Refactor: Remove OCR related code for entire page OCR (#231)

## Summary
One part of OCR refactor to move it from inference repo to unstructured
repo. This PR removes all OCR related code for entire page OCR, which
means all table related OCR still remain the same (will be moved after
table refactor to accept preprocessed OCR data)

## Test
Please see test description in
Unstructured-IO/unstructured#1579, since those
two need to work together.

## Note
The ingest test won't pass until we merge the unstructured refactor PR

---------

Co-authored-by: christinestraub <[email protected]>

Author: yuming-long

CHANGELOG.md

@@ -1,3 +1,7 @@
+## 0.7.0
+
+* Remove all OCR related code expect the table OCR code
+
 ## 0.6.6
 
 * Stop passing ocr_languages parameter into paddle to avoid invalid paddle language code error, this will be fixed until

Dockerfile

@@ -20,7 +20,6 @@ RUN python3.8 -m pip install pip==${PIP_VERSION} && \
   pip install --no-cache -r requirements/base.txt && \
   pip install --no-cache -r requirements/test.txt && \
   pip install --no-cache -r requirements/dev.txt && \
-  pip install "unstructured.PaddleOCR" && \
   dnf -y groupremove "Development Tools" && \
   dnf clean all
 

examples/layout_analysis/visualization.py

@@ -14,7 +14,6 @@ def run(f_path, scope):
         "final": None,
         "extracted": {"layout": {"color": "green", "width": 2}},
         "inferred": {"inferred_layout": {"color": "blue", "width": 2}},
-        "ocr": {"ocr_layout": {"color": "yellow", "width": 2}},
     }
 
     f_basename = os.path.splitext(os.path.basename(f_path))[0]
@@ -47,8 +46,7 @@ def run(f_path, scope):
             write_image(img, output_f_path)
 
         print(f"page_num: {idx+1} - n_total_elements: {len(page.elements)} - n_extracted_elements: "
-              f"{len(page.layout)} - n_inferred_elements: {len(page.inferred_layout)} - "
-              f"n_ocr_elements: {len(page.ocr_layout)}")
+              f"{len(page.layout)} - n_inferred_elements: {len(page.inferred_layout)}")
 
 
 if __name__ == '__main__':

test_unstructured_inference/conftest.py

@@ -107,15 +107,6 @@ def mock_embedded_text_regions():
     ]
 
 
-@pytest.fixture()
-def mock_ocr_regions():
-    return [
-        EmbeddedTextRegion(10, 10, 90, 90, text="0", source=None),
-        EmbeddedTextRegion(200, 200, 300, 300, text="1", source=None),
-        EmbeddedTextRegion(500, 320, 600, 350, text="3", source=None),
-    ]
-
-
 # TODO(alan): Make a better test layout
 @pytest.fixture()
 def mock_layout(mock_embedded_text_regions):
@@ -130,19 +121,3 @@ def mock_layout(mock_embedded_text_regions):
         )
         for r in mock_embedded_text_regions
     ]
-
-
-@pytest.fixture()
-def mock_inferred_layout(mock_embedded_text_regions):
-    return [
-        LayoutElement(
-            r.x1,
-            r.y1,
-            r.x2,
-            r.y2,
-            text=None,
-            source=None,
-            type="Text",
-        )
-        for r in mock_embedded_text_regions
-    ]