Update app.py

app.py

@@ -22,12 +22,12 @@ from appStore.region_utils import (
 # TF-IDF part (excluded from the app for now)
 # from appStore.tfidf_extraction import extract_top_keywords
 
-# Import helper modules
+# Import helper modules, including format_project_id for formatting IDs
 from appStore.rag_utils import (
     highlight_query,
     get_rag_answer,
     compute_title,
-    format_project_id  # <-- Imported formatting function
+    format_project_id
 )
 from appStore.filter_utils import (
     parse_budget,
@@ -127,14 +127,14 @@ unique_country_names = sorted(country_name_mapping.keys())
 # Define reset_filters function using session_state
 ###########################################
 def reset_filters():
-    st.session_state["region_filter"] = ["All/Not allocated"]
-    st.session_state["country_filter"] = ["All/Not allocated"]
+    st.session_state["region_filter"] = "All/Not allocated"
+    st.session_state["country_filter"] = "All/Not allocated"
     current_year = datetime.now().year
     default_start_year = current_year - 4
     st.session_state["end_year_range"] = (default_start_year, max_end_year)
-    st.session_state["crs_filter"] = ["All/Not allocated"]
+    st.session_state["crs_filter"] = "All/Not allocated"
     st.session_state["min_budget"] = min_budget_val
-    st.session_state["client_filter"] = ["All/Not allocated"]
+    st.session_state["client_filter"] = "All/Not allocated"
     st.session_state["query"] = ""
     st.session_state["show_exact_matches"] = False
     st.session_state["page"] = 1
@@ -145,19 +145,17 @@ def reset_filters():
 col1, col2, col3, col4, col5 = st.columns([1, 1, 1, 1, 1])
 
 with col1:
-    region_filter = st.multiselect("Region", options=["All/Not allocated"] + sorted(unique_sub_regions),
-                                   default=["All/Not allocated"], key="region_filter")
-if "All/Not allocated" in region_filter or not region_filter:
+    region_filter = st.selectbox("Region", ["All/Not allocated"] + sorted(unique_sub_regions), key="region_filter")
+if region_filter == "All/Not allocated":
     filtered_country_names = unique_country_names
 else:
     filtered_country_names = [
         name for name, code in country_name_mapping.items() 
-        if iso_code_to_sub_region.get(code) in region_filter
+        if iso_code_to_sub_region.get(code) == region_filter
     ]
 
 with col2:
-    country_filter = st.multiselect("Country", options=["All/Not allocated"] + filtered_country_names,
-                                    default=["All/Not allocated"], key="country_filter")
+    country_filter = st.selectbox("Country", ["All/Not allocated"] + filtered_country_names, key="country_filter")
 
 with col3:
     current_year = datetime.now().year
@@ -172,8 +170,7 @@ with col3:
 
 with col4:
     crs_options = ["All/Not allocated"] + get_crs_options(client, collection_name)
-    crs_filter = st.multiselect("CRS", options=crs_options,
-                                default=["All/Not allocated"], key="crs_filter")
+    crs_filter = st.selectbox("CRS", crs_options, key="crs_filter")
 
 with col5:
     min_budget = st.slider(
@@ -191,8 +188,7 @@ col1_2, col2_2, col3_2, col4_2, col5_2 = st.columns(5)
 
 with col1_2:
     client_options = sorted(project_data["client"].dropna().unique().tolist())
-    client_filter = st.multiselect("Client", options=["All/Not allocated"] + client_options,
-                                   default=["All/Not allocated"], key="client_filter")
+    client_filter = st.selectbox("Client", ["All/Not allocated"] + client_options, key="client_filter")
 with col2_2:
     st.empty()
 with col3_2:
@@ -200,25 +196,40 @@ with col3_2:
 with col4_2:
     st.empty()
 with col5_2:
-    st.empty()
-
+    # Plain reset button (will be moved to row 3 as well)
+    st.button("Reset Filters", on_click=reset_filters, key="reset_button_row2")
 
 ###########################################
 # Filter Controls - Row 3 (Remaining Filter)
 ###########################################
-col1_3, col2_3, col3_3 = st.columns([7, 23, 5])
+col1_3, col2_3, col3_3, col4_3, col5_3 = st.columns(5)
 with col1_3:
     # Place the "Show only exact matches" checkbox here
     show_exact_matches = st.checkbox("Show only exact matches", key="show_exact_matches")
 with col2_3:
     st.empty()
 with col3_3:
+    st.empty()
+with col4_3:
+    st.empty()
+with col5_3:
     # Right-align a more prominent reset button
     with st.container():
         st.markdown("<div style='text-align: right;'>", unsafe_allow_html=True)
         if st.button("**Reset Filters**", key="reset_button_row3"):
             reset_filters()
         st.markdown("</div>", unsafe_allow_html=True)
+
+###########################################
+# Helper function for valid project id
+###########################################
+def valid_project_id(pid_str):
+    if not pid_str:
+        return False
+    if pid_str.lower() in ["nan", "none"]:
+        return False
+    return True
+
 ###########################################
 # Main Search / Results
 ###########################################
@@ -237,7 +248,6 @@ else:
     semantic_thresholded = [r for r in semantic_all if r.score >= 0.0]
 
     # 2) Filter results based on the user’s selections
-    # (Assuming filter_results can handle a string "All/Not allocated" as meaning "no filter")
     filtered_semantic = filter_results(
         semantic_thresholded, 
         country_filter, 
@@ -263,10 +273,10 @@ else:
         get_country_name
     )
 
-    # Additional filter by client (if "All/Not allocated" is not selected)
-    if "All/Not allocated" not in client_filter:
-        filtered_semantic = [r for r in filtered_semantic if r.payload.get("metadata", {}).get("client", "Unknown Client") in client_filter]
-        filtered_lexical = [r for r in filtered_lexical if r.payload.get("metadata", {}).get("client", "Unknown Client") in client_filter]
+    # Additional filter by client
+    if client_filter != "All/Not allocated":
+        filtered_semantic = [r for r in filtered_semantic if r.payload.get("metadata", {}).get("client", "Unknown Client") == client_filter]
+        filtered_lexical = [r for r in filtered_lexical if r.payload.get("metadata", {}).get("client", "Unknown Client") == client_filter]
 
     # Remove duplicates
     filtered_semantic_no_dupe = remove_duplicates(filtered_semantic)
@@ -278,8 +288,11 @@ else:
         except (ValueError, TypeError):
             return value
 
-    # --- Reprint Query (Left aligned with "Query:") ---
-    st.markdown(f"<div style='text-align: left; font-size:2.1em; font-style: italic; font-weight: bold;'>Query: {var}</div>", unsafe_allow_html=True)
+    # --- Reprint Query (Left Aligned with "Query:") ---
+    st.markdown(
+        f"<div style='text-align: left; font-size:2.1em; font-style: italic; font-weight: bold;'>Query: {var}</div>",
+        unsafe_allow_html=True
+    )
 
     # 3) Display results
     # Lexical Search Results Branch
@@ -303,7 +316,7 @@ else:
             if "page" not in st.session_state:
                 st.session_state.page = 1
             current_page = st.session_state.page
-            # Top pagination widget (right aligned, occupying 1/7 of page width)
+            # Top pagination widget (right aligned, 1/7 width)
             col_pag_top = st.columns([6, 1])[1]
             new_page_top = col_pag_top.selectbox("Select Page", list(range(1, total_pages + 1)), index=current_page - 1, key="page_top")
             st.session_state.page = new_page_top
@@ -318,7 +331,7 @@ else:
                     metadata["title"] = compute_title(metadata)
                 title_html = highlight_query(metadata["title"], var) if var.strip() else metadata["title"]
                 title_clean = re.sub(r'<a.*?>|</a>', '', title_html)
-                # Prepend the result number and make title bold
+                # Prepend the result number
                 st.markdown(f"#### {i}. **{title_clean}**", unsafe_allow_html=True)
 
                 objective = metadata.get("objective", "None")
@@ -352,27 +365,28 @@ else:
                     new_crs_value_clean = re.sub(r'\.0$', '', str(new_crs_value))
                     crs_combined = f"{crs_key_clean}: {new_crs_value_clean}" if crs_key_clean else "Unknown"
                     
-                    # Extract and format predecessor and successor project IDs
-                    predecessor_raw = metadata.get("predecessor_id", "")
-                    successor_raw = metadata.get("successor_id", "")
-                    predecessor = safe_format_project_id(predecessor_raw)
-                    successor = safe_format_project_id(successor_raw)
-                    
-                    pred_success_ids = ""
-                    if predecessor or successor:
-                        parts = []
-                        if predecessor:
-                            parts.append(f"**Predecessor Project:** {predecessor}")
-                        if successor:
-                            parts.append(f"**Successor Project:** {successor}")
-                        pred_success_ids = "<br>" + " ".join(parts)
-                    
+                    predecessor = metadata.get("predecessor_id", "").strip()
+                    successor = metadata.get("successor_id", "").strip()
+                    parts = []
+                    if valid_project_id(predecessor):
+                        try:
+                            formatted_pred = format_project_id(int(float(predecessor)))
+                        except Exception:
+                            formatted_pred = predecessor
+                        parts.append(f"**Predecessor Project:** {formatted_pred}")
+                    if valid_project_id(successor):
+                        try:
+                            formatted_succ = format_project_id(int(float(successor)))
+                        except Exception:
+                            formatted_succ = successor
+                        parts.append(f"**Successor Project:** {formatted_succ}")
+                    extra_line = "<br>" + " | ".join(parts) if parts else ""
                     
                     additional_text = (
                         f"**Objective:** {highlight_query(objective, var)}<br>"
                         f"**Commissioned by:** {metadata.get('client', 'Unknown Client')}<br>"
                         f"**Projekt duration:** {start_year_str}-{end_year_str}<br>"
-                        f"**Budget:** Project: <b>{formatted_project_budget}</b>, Total volume: <b>{formatted_total_volume}</b>"
+                        f"**Budget:** Project: <b>{formatted_project_budget}</b>, Total volume: <b>{formatted_total_volume}</b><br>"
                         + extra_line +
                         f"<br>**Country:** {country_raw}<br>"
                         f"**Sector:** {crs_combined}"
@@ -401,7 +415,7 @@ else:
                 st.session_state.page = 1
             current_page = st.session_state.page
 
-            # Top pagination widget (right aligned, occupying 1/7 of page width)
+            # Top pagination widget (right aligned, 1/7 width)
             col_pag_top = st.columns([6, 1])[1]
             new_page_top = col_pag_top.selectbox("Select Page", list(range(1, total_pages + 1)), index=current_page - 1, key="page_top_sem")
             st.session_state.page = new_page_top
@@ -410,12 +424,12 @@ else:
             end_index = start_index + page_size
             top_results = filtered_semantic_no_dupe[start_index:end_index]
 
-            # Prominent page info with bold numbers (green if current page != 1)
+            # Prominent page info with bold numbers and green highlight if current page is not 1
             page_num = f"<b style='color: green;'>{st.session_state.page}</b>" if st.session_state.page != 1 else f"<b>{st.session_state.page}</b>"
             total_pages_str = f"<b>{total_pages}</b>"
             st.markdown(f"Showing **{len(top_results)}** Semantic Search results (Page {page_num} of {total_pages_str})", unsafe_allow_html=True)
 
-            # --- RAG Answer (Left aligned, bullet points, with bold numbers) ---
+            # --- RAG Answer (Left aligned, bullet points, bold numbers) ---
             rag_answer = get_rag_answer(var, top_results, DEDICATED_ENDPOINT, WRITE_ACCESS_TOKEN)
             bullet_lines = []
             for line in rag_answer.splitlines():
@@ -463,27 +477,29 @@ else:
                     new_crs_value = lookup_crs_value(crs_key_clean)
                     new_crs_value_clean = re.sub(r'\.0$', '', str(new_crs_value))
                     crs_combined = f"{crs_key_clean}: {new_crs_value_clean}" if crs_key_clean else "Unknown"
-
-                    # Extract and format predecessor and successor project IDs
-                    predecessor_raw = metadata.get("predecessor_id", "")
-                    successor_raw = metadata.get("successor_id", "")
-                    predecessor = safe_format_project_id(predecessor_raw)
-                    successor = safe_format_project_id(successor_raw)
                     
-                    pred_success_ids = ""
-                    if predecessor or successor:
-                        parts = []
-                        if predecessor:
-                            parts.append(f"**Predecessor Project:** {predecessor}")
-                        if successor:
-                            parts.append(f"**Successor Project:** {successor}")
-                        pred_success_ids = "<br>" + " ".join(parts)
+                    predecessor = metadata.get("predecessor_id", "").strip()
+                    successor = metadata.get("successor_id", "").strip()
+                    parts = []
+                    if valid_project_id(predecessor):
+                        try:
+                            formatted_pred = format_project_id(int(float(predecessor)))
+                        except Exception:
+                            formatted_pred = predecessor
+                        parts.append(f"**Predecessor Project:** {formatted_pred}")
+                    if valid_project_id(successor):
+                        try:
+                            formatted_succ = format_project_id(int(float(successor)))
+                        except Exception:
+                            formatted_succ = successor
+                        parts.append(f"**Successor Project:** {formatted_succ}")
+                    extra_line = "<br>" + " | ".join(parts) if parts else ""
                     
                     additional_text = (
                         f"**Objective:** {metadata.get('objective', '')}<br>"
                         f"**Commissioned by:** {metadata.get('client', 'Unknown Client')}<br>"
                         f"**Projekt duration:** {start_year_str}-{end_year_str}<br>"
-                        f"**Budget:** Project: <b>{formatted_project_budget}</b>, Total volume: <b>{formatted_total_volume}</b>"
+                        f"**Budget:** Project: <b>{formatted_project_budget}</b>, Total volume: <b>{formatted_total_volume}</b><br>"
                         + extra_line +
                         f"<br>**Country:** {country_raw}<br>"
                         f"**Sector:** {crs_combined}"
@@ -494,7 +510,7 @@ else:
                     st.markdown(additional_text, unsafe_allow_html=True)
                 st.divider()
 
-            # Bottom pagination widget (right aligned, occupying 1/7 width)
+            # Bottom pagination widget (right aligned, 1/7 width)
             col_pag_bot = st.columns([6, 1])[1]
             new_page_bot = col_pag_bot.selectbox("Select Page", list(range(1, total_pages + 1)), index=st.session_state.page - 1, key="page_bot_sem")
             st.session_state.page = new_page_bot