Rust features

#!/usr/bin/env python3
import glob
import html
import json
import os
import re
import shutil

import toml


def get_features(dirname):
    features = {}
    for feature in sorted(os.listdir('caniuse.rs/data/' + dirname)):
        with open('caniuse.rs/data/' + dirname + '/' + feature) as f:
            name = feature.split('.')[0]
            features[name] = toml.load(f)
    return features


with open('caniuse.rs/data/versions.toml') as f:
    versions = toml.load(f)

for version, data in versions.items():
    try:
        data['features'] = get_features(version)
    except FileNotFoundError:
        pass

unstable_features = get_features('unstable')


os.makedirs('target', exist_ok=True)
shutil.copy('style.css', 'target')

with open('target/data.json', 'w') as f:
    json.dump(dict(unstable=unstable_features, versions=versions), f)


def write_features(f, feature_data):
    features = {}
    for feat, data in feature_data.items():
        key = data['flag'] if 'flag' in data else feat
        # caniuse.rs sometimes has several .toml files for one feature flag e.g.
        # For the const_io feature flag it has 6 .toml files, all with the same
        # tracking_issue_id and the following titles:
        #
        # * `io::Cursor::get_ref` as `const fn`
        # * `io::Cursor::new` as `const fn`
        # * `io::Cursor::position` as `const fn`
        # * `io::empty` as `const fn`
        # * `io::repeat` as `const fn`
        # * `io::sink` as `const fn`
        #
        # That makes sense for a search-centric application.  Not so much for a
        # static-site generator since showing the same link 6 times is confusing.

        url = None
        if 'tracking_issue_id' in data:
            url = 'https://github.com/rust-lang/rust/issues/{}'.format(
                data['tracking_issue_id']
            )
        elif 'impl_pr_id' in data:
            url = 'https://github.com/rust-lang/rust/pull/{}'.format(data['impl_pr_id'])
        data['url'] = url
        data['filename'] = feat

        if key in features and features[key]['url'] != url:
            print(
                'different urls for feature {}:\n* {}: {}\n* {}: {}'.format(
                    key,
                    data['filename'],
                    data['url'],
                    features[key]['filename'],
                    features[key]['url'],
                )
            )
            data['title'] = data['flag']

        features[key] = data

    f.write('<ul>')
    for feat, data in features.items():
        f.write('<li><a')
        url = data['url']
        if url:
            f.write(f' href="{url}"')
        f.write('>')
        title = html.escape(data['title'])
        title = re.sub('`(.+?)`', lambda m: '<code>{}</code>'.format(m.group(1)), title)
        f.write(title)
        f.write('</a></li>')
    f.write('</ul>')


with open('target/index.html', 'w') as f:
    f.write('<!doctype html>')
    f.write('<meta charset=utf-8>')
    f.write('<title>Rust features</title>')
    f.write('<h1>Rust features</h1>')
    f.write('<link rel="stylesheet" href="style.css" />')

    # TODO: sort by T-lang and T-lib
    f.write('<h2>Unstable features</h2>')
    write_features(f, unstable_features)

    for version, data in reversed(list(versions.items())):
        f.write('<h2>{}</h2>'.format(version))
        if 'features' in data:
            write_features(f, data['features'])

# TODO: generate HTML