Weaviate
Batch process all your records using unstructured-ingest to store structured outputs locally on your filesystem and upload those local files to a Weaviate collection.
First you’ll need to install the weaviate dependencies as shown here.
pip install "unstructured[weaviate]"
Run Locally
The upstream connector can be any of the ones supported, but for convenience here, showing a sample command using the upstream local connector. This will push elements into a collection schema of your choice into a weaviate instance running locally.
#!/usr/bin/env bash
EMBEDDING_PROVIDER=${EMBEDDING_PROVIDER:-"langchain-huggingface"}
unstructured-ingest \
local \
--input-path example-docs/book-war-and-peace-1225p.txt \
--output-dir local-output-to-weaviate \
--strategy fast \
--chunk-elements \
--embedding-provider "$EMBEDDING_PROVIDER" \
--num-processes 2 \
--verbose \
--strategy fast \
weaviate \
--host-url http://localhost:8080 \
--class-name elements
from unstructured.ingest.connector.local import SimpleLocalConfig
from unstructured.ingest.connector.weaviate import (
SimpleWeaviateConfig,
WeaviateAccessConfig,
WeaviateWriteConfig,
)
from unstructured.ingest.interfaces import (
ChunkingConfig,
EmbeddingConfig,
PartitionConfig,
ProcessorConfig,
ReadConfig,
)
from unstructured.ingest.runner import LocalRunner
from unstructured.ingest.runner.writers.base_writer import Writer
from unstructured.ingest.runner.writers.weaviate import (
WeaviateWriter,
)
def get_writer() -> Writer:
return WeaviateWriter(
connector_config=SimpleWeaviateConfig(
access_config=WeaviateAccessConfig(),
host_url="http://localhost:8080",
class_name="elements",
),
write_config=WeaviateWriteConfig(),
)
if __name__ == "__main__":
writer = get_writer()
runner = LocalRunner(
processor_config=ProcessorConfig(
verbose=True,
output_dir="local-output-to-weaviate",
num_processes=2,
),
connector_config=SimpleLocalConfig(
input_path="example-docs/book-war-and-peace-1225p.txt",
),
read_config=ReadConfig(),
partition_config=PartitionConfig(),
chunking_config=ChunkingConfig(chunk_elements=True),
embedding_config=EmbeddingConfig(
provider="langchain-huggingface",
),
writer=writer,
writer_kwargs={},
)
runner.run()
For a full list of the options the CLI accepts check unstructured-ingest <upstream connector> weaviate --help.
NOTE: Keep in mind that you will need to have all the appropriate extras and dependencies for the file types of the documents contained in your data storage platform if you’re running this locally. You can find more information about this in the installation guide.
Sample Index Schema
To make sure the schema of the index matches the data being written to it, a sample schema json can be used:
Object description
1{
2 "class": "Elements",
3 "invertedIndexConfig": {
4 "bm25": {
5 "b": 0.75,
6 "k1": 1.2
7 },
8 "cleanupIntervalSeconds": 60,
9 "stopwords": {
10 "additions": null,
11 "preset": "en",
12 "removals": null
13 }
14 },
15 "multiTenancyConfig": {
16 "enabled": false
17 },
18 "properties": [
19 {
20 "dataType": [
21 "text"
22 ],
23 "indexFilterable": true,
24 "indexSearchable": true,
25 "name": "element_id",
26 "tokenization": "word"
27 },
28 {
29 "dataType": [
30 "text"
31 ],
32 "indexFilterable": true,
33 "indexSearchable": true,
34 "name": "text",
35 "tokenization": "word"
36 },
37 {
38 "dataType": [
39 "text"
40 ],
41 "indexFilterable": true,
42 "indexSearchable": true,
43 "name": "type",
44 "tokenization": "word"
45 },
46 {
47 "dataType": [
48 "object"
49 ],
50 "indexFilterable": true,
51 "indexSearchable": false,
52 "name": "metadata",
53 "nestedProperties": [
54 {
55 "dataType": [
56 "int"
57 ],
58 "indexFilterable": true,
59 "indexSearchable": false,
60 "name": "category_depth"
61 },
62 {
63 "dataType": [
64 "text"
65 ],
66 "indexFilterable": true,
67 "indexSearchable": true,
68 "name": "parent_id",
69 "tokenization": "word"
70 },
71 {
72 "dataType": [
73 "text"
74 ],
75 "indexFilterable": true,
76 "indexSearchable": true,
77 "name": "attached_to_filename",
78 "tokenization": "word"
79 },
80 {
81 "dataType": [
82 "text"
83 ],
84 "indexFilterable": true,
85 "indexSearchable": true,
86 "name": "filetype",
87 "tokenization": "word"
88 },
89 {
90 "dataType": [
91 "date"
92 ],
93 "indexFilterable": true,
94 "indexSearchable": false,
95 "name": "last_modified"
96 },
97 {
98 "dataType": [
99 "text"
100 ],
101 "indexFilterable": true,
102 "indexSearchable": true,
103 "name": "file_directory",
104 "tokenization": "word"
105 },
106 {
107 "dataType": [
108 "text"
109 ],
110 "indexFilterable": true,
111 "indexSearchable": true,
112 "name": "filename",
113 "tokenization": "word"
114 },
115 {
116 "dataType": [
117 "object"
118 ],
119 "indexFilterable": true,
120 "indexSearchable": false,
121 "name": "data_source",
122 "nestedProperties": [
123 {
124 "dataType": [
125 "text"
126 ],
127 "indexFilterable": true,
128 "indexSearchable": true,
129 "name": "url",
130 "tokenization": "word"
131 },
132 {
133 "dataType": [
134 "text"
135 ],
136 "indexFilterable": true,
137 "indexSearchable": true,
138 "name": "version",
139 "tokenization": "word"
140 },
141 {
142 "dataType": [
143 "date"
144 ],
145 "indexFilterable": true,
146 "indexSearchable": false,
147 "name": "date_created"
148 },
149 {
150 "dataType": [
151 "date"
152 ],
153 "indexFilterable": true,
154 "indexSearchable": false,
155 "name": "date_modified"
156 },
157 {
158 "dataType": [
159 "date"
160 ],
161 "indexFilterable": true,
162 "indexSearchable": false,
163 "name": "date_processed"
164 },
165 {
166 "dataType": [
167 "text"
168 ],
169 "indexFilterable": true,
170 "indexSearchable": true,
171 "name": "record_locator",
172 "tokenization": "word"
173 },
174 {
175 "dataType": [
176 "text"
177 ],
178 "indexFilterable": true,
179 "indexSearchable": true,
180 "name": "permissions_data",
181 "tokenization": "word"
182 }
183
184 ]
185 },
186 {
187 "dataType": [
188 "object"
189 ],
190 "indexFilterable": true,
191 "indexSearchable": false,
192 "name": "coordinates",
193 "nestedProperties": [
194 {
195 "dataType": [
196 "text"
197 ],
198 "indexFilterable": true,
199 "indexSearchable": true,
200 "name": "system",
201 "tokenization": "word"
202 },
203 {
204 "dataType": [
205 "number"
206 ],
207 "indexFilterable": true,
208 "indexSearchable": false,
209 "name": "layout_width"
210 },
211 {
212 "dataType": [
213 "number"
214 ],
215 "indexFilterable": true,
216 "indexSearchable": false,
217 "name": "layout_height"
218 },
219 {
220 "dataType": [
221 "text"
222 ],
223 "indexFilterable": true,
224 "indexSearchable": true,
225 "name": "points",
226 "tokenization": "word"
227 }
228 ]
229 },
230 {
231 "dataType": [
232 "text[]"
233 ],
234 "indexFilterable": true,
235 "indexSearchable": true,
236 "name": "languages",
237 "tokenization": "word"
238 },
239 {
240 "dataType": [
241 "text"
242 ],
243 "indexFilterable": true,
244 "indexSearchable": false,
245 "name": "page_number"
246 },
247 {
248 "dataType": [
249 "text"
250 ],
251 "indexFilterable": true,
252 "indexSearchable": true,
253 "name": "page_name",
254 "tokenization": "word"
255 },
256 {
257 "dataType": [
258 "text"
259 ],
260 "indexFilterable": true,
261 "indexSearchable": true,
262 "name": "url",
263 "tokenization": "word"
264 },
265 {
266 "dataType": [
267 "text"
268 ],
269 "indexFilterable": true,
270 "indexSearchable": true,
271 "name": "links",
272 "tokenization": "word"
273 },
274 {
275 "dataType": [
276 "text[]"
277 ],
278 "indexFilterable": true,
279 "indexSearchable": true,
280 "name": "link_urls",
281 "tokenization": "word"
282 },
283 {
284 "dataType": [
285 "text[]"
286 ],
287 "indexFilterable": true,
288 "indexSearchable": true,
289 "name": "link_texts",
290 "tokenization": "word"
291 },
292 {
293 "dataType": [
294 "text"
295 ],
296 "indexFilterable": true,
297 "indexSearchable": true,
298 "name": "sent_from",
299 "tokenization": "word"
300 },
301 {
302 "dataType": [
303 "text"
304 ],
305 "indexFilterable": true,
306 "indexSearchable": true,
307 "name": "sent_to",
308 "tokenization": "word"
309 },
310 {
311 "dataType": [
312 "text"
313 ],
314 "indexFilterable": true,
315 "indexSearchable": true,
316 "name": "subject",
317 "tokenization": "word"
318 },
319 {
320 "dataType": [
321 "text"
322 ],
323 "indexFilterable": true,
324 "indexSearchable": true,
325 "name": "section",
326 "tokenization": "word"
327 },
328 {
329 "dataType": [
330 "text"
331 ],
332 "indexFilterable": true,
333 "indexSearchable": true,
334 "name": "header_footer_type",
335 "tokenization": "word"
336 },
337 {
338 "dataType": [
339 "text[]"
340 ],
341 "indexFilterable": true,
342 "indexSearchable": true,
343 "name": "emphasized_text_contents",
344 "tokenization": "word"
345 },
346 {
347 "dataType": [
348 "text[]"
349 ],
350 "indexFilterable": true,
351 "indexSearchable": true,
352 "name": "emphasized_text_tags",
353 "tokenization": "word"
354 },
355 {
356 "dataType": [
357 "text"
358 ],
359 "indexFilterable": true,
360 "indexSearchable": true,
361 "name": "text_as_html",
362 "tokenization": "word"
363 },
364 {
365 "dataType": [
366 "text"
367 ],
368 "indexFilterable": true,
369 "indexSearchable": true,
370 "name": "regex_metadata",
371 "tokenization": "word"
372 },
373 {
374 "dataType": [
375 "number"
376 ],
377 "indexFilterable": true,
378 "indexSearchable": false,
379 "name": "detection_class_prob"
380 }
381 ]
382 }
383 ],
384 "replicationConfig": {
385 "factor": 1
386 },
387 "shardingConfig": {
388 "virtualPerPhysical": 128,
389 "desiredCount": 1,
390 "actualCount": 1,
391 "desiredVirtualCount": 128,
392 "actualVirtualCount": 128,
393 "key": "_id",
394 "strategy": "hash",
395 "function": "murmur3"
396 },
397 "vectorIndexConfig": {
398 "skip": false,
399 "cleanupIntervalSeconds": 300,
400 "maxConnections": 64,
401 "efConstruction": 128,
402 "ef": -1,
403 "dynamicEfMin": 100,
404 "dynamicEfMax": 500,
405 "dynamicEfFactor": 8,
406 "vectorCacheMaxObjects": 1000000000000,
407 "flatSearchCutoff": 40000,
408 "distance": "cosine",
409 "pq": {
410 "enabled": false,
411 "bitCompression": false,
412 "segments": 0,
413 "centroids": 256,
414 "trainingLimit": 100000,
415 "encoder": {
416 "type": "kmeans",
417 "distribution": "log-normal"
418 }
419 }
420 },
421 "vectorIndexType": "hnsw",
422 "vectorizer": "none"
423}