from pathlib import Path

from backend.app.utils.paths import (
    canonical_subdir,
    derive_subdir,
    derive_top_level_artist,
    filehash_from_url,
    hash_suffixed_name,
)

IMPORT = Path("/import")


def test_filehash_from_url_extracts_lowercased_md5():
    url = "https://c10.patreonusercontent.com/4/patreon-media/p/post/1/AbC123DeF456789012345678901234EF/1/img.png?token=x"
    assert filehash_from_url(url) == "abc123def456789012345678901234ef"


def test_filehash_from_url_none_and_no_hash():
    assert filehash_from_url(None) is None
    assert filehash_from_url("") is None
    assert filehash_from_url("https://example.test/a.png") is None


def test_filehash_from_url_query_unescaping_irrelevant():
    # The hash sits in the path before the query, so entity-escaped ampersands
    # in the query never affect the match (render-time matches escaped srcs).
    base = "https://cdn.test/p/0123456789abcdef0123456789ABCDEF/x.jpg"
    assert filehash_from_url(base + "?a=1&b=2") == "0123456789abcdef0123456789abcdef"


def test_derive_subdir_nested():
    assert derive_subdir(Path("/import/Alice/sub/x.png"), IMPORT) == "Alice/sub"


def test_derive_subdir_one_level():
    assert derive_subdir(Path("/import/Alice/x.png"), IMPORT) == "Alice"


def test_derive_subdir_root():
    assert derive_subdir(Path("/import/x.png"), IMPORT) == ""


def test_derive_subdir_outside_root():
    assert derive_subdir(Path("/elsewhere/x.png"), IMPORT) == ""


def test_hash_suffixed_name():
    h = "abcdef1234567890" + "0" * 48
    assert hash_suffixed_name("photo", h, ".png") == "photo__abcdef1234.png"


def test_derive_top_level_artist_present():
    assert derive_top_level_artist(Path("/import/Alice/x.png"), IMPORT) == "Alice"


def test_derive_top_level_artist_nested():
    assert derive_top_level_artist(Path("/import/Alice/sub/x.png"), IMPORT) == "Alice"


def test_derive_top_level_artist_root():
    assert derive_top_level_artist(Path("/import/x.png"), IMPORT) is None


# --- canonical_subdir (milestone #421) --------------------------------------


def test_canonical_subdir_replaces_the_top_segment():
    assert canonical_subdir("Conto/patreon", "conto") == "conto/patreon"
    assert canonical_subdir("Conto", "conto") == "conto"


def test_canonical_subdir_keeps_everything_below_the_artist():
    """Only the artist segment is authoritative — post folders below it are
    the downloader's business and must survive untouched."""
    assert canonical_subdir(
        "Big Bang/patreon/2026-01-02_123_A Post", "big-bang"
    ) == "big-bang/patreon/2026-01-02_123_A Post"


def test_canonical_subdir_passes_through_without_an_artist():
    """No resolved artist means nothing authoritative to canonicalise against,
    so the import folder's own name stands."""
    assert canonical_subdir("Conto/patreon", None) == "Conto/patreon"
    assert canonical_subdir("Conto", "") == "Conto"


def test_canonical_subdir_leaves_the_images_root_alone():
    """An empty subdir is a file landing at the root — there is no artist
    folder to correct, and what becomes of those is task #4247's call."""
    assert canonical_subdir("", "conto") == ""
    assert canonical_subdir("", None) == ""


def test_canonical_subdir_is_idempotent():
    once = canonical_subdir("Conto/patreon", "conto")
    assert canonical_subdir(once, "conto") == once