"""Structural accessibility checks for rendered pages.

Not a substitute for axe or a screen reader, but cheap enough to run on every page in unit tests.
"""

from html.parser import HTMLParser

_UNLABELLED_INPUT_TYPES = {"hidden", "submit", "button", "reset", "image"}


class _Collector(HTMLParser):
    def __init__(self):
        super().__init__()
        self.html_lang = None
        self.headings = []
        self.ids = []
        self.label_fors = set()
        self.controls = []
        self.images_without_alt = 0
        self._label_depth = 0

    def handle_starttag(self, tag, attrs):
        attrs = dict(attrs)
        if tag == "html":
            self.html_lang = attrs.get("lang")
        if "id" in attrs:
            self.ids.append(attrs["id"])
        if tag in {"h1", "h2", "h3", "h4", "h5", "h6"}:
            self.headings.append(int(tag[1]))
        if tag == "label":
            self._label_depth += 1
            if attrs.get("for"):
                self.label_fors.add(attrs["for"])
        if tag == "img" and "alt" not in attrs:
            self.images_without_alt += 1
        if tag in {"input", "select", "textarea"}:
            if tag == "input" and attrs.get("type", "text") in _UNLABELLED_INPUT_TYPES:
                return
            wrapped = self._label_depth > 0
            named = bool(attrs.get("aria-label") or attrs.get("aria-labelledby"))
            self.controls.append((attrs.get("id"), attrs.get("name"), wrapped or named))

    def handle_endtag(self, tag):
        if tag == "label":
            self._label_depth -= 1


def assert_accessible_html(html: str) -> None:
    page = _Collector()
    page.feed(html)
    problems = []
    if not page.html_lang:
        problems.append("<html> has no lang")
    if page.headings.count(1) != 1:
        problems.append(f"expected one <h1>, found {page.headings.count(1)}")
    for before, after in zip(page.headings, page.headings[1:], strict=False):
        if after > before + 1:
            problems.append(f"heading level skips from h{before} to h{after}")
    duplicates = {i for i in page.ids if page.ids.count(i) > 1}
    if duplicates:
        problems.append(f"duplicate ids: {sorted(duplicates)}")
    for control_id, name, labelled in page.controls:
        if not (labelled or control_id in page.label_fors):
            problems.append(f"form control without a label: id={control_id} name={name}")
    if page.images_without_alt:
        problems.append(f"{page.images_without_alt} <img> without alt")
    assert not problems, "\n".join(problems)