build(deps): update inference version (#662)

Updated to the the latest version of unstructured-inference. detectron2 now gets implemented with onnxruntime, yay!

---------

Co-authored-by: Matt Robinson <[email protected]>

Author: qued

.github/workflows/ci.yml

@@ -132,7 +132,6 @@ jobs:
     - name: Test
       run: |
         source .venv/bin/activate
-        make install-detectron2
         sudo apt-get update
         sudo apt-get install -y libmagic-dev poppler-utils libreoffice
         make install-pandoc
@@ -173,7 +172,6 @@ jobs:
         DISCORD_TOKEN: ${{ secrets.DISCORD_TOKEN }}
       run: |
         source .venv/bin/activate
-        make install-detectron2
         sudo apt-get update
         sudo apt-get install -y libmagic-dev poppler-utils libreoffice pandoc
         sudo add-apt-repository -y ppa:alex-p/tesseract-ocr5

.gitignore

@@ -151,7 +151,6 @@ tramp
 
 ## https://github.com/github/gitignore/blob/main/Global/VisualStudioCode.gitignore
 .vscode/*
-!.vscode/settings.json
 !.vscode/tasks.json
 !.vscode/launch.json
 !.vscode/extensions.json

CHANGELOG.md

@@ -1,7 +1,8 @@
-## 0.6.12-dev1
+## 0.7.0
 
 ### Enhancements
 
+* Installing `detectron2` from source is no longer required when using the `local-inference` extra.
 * Updates `.pptx` parsing to include text in tables.
 
 ### Features

Dockerfile

@@ -29,8 +29,7 @@ RUN python3.8 -m pip install pip==${PIP_VERSION} && \
   pip install --no-cache -r requirements/ingest-slack.txt && \
   pip install --no-cache -r requirements/ingest-wikipedia.txt && \
   pip install --no-cache -r requirements/local-inference.txt && \
-  scl enable devtoolset-9 bash && \
-  pip install --no-cache "detectron2@git+https://github.com/facebookresearch/detectron2.git@e2ce8dc#egg=detectron2"
+  scl enable devtoolset-9 bash
 
 COPY example-docs example-docs
 COPY unstructured unstructured

Makefile

@@ -103,7 +103,7 @@ install-detectron2: install-tensorboard
 
 ## install-local-inference: installs requirements for local inference
 .PHONY: install-local-inference
-install-local-inference: install install-unstructured-inference install-detectron2
+install-local-inference: install install-unstructured-inference
 
 .PHONY: install-pandoc
 install-pandoc:
@@ -116,9 +116,6 @@ pip-compile:
 	pip-compile --upgrade requirements/base.in
 	# Extra requirements for huggingface staging functions
 	pip-compile --upgrade requirements/huggingface.in
-	# NOTE(robinson) - We want the dependencies for detectron2 in the requirements.txt, but not
-	# the detectron2 repo itself. If detectron2 is in the requirements.txt file, an order of
-	# operations issue related to the torch library causes the install to fail
 	pip-compile --upgrade requirements/test.in
 	pip-compile --upgrade requirements/dev.in
 	pip-compile --upgrade requirements/build.in

docker/ubuntu-22/Dockerfile

@@ -16,7 +16,6 @@ SHELL ["/bin/bash", "-c"]
 RUN source ~/.bashrc && pyenv virtualenv 3.8.15 unstructured && \
     source ~/.pyenv/versions/unstructured/bin/activate && \
     make install-ci && \
-    make install-detectron2 && \
     make install-ingest-s3 && \
     make install-ingest-azure && \
     make install-ingest-github && \

docs/source/installing.rst

@@ -17,8 +17,7 @@ installation.
 	* ``libreoffice`` (MS Office docs)
 	* ``pandocs`` (EPUBs, RTFs and Open Office docs)
 
-* If you are parsing PDFs, run the following to install the ``detectron2`` model, which ``unstructured`` uses for layout detection:
-	* ``pip install "detectron2@git+https://github.com/facebookresearch/detectron2.git@e2ce8dc#egg=detectron2"``
+* Follow the instructions `here <https://github.com/Unstructured-IO/unstructured-inference#detectron2>`_ to install ``detectron2``. This is required if you would like to use custom models from the `LayoutParser Model Zoo <https://github.com/Unstructured-IO/unstructured-inference#using-models-from-the-layoutparser-model-zoo>`_.
 
 At this point, you should be able to run the following code:
 

requirements/base.txt

@@ -4,7 +4,7 @@
 #
 #    pip-compile requirements/base.in
 #
-anyio==3.6.2
+anyio==3.7.0
     # via httpcore
 argilla==1.7.0
     # via -r requirements/base.in
@@ -30,12 +30,14 @@ click==8.1.3
     #   typer
 commonmark==0.9.1
     # via rich
-cryptography==40.0.2
+cryptography==41.0.0
     # via pdfminer-six
-deprecated==1.2.13
+deprecated==1.2.14
     # via argilla
 et-xmlfile==1.1.0
     # via openpyxl
+exceptiongroup==1.1.1
+    # via anyio
 h11==0.14.0
     # via httpcore
 httpcore==0.16.3
@@ -117,6 +119,8 @@ sniffio==1.3.0
     #   anyio
     #   httpcore
     #   httpx
+tabulate==0.9.0
+    # via unstructured (setup.py)
 tqdm==4.65.0
     # via
     #   argilla
@@ -138,7 +142,7 @@ wrapt==1.14.1
     #   deprecated
 xlrd==2.0.1
     # via -r requirements/base.in
-xlsxwriter==3.1.1
+xlsxwriter==3.1.2
     # via python-pptx
 zipp==3.15.0
     # via importlib-metadata

requirements/dev.txt

@@ -4,7 +4,7 @@
 #
 #    pip-compile requirements/dev.in
 #
-anyio==3.6.2
+anyio==3.7.0
     # via
     #   -c requirements/base.txt
     #   jupyter-server
@@ -54,6 +54,11 @@ defusedxml==0.7.1
     # via nbconvert
 distlib==0.3.6
     # via virtualenv
+exceptiongroup==1.1.1
+    # via
+    #   -c requirements/base.txt
+    #   -c requirements/test.txt
+    #   anyio
 executing==1.2.0
     # via stack-data
 fastjsonschema==2.17.1
@@ -265,7 +270,7 @@ pyyaml==6.0
     #   -c requirements/test.txt
     #   jupyter-events
     #   pre-commit
-pyzmq==25.0.2
+pyzmq==25.1.0
     # via
     #   ipykernel
     #   jupyter-client

requirements/ingest-azure.txt

@@ -40,7 +40,7 @@ charset-normalizer==3.1.0
     #   -c requirements/base.txt
     #   aiohttp
     #   requests
-cryptography==40.0.2
+cryptography==41.0.0
     # via
     #   -c requirements/base.txt
     #   azure-identity