Merge branch 'aws-samples:master' into master

Author: Chuukwudi

README.md

@@ -127,7 +127,7 @@ Textractor can be cited using:
   author = {Belval, Edouard and Delteil, Thomas and Schade, Martin and Radhakrishna, Srividhya},
   title = {{Amazon Textractor}},
   url = {https://github.com/aws-samples/amazon-textract-textractor},
-  version = {1.8.4},
+  version = {1.8.5},
   year = {2024}
 }
 ```

setup.py

@@ -41,7 +41,7 @@ def _run(self, command):
 setup(
     # include data files
     name="amazon-textract-textractor",
-    version="1.8.4",
+    version="1.8.5",
     license="Apache 2.0",
     description="A package to use AWS Textract services.",
     url="https://github.com/aws-samples/amazon-textract-textractor",

textractor/__init__.py

@@ -1,3 +1,3 @@
-__version__ = "1.8.4"
+__version__ = "1.8.5"
 
 from .textractor import Textractor

textractor/utils/legacy_utils.py

@@ -1,3 +1,4 @@
+import logging
 from textractor.data.constants import (
     LAYOUT_FIGURE,
     LAYOUT_LIST,
@@ -13,39 +14,46 @@
 
 def converter(response):
     blocks_to_delete = []
-    page_block = None
-    for i, block in enumerate(response["Blocks"]):
-        if block.get("BlockType") == "PAGE":
-            page_block = block
-        elif block.get("BlockType", "").startswith("LAYOUT_FIGURE_"):
-            block["BlockType"] = LAYOUT_TEXT
-        elif (
-            block.get("BlockType", "").startswith("LAYOUT_") and
-            block.get("BlockType") not in [
-                LAYOUT_TEXT,
-                LAYOUT_TITLE,
-                LAYOUT_HEADER,
-                LAYOUT_FOOTER,
-                LAYOUT_SECTION_HEADER,
-                LAYOUT_PAGE_NUMBER,
-                LAYOUT_LIST,
-                LAYOUT_FIGURE,
-                LAYOUT_TABLE,
-                LAYOUT_KEY_VALUE,
-            ]
-        ):
-            block["BlockType"] = LAYOUT_FIGURE
-        elif block.get("BlockType") == LAYOUT_FIGURE and "CONTAINER" in block.get("EntityTypes", []):
-            blocks_to_delete.append((i, block))
-    
-    page_relationships = []
-    for relationship in page_block["Relationships"]:
-        if relationship["Type"] == "CHILD":
-            page_relationships = relationship["Ids"]
-            break
+    page_blocks = []
+    try:
+        for i, block in enumerate(response["Blocks"]):
+            if block.get("BlockType") == "PAGE":
+                page_blocks.append(block)
+            elif block.get("BlockType", "").startswith("LAYOUT_FIGURE_"):
+                block["BlockType"] = LAYOUT_TEXT
+            elif (
+                block.get("BlockType", "").startswith("LAYOUT_") and
+                block.get("BlockType") not in [
+                    LAYOUT_TEXT,
+                    LAYOUT_TITLE,
+                    LAYOUT_HEADER,
+                    LAYOUT_FOOTER,
+                    LAYOUT_SECTION_HEADER,
+                    LAYOUT_PAGE_NUMBER,
+                    LAYOUT_LIST,
+                    LAYOUT_FIGURE,
+                    LAYOUT_TABLE,
+                    LAYOUT_KEY_VALUE,
+                ]
+            ):
+                block["BlockType"] = LAYOUT_FIGURE
+            elif block.get("BlockType") == LAYOUT_FIGURE and "CONTAINER" in block.get("EntityTypes", []):
+                blocks_to_delete.append((i, block))
 
-    for i, block in blocks_to_delete[::-1]:
-        del response["Blocks"][i]
-        page_relationships.remove(block["Id"])
+        blocks_to_delete_id_set = set([b["Id"] for _, b in blocks_to_delete])
+        for page_block in page_blocks:
+            for relationship in page_block.get("Relationships", []):
+                if relationship["Type"] == "CHILD":
+                    relationship["Ids"] = [
+                        id
+                        for id in relationship["Ids"]
+                        if id not in blocks_to_delete_id_set
+                    ]
+                    break
+            
+        for i, block in blocks_to_delete[::-1]:
+            del response["Blocks"][i]
+    except Exception as ex:
+        logging.warning(f"Failed to convert the response for backward compatibility. {str(ex)}")
 
     return response