Update with rerank score

app.py

@@ -293,6 +293,11 @@ def check_keywords_in_content(database_path, table_name, input_id, keywords):
 
     return contains_keywords
 
+def clear_temp_candidates():
+    if not st.session_state["final_candidates"]:
+        print("i am cleared")
+        st.session_state["docs_res"] = []
+        
 def load_candidates(fillup):
     with st.spinner("Load the candidates, this may take a moment..."):
         # print(st.session_state["job_string"])
@@ -320,16 +325,36 @@ def load_candidates(fillup):
         # print(filter_string)
         if not fillup:
             while len(checked_candidates) < target_candidates_count:
-                # Führe eine similarity search durch und erhalte 100 Kandidaten
-                if st.session_state["search_type"]:
-                    print("hybrid")
-                    # raw_candidates = st.session_state["db"].hybrid_search(query_string, k=candidates_per_search+current_offset, filters=filter_string)
-                    raw_candidates = st.session_state["db"].hybrid_search_with_score(query_string, k=candidates_per_search+current_offset, filters=filter_string)
-                else:
+                # # Führe eine similarity search durch und erhalte 100 Kandidaten
+                # if st.session_state["search_type"]:
+                #     print("hybrid")
+                #     # raw_candidates = st.session_state["db"].hybrid_search(query_string, k=candidates_per_search+current_offset, filters=filter_string)
+                #     raw_candidates = st.session_state["db"].hybrid_search_with_score(query_string, k=candidates_per_search+current_offset, filters=filter_string)
+                # else:
+                #     print("similarity")
+                #     # raw_candidates = st.session_state["db"].similarity_search(query_string, k=candidates_per_search+current_offset, filters=filter_string)
+                #     raw_candidates = st.session_state["db"].similarity_search_with_relevance_scores(query_string, k=candidates_per_search+current_offset, filters=filter_string)
+                #"Similarity", "Hybrid", "Semantic ranking"
+                if st.session_state["search_radio"] == "Similarity":
                     print("similarity")
                     # raw_candidates = st.session_state["db"].similarity_search(query_string, k=candidates_per_search+current_offset, filters=filter_string)
                     raw_candidates = st.session_state["db"].similarity_search_with_relevance_scores(query_string, k=candidates_per_search+current_offset, filters=filter_string)
-
+                elif st.session_state["search_radio"] == "Hybrid":
+                    print("hybrid")
+                    # raw_candidates = st.session_state["db"].hybrid_search(query_string, k=candidates_per_search+current_offset, filters=filter_string)
+                    raw_candidates = st.session_state["db"].hybrid_search_with_score(query_string, k=candidates_per_search+current_offset, filters=filter_string)
+                elif st.session_state["search_radio"] == "Semantic ranking":
+                    print("Semantic ranking")
+                    print("Filter string"+filter_string)
+                    print("query"+query_string)
+                    print("offset: "+str(candidates_per_search+current_offset))
+                    # raw_candidates = st.session_state["db"].hybrid_search(query_string, k=candidates_per_search+current_offset, filters=filter_string)
+                    try:
+                        raw_candidates = st.session_state["db"].semantic_hybrid_search_with_score_and_rerank(query_string, k=50, filters=filter_string)
+                    except Exception as e:
+                        print(f"Fehler beim laden der Kandidaten: {str(e)}")
+                        raw_candidates = []
+                        st.warning("Something went wrong. Please press 'Search candidates' again or reload the page.")
                 for candidate in raw_candidates[current_offset:]:
                     candidates_id = candidate[0].metadata["source"].split("/")[-1]
                     keyword_bool = check_keywords_in_content(db_path, table_name, candidates_id, text_area_params.split(','))
@@ -358,14 +383,26 @@ def load_candidates(fillup):
             # Solange die Anzahl der überprüften Kandidaten kleiner als die Zielanzahl ist
             while len(st.session_state["docs_res"]) < target_candidates_count:
                 # Führe eine similarity search durch und erhalte 100 Kandidaten
-                if st.session_state["search_type"]:
-                    print("hybrid")
-                    # raw_candidates = st.session_state["db"].hybrid_search(query_string, k=candidates_per_search+current_offset, filters=filter_string)
-                    raw_candidates = st.session_state["db"].hybrid_search_with_score(query_string, k=candidates_per_search+current_offset, filters=filter_string)
-                else:
+                if st.session_state["search_radio"] == "Similarity":
                     print("similarity")
                     # raw_candidates = st.session_state["db"].similarity_search(query_string, k=candidates_per_search+current_offset, filters=filter_string)
                     raw_candidates = st.session_state["db"].similarity_search_with_relevance_scores(query_string, k=candidates_per_search+current_offset, filters=filter_string)
+                elif st.session_state["search_radio"] == "Hybrid":
+                    print("hybrid")
+                    # raw_candidates = st.session_state["db"].hybrid_search(query_string, k=candidates_per_search+current_offset, filters=filter_string)
+                    raw_candidates = st.session_state["db"].hybrid_search_with_score(query_string, k=candidates_per_search+current_offset, filters=filter_string)
+                elif st.session_state["search_radio"] == "Semantic ranking":
+                    print("Semantic ranking")
+                    print("Filter string"+filter_string)
+                    print("query"+query_string)
+                    print("offset: "+str(candidates_per_search+current_offset))
+                    # raw_candidates = st.session_state["db"].hybrid_search(query_string, k=candidates_per_search+current_offset, filters=filter_string)
+                    try:
+                        raw_candidates = st.session_state["db"].semantic_hybrid_search_with_score_and_rerank(query_string, k=50, filters=filter_string)
+                    except Exception as e:
+                        print(f"Fehler beim laden der Kandidaten: {str(e)}")
+                        raw_candidates = []
+                        st.warning("Something went wrong. Please press 'Search candidates' again or reload the page.")
                 temp_offset_add = 0 
                 for candidate in raw_candidates[current_offset:]:
                     candidates_id = candidate[0].metadata["source"].split("/")[-1]
@@ -425,13 +462,14 @@ col1.title("Candidate Search")
 col2.image("https://www.workgenius.com/wp-content/uploads/2023/03/WorkGenius_navy-1.svg")
 
 st.write("Please upload the job description for which you would like candidates to be proposed.")
-col_file, col_clear = st.columns([6,1])
+uploaded_file_jobdescription = st.file_uploader("Upload the job description:", type=["pdf"], key="job")
+# col_file, col_clear = st.columns([6,1])
 
-with col_file:
-    uploaded_file_jobdescription = st.file_uploader("Upload the job description:", type=["pdf"], key="job")
-with col_clear:
-    if st.button("Clear", use_container_width=True):
-        streamlit_js_eval(js_expressions="parent.window.location.reload()")
+# with col_file:
+#     uploaded_file_jobdescription = st.file_uploader("Upload the job description:", type=["pdf"], key="job")
+# with col_clear:
+#     if st.button("Clear", use_container_width=True):
+#         streamlit_js_eval(js_expressions="parent.window.location.reload()")
 
 if st.session_state["job"]:
     if not st.session_state["job_string"]:
@@ -470,8 +508,10 @@ if st.session_state["job"]:
             st.session_state["job_string"] = st.session_state["optimized_job_edited"]
             st.rerun()
 
-st.write("Switch from a similarity search (default) to a hybrid search (activated)")
-st.toggle("Switch Search", key="search_type")
+# st.write("Switch from a similarity search (default) to a hybrid search (activated)")
+# st.toggle("Switch Search", key="search_type")
+
+st.radio("Select a search variant",options=["Similarity", "Hybrid", "Semantic ranking"], key="search_radio",on_change=clear_temp_candidates)
 
 st.write("Activate the following toggles to filter according to the respective properties:")
 col_screening, col_handoff, col_placed = st.columns([1,1,1])
@@ -514,8 +554,15 @@ if (st.session_state["job_string"] and submit) or st.session_state["docs_res"]:
                     st.rerun()
             with cols_final[0]:
                 # st.subheader(doc.metadata["source"])
-                with st.expander(doc[0].metadata["name"]+" with a search score of: "+str(round(doc[1] * 100, 3))+"%"):
-                    st.write(doc[0].page_content)
+                if st.session_state["search_radio"] == "Similarity":
+                    with st.expander(doc[0].metadata["name"]+" with a similarity percentage of: "+str(round(doc[1] * 100, 3))+ "%"):
+                        st.write(doc[0].page_content)
+                elif st.session_state["search_radio"] == "Hybrid":
+                    with st.expander(doc[0].metadata["name"]+" with a hybrid search score of: "+str(round(doc[1] * 100, 3))):
+                        st.write(doc[0].page_content)
+                elif st.session_state["search_radio"] == "Semantic ranking":
+                    with st.expander(doc[0].metadata["name"]+" with a rerank score of: "+str(round(doc[2] * 100, 3))):
+                        st.write(doc[0].page_content)
         if len(st.session_state["docs_res"])>=10:    
             if st.button("Accept candidates", key="accept_candidates_btn"):       
                 print("hello")
@@ -698,7 +745,7 @@ Your Candidate-Search-Tool
                         try:
                             bullets = generate_job_bullets(st.session_state["job_string"])
                             client = Client(os.getenv("TWILIO_SID"), os.getenv("TWILIO_API"))
-                            message_body = f"Dear candidate,\n\nare you interested in the following position: \n\n"+st.session_state["job_title"]+"\n"+bullets+"\n\nThen please answer with 'yes'\n\nSincerely,\n"+"WorkGenius"
+                            message_body = f"Dear candidate,\n\nare you interested in the following position: \n"+st.session_state["job_title"]+"\n\n"+bullets+"\n\nThen please answer with 'yes'\n\nSincerely,\n"+"WorkGenius"
                             message = client.messages.create(
                                 to=st.session_state["recruiter_phone"],
                                 from_="+1 857 214 8753",