utils/extractor.py

import io
import fitz
import docx

MAX_BYTES = 10 * 1024 * 1024  # 10 MB
MAX_PAGES = 100
MAX_CHARS = MAX_PAGES * 3000  # ~3000 chars/page ceiling


def extract_pdf(data: bytes) -> str:
    doc = fitz.open(stream=data, filetype="pdf")
    pages = min(len(doc), MAX_PAGES)
    parts = []
    for i in range(pages):
        text = doc[i].get_text("text")
        parts.append(text)
    return "\n".join(parts)[:MAX_CHARS]


def extract_docx(data: bytes) -> str:
    doc = docx.Document(io.BytesIO(data))
    lines = []
    for para in doc.paragraphs:
        t = para.text.strip()
        if t:
            lines.append(t)
    return "\n".join(lines)[:MAX_CHARS]


def extract_text(data: bytes, filename: str) -> str:
    if len(data) > MAX_BYTES:
        raise ValueError("File exceeds 10 MB limit.")
    name = filename.lower()
    if name.endswith(".pdf"):
        return extract_pdf(data)
    if name.endswith(".docx"):
        return extract_docx(data)
    raise ValueError("Unsupported file type. Upload PDF or DOCX.")