text-based query with metadata information

.gitignore

@@ -1,2 +1,45 @@
 .venv/*
-.env/*
+.env/*
+# Python cache files
+__pycache__/
+*.py[cod]
+# Distribution / packaging
+.Python
+build/
+develop-eggs/
+dist/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+# Unit test / coverage reports
+.coverage
+.coverage.*
+.hypothesis/
+.tox/
+# Pytest cache
+.pytest_cache/
+# Jupyter Notebook checkpoints
+.ipynb_checkpoints
+# PyCharm project files
+.idea/
+# VS Code settings
+.vscode/
+# Environment variables
+.env
+.gradio/
+# Local configuration files
+*.local
+# MacOS specific files
+.DS_Store
+# Windows specific files
+Thumbs.db
+# Logs
+*.log

app.py

@@ -275,7 +275,7 @@ def extract_step_metadata(file_path):
 
             # Extract FILE_DESCRIPTION
             desc_match = re.search(
-                r"FILE_DESCRIPTION\s*\(\s*\((.*?)\),\s*\'(.*?)\'\);", content, re.DOTALL
+                r"FILE_DESCRIPTION\s*\(\s*\((.*?)\),\s*'([^']*)'\);", content, re.DOTALL
             )
             if desc_match:
                 metadata["Description"] = desc_match.group(1).replace("'", "")
@@ -320,11 +320,11 @@ def parse_3d_file(original_filepath: str):
     if original_filepath.endswith((".3dxml", ".3DXML")):
         meta = extract_header_from_3dxml(original_filepath)
         text = dict_to_markdown(meta)
-        return f"Parsed metadata: {text}"
+        return text
     elif original_filepath.endswith((".step", ".STEP")):
         meta = extract_step_metadata(original_filepath)
         text = dict_to_markdown(meta)
-        return f"Parsed metadata: {text}"
+        return text
     logger.warning(f"No metadata found in the file {original_filepath}")
     return "No metadata found!"
 
@@ -332,9 +332,10 @@ def parse_3d_file(original_filepath: str):
 def render_3D_metadata(
     original_filepath: str, obj_path: str, embedding_dict: dict
 ) -> Tuple[str, str]:
-    return parse_3d_file(original_filepath=original_filepath), embedding_dict.get(
-        obj_path, {}
-    ).get("description", "No description found!")
+    return (
+        embedding_dict.get(obj_path, {}).get("metadata", "No metadata found!"),
+        embedding_dict.get(obj_path, {}).get("description", "No description found!"),
+    )
 
 
 #######################################################################################################################
@@ -470,12 +471,7 @@ async def embedding_3d_object(obj_path: str) -> Dict[str, Any]:
     image_embedding = await aget_image_embedding_from_np_image(
         np_image=aggregated_image
     )
-    text_embedding = await text_embedding_model.aget_text_embedding(text=description)
-    return {
-        "description": description,
-        "image_embedding": image_embedding,
-        "text_embedding": text_embedding,
-    }
+    return {"description": description, "image_embedding": image_embedding}
 
 
 BASE_SAMPLE_DIR = "/Users/tridoan/Spartan/Datum/service-ai/poc/3D/gradio_cache/"
@@ -505,11 +501,20 @@ async def accumulate_and_embedding(input_files, file_list, embedding_dict):
         logger.info("Processing new upload file:", file_path)
         obj_path = convert_to_obj(file_path)
         embeddings = await embedding_3d_object(obj_path)
+        metadata = parse_3d_file(original_filepath=file_path)
         if obj_path not in embedding_dict:
             embedding_dict[obj_path] = {}
+        text_embedding = await text_embedding_model.aget_text_embedding(
+            text="The 3D object is: "
+            + embeddings["description"]
+            + f".\n {'n' * 20}\nMetadata: "
+            + metadata
+        )
+        # store embeddings and metadata
+        embedding_dict[obj_path]["metadata"] = metadata
         embedding_dict[obj_path]["description"] = embeddings["description"]
         embedding_dict[obj_path]["image_embedding"] = embeddings["image_embedding"]
-        embedding_dict[obj_path]["text_embedding"] = embeddings["text_embedding"]
+        embedding_dict[obj_path]["text_embedding"] = text_embedding
 
     return all_files, gr.update(choices=all_files), embedding_dict
 
@@ -679,4 +684,4 @@ with gr.Blocks() as demo:
     )
 
 if __name__ == "__main__":
-    demo.launch(share=True)
+    demo.launch(share=True, debug=True)