Add legacy check to parser

Author: Belval

textractor/parsers/response_parser.py

@@ -66,6 +66,7 @@
     LAYOUT_TABLE,
     LAYOUT_KEY_VALUE,
 )
+from textractor.utils.legacy_utils import converter
 
 THRESHOLD = 0.95
 
@@ -1542,7 +1543,6 @@ def parser_analyze_expense_response(response):
     document.response = response
     return document
 
-
 def parse(response: dict) -> Document:
     """
     Ingests response data and API Call Mode and calls the appropriate function for it.
@@ -1559,4 +1559,4 @@ def parse(response: dict) -> Document:
     if "ExpenseDocuments" in response:
         return parser_analyze_expense_response(response)
     else:
-        return parse_document_api_response(response)
+        return parse_document_api_response(converter(response))

textractor/utils/legacy_utils.py

@@ -0,0 +1,51 @@
+from textractor.data.constants import (
+    LAYOUT_FIGURE,
+    LAYOUT_LIST,
+    LAYOUT_TABLE,
+    LAYOUT_KEY_VALUE,
+    LAYOUT_TEXT,
+    LAYOUT_TITLE,
+    LAYOUT_HEADER,
+    LAYOUT_FOOTER,
+    LAYOUT_SECTION_HEADER,
+    LAYOUT_PAGE_NUMBER,
+)
+
+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
+        
+    for i, block in blocks_to_delete[::-1]:
+        del response["Blocks"][i]
+        page_relationships.remove(block["Id"])
+    
+    return response