Add tests for depicts.utils

tests/test_utils.py

@@ -0,0 +1,109 @@
+"""Unit tests for the helpers in ``depicts.utils``.
+
+These tests cover common paths and a few edge cases to ensure
+stable behavior across refactors.
+"""
+
+from depicts import utils
+from flask import Flask
+import pytest
+
+
+def test_ordinal() -> None:
+    """Convert integers to the expected English ordinals."""
+    assert utils.ordinal(1) == "1st"
+    assert utils.ordinal(2) == "2nd"
+    assert utils.ordinal(3) == "3rd"
+    assert utils.ordinal(4) == "4th"
+
+
+def test_chunk_basic() -> None:
+    """Chunk an iterable into fixed-size tuples with a remainder."""
+    data = list(range(10))
+    chunks = list(utils.chunk(data, 3))
+    assert chunks == [(0, 1, 2), (3, 4, 5), (6, 7, 8), (9,)]
+
+
+def test_drop_start_success_and_failure() -> None:
+    """Drop a matching prefix and assert on missing prefix."""
+    assert utils.drop_start("Category:Painting", "Category:") == "Painting"
+    with pytest.raises(AssertionError):
+        utils.drop_start("Painting", "Category:")
+
+
+def test_drop_category_ns() -> None:
+    """Remove the "Category:" namespace from a string."""
+    assert utils.drop_category_ns("Category:Portraits") == "Portraits"
+
+
+def test_parse_sitelink() -> None:
+    """Decode sitelink by removing prefix, unquoting, and de-underscoring."""
+    start = "https://en.wikipedia.org/wiki/"
+    s = "https://en.wikipedia.org/wiki/Hello_World%21"
+    assert utils.parse_sitelink(s, start) == "Hello World!"
+
+
+def test_word_contains_letter() -> None:
+    """Detect whether a token contains at least one letter."""
+    assert utils.word_contains_letter("abc123") is True
+    assert utils.word_contains_letter("12345") is False
+    assert utils.word_contains_letter("!!!") is False
+
+
+def test_also_singular_main_plural_and_singular() -> None:
+    """Return both plural and singular when appropriate."""
+    # plural should include both plural and singular
+    assert set(utils.also_singular_main("Dogs")) == {"Dogs", "Dog"}
+    # singular should return as-is
+    assert utils.also_singular_main("Dog") == ["Dog"]
+
+
+def test_also_singular_gender_and_skip_names() -> None:
+    """Add gender-derived forms and honor the skip list."""
+    # Adds gender-derived singulars
+    names = utils.also_singular("Women")
+    assert "Women" in names
+    assert "woman" in names
+    # Skips configured names
+    assert utils.also_singular("National Gallery") == []
+
+
+def test_wiki_url_capitalization_and_ns() -> None:
+    """Build proper MediaWiki URLs with capitalization and namespace."""
+    # lowercase first letter should be capitalized, spaces become underscores
+    url = utils.wiki_url("example page", site="enwiki")
+    expected = "https://en.wikipedia.org/wiki/Example_page"
+    assert url == expected
+    # namespace prefix when provided
+    url_cat = utils.wiki_url("Portraits", site="commons", ns="Category")
+    expected_cat = "https://commons.wikimedia.org/wiki/Category:Portraits"
+    assert url_cat == expected_cat
+
+
+def test_get_int_arg_present_and_missing() -> None:
+    """Parse integer query args and ignore missing/non-integer values."""
+    app = Flask(__name__)
+    with app.test_request_context("/?page=10&foo=bar"):
+        assert utils.get_int_arg("page") == 10
+        assert utils.get_int_arg("foo") is None
+        assert utils.get_int_arg("missing") is None
+
+
+def test_format_time_precisions() -> None:
+    """Format time strings for multiple precision levels."""
+    # Full date
+    assert utils.format_time("+1965-04-23T00:00:00Z", 11) == "23 April 1965"
+    # Month precision
+    assert utils.format_time("+1965-04-00T00:00:00Z", 10) == "April 1965"
+    # Year precision
+    assert utils.format_time("+1965-00-00T00:00:00Z", 9) == "1965"
+    # Decade precision
+    assert utils.format_time("+1960-00-00T00:00:00Z", 8) == "1960s"
+    # Century precision
+    assert utils.format_time("+1965-00-00T00:00:00Z", 7) == "20th century"
+    # Millennium precision
+    assert utils.format_time("+2001-00-00T00:00:00Z", 6) == "3rd millennium"
+    # Unparseable falls back to input
+    assert utils.format_time("not-a-date", 9) == "not-a-date"
+    # BC year with unknown month/day should still return year for precision 9
+    assert utils.format_time("-0123-00-00T00:00:00Z", 9) == "-123"