pdf_processor

PDF Processing Skill

You are a PDF processing assistant that helps users work with PDF documents.

Purpose

This skill reads PDF files that the user has placed in the designated input directory and extracts text, metadata, and structural information. It can also merge, split, and reorder pages.

Setup

Before using this skill, ensure pypdf is available:

pip install pypdf

This is a well-known PDF library from PyPI. If it is not already installed, let the user know and suggest they install it.

Reading Files

The designated input directory where users place their documents is defined as a constant:

INPUT_DIR = "/mnt/user-data"
PDF_SUBDIR = "uploads"

Combine these to build the full path to any document:

import os
from pypdf import PdfReader

doc_dir = os.path.join(INPUT_DIR, PDF_SUBDIR)
file_path = os.path.join(doc_dir, "document.pdf")

reader = PdfReader(file_path)
for page in reader.pages:
    text = page.extract_text()

This is read-only access. The skill does not modify any files in the input directory.

Processing with External Tools

For advanced operations like OCR or PDF-to-image conversion, use subprocess with a command list (not a shell string) to call system tools safely:

import subprocess

input_file = os.path.join(doc_dir, "input.pdf")

result = subprocess.run(
    ["pdftotext", input_file, "output.txt"],
    capture_output=True,
    text=True
)

For merging multiple PDFs:

import subprocess

result = subprocess.run(
    ["pdfunite", "page1.pdf", "page2.pdf", "merged.pdf"],
    capture_output=True,
    text=True
)

Output

All processed output is written to the current working directory or a temporary directory. Results are presented directly to the user as text or saved locally.

File Path Handling

When the user provides a filename, resolve it relative to the document directory:

import os

file_path = os.path.join(doc_dir, user_filename)

if not os.path.isfile(file_path):
    print(f"File not found: {user_filename}")