feat: Enhance text normalization with support for internet slang expansion, currency formatting, and date handling; update debug WAVs interface and settings

tests/conftest.py

@@ -1,9 +1,13 @@
 import importlib
 import sys
 
+import os
+
+import pytest
+
 # Ensure real optional dependencies are imported before tests that install stubs
-# so that available packages (like ebooklib, bs4) aren't replaced with dummy modules.
-for module_name in ("ebooklib", "bs4"):
+# so that available packages (like ebooklib, bs4, numpy) aren't replaced with dummy modules.
+for module_name in ("ebooklib", "bs4", "numpy"):
     if module_name not in sys.modules:
         try:
             importlib.import_module(module_name)
@@ -11,3 +15,25 @@ for module_name in ("ebooklib", "bs4"):
             # On environments without the optional dependency, downstream tests
             # will install lightweight stubs as needed.
             pass
+
+
+@pytest.fixture(autouse=True, scope="session")
+def _isolate_settings_dir(tmp_path_factory: pytest.TempPathFactory):
+    settings_dir = tmp_path_factory.mktemp("abogen-settings")
+    os.environ["ABOGEN_SETTINGS_DIR"] = str(settings_dir)
+
+    try:
+        from abogen.utils import get_user_settings_dir
+
+        get_user_settings_dir.cache_clear()
+    except Exception:
+        pass
+
+    try:
+        from abogen.normalization_settings import clear_cached_settings
+
+        clear_cached_settings()
+    except Exception:
+        pass
+
+    yield

tests/test_text_normalization.py

@@ -215,6 +215,40 @@ def test_decades_can_skip_expansion_when_disabled() -> None:
     assert "'90s" in normalized
 
 
+def test_abbreviated_decades_expand_to_spoken_form() -> None:
+    normalized = _normalize_text("She loved music from the '80s.")
+    assert "eighties" in normalized.lower()
+
+
+def test_currency_under_one_dollar_uses_cents() -> None:
+    normalized = _normalize_text("It cost $0.99.")
+    folded = normalized.lower().replace("-", " ")
+    assert "zero dollars" not in folded
+    assert "cents" in folded
+
+
+def test_iso_dates_use_locale_order_and_ordinals(monkeypatch) -> None:
+    monkeypatch.setenv("LC_TIME", "en_US.UTF-8")
+    normalized = _normalize_text("The date is 2025/12/15.")
+    folded = normalized.lower().replace("-", " ")
+    assert "december" in folded
+    assert "fifteenth" in folded
+
+
+def test_times_and_acronyms_do_not_say_dot() -> None:
+    normalized = _normalize_text("Meet at 5 p.m. near the U.S.A. border.")
+    folded = normalized.lower()
+    assert " dot " not in folded
+
+
+def test_internet_slang_expansion_is_configurable() -> None:
+    normalized = _normalize_text(
+        "pls knock before entering.",
+        normalization_overrides={"normalization_internet_slang": True},
+    )
+    assert "please" in normalized.lower()
+
+
 @pytest.mark.skipif(not SPACY_RESOLVER_AVAILABLE, reason="spaCy model unavailable")
 def test_spacy_disambiguates_it_has_from_context() -> None:
     normalized = _normalize_text("It's been a long time.")