fix: Add handling for documents missing all layout elements. (#161)

* fix: Add handling for documents missing all layout elements.

- Bounding boxes will show up as 0,0,0,0
- Fixes #160

* docs: Add information about return value of `get_bounding_box()`

* fix: Addressed coment and moved if statement

Author: holtskinner

google/cloud/documentai_toolbox/utilities/docai_utilities.py

@@ -35,14 +35,19 @@ def get_bounding_box(
     Returns:
         Tuple[int, int, int, int]:
             Bounding box coordinates in order (top, left, bottom, right).
+            Returns `0, 0, 0, 0` if `bounding_poly.normalized_vertices` is empty.
     """
+    if not bounding_poly.normalized_vertices:
+        return 0, 0, 0, 0
+
     vertices = [
         (
             int(vertex.x * page_dimension.width + 0.5),
             int(vertex.y * page_dimension.height + 0.5),
         )
         for vertex in bounding_poly.normalized_vertices
     ]
+
     top, left = vertices[0]
     bottom, right = vertices[2]
     return top, left, bottom, right

tests/unit/resources/blank_document.json

@@ -661,6 +661,16 @@ def test_export_hocr_str():
     assert actual_hocr == expected
 
 
+def test_export_hocr_str_with_blank_document():
+    wrapped_document = document.Document.from_document_path(
+        document_path="tests/unit/resources/blank_document.json"
+    )
+
+    actual_hocr = wrapped_document.export_hocr_str(title="hocr_blank")
+
+    assert actual_hocr
+
+
 def test_document_to_merged_documentai_document(get_bytes_multiple_files_mock):
     wrapped_document = document.Document.from_gcs(
         gcs_bucket_name="test-directory", gcs_prefix="documentai/output/123456789/1/"