"""R3 — compare: two picked items of a booth side by side.

Contract: docs/contracts/r3_compare.contract.md. The server half: the route,
the pair, the step and strip links, the regions, and the JS-off flag landing.
The browser half is tests/test_compare_browser.py.
"""

from __future__ import annotations

import pathlib
import re
import sys
from urllib.parse import parse_qs, urlsplit

from fastapi.testclient import TestClient

sys.path.insert(0, str(pathlib.Path(__file__).parent.parent))

from booth.app import create_app  # noqa: E402

PNG = b"\x89PNG\r\n\x1a\n"


def _booth(root: pathlib.Path, name: str, files: dict[str, bytes]) -> pathlib.Path:
    b = root / name
    b.mkdir()
    for rel, data in files.items():
        p = b / rel
        p.parent.mkdir(parents=True, exist_ok=True)
        p.write_bytes(data)
    return b


def _client(root: pathlib.Path) -> TestClient:
    return TestClient(create_app(root, ttl_hours=24, start_sweeper=False),
                      follow_redirects=False)


def _frames(body: str) -> dict[str, str]:
    """rel -> the A/B marks its filmstrip frame carries ('' for none), in
    strip order."""
    film = re.search(r'<nav class="film"[^>]*data-region="film".*?</nav>', body, re.S).group(0)
    out = {}
    for f in re.findall(r'<a class="film-f[^"]*"[^>]*>.*?</a>', film, re.S):
        rel = re.search(r'data-rel="([^"]*)"', f).group(1)
        out[rel] = "".join(re.findall(r'<span class="film-ab">([AB]+)</span>', f))
    return out


# ---- C1: the route and the pair ----------------------------------------------

def test_compare_renders_the_pair(tmp_path):
    """The tracer: four pictures, #1 against #3. Both names and both ordinals
    are printed, and the filmstrip marks #1 A and #3 B."""
    _booth(tmp_path, "g", {f"{n}.png": PNG for n in ("p", "q", "r", "s")})
    r = _client(tmp_path).get("/b/g/compare?a=p.png&b=r.png")
    assert r.status_code == 200
    body = r.text
    label_a = re.search(r'data-region="label-a".*?</div>', body, re.S).group(0)
    label_b = re.search(r'data-region="label-b".*?</div>', body, re.S).group(0)
    assert "p.png" in label_a and "#1" in label_a, label_a
    assert "r.png" in label_b and "#3" in label_b, label_b
    assert _frames(body) == {"p.png": "A", "q.png": "", "r.png": "B", "s.png": ""}
    assert list(_frames(body)) == ["p.png", "q.png", "r.png", "s.png"], "the strip is in RING order"
    # each side's flag form names its OWN item
    for key, rel in (("a", "p.png"), ("b", "r.png")):
        form = re.search(r'data-region="flag-%s".*?</form>' % key, body, re.S).group(0)
        assert f'name="target" value="{rel}"' in form, (key, form)


def _four(root: pathlib.Path) -> pathlib.Path:
    """Four pictures, a doc, a caption sidecar and a dotfile: every kind of
    thing a side can name that is not a side."""
    return _booth(root, "g", {"p.png": PNG, "q.png": PNG, "r.png": PNG, "s.png": PNG,
                              "notes.md": b"# n", "p.png.txt": b"a caption",
                              ".hidden.png": PNG})


def test_a_bad_side_is_a_404(tmp_path):
    """Missing, traversal, a NUL, a dotfile, a doc item, a non-item file: a 404
    each, on either side, never a 500."""
    _four(tmp_path)
    c = _client(tmp_path)
    bad = ["", "../g/p.png/..", "../../etc/passwd", "p.png\x00", ".hidden.png",
           "notes.md", "p.png.txt", "gone.png", "sub/"]
    for rel in bad:
        for q in ({"a": rel, "b": "q.png"}, {"a": "q.png", "b": rel}):
            r = c.get("/b/g/compare", params=q)
            assert r.status_code == 404, (q, r.status_code)
    for q in ({"b": "q.png"}, {"a": "q.png"}, {}):
        assert c.get("/b/g/compare", params=q).status_code == 404, q


def test_a_missing_param_is_404_not_422(tmp_path):
    """The review declares `f: str` and answers 422 without it; compare
    declares both sides with a default and answers 404."""
    _four(tmp_path)
    r = _client(tmp_path).get("/b/g/compare?a=p.png")
    assert r.status_code == 404


def test_an_outside_symlink_in_the_ring_is_404(tmp_path):
    """`booth_items` follows symlinks, so a link pointing OUTSIDE the booth is
    in the review ring; only the containment check refuses it."""
    from booth.items import booth_items, review_chain
    b = _four(tmp_path)
    outside = tmp_path / "elsewhere.png"
    outside.write_bytes(PNG)
    (b / "zz-link.png").symlink_to(outside)
    assert "zz-link.png" in review_chain(booth_items(b)), "the fixture must put it in the ring"
    # a SIBLING whose name shares the booth's prefix is outside too (the
    # containment check compares with the separator, never a bare prefix)
    sib = tmp_path / "g-extra"
    sib.mkdir()
    (sib / "x.png").write_bytes(PNG)
    (b / "zz-sib.png").symlink_to(sib / "x.png")
    c = _client(tmp_path)
    for rel in ("zz-link.png", "zz-sib.png"):
        assert c.get(f"/b/g/compare?a={rel}&b=p.png").status_code == 404, rel
        assert c.get(f"/b/g/compare?a=p.png&b={rel}").status_code == 404, rel


def test_no_navigation_offers_a_pair_that_404s(tmp_path):
    """An outside symlink stays in the review ring, and compare 404s it. So no
    compare link may offer it: not the strip, not a step, not the review's
    Compare control, not the JS-off flag landing (heid bug hunt, 3 of 4)."""
    b = _booth(tmp_path, "g", {"a.png": PNG, "c.png": PNG})
    outside = tmp_path / "elsewhere.png"
    outside.write_bytes(PNG)
    (b / "b-link.png").symlink_to(outside)
    c = _client(tmp_path)
    body = c.get("/b/g/compare?a=a.png&b=c.png").text
    assert list(_frames(body)) == ["a.png", "c.png"], _frames(body)
    for h in _compare_links(body):
        q = parse_qs(urlsplit(h).query)
        assert "b-link.png" not in (q["a"][0], q["b"][0]), h
    assert _step(body, "a-next") == ("c.png", "c.png")          # steps over it
    assert _compare_href(c.get("/b/g/view?f=a.png").text) == ("a.png", "c.png")
    r = c.post("/b/g/flag", data={"target": "a.png", "on": "1", "back": "compare",
                                  "a": "a.png", "b": "b-link.png"})
    assert r.headers["location"] == "/b/g/#item-a.png", r.headers["location"]


def test_hostile_booth_names_are_404_not_500(tmp_path):
    """A NUL in the booth segment makes Path.resolve raise ValueError, which
    is not an OSError: it must still be a 404 (heid bug hunt, hulda)."""
    _four(tmp_path)
    c = _client(tmp_path)
    for path in ("/b/g%00/compare?a=p.png&b=q.png", "/b/g%00/view?f=p.png", "/b/g%00/"):
        assert c.get(path).status_code == 404, path


def test_a_planted_fifo_marker_cannot_hang_a_look(tmp_path):
    """Recording a look never costs the page: a FIFO planted at `.viewed` must
    not block the open that touches it (heid bug hunt, hulda)."""
    import os
    import threading
    b = _four(tmp_path)
    os.mkfifo(b / ".viewed")
    got = []
    t = threading.Thread(target=lambda: got.append(
        _client(tmp_path).get("/b/g/compare?a=p.png&b=q.png").status_code), daemon=True)
    t.start()
    t.join(10)
    assert got == [200], "a planted FIFO held the look open"


def test_a_look_records_both_seen(tmp_path):
    """A compare GET is a look at both sides; a 404 records nothing."""
    import json
    b = _four(tmp_path)
    c = _client(tmp_path)
    assert c.get("/b/g/compare?a=p.png&b=gone.png").status_code == 404
    assert not (b / ".seen").exists() and not (b / ".viewed").exists()
    assert c.get("/b/g/compare?a=q.png&b=s.png").status_code == 200
    assert set(json.loads((b / ".seen").read_text())) == {"q.png", "s.png"}
    assert (b / ".viewed").exists()


def test_compare_carries_data_booth(tmp_path):
    """Reveal all's script and the head script's reveal restore both read
    `data-booth` off <html>, and bail without it."""
    _four(tmp_path)
    body = _client(tmp_path).get("/b/g/compare?a=p.png&b=q.png").text
    assert re.search(r'<html lang="en" data-booth="g">', body)


def test_no_data_region_repeats(tmp_path):
    """The swap keeps the FIRST fresh node per id and copies it over EVERY live
    node with that id, so a shared id would turn B's flag into A's. Unique,
    keyed by side — including when a == b."""
    _four(tmp_path)
    c = _client(tmp_path)
    for q in ("a=p.png&b=r.png", "a=q.png&b=q.png"):
        # attributes only: base.html's script names `[data-region="status"]`
        ids = re.findall(r'\sdata-region="([^"]+)"', c.get(f"/b/g/compare?{q}").text)
        assert len(ids) == len(set(ids)), (q, ids)
        assert {"flag-a", "flag-b", "label-a", "label-b", "film"} <= set(ids), ids
        assert not [i for i in ids if i.startswith("item-")], ids
    body = c.get("/b/g/compare?a=q.png&b=q.png").text
    assert _frames(body)["q.png"] == "AB", "a == b marks the one frame both ways"


def test_a_video_or_track_plays_in_its_own_stage_and_two_get_no_toggle(tmp_path):
    """C4: video and audio play in their own stage; the Fit | 1:1 toggle is
    bound only when a side is a picture, so two videos get none."""
    _booth(tmp_path, "g", {"a.webm": b"\x1aE\xdf\xa3", "b.mp3": b"ID3", "c.png": PNG})
    c = _client(tmp_path)
    body = c.get("/b/g/compare?a=a.webm&b=b.mp3").text
    assert re.search(r'<video class="cmp-media"[^>]*src="a.webm"', body)
    assert re.search(r'<audio class="cmp-media"[^>]*src="b.mp3"', body)
    assert 'id="vtoggle"' not in body
    assert 'id="vtoggle"' in c.get("/b/g/compare?a=a.webm&b=c.png").text


# ---- C3: stepping --------------------------------------------------------------

def _ring6(root: pathlib.Path) -> pathlib.Path:
    """Six media with a DOC between them: 03-notes.md takes ordinal 3, so an
    ordinal is not a ring position."""
    return _booth(root, "g", {"01.png": PNG, "02.png": PNG, "03-notes.md": b"# n",
                              "04.png": PNG, "05.png": PNG, "06.png": PNG, "07.png": PNG})


def _step(body: str, which: str) -> tuple[str, str]:
    """The (a, b) rels a step link targets."""
    href = re.search(r'<a [^>]*data-step="%s"[^>]*href="([^"]+)"' % which, body)
    href = href or re.search(r'<a [^>]*href="([^"]+)"[^>]*data-step="%s"' % which, body)
    q = parse_qs(urlsplit(href.group(1).replace("&amp;", "&")).query)
    return q["a"][0], q["b"][0]


def test_linked_steps_keep_the_distance_and_wrap(tmp_path):
    """Ring positions 2 and 5 step forward to (3, 6), then (4, 1) — wrapped —
    and back from (1, 4) to (6, 3). Named by rel, never by ordinal."""
    _ring6(tmp_path)
    c = _client(tmp_path)
    body = c.get("/b/g/compare?a=02.png&b=06.png").text
    assert _step(body, "both-next") == ("04.png", "07.png")
    body = c.get("/b/g/compare?a=04.png&b=07.png").text
    assert _step(body, "both-next") == ("05.png", "01.png")
    body = c.get("/b/g/compare?a=07.png&b=02.png").text
    assert _step(body, "both-next") == ("01.png", "04.png"), "A wraps as B does"
    body = c.get("/b/g/compare?a=01.png&b=05.png").text
    assert _step(body, "both-prev") == ("07.png", "04.png")
    # each side on its own moves only itself, and wraps the same way
    assert _step(body, "a-prev") == ("07.png", "05.png")
    assert _step(body, "a-next") == ("02.png", "05.png")
    assert _step(body, "b-prev") == ("01.png", "04.png")
    assert _step(body, "b-next") == ("01.png", "06.png")


def _compare_links(body: str) -> list[str]:
    """Every step and filmstrip href on the page, entity-decoded."""
    hrefs = re.findall(r'<a [^>]*(?:data-step="[^"]+"[^>]*href|class="film-f[^"]*"[^>]*href)="([^"]+)"', body)
    return [h.replace("&amp;", "&") for h in hrefs]


def test_the_urls_are_keyed_by_rel(tmp_path):
    """Every step and strip link names both sides by rel, url-quoted; none
    carries an ordinal or any key beyond the pair and the view state."""
    _booth(tmp_path, "g", {"a b.png": PNG, "sub dir/c#d.png": PNG, "e&f.png": PNG})
    body = _client(tmp_path).get("/b/g/compare", params={"a": "a b.png", "b": "sub dir/c#d.png"}).text
    links = _compare_links(body)
    assert len(links) == 6 + 3, links                       # six steps, three frames
    for h in links:
        u = urlsplit(h)
        assert u.path == "/b/g/compare", h
        assert " " not in h and "#" not in u.query.replace("%23", ""), h
        q = parse_qs(u.query)
        assert set(q) == {"a", "b"}, h
        assert all(v in ("a b.png", "sub dir/c#d.png", "e&f.png") for v in (q["a"][0], q["b"][0])), h


def test_view_state_rides_the_links(tmp_path):
    """`side=a&link=0` rides every step and strip link; an unknown value reads
    as the default (B active, linked) and is never an error."""
    _four(tmp_path)
    c = _client(tmp_path)
    body = c.get("/b/g/compare?a=p.png&b=r.png&side=a&link=0").text
    links = _compare_links(body)
    assert links and all(h.endswith("&side=a&link=0") for h in links), links
    assert 'class="cmp-side is-active" data-side="a"' in body
    # with A active a frame replaces A, keeping B
    frame = re.search(r'<a class="film-f[^"]*"\s+href="([^"]+)" data-rel="q.png"', body).group(1)
    assert parse_qs(urlsplit(frame.replace("&amp;", "&")).query) == {
        "a": ["q.png"], "b": ["r.png"], "side": ["a"], "link": ["0"]}
    links = _compare_links(c.get("/b/g/compare?a=p.png&b=r.png&link=0").text)
    assert links and all(h.endswith("&link=0") and "side=" not in h for h in links), links
    for odd in ("side=z&link=maybe", "side=A&link=00", "side=&link="):
        r = c.get(f"/b/g/compare?a=p.png&b=r.png&{odd}")
        assert r.status_code == 200, odd
        links = _compare_links(r.text)
        assert all(set(parse_qs(urlsplit(h).query)) == {"a", "b"} for h in links), (odd, links)
        assert 'class="cmp-side is-active" data-side="b"' in r.text, odd
        assert 'data-linked="1"' in r.text, odd


# ---- C5: judging without JS -----------------------------------------------------

def _flag(c: TestClient, form: dict, accept: str | None = None):
    headers = {"accept": accept} if accept else {}
    r = c.post("/b/g/flag", data={"target": "q.png", "on": "1", **form}, headers=headers)
    assert r.status_code == (204 if accept else 303), (form, r.status_code)
    return r


def test_a_flag_without_js_lands_on_the_same_pair(tmp_path):
    """`back=compare` lands on exactly the pair, built from the checked rels,
    with the view state mapped from a closed set and never echoed, and no
    fragment. Anything outside the ring takes the no-`back` landing; the
    in-place answer is the 204 it always was."""
    _booth(tmp_path, "g", {"p.png": PNG, "q.png": PNG, "sub dir/r s.png": PNG, "n.md": b"# n"})
    c = _client(tmp_path)
    pair = {"back": "compare", "a": "p.png", "b": "sub dir/r s.png"}
    want = "/b/g/compare?a=p.png&b=sub%20dir/r%20s.png"
    r = _flag(c, pair)
    assert r.status_code == 303 and r.headers["location"] == want
    assert _flag(c, {**pair, "side": "a", "link": "0"}).headers["location"] == want + "&side=a&link=0"
    assert _flag(c, {**pair, "link": "0"}).headers["location"] == want + "&link=0"
    for odd in ({"side": "A"}, {"link": "00"}, {"side": "b", "link": "1"},
                {"side": "a#x", "link": "0&side=a"}):
        assert _flag(c, {**pair, **odd}).headers["location"] == want, odd
    no_back = _flag(c, {"target": "q.png"}).headers["location"]
    assert no_back == "/b/g/#item-q.png"
    for bad in ({"a": "n.md"}, {"b": "gone.png"}, {"a": "../g/p.png"}, {"b": ""}):
        assert _flag(c, {**pair, **bad}).headers["location"] == no_back, bad
    r = _flag(c, pair, accept="application/json")
    assert r.status_code == 204 and "location" not in r.headers


def test_every_other_landing_is_byte_identical(tmp_path):
    """R2 INV-4: `back=view`, `back=marks` and no `back` land exactly where they
    did before compare existed."""
    _booth(tmp_path, "g", {"p.png": PNG, "q.png": PNG})
    c = _client(tmp_path)
    assert _flag(c, {}).headers["location"] == "/b/g/#item-q.png"
    assert _flag(c, {"back": "marks"}).headers["location"] == "/b/g/marks#item-q.png"
    assert _flag(c, {"back": "view", "f": "p.png"}).headers["location"] == "/b/g/view?f=p.png#rail"
    assert _flag(c, {"back": "view", "f": "gone.png"}).headers["location"] == "/b/g/#item-q.png"
    assert _flag(c, {"back": "Compare", "a": "p.png", "b": "q.png"}).headers["location"] == "/b/g/#item-q.png"


# ---- C2: picking from the review ------------------------------------------------

def _compare_href(body: str) -> tuple[str, str]:
    href = re.search(r'<a class="vbtn vcompare"[^>]*href="([^"]+)"', body).group(1)
    u = urlsplit(href.replace("&amp;", "&"))
    assert u.path == "/b/g/compare", href
    q = parse_qs(u.query)
    assert set(q) == {"a", "b"}, href
    return q["a"][0], q["b"][0]


def test_the_review_offers_compare_with_the_next_item(tmp_path):
    """The review's Compare control opens this item against the NEXT media item
    in the ring — skipping the doc — and the last wraps to the first. A ring of
    one compares the item with itself."""
    _ring6(tmp_path)
    c = _client(tmp_path)
    assert _compare_href(c.get("/b/g/view?f=02.png").text) == ("02.png", "04.png")
    assert _compare_href(c.get("/b/g/view?f=07.png").text) == ("07.png", "01.png")
    _booth(tmp_path, "one", {"only.png": PNG, "n.md": b"# n"})
    body = c.get("/b/one/view?f=only.png").text
    href = re.search(r'<a class="vbtn vcompare"[^>]*href="([^"]+)"', body).group(1)
    assert href.replace("&amp;", "&") == "/b/one/compare?a=only.png&b=only.png"
    # a doc's own page is not a review, and offers no compare
    assert 'class="vbtn vcompare"' not in c.get("/b/g/view?f=03-notes.md").text