PDF Processing Guide

Overview

This guide covers PDF transformations — merging, splitting, creating, watermarking, encrypting, extracting images, and OCRing scanned PDFs. To fill, annotate, redact or rotate: office_edit where the chat has it, else the pdf-forms skill. Reading text/tables is handled by read_file directly; do NOT use this skill for plain reading. See REFERENCE.md for advanced library usage.

Dependencies are pre-installed in the sandbox image — do NOT pip install or npm install anything. Just import or invoke directly.

• Python libraries (always present): pypdf, pdfplumber, pypdfium2, reportlab, pytesseract, pdf2image, Pillow, pandas, numpy.
• JavaScript libraries (always present): pdf-lib, pdfjs-dist.
• CLI tools (present on current sandbox images): pdftotext / pdftoppm / pdfimages (poppler-utils), qpdf, pdftk, pandoc, tesseract. Older sandbox images may be missing some of these — if a CLI tool returns not found, fall back to the equivalent Python library on this page rather than trying to install it.

Reading PDF text — use read_file

For any request to read, summarise, analyse, extract data from, or aggregate the contents of a PDF, call read_file directly:

read_file(path="path/to/document.pdf")

The harness parses PDFs natively (mupdf; a scan without a text layer is OCR'd over its first pages only, and the output says so) and returns markdown. Pagination via offset/limit is free (cached). Do NOT write extraction scripts for plain reading — that bootstrap is exactly what read_file eliminates.

When read_file is not enough — it fails or times out, the text comes back garbled, or the question is about figures, charts, layout, embedded images, attachments or form-field values — load the pdf-reading skill. It covers the content inventory, page rasterization for vision_analyze, OCR and table extraction.

The Python libraries on this page are for transformations (the next sections), not for reading.

Python Libraries

pypdf - Basic Operations

Merge PDFs

from pypdf import PdfWriter, PdfReader

writer = PdfWriter()
for pdf_file in ["doc1.pdf", "doc2.pdf", "doc3.pdf"]:
    reader = PdfReader(pdf_file)
    for page in reader.pages:
        writer.add_page(page)

with open("merged.pdf", "wb") as output:
    writer.write(output)

Split PDF

reader = PdfReader("input.pdf")
for i, page in enumerate(reader.pages):
    writer = PdfWriter()
    writer.add_page(page)
    with open(f"page_{i+1}.pdf", "wb") as output:
        writer.write(output)

Extract Metadata

reader = PdfReader("document.pdf")
meta = reader.metadata
print(f"Title: {meta.title}")
print(f"Author: {meta.author}")
print(f"Subject: {meta.subject}")
print(f"Creator: {meta.creator}")

pdfplumber - Layout-aware text + tables (fallback only)

For plain text/markdown, read_file(path) is the right call (see top of this skill). Use pdfplumber only when you need precise layout preservation or per-table structured extraction that read_file does not give you — e.g., multi-line cells, merged headers, or extracting a specific table on a specific page.

Extract Text with Layout

import pdfplumber

with pdfplumber.open("document.pdf") as pdf:
    for page in pdf.pages:
        text = page.extract_text()
        print(text)

Extract Tables

with pdfplumber.open("document.pdf") as pdf:
    for i, page in enumerate(pdf.pages):
        tables = page.extract_tables()
        for j, table in enumerate(tables):
            print(f"Table {j+1} on page {i+1}:")
            for row in table:
                print(row)

Advanced Table Extraction

import pandas as pd

with pdfplumber.open("document.pdf") as pdf:
    all_tables = []
    for page in pdf.pages:
        tables = page.extract_tables()
        for table in tables:
            if table:  # Check if table is not empty
                df = pd.DataFrame(table[1:], columns=table[0])
                all_tables.append(df)

# Combine all tables
if all_tables:
    combined_df = pd.concat(all_tables, ignore_index=True)
    combined_df.to_excel("extracted_tables.xlsx", index=False)

reportlab - Create PDFs

Basic PDF Creation

from reportlab.lib.pagesizes import letter
from reportlab.pdfgen import canvas

c = canvas.Canvas("hello.pdf", pagesize=letter)
width, height = letter

# Add text
c.drawString(100, height - 100, "Hello World!")
c.drawString(100, height - 120, "This is a PDF created with reportlab")

# Add a line
c.line(100, height - 140, 400, height - 140)

# Save
c.save()

Create PDF with Multiple Pages

from reportlab.lib.pagesizes import letter
from reportlab.platypus import SimpleDocTemplate, Paragraph, Spacer, PageBreak
from reportlab.lib.styles import getSampleStyleSheet

doc = SimpleDocTemplate("report.pdf", pagesize=letter)
styles = getSampleStyleSheet()
story = []

# Add content
title = Paragraph("Report Title", styles['Title'])
story.append(title)
story.append(Spacer(1, 12))

body = Paragraph("This is the body of the report. " * 20, styles['Normal'])
story.append(body)
story.append(PageBreak())

# Page 2
story.append(Paragraph("Page 2", styles['Heading1']))
story.append(Paragraph("Content for page 2", styles['Normal']))

# Build PDF
doc.build(story)

Fonts: any text beyond plain English

reportlab's built-in fonts (Helvetica, Times-Roman, Courier) only cover Western European letters. Romanian ș ț ă, Polish, Czech, Turkish, Greek and Cyrillic come out as black boxes. Register a TrueType font before building — Liberation Sans is installed (and is metric-compatible with Arial):

from reportlab.lib.styles import getSampleStyleSheet
from reportlab.pdfbase import pdfmetrics
from reportlab.pdfbase.ttfonts import TTFont

FONTS = "/usr/share/fonts/truetype/liberation2/LiberationSans-"
for name, file in [("Sans", "Regular"), ("Sans-Bold", "Bold"),
                   ("Sans-Italic", "Italic"), ("Sans-BoldItalic", "BoldItalic")]:
    pdfmetrics.registerFont(TTFont(name, f"{FONTS}{file}.ttf"))
pdfmetrics.registerFontFamily("Sans", normal="Sans", bold="Sans-Bold",
                              italic="Sans-Italic", boldItalic="Sans-BoldItalic")

styles = getSampleStyleSheet()
for s in styles.byName.values():
    if hasattr(s, "fontName"):  # the list styles have none
        s.fontName = "Sans-Bold" if "Bold" in s.fontName else "Sans"

Then use "Sans" / "Sans-Bold" everywhere a font is named: table styles (("FONTNAME", (0, 0), (-1, -1), "Sans")) and canvas text (c.setFont("Sans", 12)). LiberationSerif-* is the Times equivalent. Use it for every document whose text isn't plain English: names and places carry these letters even in English text.

Subscripts and Superscripts

IMPORTANT: Never use Unicode subscript/superscript characters (₀₁₂₃₄₅₆₇₈₉, ⁰¹²³⁴⁵⁶⁷⁸⁹) in ReportLab PDFs. The built-in fonts do not include these glyphs, causing them to render as solid black boxes.

Instead, use ReportLab's XML markup tags in Paragraph objects:

from reportlab.platypus import Paragraph
from reportlab.lib.styles import getSampleStyleSheet

styles = getSampleStyleSheet()

# Subscripts: use <sub> tag
chemical = Paragraph("H<sub>2</sub>O", styles['Normal'])

# Superscripts: use <super> tag
squared = Paragraph("x<super>2</super> + y<super>2</super>", styles['Normal'])

For canvas-drawn text (not Paragraph objects), manually adjust font the size and position rather than using Unicode subscripts/superscripts.

Check what you created

pdftotext output.pdf - | grep -c "■"          # must print 0
pdftoppm -png -r 80 output.pdf page && ls -1 "$PWD"/page-*.png

Each ■ is a character the font couldn't draw: switch to the TrueType font above. Then ask vision_analyze about each page image once, for the look: layout, tables, the watermark. Render with pdftoppm, not pypdfium2.

Command-Line Tools

pdftotext (poppler-utils)

# Extract text
pdftotext input.pdf output.txt

# Extract text preserving layout
pdftotext -layout input.pdf output.txt

# Extract specific pages
pdftotext -f 1 -l 5 input.pdf output.txt  # Pages 1-5

qpdf

# Merge PDFs
qpdf --empty --pages file1.pdf file2.pdf -- merged.pdf

# Split pages
qpdf input.pdf --pages . 1-5 -- pages1-5.pdf
qpdf input.pdf --pages . 6-10 -- pages6-10.pdf

# Remove password
qpdf --password=mypassword --decrypt encrypted.pdf decrypted.pdf

pdftk (if available)

# Merge
pdftk file1.pdf file2.pdf cat output merged.pdf

# Split
pdftk input.pdf burst

Common Tasks

Extract Text from Scanned PDFs

# pytesseract + pdf2image are pre-installed; tesseract-ocr is in the
# sandbox image's apt layer.  Just import and run.
import pytesseract
from pdf2image import convert_from_path

# Convert PDF to images
images = convert_from_path('scanned.pdf')

# OCR each page
text = ""
for i, image in enumerate(images):
    text += f"Page {i+1}:\n"
    text += pytesseract.image_to_string(image)
    text += "\n\n"

print(text)

Add Watermark

Use the script. It stamps diagonal text on top of every page, tables and images included, in translucent grey, so the content stays readable under it:

python scripts/watermark.py document.pdf watermarked.pdf "CONFIDENTIAL"                 # opacity 0.15
python scripts/watermark.py document.pdf watermarked.pdf "DRAFT" --opacity 0.25

For a PDF you create, build it first without a watermark, then stamp it. Don't draw the watermark in reportlab's onFirstPage/onLaterPages: those run before the page content, so tables and filled shapes cover it.

Extract Images

# Using pdfimages (poppler-utils)
pdfimages -j input.pdf output_prefix

# This extracts all images as output_prefix-000.jpg, output_prefix-001.jpg, etc.

Password Protection

from pypdf import PdfReader, PdfWriter

reader = PdfReader("input.pdf")
writer = PdfWriter()

for page in reader.pages:
    writer.add_page(page)

# Add password
writer.encrypt("userpassword", "ownerpassword")

with open("encrypted.pdf", "wb") as output:
    writer.write(output)

Quick Reference

Task	Best Tool	Command/Code
Read text / tables / aggregate data	read_file	read_file(path="document.pdf")
Merge PDFs	pypdf	writer.add_page(page)
Split PDFs	pypdf	One page per file
Layout-precise text / structured tables (fallback)	pdfplumber	page.extract_text() / page.extract_tables()
Create PDFs	reportlab	Canvas or Platypus; a TrueType font for non-English text
Watermark	scripts/watermark.py	python scripts/watermark.py in.pdf out.pdf "CONFIDENTIAL"
Command line merge	qpdf	qpdf --empty --pages ...
OCR scanned PDFs	pdf-reading skill or pytesseract	Convert to image first
Fill, annotate, redact or rotate	office_edit where the chat has it, else the pdf-forms skill	—

Next Steps

• For advanced pypdfium2 usage, see REFERENCE.md
• For JavaScript libraries (pdf-lib), see REFERENCE.md
• To fill, annotate, redact or rotate: office_edit where the chat has it, else the pdf-forms skill
• For troubleshooting guides, see REFERENCE.md