Skip to content

python.md

Latest commit

 

History

History
39 lines (28 loc) · 1.73 KB

python.md

File metadata and controls

39 lines (28 loc) · 1.73 KB
Raw

Python

from img2irc import Simple

output = Simple(width=80, mode="ansi24", smooth=True).generate("photo.png")
output.write("art.ans")
output.save_preview("preview.png")

inputs can be paths, HTTP(S) URLs, bytes, or binary file objects. Cascadia Code is embedded; font= accepts another font file or bytes.

Advanced exposes all native render settings, an ordered pipeline, and overlays:

from img2irc import Advanced

output = (
    Advanced(width=100, render="ansi24", score_fix=True, include_cells=True)
    .add_effect("contrast", 18)
    .add_replace_colour((255, 0, 0), (20, 20, 20), tolerance=8)
    .add_effect("sharpen")
    .add_overlay("hello", 2, 1, foreground=(255, 255, 255), bold=True)
    .generate("photo.png")
)

available_options() lists settings; resolved_options, validate(), and to_json() inspect, check, and serialize them. output contains encoded text, RGBA/PNG previews, scores/timings, OCR text, and optional styled cells. process(), detect_text(), nearest_colour(), and score_contours() expose individual subsystems. see the gradient example.

wheels

./scripts/build-wheel.sh --python python3.12
python3.12 -m pip install dist/<generated-wheel>.whl

the script requires Docker and x86_64 Linux, and produces a manylinux 2.28 wheel. OCR, ONNX Runtime, and OpenSSL are included; the first build compiles native dependencies and later builds reuse caches under target/. --no-ocr omits OCR, --out DIR changes the wheel directory, and IMG2IRC_BUILD_JOBS limits parallelism. install with the Python version selected for the build. Python 3.10 or later is required.

a small testing front end is also available as img2irc-tui photo.png.