Update code and flows

Author: BenConstable9

deploy_ai_search_indexes/src/deploy_ai_search_indexes/ai_search.py

@@ -425,7 +425,7 @@ def get_figure_analysis_skill(self, chunk_by_page=False) -> WebApiSkill:
         ]
 
         if chunk_by_page:
-            figure_context = "/document/page_wise_layout/*"
+            figure_context = "/document/page_wise_layout/*/figures/*"
             inputs = [
                 InputFieldMappingEntry(
                     name="figure", source="/document/page_wise_layout/*/figures/*"

deploy_ai_search_indexes/src/deploy_ai_search_indexes/image_processing.py

@@ -250,11 +250,11 @@ def get_index_projections(self) -> SearchIndexerIndexProjection:
                     name="DateLastModified", source="/document/DateLastModified"
                 ),
                 InputFieldMappingEntry(
-                    name="PageNumber", source="/document/page_wise_layout/*/#"
+                    name="PageNumber", source="/document/page_wise_layout/*/page_number"
                 ),
             ]
         else:
-            source_context = "/document/chunks/*"
+            source_context = "/document/chunk_mark_ups/*"
             mappings = [
                 InputFieldMappingEntry(
                     name="Chunk", source="/document/chunk_mark_ups/*/chunk_mark_up"

image_processing/images/Page Wise Chunking.png

@@ -197,10 +197,9 @@ async def analyse(self, record: dict) -> dict:
         - record (dict): The record containing the image, its caption, and the generated description.
         """
 
-        logging.info(f"Record: {record}")
-        figure = FigureHolder(**record["data"]["figure"])
-
         try:
+            logging.info(f"Record: {record}")
+            figure = FigureHolder(**record["data"]["figure"])
             updated_data = await self.understand_image_with_gptv(figure)
             logging.info(f"Updated Figure Data: {updated_data}")
         except RetryError as e:
@@ -230,6 +229,19 @@ async def analyse(self, record: dict) -> dict:
                     ],
                     "warnings": None,
                 }
+        except Exception as e:
+            logging.error(f"Failed to analyse image. Error: {e}")
+            logging.error(f"Failed input: {record}")
+            return {
+                "recordId": record["recordId"],
+                "data": None,
+                "errors": [
+                    {
+                        "message": "Failed to analyse image. Check the logs for more details.",
+                    }
+                ],
+                "warnings": None,
+            }
         else:
             return {
                 "recordId": record["recordId"],

image_processing/src/image_processing/figure_analysis.py

@@ -22,7 +22,7 @@ async def layout_analysis(req: func.HttpRequest) -> func.HttpResponse:
         values = req_body.get("values")
         adi_config = req.headers
 
-        page_wise = adi_config.get("page_wise", "False").lower() == "true"
+        page_wise = adi_config.get("chunk_by_page", "False").lower() == "true"
         extract_figures = adi_config.get("extract_figures", "True").lower() == "true"
         logging.info(f"Chunk by Page: {page_wise}")
     except ValueError:

image_processing/src/image_processing/function_app.py

@@ -1,24 +1,26 @@
 # Copyright (c) Microsoft Corporation. All rights reserved.
 # Licensed under the MIT License.
 
-from pydantic import BaseModel, Field
+from pydantic import BaseModel, Field, ConfigDict
 from typing import Optional
 
 
 class FigureHolder(BaseModel):
 
     """A class to hold the figure extracted from the document."""
 
-    figure_id: str
-    container: str = Field(exclude=True)
-    blob: str = Field(exclude=True)
-    caption: Optional[str] = Field(default=None)
+    figure_id: str = Field(..., alias="FigureId")
+    container: Optional[str] = Field(exclude=True, default=None)
+    blob: Optional[str] = Field(exclude=True, default=None)
+    caption: Optional[str] = Field(default=None, alias="Caption")
     offset: int
     length: int
-    page_number: Optional[int] = Field(default=None)
-    uri: str
-    description: Optional[str] = Field(default="")
-    data: Optional[str] = Field(default=None)
+    page_number: Optional[int] = Field(default=None, alias="PageNumber")
+    uri: str = Field(..., alias="Uri")
+    description: Optional[str] = Field(default="", alias="Description")
+    data: Optional[str] = Field(default=None, alias="Data")
+
+    model_config = ConfigDict(populate_by_name=True)
 
     @property
     def markdown(self) -> str:

image_processing/src/image_processing/layout_holders.py

@@ -94,18 +94,18 @@ def clean_text_and_extract_metadata(
                 logging.error("Input text is empty")
                 raise ValueError("Input text is empty")
 
-            return_record["marked_up_chunk"] = text
+            return_record["chunk_mark_up"] = text
 
             figure_ids = self.get_figure_ids(text)
 
-            return_record["sections"] = self.get_sections(text)
-            return_record["figures"] = [
+            return_record["chunk_sections"] = self.get_sections(text)
+            return_record["chunk_figures"] = [
                 figure.model_dump(by_alias=True)
                 for figure in figures
                 if figure.figure_id in figure_ids
             ]
 
-            logging.info(f"Sections: {return_record['sections']}")
+            logging.info(f"Sections: {return_record['chunk_sections']}")
 
             # Define specific patterns for each tag
             tag_patterns = {
@@ -128,7 +128,7 @@ def clean_text_and_extract_metadata(
                 logging.error("Cleaned text is empty")
                 raise ValueError("Cleaned text is empty")
             else:
-                return_record["cleaned_chunk"] = cleaned_text
+                return_record["chunk_cleaned"] = cleaned_text
         except Exception as e:
             logging.error(f"An error occurred in clean_text_and_extract_metadata: {e}")
             return ""