unstructured-ingest 0.0.0__py3-none-any.whl → 0.0.2__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 unstructured-ingest might be problematic. Click here for more details.

Files changed (44) hide show
  1. unstructured_ingest/__version__.py +1 -1
  2. unstructured_ingest/connector/notion/helpers.py +1 -1
  3. unstructured_ingest/logger.py +2 -2
  4. unstructured_ingest/v2/cli/base/cmd.py +10 -0
  5. unstructured_ingest/v2/cli/base/src.py +2 -0
  6. unstructured_ingest/v2/cli/cmds/__init__.py +2 -0
  7. unstructured_ingest/v2/cli/cmds/fsspec/fsspec.py +1 -9
  8. unstructured_ingest/v2/cli/cmds/local.py +0 -8
  9. unstructured_ingest/v2/cli/cmds/milvus.py +72 -0
  10. unstructured_ingest/v2/cli/configs/__init__.py +8 -1
  11. unstructured_ingest/v2/cli/configs/filter.py +28 -0
  12. unstructured_ingest/v2/interfaces/__init__.py +2 -1
  13. unstructured_ingest/v2/interfaces/downloader.py +9 -3
  14. unstructured_ingest/v2/interfaces/file_data.py +6 -1
  15. unstructured_ingest/v2/interfaces/process.py +3 -0
  16. unstructured_ingest/v2/logger.py +1 -1
  17. unstructured_ingest/v2/pipeline/interfaces.py +3 -1
  18. unstructured_ingest/v2/pipeline/pipeline.py +72 -2
  19. unstructured_ingest/v2/pipeline/steps/download.py +77 -13
  20. unstructured_ingest/v2/pipeline/steps/filter.py +40 -0
  21. unstructured_ingest/v2/processes/connectors/__init__.py +4 -2
  22. unstructured_ingest/v2/processes/connectors/astra.py +8 -0
  23. unstructured_ingest/v2/processes/connectors/azure_cognitive_search.py +8 -0
  24. unstructured_ingest/v2/processes/connectors/chroma.py +8 -6
  25. unstructured_ingest/v2/processes/connectors/databricks_volumes.py +9 -0
  26. unstructured_ingest/v2/processes/connectors/elasticsearch.py +23 -9
  27. unstructured_ingest/v2/processes/connectors/fsspec/fsspec.py +22 -31
  28. unstructured_ingest/v2/processes/connectors/fsspec/s3.py +13 -5
  29. unstructured_ingest/v2/processes/connectors/google_drive.py +13 -9
  30. unstructured_ingest/v2/processes/connectors/local.py +15 -15
  31. unstructured_ingest/v2/processes/connectors/milvus.py +200 -0
  32. unstructured_ingest/v2/processes/connectors/mongodb.py +10 -4
  33. unstructured_ingest/v2/processes/connectors/onedrive.py +14 -2
  34. unstructured_ingest/v2/processes/connectors/pinecone.py +10 -7
  35. unstructured_ingest/v2/processes/connectors/salesforce.py +10 -8
  36. unstructured_ingest/v2/processes/connectors/sharepoint.py +14 -8
  37. unstructured_ingest/v2/processes/connectors/sql.py +24 -9
  38. unstructured_ingest/v2/processes/connectors/weaviate.py +13 -5
  39. unstructured_ingest/v2/processes/filter.py +54 -0
  40. {unstructured_ingest-0.0.0.dist-info → unstructured_ingest-0.0.2.dist-info}/METADATA +16 -14
  41. {unstructured_ingest-0.0.0.dist-info → unstructured_ingest-0.0.2.dist-info}/RECORD +44 -39
  42. {unstructured_ingest-0.0.0.dist-info → unstructured_ingest-0.0.2.dist-info}/WHEEL +0 -0
  43. {unstructured_ingest-0.0.0.dist-info → unstructured_ingest-0.0.2.dist-info}/entry_points.txt +0 -0
  44. {unstructured_ingest-0.0.0.dist-info → unstructured_ingest-0.0.2.dist-info}/top_level.txt +0 -0
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.1
2
2
  Name: unstructured-ingest
3
- Version: 0.0.0
3
+ Version: 0.0.2
4
4
  Summary: A library that prepares raw documents for downstream ML tasks.
5
5
  Home-page: https://github.com/Unstructured-IO/unstructured-ingest
6
6
  Author: Unstructured Technologies
@@ -21,9 +21,9 @@ Classifier: Programming Language :: Python :: 3.12
21
21
  Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
22
22
  Requires-Python: >=3.9.0,<3.13
23
23
  Description-Content-Type: text/markdown
24
- Requires-Dist: pandas
25
24
  Requires-Dist: unstructured
26
25
  Requires-Dist: python-dateutil
26
+ Requires-Dist: pandas
27
27
  Provides-Extra: airtable
28
28
  Requires-Dist: pyairtable ; extra == 'airtable'
29
29
  Provides-Extra: astra
@@ -34,8 +34,8 @@ Requires-Dist: adlfs ; extra == 'azure'
34
34
  Provides-Extra: azure-cognitive-search
35
35
  Requires-Dist: azure-search-documents ; extra == 'azure-cognitive-search'
36
36
  Provides-Extra: bedrock
37
- Requires-Dist: boto3 ; extra == 'bedrock'
38
37
  Requires-Dist: langchain-community ; extra == 'bedrock'
38
+ Requires-Dist: boto3 ; extra == 'bedrock'
39
39
  Provides-Extra: biomed
40
40
  Requires-Dist: bs4 ; extra == 'biomed'
41
41
  Provides-Extra: box
@@ -54,8 +54,8 @@ Requires-Dist: unstructured[tsv] ; extra == 'csv'
54
54
  Provides-Extra: databricks-volumes
55
55
  Requires-Dist: databricks-sdk ; extra == 'databricks-volumes'
56
56
  Provides-Extra: delta-table
57
- Requires-Dist: fsspec ; extra == 'delta-table'
58
57
  Requires-Dist: deltalake ; extra == 'delta-table'
58
+ Requires-Dist: fsspec ; extra == 'delta-table'
59
59
  Provides-Extra: discord
60
60
  Requires-Dist: discord-py ; extra == 'discord'
61
61
  Provides-Extra: doc
@@ -69,24 +69,24 @@ Provides-Extra: elasticsearch
69
69
  Requires-Dist: elasticsearch[async] ; extra == 'elasticsearch'
70
70
  Provides-Extra: embed-huggingface
71
71
  Requires-Dist: huggingface ; extra == 'embed-huggingface'
72
- Requires-Dist: langchain-community ; extra == 'embed-huggingface'
73
72
  Requires-Dist: sentence-transformers ; extra == 'embed-huggingface'
73
+ Requires-Dist: langchain-community ; extra == 'embed-huggingface'
74
74
  Provides-Extra: embed-octoai
75
- Requires-Dist: openai ; extra == 'embed-octoai'
76
75
  Requires-Dist: tiktoken ; extra == 'embed-octoai'
76
+ Requires-Dist: openai ; extra == 'embed-octoai'
77
77
  Provides-Extra: embed-vertexai
78
- Requires-Dist: langchain-google-vertexai ; extra == 'embed-vertexai'
79
78
  Requires-Dist: langchain-community ; extra == 'embed-vertexai'
80
79
  Requires-Dist: langchain ; extra == 'embed-vertexai'
80
+ Requires-Dist: langchain-google-vertexai ; extra == 'embed-vertexai'
81
81
  Provides-Extra: embed-voyageai
82
- Requires-Dist: langchain-voyageai ; extra == 'embed-voyageai'
83
82
  Requires-Dist: langchain ; extra == 'embed-voyageai'
83
+ Requires-Dist: langchain-voyageai ; extra == 'embed-voyageai'
84
84
  Provides-Extra: epub
85
85
  Requires-Dist: unstructured[epub] ; extra == 'epub'
86
86
  Provides-Extra: gcs
87
- Requires-Dist: fsspec ; extra == 'gcs'
88
87
  Requires-Dist: bs4 ; extra == 'gcs'
89
88
  Requires-Dist: gcsfs ; extra == 'gcs'
89
+ Requires-Dist: fsspec ; extra == 'gcs'
90
90
  Provides-Extra: github
91
91
  Requires-Dist: pygithub >1.58.0 ; extra == 'github'
92
92
  Provides-Extra: gitlab
@@ -94,21 +94,23 @@ Requires-Dist: python-gitlab ; extra == 'gitlab'
94
94
  Provides-Extra: google-drive
95
95
  Requires-Dist: google-api-python-client ; extra == 'google-drive'
96
96
  Provides-Extra: hubspot
97
- Requires-Dist: urllib3 ; extra == 'hubspot'
98
97
  Requires-Dist: hubspot-api-client ; extra == 'hubspot'
98
+ Requires-Dist: urllib3 ; extra == 'hubspot'
99
99
  Provides-Extra: jira
100
100
  Requires-Dist: atlassian-python-api ; extra == 'jira'
101
101
  Provides-Extra: kafka
102
102
  Requires-Dist: confluent-kafka ; extra == 'kafka'
103
103
  Provides-Extra: md
104
104
  Requires-Dist: unstructured[md] ; extra == 'md'
105
+ Provides-Extra: milvus
106
+ Requires-Dist: pymilvus ; extra == 'milvus'
105
107
  Provides-Extra: mongodb
106
108
  Requires-Dist: pymongo ; extra == 'mongodb'
107
109
  Provides-Extra: msg
108
110
  Requires-Dist: unstructured[msg] ; extra == 'msg'
109
111
  Provides-Extra: notion
110
- Requires-Dist: notion-client ; extra == 'notion'
111
112
  Requires-Dist: htmlBuilder ; extra == 'notion'
113
+ Requires-Dist: notion-client ; extra == 'notion'
112
114
  Provides-Extra: odt
113
115
  Requires-Dist: unstructured[odt] ; extra == 'odt'
114
116
  Provides-Extra: onedrive
@@ -116,8 +118,8 @@ Requires-Dist: msal ; extra == 'onedrive'
116
118
  Requires-Dist: bs4 ; extra == 'onedrive'
117
119
  Requires-Dist: Office365-REST-Python-Client ; extra == 'onedrive'
118
120
  Provides-Extra: openai
119
- Requires-Dist: openai ; extra == 'openai'
120
121
  Requires-Dist: tiktoken ; extra == 'openai'
122
+ Requires-Dist: openai ; extra == 'openai'
121
123
  Requires-Dist: langchain-community ; extra == 'openai'
122
124
  Provides-Extra: opensearch
123
125
  Requires-Dist: opensearch-py ; extra == 'opensearch'
@@ -145,13 +147,13 @@ Requires-Dist: unstructured[rst] ; extra == 'rst'
145
147
  Provides-Extra: rtf
146
148
  Requires-Dist: unstructured[rtf] ; extra == 'rtf'
147
149
  Provides-Extra: s3
148
- Requires-Dist: fsspec ; extra == 's3'
149
150
  Requires-Dist: s3fs ; extra == 's3'
151
+ Requires-Dist: fsspec ; extra == 's3'
150
152
  Provides-Extra: salesforce
151
153
  Requires-Dist: simple-salesforce ; extra == 'salesforce'
152
154
  Provides-Extra: sftp
153
- Requires-Dist: fsspec ; extra == 'sftp'
154
155
  Requires-Dist: paramiko ; extra == 'sftp'
156
+ Requires-Dist: fsspec ; extra == 'sftp'
155
157
  Provides-Extra: sharepoint
156
158
  Requires-Dist: msal ; extra == 'sharepoint'
157
159
  Requires-Dist: Office365-REST-Python-Client ; extra == 'sharepoint'
@@ -1,9 +1,9 @@
1
1
  unstructured_ingest/__init__.py,sha256=U4S_2y3zgLZVfMenHRaJFBW8yqh2mUBuI291LGQVOJ8,35
2
- unstructured_ingest/__version__.py,sha256=Q49HKCptFgT2OeWLy_cQ73sq2sMFreeYIt3GaSMpXf8,42
2
+ unstructured_ingest/__version__.py,sha256=t0CFzEk7qlIWbgyEWA53ytTKmHbZ9ow2lAyjeP1bFqw,42
3
3
  unstructured_ingest/error.py,sha256=qDncnJgbf5ils956RcO2CGlAKYDT5OaEM9Clv1JVTNc,1448
4
4
  unstructured_ingest/evaluate.py,sha256=R-mKLFXbVX1xQ1tjGsLHjdP-TbSSV-925IHzggW_bIg,9793
5
5
  unstructured_ingest/interfaces.py,sha256=uS8L5mS0mXD8I4XTfVlKZxAwqnpJ4yrRqn4vxWVRhQI,31107
6
- unstructured_ingest/logger.py,sha256=P5KVgFSRN4uSSNmf5S00zr_TdlL7uAhjxn_26tcNWxI,4480
6
+ unstructured_ingest/logger.py,sha256=TrhyH7VbCWO5VVuhvL0yUyXxuem3b4pzbqj2uQHUwZk,4480
7
7
  unstructured_ingest/main.py,sha256=82G_7eG4PNhc_xIqj4Y_sFbDV9VI-nwSfsfJQMzovMk,169
8
8
  unstructured_ingest/processor.py,sha256=XKKrvbxsb--5cDzz4hB3-GfWZYyIjJ2ah8FpzQKF_DM,2760
9
9
  unstructured_ingest/cli/__init__.py,sha256=9kNcBOHuXON5lB1MJU9QewEhwPmId56vXqB29-kqEAA,302
@@ -104,7 +104,7 @@ unstructured_ingest/connector/fsspec/sftp.py,sha256=x2w8JGM81S_HXww7Aa-bTY1LjZSi
104
104
  unstructured_ingest/connector/notion/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
105
105
  unstructured_ingest/connector/notion/client.py,sha256=vU1GE64ktEAM4b-jo8UnMAwz60KSiQ6iRI3De3ixNdI,8689
106
106
  unstructured_ingest/connector/notion/connector.py,sha256=E-t7q5XAiYP9xk-1aqIqcGwdJOH8UNgiE0HcH9Oc4i4,17475
107
- unstructured_ingest/connector/notion/helpers.py,sha256=5SbQbNxIenMHyxEAMfrsVsXpNcAKPHo3gwWQVi1NUOc,20702
107
+ unstructured_ingest/connector/notion/helpers.py,sha256=jqg5-cPcrjm5G7dBF4jOjHxdJN1CI7yBeTefiX4hBoM,20702
108
108
  unstructured_ingest/connector/notion/interfaces.py,sha256=SrTT-9c0nvk0fMqVgudYF647r04AdMKi6wkIkMy7Szw,563
109
109
  unstructured_ingest/connector/notion/types/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
110
110
  unstructured_ingest/connector/notion/types/block.py,sha256=AKOY-o6CTFC-caWlkLfKskMuFemH4-Vdrhv7HnRkS8w,3009
@@ -253,25 +253,26 @@ unstructured_ingest/utils/string_and_date_utils.py,sha256=hnGglD8Z626vLhH_UV4Qyb
253
253
  unstructured_ingest/utils/table.py,sha256=aWjcowDVSClNpEAdR6PY3H7khKu4T6T3QqQE6GjmQ_M,3469
254
254
  unstructured_ingest/v2/__init__.py,sha256=U4S_2y3zgLZVfMenHRaJFBW8yqh2mUBuI291LGQVOJ8,35
255
255
  unstructured_ingest/v2/example.py,sha256=qkwmpMxUlaJXdDNKQ4LlUt3XGxgTUU3CXGGO57eW5Gs,1644
256
- unstructured_ingest/v2/logger.py,sha256=tI_PtlibnJmv1MMajHT5GSZhQ77dv30UADEWaXWgynA,4324
256
+ unstructured_ingest/v2/logger.py,sha256=akcghdHwpKM3CfoeFzir0zmc7R9Hk7zjquU-X-gwUIw,4324
257
257
  unstructured_ingest/v2/main.py,sha256=WFdLEqEXRy6E9_G-dF20MK2AtgX51Aan1sp_N67U2B8,172
258
258
  unstructured_ingest/v2/cli/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
259
259
  unstructured_ingest/v2/cli/cli.py,sha256=qHXIs-PcvMgDZhP1AR9iDMxh8FXBMJCEDksPBfiMULE,648
260
260
  unstructured_ingest/v2/cli/interfaces.py,sha256=4Bbow6QHks2a1H56tmVQ4vG3sZy-577ZbwrPmDfizmE,829
261
261
  unstructured_ingest/v2/cli/utils.py,sha256=QK-ee6FzxPf-IbaNXXWlH-GhvqeNnjK2m8ljBD1SusU,9075
262
262
  unstructured_ingest/v2/cli/base/__init__.py,sha256=zXCa7F4FMqItmzxfUIVmyI-CeGh8X85yF8lRxwX_OYQ,83
263
- unstructured_ingest/v2/cli/base/cmd.py,sha256=qVHmquVsVDoYyPByKdUTVCwAFfILMYBw5w6eTTVku-E,9308
263
+ unstructured_ingest/v2/cli/base/cmd.py,sha256=qi9N5rcyyE2nmswFaoKWbs1PonhHsMC5llqND9-rQso,9790
264
264
  unstructured_ingest/v2/cli/base/dest.py,sha256=YMbVIHmYDqvOtxZeEY93stmF2p2ImjuJts7-u-NznYw,2887
265
265
  unstructured_ingest/v2/cli/base/importer.py,sha256=nRt0QQ3qpi264-n_mR0l55C2ddM8nowTNzT1jsWaam8,1128
266
- unstructured_ingest/v2/cli/base/src.py,sha256=7LnZh9FgUX9rerBH6cizVtTWmM6R2sRkxatnGsxYHG0,2410
267
- unstructured_ingest/v2/cli/cmds/__init__.py,sha256=aOcJb2FLQaUOU-vdu4xHr5_BJQme6ADlPaRjCSHL1Io,2590
266
+ unstructured_ingest/v2/cli/base/src.py,sha256=oUPO9GPEbkYm1udfD4YQBYTfaefbhpoIN1HPnD672SQ,2460
267
+ unstructured_ingest/v2/cli/cmds/__init__.py,sha256=DWPMD6Wqus22sSoIEyTSiOJAm97aNjvdpdrXgsL4uQ0,2647
268
268
  unstructured_ingest/v2/cli/cmds/astra.py,sha256=L-GR2KSP_cFQkQm0aVcdiXmgYMJZCVKIAH794y8qT1M,2590
269
269
  unstructured_ingest/v2/cli/cmds/azure_cognitive_search.py,sha256=VTCSUYeIYKnP60lC7DeBYqoqAJnWuBZrwevCXbeIEzw,2248
270
270
  unstructured_ingest/v2/cli/cmds/chroma.py,sha256=RinNOPripk2zRYx1Rt-u-jywXbwh7JsidVia4F0-wyU,3359
271
271
  unstructured_ingest/v2/cli/cmds/databricks_volumes.py,sha256=53d9A7UunJLYZFwwwHEraVshFc3gSzUbmKjMOiv7hn4,5920
272
272
  unstructured_ingest/v2/cli/cmds/elasticsearch.py,sha256=joUfnV992fAwEDCtFVJaABwgpyQiWeDl1ZCBEudRtnk,5258
273
273
  unstructured_ingest/v2/cli/cmds/google_drive.py,sha256=mXozabpi8kjRFb0S7kw-xMGtEuFoVUxnvefwL5ZIPHc,2334
274
- unstructured_ingest/v2/cli/cmds/local.py,sha256=lGBFOVDRlrcCtPFjyk0IAYHLRWg95Kunu1Kli7t0ZK4,1899
274
+ unstructured_ingest/v2/cli/cmds/local.py,sha256=UOTYjSdNqCFxhZfN6bdxm8jRp6Ijun2K-WpQq1X83OQ,1544
275
+ unstructured_ingest/v2/cli/cmds/milvus.py,sha256=PB1ib1rFGGH_-KDi1bSIO3BIiVcqSJEHCBFFrzQrnmI,1998
275
276
  unstructured_ingest/v2/cli/cmds/mongodb.py,sha256=oyV6tacuuxm3dN-AXQgbxvYJiDYo2OOWQKRSBCUGj0E,1823
276
277
  unstructured_ingest/v2/cli/cmds/onedrive.py,sha256=DKqhQyyF-swZxs3C9G5W8ECleq8sWpDbpTuiAHXukXQ,2781
277
278
  unstructured_ingest/v2/cli/cmds/opensearch.py,sha256=7zl8dUXzxs24MDRRASKfNc14IDM798qOXRl2FZdXG1I,3064
@@ -285,32 +286,34 @@ unstructured_ingest/v2/cli/cmds/fsspec/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JC
285
286
  unstructured_ingest/v2/cli/cmds/fsspec/azure.py,sha256=ZHfchzSpGkZ99Fq1050JvHP0-aG1pZsBZxxozcFfxwI,2784
286
287
  unstructured_ingest/v2/cli/cmds/fsspec/box.py,sha256=kslkI-0-GyGSJOU7bKgrZeQRXh8HFexDq87ew8kT8kE,1338
287
288
  unstructured_ingest/v2/cli/cmds/fsspec/dropbox.py,sha256=LtcR3rCQPgzJNbV3S90HlL0LPPbW9lYEfE8BG4F-dSI,1349
288
- unstructured_ingest/v2/cli/cmds/fsspec/fsspec.py,sha256=Mgr_nto1FWxnGmbGdVlIfm-xuBGL0HEi8k3FEmQnZng,2414
289
+ unstructured_ingest/v2/cli/cmds/fsspec/fsspec.py,sha256=BlJLEu6YJUejLLFzlSCVSoZDp2pdjoTsdoFFHVwwkVY,2031
289
290
  unstructured_ingest/v2/cli/cmds/fsspec/gcs.py,sha256=3-0LYnDs0fgNrDqnHpNZKj_6rwNj9wQVaV0lGOhVFPE,2737
290
291
  unstructured_ingest/v2/cli/cmds/fsspec/s3.py,sha256=EXQzYkDtkFli2sfcj4cRDRPFac7b7z1DfQqYlGQcE6o,2279
291
292
  unstructured_ingest/v2/cli/cmds/fsspec/sftp.py,sha256=YY2xKguawMyLdcG0qDYKUgk7DT0KgyZJlV17MfwIhpo,2036
292
- unstructured_ingest/v2/cli/configs/__init__.py,sha256=5NMXm872QQZTvUFZFS06c8c1b6K940K5gxs9lbp8W6M,258
293
+ unstructured_ingest/v2/cli/configs/__init__.py,sha256=nAJ1gT3yCAzoZbEbYswE2SMfSk7TEPxa_1v3qEUsgIQ,336
293
294
  unstructured_ingest/v2/cli/configs/chunk.py,sha256=KvIhmIRIZxazCumMztAKdWs-4MK7qzOb5h6Ned_2bdU,3547
294
295
  unstructured_ingest/v2/cli/configs/embed.py,sha256=q_TwnkxKTKOsMgVYfW6xxbD8FWjU_Uh_X2BQ5-_VLGM,2725
296
+ unstructured_ingest/v2/cli/configs/filter.py,sha256=KYe65_au6m7H4VrjgugC2ain6vsUSWswNSEgcG66VPU,841
295
297
  unstructured_ingest/v2/cli/configs/partition.py,sha256=7wdI18V6c4kaXuf50Lh66n9LbtrYHYd8ffEgDQLqvSk,3931
296
298
  unstructured_ingest/v2/cli/configs/processor.py,sha256=ZHu2DBIuE8VgL3mEt73yYimw2k_PaOEtdxxFqzHfk84,3350
297
- unstructured_ingest/v2/interfaces/__init__.py,sha256=-CHWUlT4rISd-gSfcGKGYFqqSFhMY9lKsT5wxwmOThM,845
299
+ unstructured_ingest/v2/interfaces/__init__.py,sha256=Rfa8crx6De7WNOK-EjsWWwFVpsUfCc6gY8B8tQ3ae9I,899
298
300
  unstructured_ingest/v2/interfaces/connector.py,sha256=u4hE1DpTPDC04-n_IzYyn9w1gNCiPT81anrUoEh30Z8,855
299
- unstructured_ingest/v2/interfaces/downloader.py,sha256=aWlacZZrI6SGw6retnRJtZbqT5voOYq_fb326ynNOhI,2506
300
- unstructured_ingest/v2/interfaces/file_data.py,sha256=5TCMkblUW-Jvy-rS5FqRT22VzDmJqAiQRIWYarpAi64,1543
301
+ unstructured_ingest/v2/interfaces/downloader.py,sha256=zs7cxhzbWVc5L0bV4gdCTexWGMVeXTQ9jJF6PCYSAss,2790
302
+ unstructured_ingest/v2/interfaces/file_data.py,sha256=PZrPJBkNC63lNO_1nwvnAeKRxjM3CsjIY6jSO8T9bVM,1665
301
303
  unstructured_ingest/v2/interfaces/indexer.py,sha256=pMw0abNHk_tEuA4BkXX1BdAfIwHdytxj7s6tGxMvYRE,821
302
- unstructured_ingest/v2/interfaces/process.py,sha256=0ecz7mAjlY_DUi9-HhPc9zXphmGclispYwv37O8gvJ0,466
304
+ unstructured_ingest/v2/interfaces/process.py,sha256=_l4dyaM0u0XxTqQw1Ghr8k2QMpQJMFapLOLhWqSdTdo,512
303
305
  unstructured_ingest/v2/interfaces/processor.py,sha256=uHVHeKo5Gt_zFkaEXw7xgaCBDTEl2-Amh-ByA07258o,1620
304
306
  unstructured_ingest/v2/interfaces/upload_stager.py,sha256=SylhDl9pK6qa7hvfrhpabCkjwE03yIlI6oM-mQnqtho,1220
305
307
  unstructured_ingest/v2/interfaces/uploader.py,sha256=bzfx3Ei4poXKu-hsgjAB4sj4jKij9CoaRSadUM5LtGk,1083
306
308
  unstructured_ingest/v2/pipeline/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
307
- unstructured_ingest/v2/pipeline/interfaces.py,sha256=3zRcu6sc-2rnm3gALOzXA9mI39m2RUPrREZGTd9x77c,6352
308
- unstructured_ingest/v2/pipeline/pipeline.py,sha256=r8jRMZI2RF8GQIuTcjIFBDeFtMnqpOJmKhEriy6Vo5Y,11616
309
+ unstructured_ingest/v2/pipeline/interfaces.py,sha256=Z50-6XFZNajfmJbLKunLxw3RuYMzCYiUp6F0jhQwERE,6441
310
+ unstructured_ingest/v2/pipeline/pipeline.py,sha256=dqn4_O4il6gZ33mE0DVC1wQKRcXMrD_jll999NoyQgw,14283
309
311
  unstructured_ingest/v2/pipeline/utils.py,sha256=oPAitfdnITqh2O8Z0uf6VOHg9BTJhitRzNmKXqTwPxg,422
310
312
  unstructured_ingest/v2/pipeline/steps/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
311
313
  unstructured_ingest/v2/pipeline/steps/chunk.py,sha256=lfCsBo6A9u1cT57YaEjvNI79gc29nW8c-2_WZNjiO5Y,3275
312
- unstructured_ingest/v2/pipeline/steps/download.py,sha256=GA5-zTH4c7Ac8oBoc4AeDA8sQ0dYT-KUOZ4n31K3Jpg,4882
314
+ unstructured_ingest/v2/pipeline/steps/download.py,sha256=qYeuRU5jeICyuTN7E4YUdnbi6X1X2qKoooJMm4Orbdw,7499
313
315
  unstructured_ingest/v2/pipeline/steps/embed.py,sha256=VCdDBUXK6Yx8RTvRBpEFdFE7n0izvkP73w6s8Tv2sgg,3253
316
+ unstructured_ingest/v2/pipeline/steps/filter.py,sha256=mYVccl_zp0CGYFWBrSrPelvSElrXhZahebuymGuirV8,1341
314
317
  unstructured_ingest/v2/pipeline/steps/index.py,sha256=i4RcJ1oRqNp-rFdc6rvKVGcSzNhdB7woW7_W364uThQ,2269
315
318
  unstructured_ingest/v2/pipeline/steps/partition.py,sha256=q7-rpCj5Vy4BXtd7T72gxGb3xg6lmVyNmTwUfHil7Rg,3199
316
319
  unstructured_ingest/v2/pipeline/steps/stage.py,sha256=A8i6VAFY4_xFJR0uBEyBNJlQXmTMGaflXsa6Wa6U1wQ,2274
@@ -320,37 +323,39 @@ unstructured_ingest/v2/processes/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRk
320
323
  unstructured_ingest/v2/processes/chunker.py,sha256=U6zQhaUG_dii66zqL9iEEGodHENNxnpn6V3pC-e7MMI,4233
321
324
  unstructured_ingest/v2/processes/connector_registry.py,sha256=KOrvJNNRdpBPyqFwmTm42kD1xXuo7fNS_5yXjtqAz-c,2100
322
325
  unstructured_ingest/v2/processes/embedder.py,sha256=QjAsiXAjWtZzh6lJ4D5LsTMBD81zuMBkegXNWq-FZt0,3308
326
+ unstructured_ingest/v2/processes/filter.py,sha256=CfQihLV_r4yTJgAc66mmbP4_xo3wcDlro5oR_KR--bM,1986
323
327
  unstructured_ingest/v2/processes/partitioner.py,sha256=f6UQoQHVKjl8rmM5J9EcuP30RTFLSLrArGdC6qh-ffE,7645
324
328
  unstructured_ingest/v2/processes/uncompress.py,sha256=x-JZYNs1zJOtRS7xNgiMyrYoAbzKM0p18O8NAl7avCA,1631
325
- unstructured_ingest/v2/processes/connectors/__init__.py,sha256=cuPXXs7__tztof9Z8jE4b7UPQFc6p616hH1BECzBK1M,3762
326
- unstructured_ingest/v2/processes/connectors/astra.py,sha256=TSI_3GHnEh3gYAC30RTG4b2eEB07agroEFmJ38GnQY4,4903
327
- unstructured_ingest/v2/processes/connectors/azure_cognitive_search.py,sha256=PT02ZKiJuHMrmBClxqBsyDS0aXUQYLVg02Ns2qh1hD4,7935
328
- unstructured_ingest/v2/processes/connectors/chroma.py,sha256=nYzNz-8oq-DN0Z4r7lHQFmlved76IaYeRvm7-EmbGUE,6998
329
- unstructured_ingest/v2/processes/connectors/databricks_volumes.py,sha256=MTLK7SvQqWU-PXmEbGajM4f-CqGWlmlfeED6a5StEWw,3226
330
- unstructured_ingest/v2/processes/connectors/elasticsearch.py,sha256=6QBvVzPk3mWj9ZqJZN7NvhcJaOO6nSLqLwU6zggP59A,14864
331
- unstructured_ingest/v2/processes/connectors/google_drive.py,sha256=IkLVafUu280OOoqYmdfdfMB6zlpiWjs2Z5J31ZzJOj4,12681
332
- unstructured_ingest/v2/processes/connectors/local.py,sha256=maAXVKpRRXj_jseC6EPLTosMgw6ll-0lnGsDdAFLWAE,6646
333
- unstructured_ingest/v2/processes/connectors/mongodb.py,sha256=ErZWAD-su3OCRGv1h84X1PpAWleUPVZcFDEIYjtyP4E,4310
334
- unstructured_ingest/v2/processes/connectors/onedrive.py,sha256=WDDoFEfd8M_QBTpkGNI2zZGZZ_CR1rQiCsBWYOO2JoA,8311
329
+ unstructured_ingest/v2/processes/connectors/__init__.py,sha256=7QMKd8gtEJTIuK352Ho6XyoFvLLhrWIzgdu0dXwXWOE,3960
330
+ unstructured_ingest/v2/processes/connectors/astra.py,sha256=m6A34wYjnctRfIF-14bnbGIFBwht5Y8UWZ4g8R9x6a8,5241
331
+ unstructured_ingest/v2/processes/connectors/azure_cognitive_search.py,sha256=N_--5t_hxFzFEK4vERzm46gfg-Ghozb71_NmUlEYIMA,8277
332
+ unstructured_ingest/v2/processes/connectors/chroma.py,sha256=W995GLn7D85GoUhSqHQXP5QQ8OglgykA5rcNmg9Ruf4,7158
333
+ unstructured_ingest/v2/processes/connectors/databricks_volumes.py,sha256=8bVht450bxp0K4ub1XdCDmMKEooXVV4DNY5b5GWF0Ig,3636
334
+ unstructured_ingest/v2/processes/connectors/elasticsearch.py,sha256=myY2FRXtlBYhH-kbTSsn7j9UDzh36NYHqFRP-ys8am4,15358
335
+ unstructured_ingest/v2/processes/connectors/google_drive.py,sha256=-iYpwt4xxaICRlHD5Bpap7Ck5HRJcapa6uHl60E1uZ4,12702
336
+ unstructured_ingest/v2/processes/connectors/local.py,sha256=IJ5DjASp-5lPmb6J7Y8NROYjIS3sfdRhlcDAZEEGVAw,6573
337
+ unstructured_ingest/v2/processes/connectors/milvus.py,sha256=FWH4FH-zns7gh8sITg9pLYE9uKm_3GeOXJ4wjY6PMno,6776
338
+ unstructured_ingest/v2/processes/connectors/mongodb.py,sha256=XZCgkF28HCR4DtMmr8jlxb59txXgEvfCabovROUrv6Y,4602
339
+ unstructured_ingest/v2/processes/connectors/onedrive.py,sha256=_TFO-vlyCxIxMk6hv20CEsicrlh87wCrbi4I1chsMUw,8822
335
340
  unstructured_ingest/v2/processes/connectors/opensearch.py,sha256=HNRZVQsWnjLLm0yAGiIyHRbhAsBnGSXBO_VkUfIdwdE,5463
336
- unstructured_ingest/v2/processes/connectors/pinecone.py,sha256=VkPYmGmFKbgsmmrWV09roxztAv5LlTBVHizPeyPoFVc,5746
337
- unstructured_ingest/v2/processes/connectors/salesforce.py,sha256=Cz4qEtnbsD9-m1DXANxnVRZTHX2ZaUUBPVFPu5wnFRk,10832
338
- unstructured_ingest/v2/processes/connectors/sharepoint.py,sha256=SNovgGUE5tHdfX_lF5zwM_QRZK7mahHzLZKhnqfk6Tc,17696
341
+ unstructured_ingest/v2/processes/connectors/pinecone.py,sha256=0rpOh_pi4GormyIQsnEJbKVb7FeizAbLcbljpnjtpeY,5908
342
+ unstructured_ingest/v2/processes/connectors/salesforce.py,sha256=S0dEjT1UxReCC6qE9DlSQBgcSzQbOaIq7SMJqXUpNWQ,10858
343
+ unstructured_ingest/v2/processes/connectors/sharepoint.py,sha256=NRn0lbOuXqIYqZT15IVFeFQCxpCKzZC_M8pVYZeeNfo,17933
339
344
  unstructured_ingest/v2/processes/connectors/singlestore.py,sha256=upF2O4hJ2uiBhDRrpQ8CSJUvzmqu2j5H1b_QbReHJpw,5168
340
- unstructured_ingest/v2/processes/connectors/sql.py,sha256=T0rpCbhEipWlezoJOMiUewcZuk6Had6TkmsDT-PeOL0,8360
345
+ unstructured_ingest/v2/processes/connectors/sql.py,sha256=mbhBI2tcX4q1YJwR3Nr7HGbr-rb8ppUYq2JcN88We3U,9076
341
346
  unstructured_ingest/v2/processes/connectors/utils.py,sha256=nmpZZCeX0O7rGrwHSWM_heBgpZK9tKT6EV1Moer-z40,576
342
- unstructured_ingest/v2/processes/connectors/weaviate.py,sha256=7H3s44zVKbN3_eR35sbKTKSDOt6ZIIQkX-4t65LuJ6c,8254
347
+ unstructured_ingest/v2/processes/connectors/weaviate.py,sha256=HtJuOUhBs_HA7uOXlEIuYtx0elb0ecsCvP8N822tOMQ,8564
343
348
  unstructured_ingest/v2/processes/connectors/fsspec/__init__.py,sha256=TtdeImM7Ypl_n6sl7I1JqX6bGSG0t_FqvCqE3Cy24og,1846
344
349
  unstructured_ingest/v2/processes/connectors/fsspec/azure.py,sha256=RN7zoifocIWVgoP9aMDMz4TP-Z9KhE-HbCCBq33fY90,4674
345
350
  unstructured_ingest/v2/processes/connectors/fsspec/box.py,sha256=UnD-F9g7yOOBStrAqeKq6GuQjEyHdwOA3jYLj8YZIRM,4088
346
351
  unstructured_ingest/v2/processes/connectors/fsspec/dropbox.py,sha256=I6mPG9EIso9TcIczCw5Y14Yqd-EhTQ2CLw1MJx1V3dY,4420
347
- unstructured_ingest/v2/processes/connectors/fsspec/fsspec.py,sha256=gNgrRqKqk9YpBRGqGPvBUuEcBv1jN59fmBBj6NrB4sA,12394
352
+ unstructured_ingest/v2/processes/connectors/fsspec/fsspec.py,sha256=zKrwKTVGnhnitD8h_Url5HRFsJZjM66o3jWrzAm-_UA,12153
348
353
  unstructured_ingest/v2/processes/connectors/fsspec/gcs.py,sha256=RYZq_8hKF7bRxuB5Gozv5AzB3_nTuuooE4UfRjXwEFU,4443
349
- unstructured_ingest/v2/processes/connectors/fsspec/s3.py,sha256=7lOm5hjb0LBkbe-OWXnV3wDC-3mM_GWwwmdKW0xzh8c,5333
354
+ unstructured_ingest/v2/processes/connectors/fsspec/s3.py,sha256=PXK9a5O3woDuBWSf4R5XLQI5mzHtap8wAKpHI8Rh5gQ,5462
350
355
  unstructured_ingest/v2/processes/connectors/fsspec/sftp.py,sha256=J7Ej-j7dtXAluHunwynUfHlNsYwymb-LsrGUFcljcsA,5700
351
356
  unstructured_ingest/v2/processes/connectors/fsspec/utils.py,sha256=jec_Qfe2hbfahBuY-u8FnvHuv933AI5HwPFjOL3kEEY,456
352
- unstructured_ingest-0.0.0.dist-info/METADATA,sha256=XkFXINFQaqCmpnfNfymilYFVadQCXgkbXUm236ko_so,21501
353
- unstructured_ingest-0.0.0.dist-info/WHEEL,sha256=GJ7t_kWBFywbagK5eo9IoUwLW6oyOeTKmQ-9iHFVNxQ,92
354
- unstructured_ingest-0.0.0.dist-info/entry_points.txt,sha256=gUAAFnjFPnBgThJSEbw0N5ZjxtaKlT1s9e05_arQrNw,70
355
- unstructured_ingest-0.0.0.dist-info/top_level.txt,sha256=QaTxTcjfM5Hr9sZJ6weOJvSe5ESQc0F8AWkhHInTCf8,20
356
- unstructured_ingest-0.0.0.dist-info/RECORD,,
357
+ unstructured_ingest-0.0.2.dist-info/METADATA,sha256=a68Sz8-m1-ZRFz0p4yic64BhgwTuMdIMmCuPECdhWwA,21568
358
+ unstructured_ingest-0.0.2.dist-info/WHEEL,sha256=GJ7t_kWBFywbagK5eo9IoUwLW6oyOeTKmQ-9iHFVNxQ,92
359
+ unstructured_ingest-0.0.2.dist-info/entry_points.txt,sha256=gUAAFnjFPnBgThJSEbw0N5ZjxtaKlT1s9e05_arQrNw,70
360
+ unstructured_ingest-0.0.2.dist-info/top_level.txt,sha256=QaTxTcjfM5Hr9sZJ6weOJvSe5ESQc0F8AWkhHInTCf8,20
361
+ unstructured_ingest-0.0.2.dist-info/RECORD,,