Run3 pdf text extraction for classification

Skill cxcscmu/SkillLearnBench/skills/b3-teacher-feedback-claude-opus-4-6/organize-messy-files/run3_pdf-text-extraction-for-classification

[COLM'26] SkillLearnBench is the first benchmark for evaluating continual learning methods that automatically generate agent skills.

Install

npx -y skills add cxcscmu/SkillLearnBench --skill run3_pdf-text-extraction-for-classification

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

What its author says it does

Copied from the file, not written here

How to extract text from PDF files for content-based classification. Use this when you need to read PDF content to determine what subject/topic a paper belongs to. Covers both text-based and scanned PDFs.

SKILL.md

2.7 KB, as published. Nobody here has run it

PDF Text Extraction for Classification

Primary Method: PyMuPDF (fitz)

PyMuPDF is the most reliable library for extracting text from PDFs. Install with pip install PyMuPDF.

import fitz  # PyMuPDF

def extract_text_from_pdf(filepath, max_pages=20):
    """Extract text from a PDF, reading up to max_pages pages."""
    text = ""
    try:
        doc = fitz.open(filepath)
        num_pages = min(len(doc), max_pages)
        for page_num in range(num_pages):
            page = doc[page_num]
            page_text = page.get_text()
            text += page_text + "\n"
        doc.close()
    except Exception as e:
        print(f"Error reading {filepath}: {e}")
    return text

Key Details

Always read at least 15-20 pages — some papers have long introductions and the key subject matter doesn't appear until later pages. Reading only 3-5 pages is NOT enough.
Extract ALL available text — for short papers (< 20 pages), read the entire document.
Fallback for empty text — if get_text() returns empty or very short strings (< 50 chars total), the PDF may be scanned/image-based. In that case, try OCR or fall back to filename analysis.

Fallback: pdfminer

from pdfminer.high_level import extract_text

def extract_text_pdfminer(filepath):
    try:
        return extract_text(filepath)
    except Exception:
        return ""

Fallback: OCR with pdf2image + pytesseract

For scanned PDFs where text extraction returns nothing:

from pdf2image import convert_from_path
import pytesseract

def ocr_pdf(filepath, max_pages=5):
    text = ""
    try:
        images = convert_from_path(filepath, last_page=max_pages)
        for img in images:
            text += pytesseract.image_to_string(img) + "\n"
    except Exception:
        pass
    return text

Reading PPTX and DOCX files

from pptx import Presentation
import docx

def extract_text_from_pptx(filepath):
    text = ""
    try:
        prs = Presentation(filepath)
        for slide in prs.slides:
            for shape in slide.shapes:
                if hasattr(shape, "text"):
                    text += shape.text + "\n"
    except Exception:
        pass
    return text

def extract_text_from_docx(filepath):
    text = ""
    try:
        doc = docx.Document(filepath)
        for para in doc.paragraphs:
            text += para.text + "\n"
    except Exception:
        pass
    return text