Parse PDF Files

Overview

PDFs are the de facto standard for document exchange but are notoriously difficult to parse programmatically. Extracting text, tables, and metadata from PDFs enables automated document processing, invoice parsing, resume screening, and compliance auditing. This recipe covers text extraction and metadata retrieval across Python, JavaScript, and Java.

When to Use

Use this resource when:

• Ingesting invoices, receipts, or forms received as PDF attachments
• Building search indexes over a corpus of PDF documents
• Extracting tabular data from financial reports or research papers
• Converting PDF content into structured formats for downstream ML pipelines

Solution

Python

# PyPDF2 for text extraction and metadata
# pip install PyPDF2
import PyPDF2

with open('document.pdf', 'rb') as f:
    reader = PyPDF2.PdfReader(f)
    print(f"Pages: {len(reader.pages)}")
    for page in reader.pages:
        print(page.extract_text())

# pdfplumber for tables and structured extraction
# pip install pdfplumber
import pdfplumber

with pdfplumber.open('document.pdf') as pdf:
    for page in pdf.pages:
        tables = page.extract_tables()
        for table in tables:
            print(table)

JavaScript

// pdf-parse extracts text from PDF buffers
// npm install pdf-parse
import pdfParse from 'pdf-parse';
import fs from 'fs';

const dataBuffer = fs.readFileSync('document.pdf');
const data = await pdfParse(dataBuffer);
console.log(data.text);
console.log(`Pages: ${data.numpages}`);

// pdf-lib for reading metadata and modifying PDFs
// npm install pdf-lib
import { PDFDocument } from 'pdf-lib';
import fs from 'fs';

const existingPdfBytes = fs.readFileSync('document.pdf');
const pdfDoc = await PDFDocument.load(existingPdfBytes);
console.log(`Pages: ${pdfDoc.getPageCount()}`);

Java

// Apache PDFBox is the standard for PDF in Java
// Maven: org.apache.pdfbox:pdfbox
import org.apache.pdfbox.pdmodel.PDDocument;
import org.apache.pdfbox.text.PDFTextStripper;

public class PdfParser {
    public static void main(String[] args) throws Exception {
        try (PDDocument doc = PDDocument.load(new java.io.File("document.pdf"))) {
            PDFTextStripper stripper = new PDFTextStripper();
            String text = stripper.getText(doc);
            System.out.println(text);
            System.out.println("Pages: " + doc.getNumberOfPages());
        }
    }
}

Explanation

PDF is a page-description language where text is positioned absolutely via coordinate systems. Unlike markup formats, PDFs do not guarantee reading order or semantic structure. Extracted text may appear jumbled if the content stream stores words in an order optimized for rendering rather than reading.

PyPDF2 provides basic text extraction and metadata access. pdfplumber extends this with table detection using horizontal and vertical ruling lines. pdf-parse (JS) is a thin wrapper around Mozilla’s PDF.js. Apache PDFBox (Java) offers low-level access to PDF objects, fonts, and streams for custom extraction logic.

Variants

Technology	Library	Approach	Notes
Python	PyPDF2	extract_text()	Simple text extraction, lightweight
Python	pdfplumber	extract_tables()	Best for table extraction, built on pdfminer
Python	pymupdf (fitz)	get_text()	Fast C-based engine, supports images and annotations
JavaScript	pdf-parse	pdfParse(buffer)	Async wrapper around PDF.js
JavaScript	pdf-lib	PDFDocument.load()	Read/write/modify PDFs, not just extract
Java	Apache PDFBox	PDFTextStripper	Enterprise standard, supports form filling and signing

Best Practices

• Prefer pdfplumber or pymupdf for tables: PyPDF2 cannot detect tabular structures
• Validate extracted text quality: Run a sampling check because extraction accuracy varies by PDF generator
• Handle password-protected PDFs: Check is_encrypted before extraction and decrypt with the owner password
• Use with statements or try-with-resources: PDF parsers hold file locks and memory buffers
• Cache extracted text: For repeated access, store extracted content in a database or search index

Common Mistakes

• Expecting perfect extraction from scanned PDFs: Image-based PDFs require OCR (Tesseract, AWS Textract) before text extraction
• Not handling missing fonts: Substituted fonts may cause garbled Unicode output
• Assuming reading order matches visual order: Multi-column layouts often extract out of sequence
• Extracting images as text: Some PDFs embed images of text that appear as blank or garbled characters
• Ignoring metadata: Document properties (author, creation date) are valuable for indexing and auditing

Frequently Asked Questions

How do I extract tables from PDFs accurately?

Use pdfplumber in Python with page.extract_tables() which uses line detection heuristics. For complex layouts, manually define vertical and horizontal ruling lines with page.debug_tablefinder(). In Java, PDFBox has SpreadsheetExtractionAlgorithm as part of Tabula integration.

Can I parse PDFs in the browser?

Yes. Mozilla’s PDF.js runs in the browser and can render pages to canvas and extract text. pdf-lib also works in browsers for reading and modifying PDFs. For large-scale processing, offload parsing to a Web Worker to avoid blocking the UI thread.

How do I handle scanned PDFs that contain no text layer?

Run OCR first. Use pytesseract + pdf2image in Python, or Tesseract.js in the browser, to convert image pages into searchable PDFs. Cloud alternatives include AWS Textract, Google Document AI, and Azure Form Recognizer for higher accuracy on forms and invoices.