Investigate using libappstream to parse AppStream

[barthalion]

          > Website is likely also fine https://github.com/flathub-infra/website/blob/main/backend/app/utils.py#L216

Hmm... A bit OT, but did you know that you could get appstream2dict relatively efficiently via AppStream directly?

import gi
import yaml

gi.require_version('AppStream', '1.0')
from gi.repository import AppStream

def appstream2dict(xml_fname) -> dict[str, dict]:
    mdata = AppStream.Metadata()
    mdata.set_locale('ALL')
    mdata.set_format_style(AppStream.FormatStyle.CATALOG)
    mdata.set_parse_flags(AppStream.ParseFlags.IGNORE_MEDIABASEURL)

    with lzma.open(xml_fname, 'r') as f:
        xml_catalog_data = str(f.read(), 'utf-8')

    mdata.clear_components()
    mdata.parse_data(xml_catalog_data, AppStream.FormatKind.YAML)

    return yaml.safe_load(mdata.components_to_catalog(AppStream.FormatKind.YAML))

You might not want that for various reasons, but it would be dramatically less code ^^

Originally posted by @ximion in flathub-infra/flat-manager-hooks#101 (comment)

[ximion]

Advantages

• Always parse with the latest AppStream version and have any differences between specification revisions automatically normalized into one canonical form
• Very fast XML parser
• Lots of utility functions readily available to validate, transform markup or query additional information
• Extremely well tested
• Could reduce the amount of custom code drastically

Disadvantages

• PyGObject dependency, GIR and dependency on a C library is annoying to handle and could make upgrading a bit harder
• Some porting effort
• Drags in a few dependencies (xmlb, libyaml, curl, potentially snowball)