pdf-squish

A pure Rust TUI tool that compresses PDF files while preserving quality.

Recompresses streams with maximum deflate, removes unused objects and metadata, and deduplicates identical streams — all without touching image codecs.

Features

• Stream recompression — Decompress and re-encode all FlateDecode/LZW streams at zlib level 9
• Safe — Never touches JPEG, JBIG2, or JPX image data
• Unused object cleanup — Removes unreferenced objects from the PDF tree
• Metadata stripping — Removes XMP, Info dictionary, and per-page metadata
• Stream deduplication — SHA-256 hashing to merge identical streams
• Interactive TUI — File browser, quality selector, real-time progress

Installation

git clone https://github.com/Dann1y/pdf_squish.git
cd pdf_squish
cargo build --release

The binary will be at target/release/pdf-squish.

Usage

# Launch TUI
cargo run --release

# Pre-fill input path
cargo run --release -- /path/to/file.pdf

# Glob pattern
cargo run --release -- "/path/to/docs/*.pdf"

TUI Controls

Key	Action
Tab / Shift+Tab	Cycle focus: Input → Quality → File List
Enter	Load files / Start compression
1 2 3	Select Light / Standard / Aggressive
j k / ↑ ↓	Navigate file list
d	Remove selected file
r	Reset after completion
q	Quit

Compression Levels

Strategy	Light	Standard	Aggressive
Stream recompression (zlib best)	✓	✓	✓
Empty stream removal	✓	✓	✓
Unused object cleanup		✓	✓
Metadata stripping		✓	✓
Object renumbering		✓	✓
Duplicate stream dedup			✓

How It Works

PDF files store content as streams — compressed binary data for text, fonts, vector graphics, etc. Many PDF generators use suboptimal compression settings.

pdf-squish:

1. Parses the PDF object tree via lopdf
2. Identifies recompressible streams (FlateDecode, LZW, uncompressed)
3. Decompresses each stream, then re-encodes with flate2 at maximum compression (zlib level 9)
4. Optionally removes dead objects, metadata, and duplicate streams
5. Writes the optimized PDF as {filename}_compressed.pdf

The original file is never modified.

Dependencies

• ratatui + crossterm — TUI
• lopdf — PDF manipulation
• flate2 — Compression
• tokio — Async runtime
• sha2 — Deduplication hashing

License

MIT