main.py

from fastapi import FastAPI, File, UploadFile, HTTPException, BackgroundTasks
from pydantic import BaseModel
import fitz  # PyMuPDF
import tempfile, os, json, logging
from transformers import pipeline, T5Tokenizer, T5ForConditionalGeneration

# Initialize FastAPI app and logging
app = FastAPI()
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)

# Load model and tokenizer
model = T5ForConditionalGeneration.from_pretrained("google/flan-t5-base")
tokenizer = T5Tokenizer.from_pretrained("google/flan-t5-base", legacy=False)
qa_pipeline = pipeline("text2text-generation", model=model, tokenizer=tokenizer, device=-1)  # CPU-based inference

# Define the structured response model
class Education(BaseModel):
    degree: str
    university: str
    graduation_year: str

class ExtractedInfo(BaseModel):
    work_experience: str
    education: Education
    professional_course_detail: str
    software_usage: str
    safety_course_detail: str
    hse_description: str
    good_conduct_certificate: str

def extract_text_from_pdf(pdf_path: str) -> str:
    """Extracts text from the uploaded PDF."""
    with fitz.open(pdf_path) as doc:
        return "".join(page.get_text() for page in doc).strip()

def chunk_text(text: str, max_tokens: int = 512) -> list:
    """Splits the text into manageable chunks that fit within the token limit."""
    tokens = tokenizer.encode(text, add_special_tokens=False)
    return [tokenizer.decode(tokens[i:i + max_tokens], skip_special_tokens=True)
            for i in range(0, len(tokens), max_tokens)]

def process_chunk(chunk: str) -> dict:
    """Uses the model to extract structured JSON data from a text chunk."""
    prompt = f"""

    Extract the following information in JSON format:

    {{

        "work_experience": "<Summarized work experience>",

        "education": {{

            "degree": "<Degree obtained>",

            "university": "<University attended>",

            "graduation_year": "<Year of graduation>"

        }},

        "professional_course_detail": "<Professional courses>",

        "software_usage": "<Software tools used>",

        "safety_course_detail": "<Safety courses completed>",

        "hse_description": "<HSE practices>",

        "good_conduct_certificate": "<Good conduct certificate>"

    }}

    Resume text: {chunk}

    """

    try:
        response = qa_pipeline(prompt, max_new_tokens=150)
        generated_text = response[0]["generated_text"]
        # Extract JSON from the generated text
        json_start = generated_text.find("{")
        json_end = generated_text.rfind("}") + 1
        return json.loads(generated_text[json_start:json_end])
    except (json.JSONDecodeError, IndexError) as e:
        logger.warning(f"Failed to parse JSON: {e}")
        return {}

def merge_outputs(chunks: list) -> dict:
    """Combines multiple chunk outputs into a single structured result."""
    merged = {
        "work_experience": "",
        "education": {"degree": "", "university": "", "graduation_year": ""},
        "professional_course_detail": "",
        "software_usage": "",
        "safety_course_detail": "",
        "hse_description": "",
        "good_conduct_certificate": ""
    }

    for chunk in chunks:
        chunk_output = process_chunk(chunk)
        for key, value in chunk_output.items():
            if isinstance(value, dict):
                merged[key].update(value)
            elif not merged[key]:
                merged[key] = value

    return merged

@app.post("/process_cv/", response_model=ExtractedInfo)
async def process_cv(file: UploadFile = File(...), background_tasks: BackgroundTasks = BackgroundTasks()):
    """Processes a PDF resume and returns structured information in JSON format."""
    if not file.filename.endswith(".pdf"):
        raise HTTPException(status_code=400, detail="Only PDF files are allowed.")

    temp_path = tempfile.mktemp(suffix=".pdf")
    with open(temp_path, "wb") as f:
        f.write(await file.read())

    try:
        # Extract text from the uploaded PDF
        text = extract_text_from_pdf(temp_path)
        if not text:
            raise HTTPException(status_code=400, detail="No extractable text found in the PDF.")

        # Process the text in chunks and merge the output
        chunks = chunk_text(text)
        structured_data = merge_outputs(chunks)
        return ExtractedInfo(**structured_data)
    finally:
        os.remove(temp_path)
        background_tasks.add_task(clean_temp_files)

def clean_temp_files():
    """Cleans up temporary PDF files."""
    for filename in os.listdir(tempfile.gettempdir()):
        if filename.endswith(".pdf"):
            try:
                os.remove(os.path.join(tempfile.gettempdir(), filename))
            except Exception as e:
                logger.warning(f"Failed to delete {filename}: {e}")