Update app.py

app.py

@@ -116,7 +116,6 @@ def preprocess_text(text):
     return text.strip()
 
 def sanitize_json_text(text):
-    # Remove control characters but keep newlines intact
     text = re.sub(r'[\x00-\x08\x0B\x0C\x0E-\x1F]', '', text)
     return text.strip()
 
@@ -156,7 +155,7 @@ def delete_record(container, record):
     try:
         doc_id = record["id"]
         partition_key_value = record.get("pk", doc_id)
-        st.write(f"Deleting {doc_id} with partition key {partition_key_value}")  # Debug log
+        st.write(f"Deleting {doc_id} with partition key {partition_key_value}")
         container.delete_item(item=doc_id, partition_key=partition_key_value)
         return True, f"Record {doc_id} deleted. 🗑️"
     except exceptions.CosmosResourceNotFoundError:
@@ -347,14 +346,16 @@ def update_file_management_section():
 # =============================================================================
 # ───────────── UI FUNCTIONS ─────────────
 # =============================================================================
-def edit_all_documents(container):
-    st.markdown("### 📑 All Documents")
+def edit_all_documents(container, search_keyword=None):
+    st.markdown("### 📑 All Documents" + (f" (Filtered: '{search_keyword}')" if search_keyword else ""))
     documents = get_documents(container)
+    if search_keyword:
+        # Filter documents by search keyword
+        documents = [doc for doc in documents if vector_keyword_search(search_keyword, doc)]
     if not documents:
-        st.info("No documents in this container.")
+        st.info("No documents match the current filter." if search_keyword else "No documents in this container.")
         return
     
-    # Initialize saved_docs if not present
     if 'saved_docs' not in st.session_state:
         st.session_state.saved_docs = {}
     
@@ -365,7 +366,6 @@ def edit_all_documents(container):
         header = f"{doc.get('name', 'Unnamed')} - {formatted_ts}"
         with st.expander(header):
             doc_key = f"editor_{doc['id']}"
-            # Use saved_docs for initial value if available, else original doc
             initial_value = st.session_state.saved_docs.get(doc['id'], json.dumps(doc, indent=2))
             edited_content = st.text_area("Edit JSON", value=initial_value, height=300, key=doc_key)
             col_save, col_delete = st.columns(2)
@@ -382,7 +382,7 @@ def edit_all_documents(container):
                         if success:
                             st.success(f"Saved {doc['id']}")
                             st.session_state.saved_docs[doc['id']] = json.dumps(updated_doc, indent=2)
-                            st.rerun()  # Refresh UI to reflect saved state
+                            st.rerun()
                         else:
                             st.error(message)
                     except json.JSONDecodeError as e:
@@ -472,39 +472,29 @@ def new_links_record(container):
     else:
         st.error(f"Error creating links record: {message}")
 
-def vector_keyword_search(keyword, container):
-    try:
-        query = f"SELECT * FROM c WHERE CONTAINS(c.content, '{keyword}')"
-        results = list(container.query_items(query=query, enable_cross_partition_query=True))
-        return results
-    except Exception as e:
-        st.error(f"Search error: {str(e)}")
-        return []
-
-def display_search_results(keyword, container):
-    results = vector_keyword_search(keyword, container)
-    st.markdown("### 🔍 Search Results")
-    for res in results:
-        doc_id = res.get("id", "")
-        with st.expander(f"Result {doc_id}"):
-            edited = st.text_area("Edit Document", value=json.dumps(res, indent=2), key=f"search_{doc_id}")
-            if st.button(f"💾 Save changes", key=f"save_search_{doc_id}"):
-                try:
-                    updated_doc = json.loads(edited)
-                    success, message = update_record(container, updated_doc)
-                    if success:
-                        st.success(f"Updated {doc_id}!")
-                        st.rerun()
-                    else:
-                        st.error(message)
-                except Exception as e:
-                    st.error(f"Error saving: {str(e)}")
+def vector_keyword_search(keyword, doc):
+    # Check if keyword exists in any string field of the document
+    keyword = keyword.lower()
+    for key, value in doc.items():
+        if isinstance(value, str) and keyword in value.lower():
+            return True
+    return False
 
 def search_documents_ui(container):
     with st.sidebar.form("search_form"):
         keyword = st.text_input("Search Keyword", key="search_keyword")
-        if st.form_submit_button("🔍 Search") and keyword:
-            display_search_results(keyword, container)
+        col1, col2 = st.columns(2)
+        with col1:
+            search_submitted = st.form_submit_button("🔍 Search")
+        with col2:
+            clear_submitted = st.form_submit_button("🗑️ Clear")
+        if search_submitted and keyword:
+            st.session_state.search_keyword = keyword
+            st.rerun()
+        if clear_submitted:
+            if 'search_keyword' in st.session_state:
+                del st.session_state.search_keyword
+            st.rerun()
 
 def validate_and_preprocess_image(file_data, target_size=(576, 1024)):
     try:
@@ -633,6 +623,8 @@ def main():
         st.session_state.selected_database = selected_db
         st.session_state.selected_container = None
         st.session_state.current_container = None
+        if 'search_keyword' in st.session_state:
+            del st.session_state.search_keyword
         st.rerun()
 
     # Containers Section
@@ -653,6 +645,8 @@ def main():
         if selected_container != st.session_state.get("selected_container"):
             st.session_state.selected_container = selected_container
             st.session_state.current_container = database.get_container_client(selected_container)
+            if 'search_keyword' in st.session_state:
+                del st.session_state.search_keyword
             st.rerun()
 
         # Actions Section
@@ -677,9 +671,10 @@ def main():
                 new_links_record(st.session_state.current_container)
             search_documents_ui(st.session_state.current_container)
 
-    # Central Area: Editable Documents
+    # Central Area: Editable Documents with Search Filter
     if st.session_state.current_container:
-        edit_all_documents(st.session_state.current_container)
+        search_keyword = st.session_state.get('search_keyword', None)
+        edit_all_documents(st.session_state.current_container, search_keyword)
     else:
         st.info("Select a database and container to view and edit documents.")
 
@@ -694,6 +689,8 @@ def main():
         st.session_state.selected_database = None
         st.session_state.selected_container = None
         st.session_state.current_container = None
+        if 'search_keyword' in st.session_state:
+            del st.session_state.search_keyword
         st.rerun()
 
 if __name__ == "__main__":