tests/test_buckets.py

from src import buckets, config


def test_parse_targets():
    assert buckets.parse_targets("a.com, b.org\nc.net  d.io") == ["a.com", "b.org", "c.net", "d.io"]
    assert buckets.parse_targets("") == []
    assert buckets.parse_targets("  Example.COM ") == ["example.com"]


def test_validate_targets():
    assert buckets.validate_targets([], [])[0] is False
    assert buckets.validate_targets([], ["example.com"])[0] is True
    assert buckets.validate_targets(["bad host"], [])[0] is False  # space
    assert buckets.validate_targets([], ["nodot"])[0] is False     # no dot


def test_validate_languages():
    assert buckets.validate_languages([])[0] is True          # optional filter
    assert buckets.validate_languages(["deu", "eng"])[0] is True
    ok, msg = buckets.validate_languages(["eng", "xxx"])
    assert not ok and "xxx" in msg


def test_language_choices():
    choices = config.language_choices()
    assert ("German (deu)", "deu") in choices
    assert ("English (eng)", "eng") in choices
    codes = [c for _, c in choices]
    assert len(codes) == len(set(codes))                       # no duplicate codes
    assert set(codes) == set(config.CONTENT_LANGUAGE_CODES)
    assert all(len(c) == 3 and c.isalpha() and c.islower() for c in codes)


def test_validate_crawls():
    assert buckets.validate_crawls([])[0] is False
    ok, msg = buckets.validate_crawls(["CC-MAIN-2026-25"])
    assert ok and msg == ""
    ok, msg = buckets.validate_crawls([f"c{i}" for i in range(config.COST_GUARD_MAX_CRAWLS + 1)])
    assert ok and "⚠️" in msg  # allowed but warned


def test_validate_target_bucket_format_errors_no_network():
    # These return before any network call.
    assert buckets.validate_target_bucket("justname", "p")[0] is False
    assert buckets.validate_target_bucket("a/b/c", "p")[0] is False
    assert buckets.validate_target_bucket("me/bucket", "../escape")[0] is False
    assert buckets.validate_target_bucket("me/bucket", "/abs")[0] is False