Skip to content

CLI.md

Latest commit

 

History

History
150 lines (114 loc) · 7.33 KB

CLI.md

File metadata and controls

150 lines (114 loc) · 7.33 KB
Raw

cum Command Line Interface 🧹

The standalone cum binary exposes all watermark removal and inspection capabilities from the cum-rs library directly in your terminal.

📦 Installation

cargo install cum-rs --features rust-binary

The resulting binary is named cum.

🛠 Subcommands

cum clean: Remove watermarks

cum clean [OPTIONS] [FILE]

Cleans a file or inline text. Output goes to stdout by default; redirect with --output to write a file.

Argument Default Description
[FILE] - Path to file (PNG, JPEG, WebP, SVG, PDF, DOCX, ODT, HTML, Markdown, plain text).
-t, --text <TEXT> - Inline text to clean (alternative to a file path).
--stdin - Read bytes from stdin.
-o, --output <OUT> stdout Write cleaned output to this path.
-a, --aggressive true Also replace Cyrillic / fullwidth-Latin confusable letters.
-m, --media <MEDIA> auto Force a media type: text, png, jpeg, webp, svg, pdf, docx, odt, html, markdown.
-j, --json off Print stats as JSON to stderr.
-q, --quiet off Suppress progress output.

Examples

# Clean a text file: output to stdout
cum clean report.md

# Clean a PNG in-place
cum clean logo.png --output logo_clean.png

# Clean inline text
cum clean --text "Hello​ world"   # contains ZWSP

# Pipe from another command
cat suspicious.txt | cum clean --stdin

# JSON stats while writing cleaned file
cum clean --json -o clean.txt dirty.txt

cum inspect - Detect watermarks

cum inspect [OPTIONS] [FILE]

Inspects a file or text for watermark carriers without modifying it. Prints a per-codepoint table to stdout.

Argument Default Description
[FILE] - File to inspect.
-t, --text <TEXT> - Inline text.
--stdin - Read from stdin.
-a, --aggressive true Include Cyrillic / fullwidth confusable matches.
-m, --media <MEDIA> auto Force media type.
-j, --json off Emit findings as JSON.

Examples

# Inspect a text file
cum inspect article.txt

# Machine-readable JSON output
cum inspect --json image.png

# Inspect inline text
cum inspect --text "C​laude"   # ZWSP between C and l

# Pipe a downloaded webpage
curl -s https://example.com | cum inspect --stdin --media html

cum enhance - Stochastic synonym replacement

cum enhance [OPTIONS] [FILE]

Applies a best-effort stochastic synonym replacement to plain-text input to defeat Layer B statistical watermarks (SynthID-Text, KGW). By default, each non-stop word is replaced with a 50% probability using a curated synonym table and the system dictionary.

Argument Default Description
[FILE] - Path to the plain-text file to enhance.
-t, --text <TEXT> - Inline text to enhance.
--stdin - Read plain-text input from stdin.
-o, --output <OUT> stdout Write the enhanced output to this path.
-p, --probability 0.5 Per-word substitution probability [0.0, 1.0].

Examples

# Enhance inline text with a 80% substitution chance
cum enhance --text "The chaos governs the universe" -p 0.8

# Enhance a text file and save the result
cum enhance report.txt --output enhanced_report.txt

# Pipe via stdin
cat essay.md | cum enhance --stdin -p 0.6

Supported Formats

Extension Detection Layer
.txt, .md extension Unicode Layer A
.html, .htm extension / magic Unicode + HTML meta
.png magic bytes PNG chunk strip
.jpg, .jpeg magic bytes JPEG APP1/APP11/APP13
.webp magic bytes RIFF EXIF/XMP
.svg magic bytes XML metadata
.pdf magic bytes XMP/Info byte-scan
.docx magic bytes (ZIP PK) docProps strip
.odt magic bytes (ZIP PK) meta.xml strip

What Gets Removed

Watermark class Examples Action
Invisible controls ZWSP (U+200B), BOM (U+FEFF), WJ (U+2060) Strip
Bidirectional format controls LRE, RLE, PDF, LRI, RLI, FSI, PDI Strip
Tag characters U+E0001-U+E007F Strip
Variation selectors FE00-FE0F, E0100-E01EF (non-emoji) Strip
Private-use E000-F8FF, F0000-10FFFD Strip
Space homoglyphs En-Quad, Hair Space, Narrow NBSP, ... Replace → U+0020
Cyrillic / fullwidth confusables Cyr А→A, FF21→A, ... Replace (aggressive)
PNG metadata chunks iTXt, tEXt, zTXt, eXIf, C2PA (caBX/JUMB) Strip chunk
JPEG metadata segments APP1 (EXIF/XMP), APP11 (JUMBF), APP13 (IPTC) Strip segment
WebP RIFF chunks EXIF, XMP, C2PA Strip chunk
SVG embedded metadata <metadata>, <x:xmpmeta>, data-ai-* Strip
PDF metadata XMP xpacket, /Author, /Creator, /Producer Byte-scan strip
DOCX/ODT metadata docProps/core.xml, docProps/app.xml, customXml/ ZIP repack
HTML metadata <meta name="generator">, data-ai-*, JSON-LD C2PA Strip
Markdown front-matter ai-generated, model, generator, watermark-id Strip keys