|
764
|
+
"type": "array",
|
|
765
|
+
"items": {
|
|
766
|
+
"type": "array",
|
|
767
|
+
"items": [
|
|
768
|
+
{
|
|
769
|
+
"type": "number"
|
|
770
|
+
},
|
|
771
|
+
{
|
|
772
|
+
"type": "number"
|
|
773
|
+
},
|
|
774
|
+
{
|
|
775
|
+
"type": "number"
|
|
776
|
+
},
|
|
777
|
+
{
|
|
778
|
+
"type": "number"
|
|
779
|
+
},
|
|
780
|
+
{
|
|
781
|
+
"type": "string"
|
|
782
|
+
},
|
|
783
|
+
{
|
|
784
|
+
"type": "string"
|
|
785
|
+
}
|
|
786
|
+
]
|
|
787
|
+
}
|
|
788
|
+
},
|
|
789
|
+
"header": {
|
|
790
|
+
"title": "Header",
|
|
791
|
+
"default": [
|
|
792
|
+
"x0",
|
|
793
|
+
"y0",
|
|
794
|
+
"x1",
|
|
795
|
+
"y1",
|
|
796
|
+
"font",
|
|
797
|
+
"text"
|
|
798
|
+
],
|
|
799
|
+
"type": "array",
|
|
800
|
+
"items": [
|
|
801
|
+
{
|
|
802
|
+
"enum": [
|
|
803
|
+
"x0"
|
|
804
|
+
],
|
|
805
|
+
"type": "string"
|
|
806
|
+
},
|
|
807
|
+
{
|
|
808
|
+
"enum": [
|
|
809
|
+
"y0"
|
|
810
|
+
],
|
|
811
|
+
"type": "string"
|
|
812
|
+
},
|
|
813
|
+
{
|
|
814
|
+
"enum": [
|
|
815
|
+
"x1"
|
|
816
|
+
],
|
|
817
|
+
"type": "string"
|
|
818
|
+
},
|
|
819
|
+
{
|
|
820
|
+
"enum": [
|
|
821
|
+
"y1"
|
|
822
|
+
],
|
|
823
|
+
"type": "string"
|
|
824
|
+
},
|
|
825
|
+
{
|
|
826
|
+
"enum": [
|
|
827
|
+
"font"
|
|
828
|
+
],
|
|
829
|
+
"type": "string"
|
|
830
|
+
},
|
|
831
|
+
{
|
|
832
|
+
"enum": [
|
|
833
|
+
"text"
|
|
834
|
+
],
|
|
835
|
+
"type": "string"
|
|
836
|
+
}
|
|
837
|
+
]
|
|
838
|
+
}
|
|
839
|
+
},
|
|
840
|
+
"required": [
|
|
841
|
+
"data"
|
|
842
|
+
]
|
|
843
|
+
},
|
|
844
|
+
"BaseCell": {
|
|
845
|
+
"title": "BaseCell",
|
|
846
|
+
"type": "object",
|
|
847
|
+
"properties": {
|
|
848
|
+
"bounding-box": {
|
|
849
|
+
"$ref": "#/definitions/BoundingBoxContainer"
|
|
850
|
+
},
|
|
851
|
+
"cells": {
|
|
852
|
+
"$ref": "#/definitions/CellsContainer"
|
|
853
|
+
},
|
|
854
|
+
"prov": {
|
|
855
|
+
"title": "Prov",
|
|
856
|
+
"type": "array",
|
|
857
|
+
"items": {
|
|
858
|
+
"$ref": "#/definitions/Prov"
|
|
859
|
+
}
|
|
860
|
+
},
|
|
861
|
+
"text": {
|
|
862
|
+
"title": "Text",
|
|
863
|
+
"type": "string"
|
|
864
|
+
},
|
|
865
|
+
"type": {
|
|
866
|
+
"title": "Type",
|
|
867
|
+
"x-es": {
|
|
868
|
+
"type": "keyword",
|
|
869
|
+
"ignore_above": 8191
|
|
870
|
+
},
|
|
871
|
+
"type": "string"
|
|
872
|
+
}
|
|
873
|
+
},
|
|
874
|
+
"required": [
|
|
875
|
+
"bounding-box",
|
|
876
|
+
"cells",
|
|
877
|
+
"text",
|
|
878
|
+
"type"
|
|
879
|
+
]
|
|
880
|
+
},
|
|
881
|
+
"TableCell": {
|
|
882
|
+
"title": "TableCell",
|
|
883
|
+
"type": "object",
|
|
884
|
+
"properties": {
|
|
885
|
+
"bbox": {
|
|
886
|
+
"title": "Bbox",
|
|
887
|
+
"type": "array",
|
|
888
|
+
"items": [
|
|
889
|
+
{
|
|
890
|
+
"type": "number"
|
|
891
|
+
},
|
|
892
|
+
{
|
|
893
|
+
"type": "number"
|
|
894
|
+
},
|
|
895
|
+
{
|
|
896
|
+
"type": "number"
|
|
897
|
+
},
|
|
898
|
+
{
|
|
899
|
+
"type": "number"
|
|
900
|
+
}
|
|
901
|
+
]
|
|
902
|
+
},
|
|
903
|
+
"spans": {
|
|
904
|
+
"title": "Spans",
|
|
905
|
+
"type": "array",
|
|
906
|
+
"items": {
|
|
907
|
+
"type": "array",
|
|
908
|
+
"items": [
|
|
909
|
+
{
|
|
910
|
+
"type": "integer"
|
|
911
|
+
},
|
|
912
|
+
{
|
|
913
|
+
"type": "integer"
|
|
914
|
+
}
|
|
915
|
+
]
|
|
916
|
+
}
|
|
917
|
+
},
|
|
918
|
+
"text": {
|
|
919
|
+
"title": "Text",
|
|
920
|
+
"type": "string"
|
|
921
|
+
},
|
|
922
|
+
"type": {
|
|
923
|
+
"title": "Type",
|
|
924
|
+
"type": "string"
|
|
925
|
+
}
|
|
926
|
+
},
|
|
927
|
+
"required": [
|
|
928
|
+
"text",
|
|
929
|
+
"type"
|
|
930
|
+
]
|
|
931
|
+
},
|
|
932
|
+
"Table": {
|
|
933
|
+
"title": "Table",
|
|
934
|
+
"type": "object",
|
|
935
|
+
"properties": {
|
|
936
|
+
"num_cols": {
|
|
937
|
+
"title": "Num Cols",
|
|
938
|
+
"type": "integer"
|
|
939
|
+
},
|
|
940
|
+
"num_rows": {
|
|
941
|
+
"title": "Num Rows",
|
|
942
|
+
"type": "integer"
|
|
943
|
+
},
|
|
944
|
+
"bounding_box": {
|
|
945
|
+
"$ref": "#/definitions/BoundingBoxContainer"
|
|
946
|
+
},
|
|
947
|
+
"cells": {
|
|
948
|
+
"$ref": "#/definitions/CellsContainer"
|
|
949
|
+
},
|
|
950
|
+
"data": {
|
|
951
|
+
"title": "Data",
|
|
952
|
+
"type": "array",
|
|
953
|
+
"items": {
|
|
954
|
+
"type": "array",
|
|
955
|
+
"items": {
|
|
956
|
+
"$ref": "#/definitions/TableCell"
|
|
957
|
+
}
|
|
958
|
+
}
|
|
959
|
+
},
|
|
960
|
+
"model": {
|
|
961
|
+
"title": "Model",
|
|
962
|
+
"type": "string"
|
|
963
|
+
},
|
|
964
|
+
"prov": {
|
|
965
|
+
"$ref": "#/definitions/Prov"
|
|
966
|
+
},
|
|
967
|
+
"text": {
|
|
968
|
+
"title": "Text",
|
|
969
|
+
"type": "string"
|
|
970
|
+
},
|
|
971
|
+
"type": {
|
|
972
|
+
"title": "Type",
|
|
973
|
+
"type": "string"
|
|
974
|
+
}
|
|
975
|
+
},
|
|
976
|
+
"required": [
|
|
977
|
+
"num_cols",
|
|
978
|
+
"num_rows",
|
|
979
|
+
"data",
|
|
980
|
+
"text",
|
|
981
|
+
"type"
|
|
982
|
+
]
|
|
983
|
+
},
|
|
984
|
+
"BitmapObject": {
|
|
985
|
+
"title": "BitmapObject",
|
|
986
|
+
"type": "object",
|
|
987
|
+
"properties": {
|
|
988
|
+
"type": {
|
|
989
|
+
"title": "Type",
|
|
990
|
+
"type": "string"
|
|
991
|
+
},
|
|
992
|
+
"bounding_box": {
|
|
993
|
+
"$ref": "#/definitions/BoundingBoxContainer"
|
|
994
|
+
},
|
|
995
|
+
"prov": {
|
|
996
|
+
"$ref": "#/definitions/Prov"
|
|
997
|
+
}
|
|
998
|
+
},
|
|
999
|
+
"required": [
|
|
1000
|
+
"type",
|
|
1001
|
+
"bounding_box",
|
|
1002
|
+
"prov"
|
|
1003
|
+
]
|
|
1004
|
+
},
|
|
1005
|
+
"PageDimensions": {
|
|
1006
|
+
"title": "PageDimensions",
|
|
1007
|
+
"type": "object",
|
|
1008
|
+
"properties": {
|
|
1009
|
+
"height": {
|
|
1010
|
+
"title": "Height",
|
|
1011
|
+
"type": "number"
|
|
1012
|
+
},
|
|
1013
|
+
"page": {
|
|
1014
|
+
"title": "Page",
|
|
1015
|
+
"type": "integer"
|
|
1016
|
+
},
|
|
1017
|
+
"width": {
|
|
1018
|
+
"title": "Width",
|
|
1019
|
+
"type": "number"
|
|
1020
|
+
}
|
|
1021
|
+
},
|
|
1022
|
+
"required": [
|
|
1023
|
+
"height",
|
|
1024
|
+
"page",
|
|
1025
|
+
"width"
|
|
1026
|
+
]
|
|
1027
|
+
},
|
|
1028
|
+
"S3Resource": {
|
|
1029
|
+
"title": "S3Resource",
|
|
1030
|
+
"type": "object",
|
|
1031
|
+
"properties": {
|
|
1032
|
+
"mime": {
|
|
1033
|
+
"title": "Mime",
|
|
1034
|
+
"type": "string"
|
|
1035
|
+
},
|
|
1036
|
+
"path": {
|
|
1037
|
+
"title": "Path",
|
|
1038
|
+
"type": "string"
|
|
1039
|
+
},
|
|
1040
|
+
"page": {
|
|
1041
|
+
"title": "Page",
|
|
1042
|
+
"type": "integer"
|
|
1043
|
+
}
|
|
1044
|
+
},
|
|
1045
|
+
"required": [
|
|
1046
|
+
"mime",
|
|
1047
|
+
"path"
|
|
1048
|
+
]
|
|
1049
|
+
},
|
|
1050
|
+
"S3Data": {
|
|
1051
|
+
"title": "S3Data",
|
|
1052
|
+
"type": "object",
|
|
1053
|
+
"properties": {
|
|
1054
|
+
"pdf_document": {
|
|
1055
|
+
"title": "Pdf Document",
|
|
1056
|
+
"type": "array",
|
|
1057
|
+
"items": {
|
|
1058
|
+
"$ref": "#/definitions/S3Resource"
|
|
1059
|
+
}
|
|
1060
|
+
},
|
|
1061
|
+
"pdf_pages": {
|
|
1062
|
+
"title": "Pdf Pages",
|
|
1063
|
+
"type": "array",
|
|
1064
|
+
"items": {
|
|
1065
|
+
"$ref": "#/definitions/S3Resource"
|
|
1066
|
+
}
|
|
1067
|
+
}
|
|
1068
|
+
}
|
|
1069
|
+
}
|
|
1070
|
+
}
|
|
1071
|
+
}
|