pdf_utils.py

from langchain_core.output_parsers import StrOutputParser
from langchain_core.prompts import PromptTemplate
from langchain_huggingface import HuggingFacePipeline
from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline
import torch
from fpdf import FPDF
import re

# --- MODEL SETUP ---
model_id = "deepseek-ai/deepseek-coder-1.3b-instruct"

tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    dtype=torch.float16,
    device_map="auto",
    offload_folder="./offload"
)

pipe = pipeline("text-generation", model=model, tokenizer=tokenizer)
llm = HuggingFacePipeline(
    pipeline=pipe,
    model_kwargs={
        "max_new_tokens": 2000,
        "do_sample": True,
        "temperature": 0.2,
        "repetition_penalty": 1.05,
        "eos_token_id": tokenizer.eos_token_id,
    }
)

# --- PROMPT ---
test_prompt = PromptTemplate.from_template(
    """You are an expert QA engineer.

Generate EXACTLY 10 numbered test cases:
1–5 functional
6–10 edge cases

Rules:
- Output ONLY the numbered list
- Each test must be one sentence
- No explanations, no markdown

Code:
{code}

Test cases:
"""
)


# --- 🔒 CRITICAL FIX: SAFE TEXT WRAPPING FOR PDF ---
def wrap_text_for_pdf(text, max_chars=90):
    wrapped_lines = []
    for line in text.split("\n"):
        if len(line) <= max_chars:
            wrapped_lines.append(line)
        else:
            for i in range(0, len(line), max_chars):
                wrapped_lines.append(line[i:i + max_chars])
    return "\n".join(wrapped_lines)

# --- TEST CASE GENERATION ---
def test_case(code):
    test_chain = test_prompt | llm | StrOutputParser()
    test_cases = test_chain.invoke({"code": code})

    print("\n[LOG] 1. LLM Raw Output Length:", len(test_cases))

    # Clean markdown
    test_cases = re.sub(r"```.*?```", "", test_cases, flags=re.DOTALL)
    test_cases = re.sub(r"```", "", test_cases).strip()

    if not test_cases:
        test_cases = "Error: Test case generation failed."

    print("\n[LOG] 2. Cleaned Text:", test_cases)

    # 🔒 APPLY WRAPPING HERE
    safe_text = wrap_text_for_pdf(test_cases)

    pdf = FPDF()
    pdf.add_page()

    try:
        pdf.add_font(
            "DejaVuSans",
            style="",
            fname="/usr/share/fonts/truetype/dejavu/DejaVuSans.ttf",
        )
        pdf.set_font("DejaVuSans", size=10)
    except Exception as e:
        print(f"[LOG] Font fallback: {e}")
        pdf.set_font("Arial", size=10)

    pdf.multi_cell(0, 8, txt="--- Generated Test Cases ---", align="C")
    pdf.ln(2)
    pdf.multi_cell(0, 8, txt=safe_text)

    try:
        pdf_bytes = bytes(pdf.output(dest="S"))

        print("\n[LOG] 3. PDF Bytes Length:", len(pdf_bytes))

        if len(pdf_bytes) < 100:
            raise ValueError("PDF too small")

        return pdf_bytes

    except Exception as e:
        print(f"[FATAL] PDF failed: {e}")
        error_pdf = FPDF()
        error_pdf.add_page()
        error_pdf.set_font("Arial", size=12)
        error_pdf.multi_cell(0, 10, txt=f"PDF generation failed: {e}")
        return bytes(error_pdf.output(dest="S"))