refactor: layout and model (#56)

Updated so that YoloX uses the same code paths as other models.

* Removed code that was duplicating existing functions
* Integrated YoloX model into existing pipeline
    * This means YoloX now has an associated UnstructuredModel called UnstructuredYoloXModel
    * The output of YoloX inference is now a layoutparser Layout object, like the other UnstructuredModels so that it can integrate with the DocumentLayout, PageLayout, and LayoutElement classes
    * Other functionality is provided by the DocumentLayout, PageLayout, and LayoutElement classes
* Removed jsons dependency that was used in testing and updated deps. The dependency had a cool feature that allowed it to construct non-json-serializable classes from a json, but since that was being used to check content, content could be checked through another method.
* One piece of functionality that was in the code prior to this change and isn't now is the ability to create test artifacts that can be viewed for debugging purposes. This can be added back in, but artifacts must be created from the new outputs of the model (either a layoutparser Layout object or PageLayout object).
This PR also allows the capability to create a PageLayout with a fixed layout, i.e. the layout and location of text elements is specified at construction rather than inferred by model. This is useful for use cases where the layout of a particular form is known.

Author: qued

.gitignore

@@ -134,3 +134,6 @@ dmypy.json
 
 # Mac stuff
 .DS_Store
+
+# VSCode
+.vscode/

CHANGELOG.md

@@ -1,5 +1,6 @@
-## 0.2.8-dev0
+## 0.2.8
 
+* Refactored YoloX inference code to integrate better with framework
 * Improved testing time
 
 ## 0.2.7
@@ -14,9 +15,6 @@
 ## 0.2.5
 
 * Add YoloX model for images and PDFs
-
-## 0.2.5-dev0
-
 * Add generic model interface
 
 ## 0.2.4

pytest.ini

@@ -1,3 +1,3 @@
 [pytest]
 markers =
-    slow: marks tests as slow (deselect with '-m "not long"')
+    slow: marks tests as slow (deselect with '-m "not long"')

requirements/base.txt

@@ -28,7 +28,7 @@ cycler==0.11.0
     # via matplotlib
 effdet==0.3.0
     # via layoutparser
-fastapi==0.89.1
+fastapi==0.92.0
     # via unstructured-inference (setup.py)
 filelock==3.9.0
     # via
@@ -40,7 +40,7 @@ fonttools==4.38.0
     # via matplotlib
 h11==0.14.0
     # via uvicorn
-huggingface-hub==0.12.0
+huggingface-hub==0.12.1
     # via
     #   timm
     #   transformers
@@ -51,15 +51,15 @@ idna==3.4
     # via
     #   anyio
     #   requests
+importlib-resources==5.12.0
+    # via matplotlib
 iopath==0.1.10
     # via layoutparser
-jsons==1.6.3
-    # via unstructured-inference (setup.py)
 kiwisolver==1.4.4
     # via matplotlib
 layoutparser[layoutmodels,tesseract]==0.3.4
     # via unstructured-inference (setup.py)
-matplotlib==3.6.3
+matplotlib==3.7.0
     # via pycocotools
 mpmath==1.2.1
     # via sympy
@@ -96,7 +96,7 @@ pdf2image==1.16.2
     # via layoutparser
 pdfminer-six==20221105
     # via pdfplumber
-pdfplumber==0.7.6
+pdfplumber==0.8.0
     # via layoutparser
 pillow==9.4.0
     # via
@@ -108,13 +108,13 @@ pillow==9.4.0
     #   torchvision
 portalocker==2.7.0
     # via iopath
-protobuf==4.21.12
+protobuf==4.22.0
     # via onnxruntime
 pycocotools==2.0.6
     # via effdet
 pycparser==2.21
     # via cffi
-pydantic==1.10.4
+pydantic==1.10.5
     # via fastapi
 pyparsing==3.0.9
     # via matplotlib
@@ -150,7 +150,7 @@ six==1.16.0
     #   python-multipart
 sniffio==1.3.0
     # via anyio
-starlette==0.22.0
+starlette==0.25.0
     # via fastapi
 sympy==1.11.1
     # via onnxruntime
@@ -174,21 +174,21 @@ tqdm==4.64.1
     #   huggingface-hub
     #   iopath
     #   transformers
-transformers==4.26.0
+transformers==4.26.1
     # via unstructured-inference (setup.py)
-typing-extensions==4.4.0
+typing-extensions==4.5.0
     # via
     #   huggingface-hub
     #   iopath
     #   pydantic
     #   starlette
     #   torch
     #   torchvision
-typish==1.9.3
-    # via jsons
 urllib3==1.26.14
     # via requests
 uvicorn==0.20.0
     # via unstructured-inference (setup.py)
 wand==0.6.11
     # via pdfplumber
+zipp==3.14.0
+    # via importlib-resources

requirements/dev.txt

@@ -6,6 +6,10 @@
 #
 anyio==3.6.2
     # via jupyter-server
+appnope==0.1.3
+    # via
+    #   ipykernel
+    #   ipython
 argon2-cffi==21.3.0
     # via
     #   jupyter-server
@@ -49,7 +53,13 @@ idna==3.4
     # via
     #   anyio
     #   jsonschema
-ipykernel==6.21.1
+importlib-metadata==6.0.0
+    # via
+    #   jupyter-client
+    #   nbconvert
+importlib-resources==5.12.0
+    # via jsonschema
+ipykernel==6.21.2
     # via
     #   ipywidgets
     #   jupyter
@@ -59,7 +69,7 @@ ipykernel==6.21.1
     #   qtconsole
 ipython==8.10.0
     # via
-    #   -r dev.in
+    #   -r requirements/dev.in
     #   ipykernel
     #   ipywidgets
     #   jupyter-console
@@ -87,8 +97,8 @@ jsonschema[format-nongpl]==4.17.3
     #   jupyter-events
     #   nbformat
 jupyter==1.0.0
-    # via -r dev.in
-jupyter-client==8.0.2
+    # via -r requirements/dev.in
+jupyter-client==8.0.3
     # via
     #   ipykernel
     #   jupyter-console
@@ -97,12 +107,13 @@ jupyter-client==8.0.2
     #   nbclient
     #   notebook
     #   qtconsole
-jupyter-console==6.4.4
+jupyter-console==6.5.1
     # via jupyter
 jupyter-core==5.2.0
     # via
     #   ipykernel
     #   jupyter-client
+    #   jupyter-console
     #   jupyter-server
     #   nbclassic
     #   nbclient
@@ -112,7 +123,7 @@ jupyter-core==5.2.0
     #   qtconsole
 jupyter-events==0.6.3
     # via jupyter-server
-jupyter-server==2.2.1
+jupyter-server==2.3.0
     # via
     #   nbclassic
     #   notebook-shim
@@ -130,7 +141,7 @@ matplotlib-inline==0.1.6
     # via
     #   ipykernel
     #   ipython
-mistune==2.0.4
+mistune==2.0.5
     # via nbconvert
 nbclassic==0.5.1
     # via notebook
@@ -174,8 +185,10 @@ pexpect==4.8.0
 pickleshare==0.7.5
     # via ipython
 pip-tools==6.12.2
-    # via -r dev.in
-platformdirs==2.6.2
+    # via -r requirements/dev.in
+pkgutil-resolve-name==1.3.10
+    # via jsonschema
+platformdirs==3.0.0
     # via jupyter-core
 prometheus-client==0.16.0
     # via
@@ -210,14 +223,15 @@ python-dateutil==2.8.2
     # via
     #   arrow
     #   jupyter-client
-python-json-logger==2.0.4
+python-json-logger==2.0.6
     # via jupyter-events
 pyyaml==6.0
     # via jupyter-events
 pyzmq==25.0.0
     # via
     #   ipykernel
     #   jupyter-client
+    #   jupyter-console
     #   jupyter-server
     #   nbclassic
     #   notebook
@@ -247,7 +261,7 @@ six==1.16.0
     #   rfc3339-validator
 sniffio==1.3.0
     # via anyio
-soupsieve==2.3.2.post1
+soupsieve==2.4
     # via beautifulsoup4
 stack-data==0.6.2
     # via ipython
@@ -259,6 +273,10 @@ terminado==0.17.1
     #   notebook
 tinycss2==1.2.1
     # via nbconvert
+tomli==2.0.1
+    # via
+    #   build
+    #   pyproject-hooks
 tornado==6.2
     # via
     #   ipykernel
@@ -274,6 +292,7 @@ traitlets==5.9.0
     #   ipython
     #   ipywidgets
     #   jupyter-client
+    #   jupyter-console
     #   jupyter-core
     #   jupyter-events
     #   jupyter-server
@@ -300,6 +319,10 @@ wheel==0.38.4
     # via pip-tools
 widgetsnbextension==4.0.5
     # via ipywidgets
+zipp==3.14.0
+    # via
+    #   importlib-metadata
+    #   importlib-resources
 
 # The following packages are considered to be unsafe in a requirements file:
 # pip

requirements/test.txt

@@ -6,6 +6,8 @@
 #
 anyio==3.6.2
     # via httpcore
+appdirs==1.4.4
+    # via label-studio-tools
 attrs==22.2.0
     # via pytest
 black==23.1.0
@@ -37,7 +39,7 @@ httpcore==0.16.3
     # via httpx
 httpx==0.23.3
     # via -r requirements/test.in
-huggingface-hub==0.12.0
+huggingface-hub==0.12.1
     # via -r requirements/test.in
 idna==3.4
     # via
@@ -47,15 +49,19 @@ idna==3.4
     #   yarl
 iniconfig==2.0.0
     # via pytest
-label-studio-sdk==0.0.18
+label-studio-sdk==0.0.19
     # via -r requirements/test.in
-lxml==4.9.2
+label-studio-tools==0.0.2
     # via label-studio-sdk
+lxml==4.9.2
+    # via
+    #   label-studio-sdk
+    #   label-studio-tools
 mccabe==0.7.0
     # via flake8
 multidict==6.0.4
     # via yarl
-mypy==0.991
+mypy==1.0.1
     # via -r requirements/test.in
 mypy-extensions==1.0.0
     # via
@@ -72,13 +78,13 @@ pdf2image==1.16.2
     # via -r requirements/test.in
 pillow==9.4.0
     # via pdf2image
-platformdirs==2.6.2
+platformdirs==3.0.0
     # via black
 pluggy==1.0.0
     # via pytest
 pycodestyle==2.10.0
     # via flake8
-pydantic==1.10.4
+pydantic==1.10.5
     # via label-studio-sdk
 pyflakes==3.0.1
     # via flake8
@@ -111,7 +117,7 @@ tomli==2.0.1
     #   pytest
 tqdm==4.64.1
     # via huggingface-hub
-typing-extensions==4.4.0
+typing-extensions==4.5.0
     # via
     #   black
     #   huggingface-hub

setup.py

@@ -58,8 +58,7 @@
         # ref: https://github.com/opencv/opencv-python/issues/772
         "opencv-python==4.6.0.66",
         "onnxruntime",
-        "jsons",
-        "transformers"
+        "transformers",
     ],
     extras_require={},
 )