rexgen

A fast CLI to generate text that matches regex-like patterns. Inspired by the original TDG tool, rexgen focuses on a cleaner pattern syntax, intuitive CLI flags, and a high-performance Rust core.

What it does

• Generates random strings that match regex-like expressions (for example \d{5}, [A-Z]{2,4}, groups, ranges, repetition).
• Processes templates with placeholders like {{ ... }}, replacing each with generated content.
• Supports deterministic output via --seed and bulk generation via --count.
• Allows reusable named patterns loaded from external library files.

Why a new version

• Keep the expressive, regex-like feel of TDG while making templates simpler and the CLI more intuitive.
• Rust for speed, portability, and a single static binary.

CLI

• Modes (exactly one required)
  • pattern <expr> Generate strings from a single pattern expression (no free text). Default behavior: pattern is assumed if the mode is omitted.
  • template <file> Processes the provided template file containing {{ ... }} placeholders.
  • list List available pattern names (with namespace) from loaded libraries; marks entries that contain placeholders.
• Output
  • --count, -c <n> Number of items to produce (default 1). Each item is one full pattern expansion or a full template render.
  • --output, -o <path> Write results to a file instead of stdout.
  • --separator <sep> Separator to join multiple outputs when --count > 1 (default: newline). Note: this controls output joining and is unrelated to [lib].delims in .rxlib files.
• Randomness
  • --seed, -s <n> Seed for deterministic generation (same inputs => same outputs).
• Libraries (named patterns)
  • --lib, -l <file> Load a pattern library file. Repeatable.
  • --lib-dir, -I <dir> Add a directory to the library search path. Repeatable.
• Behavior & diagnostics
  • --strict Treat unknown named-patterns as fatal (default: unknown treated as literal where safe).
  • --verbose, -v Increase verbosity; -vv for more. --quiet, -q to suppress non-errors.
  • --help, -h Show help. --version, -V Show version.
  • --jobs <n> Set the parallel worker count while keeping deterministic ordering with --seed.

CLI examples

• rexgen "My SSN is '{{@us:ssn}}'." --lib us-data.rxlib --count 100 --seed 42 Generate 100 SSN-like values from a named entry, deterministically.
• rexgen '[1-9]\d\d-\d\d-\d\d\d\d' --count 5 Generate 5 SSN-like values from a raw expression.
• rexgen template template.txt --lib names.rxlib --count 10 --output out.txt Render template.txt ten times using names from names.rxlib.
• rexgen pattern '[A-Z]{3}' --count 5 --separator ', ' Inline pattern rendered five times, joined by a comma+space.
• rexgen list --lib us-data.rxlib List all named patterns in the provided lib.
• rexgen list --lib-dir ./patterns List all named patterns in the provided directory.

Design

• Pattern engine: familiar regex-like tokens with generation semantics, not full PCRE.
• Templating: minimal—plain text plus {{ pattern }} placeholders.
• Libraries: rxlib (TOML) files with [lib] metadata and a single [patterns] table. --lib loads files; --lib-dir adds search paths; --list enumerates loaded names.
• Output: stdout by default; use --output to write to a file.

Language Reference

• Full pattern and template syntax, escaping rules, named references, and library metadata are documented in docs/rexgen-language.md.

Template and escaping rules

• See Language Reference for placeholder delimiters and escaping details.

Determinism and parallelism

• --seed guarantees identical outputs for identical inputs.
• Parallel generation (--jobs) remains deterministic: worker RNGs derive from the master seed so ordering and content are stable across runs and platforms.

Output joining

• When --count > 1, outputs are joined with --separator (default newline) and no trailing separator is written.

Namespaces

• namespace and pattern keys are case-sensitive. Use {{@ns:name}} inside placeholders to reference across namespaces.

Imports and search order

• Import resolution: relative to the importing file first, then each --lib-dir in order, then the current working directory.
• Default search dirs: current working directory and ./rexgen-patterns.
• Import cycles are detected and reported with the chain of files forming the cycle.

Validation on load

• Validate duplicate names per namespace, invalid names, missing imports, and malformed files with clear errors (include file and line when possible).

rxlib format (TOML)

• See Language Reference for library file structure ([lib], [patterns]), placeholder delimiters ([lib].delims), and inline placeholders in values.