PDFMiner

Common errors

• AttributeError: '_io.BytesIO' object has no attribute 'catalog'

  cause This error typically occurs when a file-like object (like `io.BytesIO`) is passed directly to `PDFPage.create_pages()` or similar functions, but the API expects a `PDFDocument` object that has already been parsed by a `PDFParser`. This indicates incorrect API usage.
  fix Ensure you correctly parse the file first using `PDFParser` to create a `PDFDocument` instance, and then pass the `PDFDocument` object to `PDFPage.create_pages()`. Refer to the quickstart example for correct API flow.

• ModuleNotFoundError: No module named 'pdfminer.six'

  cause This usually means you have installed the original `pdfminer` package but are attempting to import modules or use `high_level` functions specific to the `pdfminer.six` fork. Or, `pdfminer.six` was not installed at all.
  fix If you intend to use `pdfminer.six` (recommended), ensure you install it with `pip install pdfminer.six`. If you're sticking to the original `pdfminer`, use its specific import paths and API patterns. The original `pdfminer` does not expose a `pdfminer.high_level` module.

• UnicodeEncodeError: 'charmap' codec can't encode character...

  cause Encoding issues are common when handling diverse text content in PDFs, especially across different operating systems or locales, or when writing to files without specifying the correct encoding.
  fix Always specify `encoding='utf-8'` when creating output files or `StringIO` objects if you expect Unicode characters. For `TextConverter`, ensure the `outfp` (output file pointer) is opened with `encoding='utf-8'` or handle character sets explicitly.

Warnings

• breaking The original `pdfminer` project (euske/pdfminer) is no longer actively maintained since 2020. While the latest version `20191125` supports Python 3, new features, bug fixes, and community support are primarily found in its actively maintained fork, `pdfminer.six`.
  Fix: Consider migrating to `pdfminer.six` for an actively developed and supported version (`pip install pdfminer.six`).
• gotcha PDFMiner struggles with text extraction from PDFs with complex layouts (e.g., multi-column, nested tables) and cannot extract text from scanned PDFs (images) without external Optical Character Recognition (OCR) tools.
  Fix: For complex layouts, extensive customization of `LAParams` or post-processing may be required. For scanned PDFs, integrate with OCR libraries like Tesseract or pre-process with tools that extract images for OCR.
• gotcha Output may contain raw character IDs like `(cid:x)` instead of readable text, especially for non-standard fonts or encoding issues. This happens when the font is not properly mapped to Unicode.
  Fix: Verify if text can be copy-pasted correctly from a PDF viewer. If it can, try adjusting `LAParams` or exploring `pdfminer.six` which might have better font/encoding handling. Otherwise, it might be an inherent limitation of the PDF itself.

Install

• pip install pdfminer Install stable version

Imports

• PDFResourceManager

  from pdfminer.pdfinterp import PDFResourceManager

• PDFPageInterpreter

  from pdfminer.pdfinterp import PDFPageInterpreter

• PDFPage

  from pdfminer.pdfpage import PDFPage

• PDFParser

  from pdfminer.pdfparser import PDFParser

• PDFDocument

  from pdfminer.pdfdocument import PDFDocument

• TextConverter

  from pdfminer.converter import TextConverter

• LAParams

  from pdfminer.layout import LAParams

Quickstart

This quickstart demonstrates how to extract text from a PDF file using PDFMiner's core components. It initializes a resource manager, a text converter, and a page interpreter to process the PDF document page by page. A dummy `dummy.pdf` file is created if not found, allowing the code to be runnable for demonstration purposes. This reflects the more verbose API usage typical of the original PDFMiner, as opposed to the simplified `high_level` API found in `pdfminer.six`.

import os
from io import StringIO

from pdfminer.converter import TextConverter
from pdfminer.layout import LAParams
from pdfminer.pdfinterp import PDFResourceManager, PDFPageInterpreter
from pdfminer.pdfpage import PDFPage
from pdfminer.pdfparser import PDFParser

def extract_text_from_pdf(pdf_path):
    # Ensure a dummy PDF exists for demonstration, or replace with a real path
    if not os.path.exists(pdf_path):
        print(f"Error: PDF file not found at {pdf_path}. Creating a dummy PDF for demonstration.")
        # In a real scenario, you'd handle the missing file appropriately.
        # For a runnable example, we'll create a simple dummy file.
        try:
            from reportlab.pdfgen import canvas
            c = canvas.Canvas(pdf_path)
            c.drawString(100, 750, "Hello, PDFMiner!")
            c.drawString(100, 730, "This is a dummy PDF for testing.")
            c.save()
            print(f"Dummy PDF created at {pdf_path}")
        except ImportError:
            print("Please install reportlab (`pip install reportlab`) to create dummy PDF, or provide a real PDF.")
            return ""

    rsrcmgr = PDFResourceManager()
    retstr = StringIO()
    laparams = LAParams()
    device = TextConverter(rsrcmgr, retstr, laparams=laparams)
    
    with open(pdf_path, 'rb') as fp:
        parser = PDFParser(fp)
        document = PDFDocument(parser)
        interpreter = PDFPageInterpreter(rsrcmgr, device)
        for page in PDFPage.create_pages(document):
            interpreter.process_page(page)
        text = retstr.getvalue()
    
    device.close()
    retstr.close()
    return text

if __name__ == '__main__':
    pdf_file = 'dummy.pdf'
    extracted_content = extract_text_from_pdf(pdf_file)
    print("\n--- Extracted Text ---")
    print(extracted_content)