Skip to content

Commit c52a2d1

Browse files
authored
version 0.0.80; bump to unstructured 0.15.10 (#458)
### Summary Bumps to `unstructured==0.15.10`.
1 parent 3c3b75a commit c52a2d1

File tree

4 files changed

+101
-107
lines changed

4 files changed

+101
-107
lines changed

CHANGELOG.md

+3-2
Original file line numberDiff line numberDiff line change
@@ -1,6 +1,7 @@
1-
## 0.0.80-dev0
1+
## 0.0.80
22

3-
* Add `include_slide_notes` parameter, indicating whether slide notes in `ppt` and `pptx` files should be partitioned. Default is `True`. Now, when slide notes are present in the file, they will be included alongside other elements, which may shift the index numbers of non-note elements.
3+
* Bump to `unstructured` 0.15.10
4+
* Add `include_slide_notes` parameter, indicating whether slide notes in `ppt` and `pptx` files should be partitioned. Default is `True`. Now, when slide notes are present in the file, they will be included alongside other elements, which may shift the index numbers of non-note elements.
45

56
## 0.0.79
67

requirements/base.in

+6-2
Original file line numberDiff line numberDiff line change
@@ -1,10 +1,14 @@
11
-c constraints.in
2-
unstructured[all-docs]>=0.8.1
2+
unstructured[all-docs]
33
# Pinning click due to a unicode issue in black
44
# can remove after black drops support for Python 3.6
55
# ref: https://github.com/psf/black/issues/2964
66
click==8.1.3
7-
fastapi
7+
# NOTE(robinson) - fastapi>=0.114.0 causes the test listed below to fail, though it
8+
# works if data if chunking strategy and new_after_n_chars are explicitly set. Pinning
9+
# for now to preserve behavior
10+
# test_parallel_mode_preserves_uniqueness_of_hashes_when_asssembling_page_splits
11+
fastapi<0.114.0
812
uvicorn
913
ratelimit
1014
requests

requirements/base.txt

+35-35
Original file line numberDiff line numberDiff line change
@@ -20,13 +20,13 @@ beautifulsoup4==4.12.3
2020
# via unstructured
2121
cachetools==5.5.0
2222
# via google-auth
23-
certifi==2024.7.4
23+
certifi==2024.8.30
2424
# via
2525
# httpcore
2626
# httpx
2727
# requests
2828
# unstructured-client
29-
cffi==1.17.0
29+
cffi==1.17.1
3030
# via cryptography
3131
chardet==5.2.0
3232
# via unstructured
@@ -43,17 +43,17 @@ click==8.1.3
4343
# uvicorn
4444
coloredlogs==15.0.1
4545
# via onnxruntime
46-
contourpy==1.2.1
46+
contourpy==1.3.0
4747
# via matplotlib
48-
cryptography==43.0.0
48+
cryptography==43.0.1
4949
# via pdfminer-six
5050
cycler==0.12.1
5151
# via matplotlib
5252
dataclasses-json==0.6.7
5353
# via
5454
# unstructured
5555
# unstructured-client
56-
deepdiff==7.0.1
56+
deepdiff==8.0.1
5757
# via unstructured-client
5858
deprecated==1.2.14
5959
# via pikepdf
@@ -65,9 +65,9 @@ et-xmlfile==1.1.0
6565
# via openpyxl
6666
exceptiongroup==1.2.2
6767
# via anyio
68-
fastapi==0.112.1
68+
fastapi==0.113.0
6969
# via -r requirements/base.in
70-
filelock==3.15.4
70+
filelock==3.16.0
7171
# via
7272
# huggingface-hub
7373
# torch
@@ -78,35 +78,35 @@ flatbuffers==24.3.25
7878
# via onnxruntime
7979
fonttools==4.53.1
8080
# via matplotlib
81-
fsspec==2024.6.1
81+
fsspec==2024.9.0
8282
# via
8383
# huggingface-hub
8484
# torch
85-
google-api-core[grpc]==2.19.1
85+
google-api-core[grpc]==2.19.2
8686
# via google-cloud-vision
8787
google-auth==2.34.0
8888
# via
8989
# google-api-core
9090
# google-cloud-vision
9191
google-cloud-vision==3.7.4
9292
# via unstructured
93-
googleapis-common-protos==1.63.2
93+
googleapis-common-protos==1.65.0
9494
# via
9595
# google-api-core
9696
# grpcio-status
97-
grpcio==1.65.5
97+
grpcio==1.66.1
9898
# via
9999
# google-api-core
100100
# grpcio-status
101-
grpcio-status==1.65.5
101+
grpcio-status==1.66.1
102102
# via google-api-core
103103
h11==0.14.0
104104
# via
105105
# httpcore
106106
# uvicorn
107107
httpcore==1.0.5
108108
# via httpx
109-
httpx==0.27.0
109+
httpx==0.27.2
110110
# via unstructured-client
111111
huggingface-hub==0.24.6
112112
# via
@@ -116,7 +116,7 @@ huggingface-hub==0.24.6
116116
# unstructured-inference
117117
humanfriendly==10.0
118118
# via coloredlogs
119-
idna==3.7
119+
idna==3.8
120120
# via
121121
# anyio
122122
# httpx
@@ -130,7 +130,7 @@ joblib==1.4.2
130130
# via nltk
131131
jsonpath-python==1.0.6
132132
# via unstructured-client
133-
kiwisolver==1.4.5
133+
kiwisolver==1.4.7
134134
# via matplotlib
135135
langdetect==1.0.9
136136
# via unstructured
@@ -191,15 +191,15 @@ onnx==1.16.2
191191
# via
192192
# unstructured
193193
# unstructured-inference
194-
onnxruntime==1.19.0
194+
onnxruntime==1.19.2
195195
# via unstructured-inference
196196
opencv-python==4.10.0.84
197197
# via
198198
# layoutparser
199199
# unstructured-inference
200200
openpyxl==3.1.5
201201
# via unstructured
202-
ordered-set==4.1.0
202+
orderly-set==5.2.2
203203
# via deepdiff
204204
packaging==24.1
205205
# via
@@ -225,28 +225,28 @@ pdfminer-six==20231228
225225
# unstructured
226226
pdfplumber==0.11.4
227227
# via layoutparser
228-
pikepdf==9.1.1
228+
pi-heif==0.18.0
229+
# via unstructured
230+
pikepdf==9.2.1
229231
# via unstructured
230232
pillow==10.4.0
231233
# via
232234
# layoutparser
233235
# matplotlib
234236
# pdf2image
235237
# pdfplumber
238+
# pi-heif
236239
# pikepdf
237-
# pillow-heif
238240
# python-pptx
239241
# torchvision
240242
# unstructured-pytesseract
241-
pillow-heif==0.18.0
242-
# via unstructured
243243
portalocker==2.10.1
244244
# via iopath
245245
proto-plus==1.24.0
246246
# via
247247
# google-api-core
248248
# google-cloud-vision
249-
protobuf==5.27.3
249+
protobuf==5.28.0
250250
# via
251251
# google-api-core
252252
# google-cloud-vision
@@ -271,13 +271,13 @@ pycparser==2.22
271271
# via cffi
272272
pycryptodome==3.20.0
273273
# via -r requirements/base.in
274-
pydantic==2.8.2
274+
pydantic==2.9.1
275275
# via fastapi
276-
pydantic-core==2.20.1
276+
pydantic-core==2.23.3
277277
# via pydantic
278278
pypandoc==1.13
279279
# via unstructured
280-
pyparsing==3.1.2
280+
pyparsing==3.1.4
281281
# via matplotlib
282282
pypdf==4.3.1
283283
# via
@@ -312,7 +312,7 @@ pyyaml==6.0.2
312312
# omegaconf
313313
# timm
314314
# transformers
315-
rapidfuzz==3.9.6
315+
rapidfuzz==3.9.7
316316
# via
317317
# unstructured
318318
# unstructured-inference
@@ -335,11 +335,11 @@ requests-toolbelt==1.0.0
335335
# via unstructured-client
336336
rsa==4.9
337337
# via google-auth
338-
safetensors==0.4.4
338+
safetensors==0.4.5
339339
# via
340340
# timm
341341
# transformers
342-
scipy==1.14.0
342+
scipy==1.14.1
343343
# via layoutparser
344344
six==1.16.0
345345
# via
@@ -352,27 +352,27 @@ sniffio==1.3.1
352352
# httpx
353353
soupsieve==2.6
354354
# via beautifulsoup4
355-
starlette==0.38.2
355+
starlette==0.38.5
356356
# via fastapi
357357
sympy==1.13.2
358358
# via
359359
# onnxruntime
360360
# torch
361361
tabulate==0.9.0
362362
# via unstructured
363-
timm==1.0.8
363+
timm==1.0.9
364364
# via
365365
# effdet
366366
# unstructured-inference
367367
tokenizers==0.19.1
368368
# via transformers
369-
torch==2.4.0
369+
torch==2.4.1
370370
# via
371371
# effdet
372372
# timm
373373
# torchvision
374374
# unstructured-inference
375-
torchvision==0.19.0
375+
torchvision==0.19.1
376376
# via
377377
# effdet
378378
# timm
@@ -383,7 +383,7 @@ tqdm==4.66.5
383383
# nltk
384384
# transformers
385385
# unstructured
386-
transformers==4.44.1
386+
transformers==4.44.2
387387
# via unstructured-inference
388388
typing-extensions==4.12.2
389389
# via
@@ -409,9 +409,9 @@ typing-inspect==0.9.0
409409
# unstructured-client
410410
tzdata==2024.1
411411
# via pandas
412-
unstructured[all-docs]==0.15.7
412+
unstructured[all-docs]==0.15.10
413413
# via -r requirements/base.in
414-
unstructured-client==0.25.5
414+
unstructured-client==0.25.8
415415
# via unstructured
416416
unstructured-inference==0.7.36
417417
# via unstructured

0 commit comments

Comments
 (0)