Update app.py

app.py

@@ -76,7 +76,7 @@ CSS = """
 
 
 # def get_vectordb(text, images, tables):
-def get_vectordb(text, images):
+def get_vectordb(text, images, img_doc_files):
     client = chromadb.EphemeralClient()
     loader = ImageLoader()
     sentence_transformer_ef = embedding_functions.SentenceTransformerEmbeddingFunction(
@@ -99,9 +99,9 @@ def get_vectordb(text, images):
         metadata={"hnsw:space": "cosine"},
     )
     descs = []
-    for image in images:
+    for i in range(len(images)):
         try:
-            descs.append(get_image_description(image)[0])
+            descs.append(img_doc_files[i]+"\n"+get_image_description(images[i])[0])
         except:
             descs.append("Could not generate image description due to some error")
 
@@ -152,6 +152,7 @@ def extract_data_from_pdfs(
     all_text = ""
 
     images = []
+    img_docs=[]
     for doc in docs:
         if do_ocr == "Get Text With OCR":
             pdf_doc = DocumentFile.from_pdf(doc)
@@ -163,11 +164,12 @@ def extract_data_from_pdfs(
 
         if include_images == "Include Images":
             images.extend(extract_images([doc]))
+            img_docs.append(doc.split("/")[-1])
 
     progress(
         0.6, "Generating image descriptions and inserting everything into vectorDB"
     )
-    vectordb = get_vectordb(all_text, images)
+    vectordb = get_vectordb(all_text, images, img_docs)
 
     progress(1, "Completed")
     session["processed"] = True
@@ -466,4 +468,4 @@ with gr.Blocks(css=CSS, theme=gr.themes.Soft(text_size=sizes.text_md)) as demo:
 
     next_p1.click(check_validity_and_llm, session_states, tabs)
 if __name__ == "__main__":
-    demo.launch()
+    demo.launch()