Update app.py

app.py

@@ -6,19 +6,38 @@ import json
 from datetime import datetime
 
 import streamlit as st
-from transformers import pipeline
+import requests
 from docx import Document
 from PyPDF2 import PdfReader
 
 # ------------------------------------------------------------------------------
-# Model Pipeline Setup
+# Groq API Setup
 # ------------------------------------------------------------------------------
 
-# Use an instruction-tuned model (e.g., Flan-T5-xl)
-model_name = "google/flan-t5-xl"
-instruct_pipeline = pipeline(
-    "text2text-generation", model=model_name, max_length=1024, truncation=True
-)
+# Your Groq API key is expected to be saved as a secret in the environment
+GROQ_API_KEY = os.getenv("GROQ_API")
+# Replace the following endpoint with the actual Groq API endpoint if different.
+GROQ_ENDPOINT = "https://api.groq.ai/v1/generate"
+
+def generate_text(prompt: str, max_length: int = 1024) -> str:
+    """
+    Call the Groq API with a prompt and return the generated text.
+    """
+    headers = {
+        "Authorization": f"Bearer {GROQ_API_KEY}",
+        "Content-Type": "application/json"
+    }
+    payload = {
+        "prompt": prompt,
+        "max_length": max_length
+    }
+    response = requests.post(GROQ_ENDPOINT, headers=headers, json=payload)
+    if response.ok:
+        # Adjust according to the actual API response structure.
+        return response.json().get("generated_text", "").strip()
+    else:
+        st.error(f"Error in Groq API call: {response.text}")
+        return ""
 
 # ------------------------------------------------------------------------------
 # Helper Functions for File Reading
@@ -49,7 +68,7 @@ def extract_resume_text(file_obj) -> str:
     Supports PDF, DOCX, and plain text.
     """
     file_bytes = file_obj.read()
-    file_obj.seek(0)  # Reset pointer for potential further use
+    file_obj.seek(0)  # Reset pointer for further use if needed
     filename = file_obj.name if hasattr(file_obj, "name") else "resume.txt"
     ext = os.path.splitext(filename)[-1].lower()
 
@@ -61,14 +80,14 @@ def extract_resume_text(file_obj) -> str:
         return read_plain(file_bytes)
 
 # ------------------------------------------------------------------------------
-# Core AI Functions
+# Core AI Functions Using Groq API
 # ------------------------------------------------------------------------------
 
 def parse_resume(resume_text: str) -> str:
     """
     Extract candidate details from the resume text.
-    The model returns a JSON with keys: first_name, last_name, location,
-    work_experience, school_experience, and skills.
+    The prompt instructs the model to output a JSON with keys:
+    first_name, last_name, location, work_experience, school_experience, and skills.
     """
     prompt = (
         "You are an expert resume parser. Extract the following keys from the resume text: "
@@ -76,8 +95,7 @@ def parse_resume(resume_text: str) -> str:
         "Return the answer strictly in JSON format with no extra text.\n\n"
         f"Resume:\n{resume_text}"
     )
-    result = instruct_pipeline(prompt, max_length=512)[0]["generated_text"]
-    return result.strip()
+    return generate_text(prompt, max_length=512)
 
 def generate_cover_letter(candidate_json: str, job_description: str, date_str: str) -> str:
     """
@@ -95,8 +113,7 @@ def generate_cover_letter(candidate_json: str, job_description: str, date_str: s
             candidate_json=candidate_json, job_description=job_description, date=date_str
         )
     )
-    result = instruct_pipeline(prompt, max_length=1024)[0]["generated_text"]
-    return result.strip()
+    return generate_text(prompt, max_length=1024)
 
 def generate_resume(first_name: str, last_name: str, location: str,
                     work_experience: str, school_experience: str, skills: str) -> str:
@@ -119,8 +136,7 @@ def generate_resume(first_name: str, last_name: str, location: str,
         "personal details, work experience, education, and skills. Ensure the resume is concise and professional.\n\n"
         "Candidate Information:\n" + candidate_str + "\n\nResume:"
     )
-    result = instruct_pipeline(prompt, max_length=1024)[0]["generated_text"]
-    return result.strip()
+    return generate_text(prompt, max_length=1024)
 
 # ------------------------------------------------------------------------------
 # Streamlit UI
@@ -128,7 +144,7 @@ def generate_resume(first_name: str, last_name: str, location: str,
 
 st.set_page_config(page_title="AI-Powered Job Application Assistant", layout="wide")
 st.title("AI-Powered Job Application Assistant")
-st.markdown("Generate a **Cover Letter** from your resume or **Create a Resume** from scratch using AI.")
+st.markdown("Generate a **Cover Letter** from your resume or **Create a Resume** from scratch using AI via Groq API.")
 
 # Create tabs for the two functionalities
 tabs = st.tabs(["Cover Letter Generator", "Resume Creator"])