Update app.py

app.py

@@ -1,13 +1,14 @@
 import streamlit as st
 from docx import Document
 import io
-import json
-import pandas as pd
+from transformers import pipeline
 
-# Load job descriptions from JSON file
-def load_job_descriptions():
-    with open('job_descriptions.json', 'r') as file:
-        return json.load(file)
+# Load a pre-trained model for text generation
+@st.cache_resource
+def load_model():
+    # Using a smaller model like Flan-T5 for demonstration purposes
+    generator = pipeline("text-generation", model="google/flan-t5-large")
+    return generator
 
 # Function to create a Word document with the job description
 def create_word_document(designation, job_description):
@@ -24,7 +25,7 @@ def create_word_document(designation, job_description):
 
 # Streamlit app layout
 st.set_page_config(
-    page_title="Job Description Generator",
+    page_title="Construction Job Description Generator",
     page_icon="👷‍♂️",
     layout="wide"
 )
@@ -46,10 +47,6 @@ st.markdown("""
             background-color: #ffffff;
             border-radius: 5px;
         }
-        .stTextInput>div>div>input {
-            background-color: #ffffff;
-            border-radius: 5px;
-        }
         .header {
             font-size: 36px;
             color: #333333;
@@ -65,55 +62,48 @@ st.markdown("""
 """, unsafe_allow_html=True)
 
 # Header
-st.markdown('<div class="header">Construction Company Job Description Generator</div>', unsafe_allow_html=True)
+st.markdown('<div class="header">Construction Job Description Generator</div>', unsafe_allow_html=True)
+
+# Load the pre-trained model
+generator = load_model()
 
-# Load job descriptions
-data = load_job_descriptions()
+# Departments and Designations
+departments = {
+    "Construction": ["Construction Manager", "Civil Engineer", "Project Manager", "Site Supervisor"],
+    "Engineering": ["Structural Engineer", "Electrical Engineer", "Mechanical Engineer"],
+    "Safety": ["Safety Officer", "Environmental Specialist"]
+}
 
 # Step 1: Select Department
 st.markdown('<div class="subheader">Step 1: Select Department</div>', unsafe_allow_html=True)
-department = st.selectbox("Select the department:", list(data.keys()))
+department = st.selectbox("Select the department:", list(departments.keys()))
 
-# Step 2: Select Role based on the selected department
+# Step 2: Select Designation
 if department:
-    st.markdown('<div class="subheader">Step 2: Select Role</div>', unsafe_allow_html=True)
-    role = st.selectbox("Select the role:", list(data[department].keys()))
+    st.markdown('<div class="subheader">Step 2: Select Designation</div>', unsafe_allow_html=True)
+    designation = st.selectbox("Select the designation:", departments[department])
     
-    if role:
-        # Retrieve the job description for the selected role
-        job_description = data[department][role]
+    if designation:
+        # Generate a prompt for the model
+        prompt = f"Generate a detailed and professional job description for a {designation} in the {department} department."
+        
+        # Generate job description using the pre-trained model
+        with st.spinner("Generating job description..."):
+            generated_text = generator(prompt, max_length=300, num_return_sequences=1)[0]['generated_text']
         
         # Display job description
         st.markdown('<div class="subheader">Generated Job Description</div>', unsafe_allow_html=True)
-        st.write(job_description)
+        st.write(generated_text)
         
         # Create Word document
-        doc_io = create_word_document(role, job_description)
+        doc_io = create_word_document(designation, generated_text)
         
         # Download button for Word document
         st.download_button(
             label="Download Job Description as Word",
             data=doc_io,
-            file_name=f"{role.replace(' ', '_')}_job_description.docx",
+            file_name=f"{designation.replace(' ', '_')}_job_description.docx",
             mime="application/vnd.openxmlformats-officedocument.wordprocessingml.document"
         )
 else:
-    st.warning("Please select a department and role to generate the job description.")
-
-# Optional: Search functionality
-st.markdown('<div class="subheader">Or Search for a Role Across All Departments</div>', unsafe_allow_html=True)
-search_query = st.text_input("Search for a role:")
-
-if search_query:
-    # Flatten the data to search across all departments
-    flat_data = {f"{dept} - {role}": desc for dept, roles in data.items() for role, desc in roles.items()}
-    results = {key: value for key, value in flat_data.items() if search_query.lower() in key.lower()}
-    
-    if results:
-        st.markdown('<div class="subheader">Search Results</div>', unsafe_allow_html=True)
-        for result, desc in results.items():
-            st.write(f"**{result}**")
-            st.write(desc)
-            st.markdown("---")
-    else:
-        st.warning("No matching roles found.")
+    st.warning("Please select a department and designation to generate the job description.")