PDF — Portable Document Format

---
name: pdf
description: "Understand PDF structure, standards (ISO 32000-2), and practical tasks like viewing, parsing, and converting PDFs across platforms."
metadata:
  tags: "file-formats, pdf, document-management, parsing, conversion, digital-signatures, accessibility"
  source: "https://skilldb.dev/skills/file-formats-skills/pdf"
  pack: "file-formats-skills"
  category: "Technology & Engineering"
---

# PDF — Portable Document Format

## When to use this skill
Use when the user says things like:
- "Explain PDF structure and key specs."
- "Explain how to parse and extract content from PDFs."
- "Show how to convert PDFs to other formats."
- "Explain PDF accessibility and tagging (PDF/UA)."


You are a file format specialist with deep knowledge of the PDF specification, its object-based structure, rendering model, and the practical realities of creating, parsing, manipulating, and converting PDFs across platforms and programming languages.

## Overview

PDF is a file format developed by Adobe Systems for presenting documents in a manner independent of application software, hardware, and operating systems. Each PDF file encapsulates a complete description of a fixed-layout flat document, including text, fonts, vector graphics, raster images, and other information needed for display. PDF has become the de facto standard for electronic document distribution.

## Technical Specifications

- **File extension:** `.pdf`
- **MIME type:** `application/pdf`
- **Current version:** PDF 2.0 (ISO 32000-2:2020)
- **Magic bytes:** `%PDF-` at file start
- **Max file size:** No theoretical limit; practical limits depend on reader software
- **Character encoding:** Supports Unicode via CIDFont and ToUnicode mappings
- **Color spaces:** RGB, CMYK, Grayscale, ICC profiles, spot colors
- **Compression:** Supports FlateDecode (zlib), DCTDecode (JPEG), JBIG2, CCITT, and others
- **Structure:** Header, body (objects), cross-reference table, trailer

A PDF file is built from a hierarchy of objects: integers, strings, arrays, dictionaries, streams, and indirect references. Page content is described using a PostScript-derived page description language with operators for drawing paths, placing text, and embedding images. Fonts can be embedded fully or as subsets.

### PDF/A, PDF/X, PDF/E, PDF/UA

- **PDF/A** (ISO 19005): Long-term archival; requires embedded fonts, no encryption
- **PDF/X** (ISO 15930): Print production exchange
- **PDF/E** (ISO 24517): Engineering documents
- **PDF/UA** (ISO 14289): Universal accessibility

## How to Work With It

### Opening
Any modern web browser (Chrome, Firefox, Edge) renders PDFs natively. Dedicated readers include Adobe Acrobat Reader, Foxit Reader, SumatraPDF (Windows), Preview (macOS), Okular and Evince (Linux).

### Creating
- **From applications:** Print to PDF is built into Windows 10+, macOS, and most Linux desktops
- **Programmatically:** Libraries like `reportlab` (Python), `iText` (Java/.NET), `PDFKit` (Node.js), `FPDF`/`TCPDF` (PHP), `Prawn` (Ruby)
- **From LaTeX:** `pdflatex`, `xelatex`, or `lualatex` compile `.tex` directly to PDF

### Parsing and Extracting
- **Text extraction:** `pdftotext` (poppler-utils), `PyPDF2`/`pypdf`, `pdfminer.six`, Apache Tika
- **Structured data:** `tabula-py` or Camelot for tables
- **Metadata:** `exiftool`, `pdfinfo`, or any PDF library
- **Images:** `pdfimages` (poppler-utils)

### Converting
- **To images:** `pdftoppm`, Ghostscript, ImageMagick
- **To Word:** LibreOffice, Adobe Acrobat Pro, online converters
- **To HTML:** `pdf2htmlEX`, `pdftohtml`
- **From HTML:** `wkhtmltopdf`, Puppeteer/Playwright `page.pdf()`, WeasyPrint

### Editing
Direct editing is possible with Adobe Acrobat Pro, LibreOffice Draw, or `qpdf` / `pdftk` for structural manipulation (merging, splitting, rotating, encrypting).

## Common Use Cases

- Official documents, contracts, and legal filings
- Scientific papers and journal articles
- Print-ready artwork and prepress workflows (PDF/X)
- Government forms (with interactive form fields)
- eBooks and digital manuals
- Invoices and financial statements
- Long-term archival of records (PDF/A)

## Pros & Cons

### Pros
- Universally supported across all platforms and devices
- Preserves exact layout regardless of viewer
- Supports encryption, digital signatures, and permissions
- Can embed fonts ensuring consistent rendering
- ISO-standardized with multiple profiles for specialized needs
- Compact with built-in compression

### Cons
- Not designed for easy editing; reflowing content is difficult
- Text extraction can be unreliable, especially from scanned documents
- Large documents with many images can become very large
- Accessibility depends on proper tagging (often missing)
- Creating accessible, tagged PDFs requires deliberate effort
- Complex internal structure makes programmatic generation non-trivial

## Compatibility

| Platform | Native Support |
|----------|---------------|
| Windows | Built-in viewer (Edge), Print to PDF |
| macOS | Preview, built-in QuickLook |
| Linux | Evince, Okular, poppler-based tools |
| iOS/Android | Built-in viewers in both platforms |
| Web | All major browsers render inline |

## Practical Usage

### Generating PDFs from HTML (the Modern Approach)
For most developers, HTML-to-PDF is the most practical generation method:

```javascript
// Puppeteer (Node.js) — pixel-perfect rendering via headless Chrome
const browser = await puppeteer.launch();
const page = await browser.newPage();
await page.goto('https://example.com/invoice/123');
await page.pdf({ path: 'invoice.pdf', format: 'A4', printBackground: true });

// WeasyPrint (Python) — CSS-based, no browser needed
// weasyprint https://example.com/report.html report.pdf
```

```bash
# wkhtmltopdf — lightweight CLI
wkhtmltopdf --enable-local-file-access report.html report.pdf
```

### Merging, Splitting, and Manipulating PDFs
```bash
# Merge multiple PDFs
qpdf --empty --pages file1.pdf file2.pdf file3.pdf -- merged.pdf

# Extract pages 5-10
qpdf input.pdf --pages input.pdf 5-10 -- extract.pdf

# Remove password protection (if you know the password)
qpdf --decrypt --password=secret protected.pdf decrypted.pdf

# Compress / linearize for web delivery
qpdf --linearize input.pdf web-optimized.pdf
```

### Text Extraction Strategies
Different tools suit different PDF types:

```bash
# Simple text PDFs — pdftotext is fast and reliable
pdftotext -layout input.pdf output.txt

# Table extraction — tabula for structured data
java -jar tabula.jar -o output.csv -p all input.pdf

# Scanned PDFs — OCR with Tesseract
ocrmypdf input-scanned.pdf output-searchable.pdf
```

### Agent and Automation Workflows
When processing PDFs programmatically, choose your approach based on the content type: `pypdf` for simple text extraction and manipulation, `pdfminer.six` for layout-aware extraction, `tabula-py` or `camelot` for tables, and `ocrmypdf` for scanned documents. For generating reports, prefer HTML-to-PDF over low-level PDF construction.

## Anti-Patterns

**Treating PDF as a data format.** PDF is a presentation format, not a data interchange format. Extracting structured data from PDFs is inherently fragile. If you control the data source, export as CSV, JSON, or a database — not PDF.

**Generating PDFs with low-level libraries when HTML would suffice.** Writing PDF construction code with reportlab or iText is time-consuming and brittle. Unless you need precise control over the PDF object structure, use an HTML-to-PDF renderer and style with CSS.

**Assuming text extraction will preserve reading order.** PDF stores text as positioned glyphs, not logical paragraphs. Columns, headers, footers, and sidebars can interleave unpredictably during extraction. Always validate extracted text against the visual layout.

**Creating PDFs without embedded fonts.** If fonts are not embedded, the viewer substitutes system fonts, which shifts layout and breaks formatting. Always embed fonts (or font subsets) when generating PDFs.

**Ignoring accessibility.** Untagged PDFs are inaccessible to screen readers. If your PDFs will be consumed by the public, use tagged PDF structure (PDF/UA) with proper heading hierarchy, alt text, and reading order.

## Related Formats

- **PostScript (.ps):** PDF's predecessor; page description language
- **XPS (.xps):** Microsoft's alternative fixed-layout format
- **DjVu (.djvu):** Optimized for scanned documents
- **PDF/A:** Archival subset of PDF
- **EPUB:** Reflowable alternative for ebooks