Update app.py

app.py

@@ -390,44 +390,38 @@ def evaluate_answers():
     else: # --- NON-HR (Analytics, Finance) Evaluation Logic ---
         base_assessment_criteria_qualitative_non_hr = """
         For the OVERALL qualitative summary, assess responses based on:
-            - Conceptual Understanding (effort and relevance more than perfect accuracy for the level)
-            - Communication Clarity (can the core idea be understood?)
-            - Depth of Explanation (relative to expected level)
-            - Use of Examples (if any, and if appropriate for the level)
-            - Logical Flow (is there a basic structure or train of thought?)
+        - Conceptual Understanding (effort and relevance more than perfect accuracy for the level)
+        - Communication Clarity (can the core idea be understood?)
+        - Depth of Explanation (relative to expected level)
+        - Use of Examples (if any, and if appropriate for the level)
+        - Logical Flow (is there a basic structure or train of thought?)
         """
         per_question_scoring_guidelines_non_hr = f"""
-            For EACH question and its answer, provide a score from 0 to 5 points.
-            The candidate is at a {level_string} level.
-            Scoring should reflect your role as a balanced evaluator — neither too lenient nor too harsh. Use wisdom and professional judgment.
-
-            Consider the following when assigning the score:
-            - Effort: Was there a genuine attempt to answer meaningfully?
-            - Relevance: Is the response at least partially on-topic?
-            - Clarity: Is the main idea understandable?
-            - Logic: Does the response have a clear train of thought?
-            - Examples: Were examples used when appropriate?
-            
-            Scoring Rubric:
-            - **5 points:** Clear, relevant, and reasonably accurate with good structure. Shows understanding and effort.
-            - **4 points:** Mostly correct with minor gaps or unclear phrasing. Shows some grasp and engagement.
-            - **3 points:** Partially correct or somewhat vague, but a fair attempt is made.
-            - **1–2 points:** Minimal relevance or weak structure, but not entirely blank or off-topic.
-            - **0 points:** Candidate says “I don’t know,” “not sure,” or gives unrelated/placeholder text.
+        For EACH question and its answer, provide a score from 0 to 5 points.
+        The candidate is at a {level_string} level.
+        Consider the following when assigning the per-question score:
+        - Effort: Did the candidate attempt a meaningful answer, even if partially incorrect?
+        - Relevance: Is the response at least partially related to the question topic?
+        - Clarity of thought for the candidate's level.
+        - Basic logical structure.
+        - Use of examples, if any were given and appropriate.
         """
-
         if level_string == "beginner":
             level_specific_instructions_non_hr = """
-                You are a **supportive but realistic evaluator** for a **BEGINNER/FRESHER**. Encourage growth and confidence, but give honest scores based on visible effort, relevance, and basic understanding.
+            You are an **extremely understanding, encouraging, and supportive** interview evaluator for a **BEGINNER/FRESHER**. Your primary goal is to **build confidence**.
+            **Scoring Guidelines for Beginners (0-5 points per question):**
+            - **5 points:** Accurate, clear, and well-structured answer. Shows clear effort and basic understanding.
+            - **4 points:** Mostly correct with minor gaps or unclear phrasing.Good attempt, relevant, shows some understanding or key terms (e.g., one/two relevant words).
+            - **3 points:** Partially correct with evident effort, but lacks clarity or completeness.
+            - **1-2 points:** Minimal effort, mostly irrelevant, but an attempt beyond silence.
+            - **0 points:** Candidate explicitly says "I don’t know", "I'm not sure", or provides placeholder/non-answers. No relevant effort or understanding shown.Incorrect or unrelated.
+            Provide VERY positive feedback.
             """
         elif level_string == "intermediate":
-            level_specific_instructions_non_hr = """
-                You are a **professional and balanced evaluator** for an **INTERMEDIATE** candidate. Score honestly based on clarity, partial correctness, and application effort.
-            """
-        else:  # advanced
-            level_specific_instructions_non_hr = """
-                You are a **discerning but fair evaluator** for an **ADVANCED** candidate. Expect depth and accuracy, but score reasonably when structure and effort are evident.
-            """        
+            level_specific_instructions_non_hr = """Supportive evaluator for **INTERMEDIATE**. Scoring (0-5): 5=Correct/Clear; 3-4=Mostly correct; 1-2=Partial/Gaps; 0=Incorrect."""
+        else: # Advanced
+            level_specific_instructions_non_hr = """Discerning evaluator for **ADVANCED**. Scoring (0-5): 5=Accurate/Comprehensive; 3-4=Correct lacks nuance; 1-2=Inaccurate; 0=Fundamentally incorrect."""
+        
         evaluation_prompt_template_non_hr = f"""
         {level_specific_instructions_non_hr} 
         {per_question_scoring_guidelines_non_hr}
@@ -711,24 +705,53 @@ st.sidebar.markdown(f"**Selected Domain:** {st.session_state['selected_domain']}
 num_qs = st.sidebar.slider("Number of Questions:", 1, 10, 3)
 
 if st.session_state["selected_domain"] == "Soft Skills":
+    soft_skill_mode = st.sidebar.radio(
+        "Choose Soft Skills Mode:",
+        ("Resume-Based", "HR Round")
+    )
+    if soft_skill_mode == "Resume-Based":
+        uploaded_file = st.sidebar.file_uploader("Upload Resume:", type=["pdf", "txt"])
+        if uploaded_file:
+            input_text = extract_pdf_text(uploaded_file)
+    else:
+        input_text = "General HR Round"
+
     if st.sidebar.button("Generate Questions"):
-        st.session_state["generated_questions"] = sample(hr_questions, num_qs)
+        if soft_skill_mode == "HR Round":
+            st.session_state["generated_questions"] = sample(hr_questions, num_qs)
+        else:
+            if not input_text.strip():
+                st.warning("⚠️ Please upload a resume.")
+                st.stop()
+            prompt = f"Ask {num_qs} HR-style interview questions based on this resume: {input_text}"
+            model = genai.GenerativeModel('gemini-1.5-pro-latest')
+            response = model.generate_content([prompt])
+            questions = [q.strip("* ") for q in response.text.strip().split("\n") if q.strip()]
+            st.session_state["generated_questions"] = questions[:num_qs]
+
         st.session_state["current_question_index"] = 0
         st.rerun()
+    
 else:
-    section_choice = st.sidebar.radio("Choose Input Type:", ("Resume", "Job Description", "Skills"))
-    difficulty = st.sidebar.selectbox("Select Difficulty Level:", ["Beginner", "Intermediate", "Advanced"])
+    section_choice = st.sidebar.radio(
+        "Choose Input Type:",
+        ("Resume", "Job Description", "Skills", "Company Specific") if st.session_state["selected_domain"] == "Finance" else ("Resume", "Job Description", "Skills")
+    )
+    #difficulty = st.sidebar.selectbox("Select Difficulty Level:", ["Beginner", "Intermediate", "Advanced"])
     input_text = ""
 
     if section_choice == "Resume":
+        difficulty = st.sidebar.selectbox("Select Difficulty Level:", ["Beginner", "Intermediate", "Advanced"])
         uploaded_file = st.sidebar.file_uploader("Upload Resume:", type=["pdf", "txt"])
         if uploaded_file:
             input_text = extract_pdf_text(uploaded_file)
 
     elif section_choice == "Job Description":
+        difficulty = st.sidebar.selectbox("Select Difficulty Level:", ["Beginner", "Intermediate", "Advanced"])
         input_text = st.sidebar.text_area("Paste Job Description:")
 
     elif section_choice == "Skills":
+        difficulty = st.sidebar.selectbox("Select Difficulty Level:", ["Beginner", "Intermediate", "Advanced"])
         input_text = ""
 
         if st.session_state["selected_domain"] == "Finance":
@@ -737,21 +760,27 @@ else:
 
             difficulty = st.session_state.get("difficulty", "Beginner")
 
-            if selected_level != "Level-1":
+            if selected_level == "Level-1":
+                excel_filename = "CIBOP Mock Questions.xlsx"
+                module_prefix = "MODULE 1"
+            elif selected_level == "Level-2":
+                excel_filename = "CIBOP Level2.xlsx"
+                module_prefix = "MODULE 2"
+            else:
                 st.sidebar.warning(f"🚧 {selected_level} content is still under development. Please select Level-1 to continue.")
                 st.stop()
 
             # Map difficulty level to column in Excel
             column_map = {
-                "Beginner": "MODULE 1-EASY",
-                "Intermediate": "MODULE 1-MEDIUM",
-                "Advanced": "MODULE 1-DIFFICULT"
+            "Beginner": f"{module_prefix}-EASY",
+            "Intermediate": f"{module_prefix}-MEDIUM",
+            "Advanced": f"{module_prefix}-DIFFICULT"
             }
 
             selected_column = column_map[difficulty]
 
             # Load Excel and questions
-            excel_path = os.path.join("data", "CIBOP Mock Questions.xlsx")
+            excel_path = os.path.join("data", excel_filename)
             try:
                 df = pd.read_excel(excel_path, engine="openpyxl")
                 questions_from_excel = df[selected_column].dropna().astype(str).tolist()
@@ -762,16 +791,57 @@ else:
 
             st.sidebar.success(f"✅ Loaded {difficulty}-level questions from {selected_level}")
 
-        else:
-            # For Analytics or any other domain
-            skills = {
+    elif section_choice == "Company Specific" and st.session_state["selected_domain"] == "Finance":
+        excel_path = os.path.join("data", "Company Specific.xlsx")
+        try:
+                # Load Excel and get sheet names (company names)
+            xls = pd.ExcelFile(excel_path, engine="openpyxl")
+            company_names = xls.sheet_names
+        except Exception as e:
+            st.sidebar.error(f"❌ Error loading company-specific Excel: {e}")
+            st.stop()
+
+        selected_company = st.sidebar.selectbox("Select Company:", company_names)
+
+        try:
+            # Load the selected company's sheet
+            df = pd.read_excel(excel_path, sheet_name=selected_company, engine="openpyxl")
+
+            if "Job Role" not in df.columns:
+                st.sidebar.error(f"❌ 'JobRole' column not found in sheet '{selected_company}'.")
+                st.stop()
+
+            job_roles = sorted(df["Job Role"].dropna().unique())
+            selected_job_role = st.sidebar.selectbox("Select Job Role:", job_roles)
+
+            filtered_df = df[df["Job Role"] == selected_job_role]
+
+            if "Question" in filtered_df.columns:
+                questions_from_excel = filtered_df["Question"].dropna().astype(str).tolist()
+            else:
+                question_cols = [col for col in filtered_df.columns if col != "JobRole"]
+                if not question_cols:
+                    st.sidebar.error(f"❌ No question column found in '{selected_company}' sheet.")
+                    st.stop()
+                questions_from_excel = filtered_df[question_cols[0]].dropna().astype(str).tolist()
+
+            input_text = f"{selected_company} - {selected_job_role}"
+            st.sidebar.success(f"✅ Loaded {len(questions_from_excel)} questions for {selected_company} / {selected_job_role}")
+
+        except Exception as e:
+            st.sidebar.error(f"❌ Error reading sheet '{selected_company}': {e}")
+            st.stop()
+
+    else:
+        # For Analytics or any other domain
+        skills = {
                 "Analytics": ["Python", "SQL", "Machine Learning", "Statistics", "Business Analytics"]
             }
-            skill_list = skills.get(st.session_state["selected_domain"], [])
-            if skill_list:
-                selected_skill = st.sidebar.selectbox("Select a Skill:", skill_list, key="skill_select")
-                input_text = selected_skill
-                st.sidebar.markdown(f"✅ Selected Skill: **{selected_skill}**")
+        skill_list = skills.get(st.session_state["selected_domain"], [])
+        if skill_list:
+            selected_skill = st.sidebar.selectbox("Select a Skill:", skill_list, key="skill_select")
+            input_text = selected_skill
+            st.sidebar.markdown(f"✅ Selected Skill: **{selected_skill}**")
 
 
     if st.sidebar.button("Generate Questions"):
@@ -779,7 +849,7 @@ else:
             st.warning("⚠️ Please provide input based on the selected method.")
             st.stop()
 
-        if st.session_state["selected_domain"] == "Finance" and section_choice == "Skills":
+        if st.session_state["selected_domain"] == "Finance" and section_choice in ["Skills","Company Specific"]:
             st.session_state["generated_questions"] = sample(questions_from_excel, min(num_qs, len(questions_from_excel)))
         else:
             prompt = f"Ask {num_qs} direct and core-level {difficulty} interview questions related to {input_text}. Do not include intros or numbering."