Files
autotagger-win/autotag
T
evazion d43ce9c4e7 autotag: add option for taking list of files from a text file.
Add an option for taking the list of input files from a text file:

  ./autotag -i files.txt

This is useful in conjunction with the `find` command:

  find ~/images/ -name '*.jpg' | ./autotag -i -
2022-06-22 17:16:22 -05:00

65 lines
3.0 KiB
Python
Executable File

#!/usr/bin/env python
import json
import click
import itertools
from autotagger import Autotagger
from pathlib import Path
from more_itertools import ichunked
@click.command(help="Automatically generate tags for a list of images.", context_settings=dict(max_content_width=140))
@click.option("-t", "--threshold", default=0.01, type=float, show_default=True, help="The minimum tag confidence level.")
@click.option("-n", "--limit", default=50, type=int, show_default=True, help="The maximum number of tags to return per image.")
@click.option("-b", "--batch", "bs", default=128, type=int, show_default=True, help="The number of images to process per batch.")
@click.option("-c", "--csv", is_flag=True, help="Output CSV instead of JSON.")
@click.option("-i", "--input-file", type=click.File(), help="Read list of image filenames from text file.")
@click.option("-g/-f", "--group-tags/--flatten-tags", default=True, show_default=True, help="Output rows in {filename, tags} format or {filename, tag, score} format.")
@click.option("-N", "--name-only", is_flag=True, help="Output only the filename without the full path or extension.")
@click.option("-m", "--model", default="models/model.pth", type=click.Path(exists=True), show_default=True, help="The model to use.")
@click.argument("files", nargs=-1, type=click.Path(exists=True, allow_dash=True, path_type=Path))
def main(files, threshold, limit, bs, csv, input_file, group_tags, name_only, model):
autotagger = Autotagger(model)
if input_file:
paths = (Path(line.rstrip()) for line in input_file)
elif len(files) > 0:
paths = get_filepaths(files)
else:
click.echo(click.get_current_context().get_help())
click.get_current_context().exit()
for filepaths in ichunked(paths, bs):
filepaths = list(filepaths)
files = [click.open_file(filepath, "rb") for filepath in filepaths]
predictions = autotagger.predict(files, threshold=threshold, limit=limit, bs=bs)
for filepath, tags in zip(filepaths, predictions):
output_result(filepath, tags, csv, group_tags, name_only)
def output_result(filepath, tags, csv, group_tags, name_only):
name = filepath.stem if name_only else str(filepath)
if csv and group_tags:
tag_names = " ".join(sorted(tags.keys()))
click.echo(f"{name},{tag_names}")
elif csv and not group_tags:
for tag, score in tags.items():
click.echo(f"{name},{tag},{score}")
elif group_tags:
data = { "filename": name, "tags": tags }
click.echo(json.dumps(data))
else:
for tag, score in tags.items():
data = { "filename": name, "tag": tag, "score": score }
click.echo(json.dumps(data))
def get_filepaths(paths):
files = (recurse_dir(path) if path.is_dir() else iter([path]) for path in paths)
return itertools.chain(*files)
def recurse_dir(directory):
return (path for path in directory.glob("**/*") if not path.is_dir())
if __name__ == "__main__":
main()