Update code

Author: BenConstable9

.pre-commit-config.yaml

@@ -57,6 +57,6 @@ repos:
       - id: compile-image-processing-requirements
         name: compile-image-processing-requirements
         language: system
-        entry: bash -c 'cd image_processing && exec uv pip compile'
+        entry: bash -c 'cd image_processing && exec uv pip compile --all-extras pyproject.toml'
         pass_filenames: false
-        files: ^image_processing/
+        files: ^image_processing/

deploy_ai_search/src/deploy_ai_search/rag_documents.py

@@ -101,11 +101,10 @@ def get_index_fields(self) -> list[SearchableField]:
                 name="ChunkFigures",
                 collection=True,
                 fields=[
-                    SearchableField(
+                    SimpleField(
                         name="FigureId",
                         type=SearchFieldDataType.String,
-                        collection=True,
-                        searchable=False,
+                        filterable=True,
                     ),
                     SimpleField(
                         name="Caption",

image_processing/src/image_processing/function_app.py

@@ -8,7 +8,7 @@
 from figure_analysis import FigureAnalysis
 from layout_and_figure_merger import LayoutAndFigureMerger
 from layout_analysis import process_layout_analysis
-from image_processing.mark_up_cleaner import process_mark_up_cleaner
+from image_processing.mark_up_cleaner import MarkUpCleaner
 from semantic_text_chunker import process_semantic_text_chunker, SemanticTextChunker
 
 logging.basicConfig(level=logging.DEBUG)
@@ -136,8 +136,10 @@ async def mark_up_cleaner(req: func.HttpRequest) -> func.HttpResponse:
 
         record_tasks = []
 
+        mark_up_cleaner = MarkUpCleaner()
+
         for value in values:
-            record_tasks.append(asyncio.create_task(process_mark_up_cleaner(value)))
+            record_tasks.append(asyncio.create_task(mark_up_cleaner.clean(value)))
 
         results = await asyncio.gather(*record_tasks)
         logging.debug("Results: %s", results)