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 +1,104 @@
|
|
|
1
|
+
{
|
|
2
|
+
"mappings": {
|
|
3
|
+
"dynamic": false,
|
|
4
|
+
"_size": {
|
|
5
|
+
"enabled": true
|
|
6
|
+
},
|
|
7
|
+
"_meta": {
|
|
8
|
+
"$ref": "ccs:schemas#/Document"
|
|
9
|
+
},
|
|
10
|
+
"properties": {
|
|
11
|
+
"description": {
|
|
12
|
+
"type": "object",
|
|
13
|
+
"properties": {
|
|
14
|
+
"abstract": {
|
|
15
|
+
"type": "text"
|
|
16
|
+
},
|
|
17
|
+
"affiliations": {
|
|
18
|
+
"type": "keyword"
|
|
19
|
+
},
|
|
20
|
+
"authors": {
|
|
21
|
+
"type": "keyword"
|
|
22
|
+
},
|
|
23
|
+
"title": {
|
|
24
|
+
"type": "text"
|
|
25
|
+
}
|
|
26
|
+
}
|
|
27
|
+
},
|
|
28
|
+
"figures": {
|
|
29
|
+
"type": "object",
|
|
30
|
+
"properties": {
|
|
31
|
+
"text": {
|
|
32
|
+
"type": "text"
|
|
33
|
+
},
|
|
34
|
+
"type": {
|
|
35
|
+
"type": "keyword"
|
|
36
|
+
},
|
|
37
|
+
"prov": {
|
|
38
|
+
"type": "object",
|
|
39
|
+
"properties": {
|
|
40
|
+
"page": {
|
|
41
|
+
"type": "integer"
|
|
42
|
+
}
|
|
43
|
+
}
|
|
44
|
+
}
|
|
45
|
+
}
|
|
46
|
+
},
|
|
47
|
+
"file-info": {
|
|
48
|
+
"type": "object",
|
|
49
|
+
"properties": {
|
|
50
|
+
"filename": {
|
|
51
|
+
"type": "text"
|
|
52
|
+
}
|
|
53
|
+
}
|
|
54
|
+
},
|
|
55
|
+
"main-text": {
|
|
56
|
+
"type": "object",
|
|
57
|
+
"properties": {
|
|
58
|
+
"text": {
|
|
59
|
+
"type": "text"
|
|
60
|
+
},
|
|
61
|
+
"type": {
|
|
62
|
+
"type": "keyword"
|
|
63
|
+
},
|
|
64
|
+
"name": {
|
|
65
|
+
"type": "keyword"
|
|
66
|
+
},
|
|
67
|
+
"prov": {
|
|
68
|
+
"type": "object",
|
|
69
|
+
"properties": {
|
|
70
|
+
"page": {
|
|
71
|
+
"type": "integer"
|
|
72
|
+
}
|
|
73
|
+
}
|
|
74
|
+
}
|
|
75
|
+
}
|
|
76
|
+
},
|
|
77
|
+
"_name": {
|
|
78
|
+
"type": "keyword"
|
|
79
|
+
},
|
|
80
|
+
"tables": {
|
|
81
|
+
"type": "object",
|
|
82
|
+
"properties": {
|
|
83
|
+
"text": {
|
|
84
|
+
"type": "text"
|
|
85
|
+
},
|
|
86
|
+
"type": {
|
|
87
|
+
"type": "keyword"
|
|
88
|
+
},
|
|
89
|
+
"prov": {
|
|
90
|
+
"type": "object",
|
|
91
|
+
"properties": {
|
|
92
|
+
"page": {
|
|
93
|
+
"type": "integer"
|
|
94
|
+
}
|
|
95
|
+
}
|
|
96
|
+
}
|
|
97
|
+
}
|
|
98
|
+
},
|
|
99
|
+
"type": {
|
|
100
|
+
"type": "keyword"
|
|
101
|
+
}
|
|
102
|
+
}
|
|
103
|
+
}
|
|
104
|
+
}
|
|
@@ -0,0 +1,256 @@
|
|
|
1
|
+
{
|
|
2
|
+
"settings": {
|
|
3
|
+
"analysis": {
|
|
4
|
+
"normalizer": {
|
|
5
|
+
"lowercase_asciifolding": {
|
|
6
|
+
"type": "custom",
|
|
7
|
+
"filter": [
|
|
8
|
+
"lowercase",
|
|
9
|
+
"asciifolding"
|
|
10
|
+
]
|
|
11
|
+
}
|
|
12
|
+
}
|
|
13
|
+
}
|
|
14
|
+
},
|
|
15
|
+
"mappings": {
|
|
16
|
+
"dynamic": false,
|
|
17
|
+
"_size": {
|
|
18
|
+
"enabled": true
|
|
19
|
+
},
|
|
20
|
+
"_meta": {
|
|
21
|
+
"version": "1.0",
|
|
22
|
+
"$ref": "ccs:schemas#/Document"
|
|
23
|
+
},
|
|
24
|
+
"properties": {
|
|
25
|
+
"_name": {
|
|
26
|
+
"type": "text"
|
|
27
|
+
},
|
|
28
|
+
"identifiers": {
|
|
29
|
+
"properties": {
|
|
30
|
+
"_name": {
|
|
31
|
+
"ignore_above": 8191,
|
|
32
|
+
"type": "keyword"
|
|
33
|
+
},
|
|
34
|
+
"type": {
|
|
35
|
+
"ignore_above": 8191,
|
|
36
|
+
"type": "keyword"
|
|
37
|
+
},
|
|
38
|
+
"value": {
|
|
39
|
+
"ignore_above": 8191,
|
|
40
|
+
"type": "keyword"
|
|
41
|
+
}
|
|
42
|
+
}
|
|
43
|
+
},
|
|
44
|
+
"description": {
|
|
45
|
+
"properties": {
|
|
46
|
+
"abstract": {
|
|
47
|
+
"type": "text"
|
|
48
|
+
},
|
|
49
|
+
"affiliations": {
|
|
50
|
+
"properties": {
|
|
51
|
+
"name": {
|
|
52
|
+
"type": "text",
|
|
53
|
+
"fields": {
|
|
54
|
+
"lower": {
|
|
55
|
+
"normalizer": "lowercase_asciifolding",
|
|
56
|
+
"type": "keyword"
|
|
57
|
+
},
|
|
58
|
+
"keyword": {
|
|
59
|
+
"type": "keyword"
|
|
60
|
+
}
|
|
61
|
+
}
|
|
62
|
+
},
|
|
63
|
+
"id": {
|
|
64
|
+
"ignore_above": 8191,
|
|
65
|
+
"type": "keyword"
|
|
66
|
+
},
|
|
67
|
+
"source": {
|
|
68
|
+
"ignore_above": 8191,
|
|
69
|
+
"type": "keyword"
|
|
70
|
+
}
|
|
71
|
+
}
|
|
72
|
+
},
|
|
73
|
+
"authors": {
|
|
74
|
+
"properties": {
|
|
75
|
+
"name": {
|
|
76
|
+
"type": "text",
|
|
77
|
+
"fields": {
|
|
78
|
+
"lower": {
|
|
79
|
+
"normalizer": "lowercase_asciifolding",
|
|
80
|
+
"type": "keyword"
|
|
81
|
+
},
|
|
82
|
+
"keyword": {
|
|
83
|
+
"type": "keyword"
|
|
84
|
+
}
|
|
85
|
+
}
|
|
86
|
+
},
|
|
87
|
+
"affiliations": {
|
|
88
|
+
"properties": {
|
|
89
|
+
"name": {
|
|
90
|
+
"type": "text",
|
|
91
|
+
"fields": {
|
|
92
|
+
"lower": {
|
|
93
|
+
"normalizer": "lowercase_asciifolding",
|
|
94
|
+
"type": "keyword"
|
|
95
|
+
},
|
|
96
|
+
"keyword": {
|
|
97
|
+
"type": "keyword"
|
|
98
|
+
}
|
|
99
|
+
}
|
|
100
|
+
},
|
|
101
|
+
"id": {
|
|
102
|
+
"ignore_above": 8191,
|
|
103
|
+
"type": "keyword"
|
|
104
|
+
},
|
|
105
|
+
"source": {
|
|
106
|
+
"ignore_above": 8191,
|
|
107
|
+
"type": "keyword"
|
|
108
|
+
}
|
|
109
|
+
}
|
|
110
|
+
}
|
|
111
|
+
}
|
|
112
|
+
},
|
|
113
|
+
"title": {
|
|
114
|
+
"type": "text"
|
|
115
|
+
},
|
|
116
|
+
"subjects": {
|
|
117
|
+
"type": "text",
|
|
118
|
+
"fields": {
|
|
119
|
+
"keyword": {
|
|
120
|
+
"ignore_above": 8191,
|
|
121
|
+
"type": "keyword"
|
|
122
|
+
}
|
|
123
|
+
}
|
|
124
|
+
},
|
|
125
|
+
"publication_date": {
|
|
126
|
+
"type": "date"
|
|
127
|
+
},
|
|
128
|
+
"languages": {
|
|
129
|
+
"ignore_above": 8191,
|
|
130
|
+
"type": "keyword"
|
|
131
|
+
},
|
|
132
|
+
"publishers": {
|
|
133
|
+
"ignore_above": 8191,
|
|
134
|
+
"type": "keyword"
|
|
135
|
+
},
|
|
136
|
+
"url_refs": {
|
|
137
|
+
"ignore_above": 8191,
|
|
138
|
+
"type": "keyword"
|
|
139
|
+
},
|
|
140
|
+
"references": {
|
|
141
|
+
"properties": {
|
|
142
|
+
"_name": {
|
|
143
|
+
"ignore_above": 8191,
|
|
144
|
+
"type": "keyword"
|
|
145
|
+
},
|
|
146
|
+
"type": {
|
|
147
|
+
"ignore_above": 8191,
|
|
148
|
+
"type": "keyword"
|
|
149
|
+
},
|
|
150
|
+
"value": {
|
|
151
|
+
"ignore_above": 8191,
|
|
152
|
+
"type": "keyword"
|
|
153
|
+
}
|
|
154
|
+
}
|
|
155
|
+
},
|
|
156
|
+
"logs": {
|
|
157
|
+
"properties": {
|
|
158
|
+
"date": {
|
|
159
|
+
"type": "date"
|
|
160
|
+
},
|
|
161
|
+
"agent": {
|
|
162
|
+
"ignore_above": 8191,
|
|
163
|
+
"type": "keyword"
|
|
164
|
+
},
|
|
165
|
+
"comment": {
|
|
166
|
+
"type": "text"
|
|
167
|
+
},
|
|
168
|
+
"type": {
|
|
169
|
+
"ignore_above": 8191,
|
|
170
|
+
"type": "keyword"
|
|
171
|
+
}
|
|
172
|
+
}
|
|
173
|
+
}
|
|
174
|
+
}
|
|
175
|
+
},
|
|
176
|
+
"figures": {
|
|
177
|
+
"properties": {
|
|
178
|
+
"text": {
|
|
179
|
+
"type": "text"
|
|
180
|
+
},
|
|
181
|
+
"type": {
|
|
182
|
+
"ignore_above": 8191,
|
|
183
|
+
"type": "keyword"
|
|
184
|
+
},
|
|
185
|
+
"prov": {
|
|
186
|
+
"properties": {
|
|
187
|
+
"page": {
|
|
188
|
+
"type": "integer"
|
|
189
|
+
}
|
|
190
|
+
}
|
|
191
|
+
}
|
|
192
|
+
}
|
|
193
|
+
},
|
|
194
|
+
"file-info": {
|
|
195
|
+
"properties": {
|
|
196
|
+
"filename-prov": {
|
|
197
|
+
"ignore_above": 8191,
|
|
198
|
+
"type": "keyword"
|
|
199
|
+
},
|
|
200
|
+
"filename": {
|
|
201
|
+
"ignore_above": 8191,
|
|
202
|
+
"type": "keyword"
|
|
203
|
+
},
|
|
204
|
+
"document-hash": {
|
|
205
|
+
"ignore_above": 8191,
|
|
206
|
+
"type": "keyword"
|
|
207
|
+
}
|
|
208
|
+
}
|
|
209
|
+
},
|
|
210
|
+
"main-text": {
|
|
211
|
+
"properties": {
|
|
212
|
+
"text": {
|
|
213
|
+
"type": "text"
|
|
214
|
+
},
|
|
215
|
+
"type": {
|
|
216
|
+
"ignore_above": 8191,
|
|
217
|
+
"type": "keyword"
|
|
218
|
+
},
|
|
219
|
+
"name": {
|
|
220
|
+
"ignore_above": 8191,
|
|
221
|
+
"type": "keyword"
|
|
222
|
+
},
|
|
223
|
+
"prov": {
|
|
224
|
+
"properties": {
|
|
225
|
+
"page": {
|
|
226
|
+
"type": "integer"
|
|
227
|
+
}
|
|
228
|
+
}
|
|
229
|
+
}
|
|
230
|
+
}
|
|
231
|
+
},
|
|
232
|
+
"tables": {
|
|
233
|
+
"properties": {
|
|
234
|
+
"text": {
|
|
235
|
+
"type": "text"
|
|
236
|
+
},
|
|
237
|
+
"type": {
|
|
238
|
+
"ignore_above": 8191,
|
|
239
|
+
"type": "keyword"
|
|
240
|
+
},
|
|
241
|
+
"prov": {
|
|
242
|
+
"properties": {
|
|
243
|
+
"page": {
|
|
244
|
+
"type": "integer"
|
|
245
|
+
}
|
|
246
|
+
}
|
|
247
|
+
}
|
|
248
|
+
}
|
|
249
|
+
},
|
|
250
|
+
"type": {
|
|
251
|
+
"ignore_above": 8191,
|
|
252
|
+
"type": "keyword"
|
|
253
|
+
}
|
|
254
|
+
}
|
|
255
|
+
}
|
|
256
|
+
}
|