scriptconv 0.0.4a20__tar.gz → 0.0.4a22__tar.gz

{scriptconv-0.0.4a20/scriptconv.egg-info → scriptconv-0.0.4a22}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: scriptconv
-Version: 0.0.4a20
+Version: 0.0.4a22
 Summary: Zero-dependency script & phoneme-notation core — ISO-15924 detection & metadata, IPA↔ARPABET/X-SAMPA/Lexique/Kirshenbaum/Cotovía/RFE, Buckwalter↔Arabic, Hangul→jamo, kana
 License: Apache-2.0
 Project-URL: Homepage, https://github.com/TigreGotico/scriptconv
@@ -69,6 +69,7 @@ Requires-Dist: pycotovia; extra == "gl"
 Provides-Extra: he
 Requires-Dist: scriptconv[phonemizers]; extra == "he"
 Requires-Dist: phonikud-onnx; extra == "he"
+Requires-Dist: phonikud; extra == "he"
 Provides-Extra: fa
 Requires-Dist: scriptconv[phonemizers]; extra == "fa"
 Requires-Dist: persian_phonemizer; extra == "fa"
@@ -124,6 +125,7 @@ Requires-Dist: regex; extra == "test"
 Requires-Dist: ovos-number-parser; extra == "test"
 Requires-Dist: bifonia>=0.1.1; extra == "test"
 Requires-Dist: stressonnx>=0.0.3a2; extra == "test"
+Requires-Dist: tomli; python_version < "3.11" and extra == "test"
 Dynamic: license-file
 
 <div align="center">

{scriptconv-0.0.4a20 → scriptconv-0.0.4a22}/pyproject.toml

@@ -49,7 +49,7 @@ eu = ["scriptconv[phonemizers]", "euskaphone", "ahotts_g2p"]
 pt-phonemizers = ["scriptconv[phonemizers]", "tugaphone", "g2p_barranquenho"]
 mwl = ["scriptconv[phonemizers]", "mwl_phonemizer"]
 gl = ["scriptconv[phonemizers]", "pycotovia"]
-he = ["scriptconv[phonemizers]", "phonikud-onnx"]
+he = ["scriptconv[phonemizers]", "phonikud-onnx", "phonikud"]
 fa = ["scriptconv[phonemizers]", "persian_phonemizer"]
 vi = ["scriptconv[phonemizers]", "viphoneme"]
 tashkeel = ["scriptconv[phonemizers]", "text2tashkeel"]
@@ -61,7 +61,7 @@ zh-phonemizers = ["scriptconv[phonemizers]", "jieba", "pypinyin>=0.50,<1", "xpin
 en-phonemizers = ["scriptconv[phonemizers]", "g2p_en", "openphonemizer", "deep-phonemizer", "requests"]
 ar-phonemizers = ["scriptconv[phonemizers]", "arbtok", "ovos-number-parser"]
 shami = ["scriptconv[phonemizers]", "regex"]
-test = ["pytest", "pytest-timeout", "pykakasi>=2.3,<3", "pypinyin>=0.50,<1", "quebra-frases", "langcodes", "regex", "ovos-number-parser", "bifonia>=0.1.1", "stressonnx>=0.0.3a2"]
+test = ["pytest", "pytest-timeout", "pykakasi>=2.3,<3", "pypinyin>=0.50,<1", "quebra-frases", "langcodes", "regex", "ovos-number-parser", "bifonia>=0.1.1", "stressonnx>=0.0.3a2", "tomli; python_version < '3.11'"]
 
 [project.urls]
 Homepage = "https://github.com/TigreGotico/scriptconv"

{scriptconv-0.0.4a20 → scriptconv-0.0.4a22}/scriptconv/phonemizers/__init__.py

@@ -10,6 +10,7 @@ from scriptconv.phonemizers.enums import Alphabet, Phonemizer
 from scriptconv.phonemizers.base import (
     BasePhonemizer,
     GraphemePhonemizer,
+    MissingLanguageError,
     UnicodeCodepointPhonemizer,
     TextChunks,
     RawPhonemizedChunks,
@@ -30,6 +31,7 @@ __all__ = [
     "Phonemizer",
     "BasePhonemizer",
     "GraphemePhonemizer",
+    "MissingLanguageError",
     "UnicodeCodepointPhonemizer",
     "TextChunks",
     "RawPhonemizedChunks",

{scriptconv-0.0.4a20 → scriptconv-0.0.4a22}/scriptconv/phonemizers/base.py

@@ -35,6 +35,16 @@ RawPhonemizedChunks = List[Tuple[str, str, bool]]
 PhonemizedChunks = list[list[str]]
 
 
+class MissingLanguageError(ValueError):
+    """No language reached :meth:`BasePhonemizer.match_lang` at all.
+
+    Raised for ``None``/``""``/``"und"`` targets, which is a distinct failure
+    from an unsupported-but-present tag: it means whatever called into the
+    phonemizer never determined a language for the text, an upstream
+    config/data defect rather than a phonemizer capability gap.
+    """
+
+
 def _primary_subtag(lang: str) -> str:
     """Lowercase, ``_``→``-`` normalized primary language subtag.
 
@@ -165,8 +175,13 @@ class BasePhonemizer(metaclass=abc.ABCMeta):
             str: The validated language code.
 
         Raises:
+            MissingLanguageError: If no language reached the matcher at all
+                (``None``/``""``/``"und"``).
             ValueError: If the language code is unsupported.
         """
+        if target_lang in (None, "", "und"):
+            raise MissingLanguageError(
+                f"no language provided to phonemize: {target_lang!r}")
         lang, score = BasePhonemizer._match_lang(target_lang, valid_langs)
         if score > 10:
             # raise an error for unsupported language

{scriptconv-0.0.4a20 → scriptconv-0.0.4a22}/scriptconv/phonemizers/he.py

@@ -10,7 +10,7 @@ class PhonikudPhonemizer(BasePhonemizer):
         except ImportError as e:
             raise ImportError(
                 "phonikud is required for the Hebrew phonemizer. "
-                "Install it with 'pip install phonikud-onnx' "
+                "Install it with 'pip install phonikud' "
                 "(or 'pip install scriptconv[he]')."
             ) from e
         self.g2p = phonemize

{scriptconv-0.0.4a20 → scriptconv-0.0.4a22}/scriptconv/phonemizers/mul.py

@@ -325,6 +325,19 @@ class EspeakPhonemizer(BasePhonemizer):
         """
         if target_lang.lower() == "en-gb":
             return "en-gb-x-rp"
+        # espeak-ng has no bare "zh"/region-tagged Chinese codes, only "cmn" (Mandarin,
+        # any script) and "yue" (Cantonese); map BCP-47 Chinese tags onto those voices
+        # directly instead of falling through to tag_distance, which scores them below
+        # the match threshold and raises. Hong Kong/Macau are the only regions where
+        # Cantonese, not Mandarin, is the spoken vernacular.
+        ZH_ALIASES = {
+            "zh": "cmn", "zh-cn": "cmn", "zh-sg": "cmn",
+            "zh-hans": "cmn", "zh-hans-cn": "cmn", "zh-hans-sg": "cmn",
+            "zh-tw": "cmn", "zh-hant": "cmn", "zh-hant-tw": "cmn",
+            "zh-hk": "yue", "zh-mo": "yue",
+        }
+        if target_lang.lower() in ZH_ALIASES:
+            return ZH_ALIASES[target_lang.lower()]
         if target_lang in cls.ESPEAK_LANGS:
             return target_lang
         if target_lang.lower().split("-")[0] in cls.ESPEAK_LANGS:

{scriptconv-0.0.4a20 → scriptconv-0.0.4a22}/scriptconv/version.py

@@ -2,7 +2,7 @@
 VERSION_MAJOR = 0
 VERSION_MINOR = 0
 VERSION_BUILD = 4
-VERSION_ALPHA = 20
+VERSION_ALPHA = 22
 # END_VERSION_BLOCK
 
 VERSION_STR = f"{VERSION_MAJOR}.{VERSION_MINOR}.{VERSION_BUILD}"

{scriptconv-0.0.4a20 → scriptconv-0.0.4a22/scriptconv.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: scriptconv
-Version: 0.0.4a20
+Version: 0.0.4a22
 Summary: Zero-dependency script & phoneme-notation core — ISO-15924 detection & metadata, IPA↔ARPABET/X-SAMPA/Lexique/Kirshenbaum/Cotovía/RFE, Buckwalter↔Arabic, Hangul→jamo, kana
 License: Apache-2.0
 Project-URL: Homepage, https://github.com/TigreGotico/scriptconv
@@ -69,6 +69,7 @@ Requires-Dist: pycotovia; extra == "gl"
 Provides-Extra: he
 Requires-Dist: scriptconv[phonemizers]; extra == "he"
 Requires-Dist: phonikud-onnx; extra == "he"
+Requires-Dist: phonikud; extra == "he"
 Provides-Extra: fa
 Requires-Dist: scriptconv[phonemizers]; extra == "fa"
 Requires-Dist: persian_phonemizer; extra == "fa"
@@ -124,6 +125,7 @@ Requires-Dist: regex; extra == "test"
 Requires-Dist: ovos-number-parser; extra == "test"
 Requires-Dist: bifonia>=0.1.1; extra == "test"
 Requires-Dist: stressonnx>=0.0.3a2; extra == "test"
+Requires-Dist: tomli; python_version < "3.11" and extra == "test"
 Dynamic: license-file
 
 <div align="center">

{scriptconv-0.0.4a20 → scriptconv-0.0.4a22}/scriptconv.egg-info/SOURCES.txt

@@ -498,6 +498,7 @@ tests/test_diacritics_graph.py
 tests/test_errors_policy.py
 tests/test_examples.py
 tests/test_graph.py
+tests/test_he_extra_declares_phonikud.py
 tests/test_notation.py
 tests/test_phonemizers_africa.py
 tests/test_phonemizers_base.py

{scriptconv-0.0.4a20 → scriptconv-0.0.4a22}/scriptconv.egg-info/requires.txt

@@ -48,6 +48,7 @@ gruut
 [he]
 scriptconv[phonemizers]
 phonikud-onnx
+phonikud
 
 [ja]
 pykakasi<3,>=2.3
@@ -112,6 +113,9 @@ ovos-number-parser
 bifonia>=0.1.1
 stressonnx>=0.0.3a2
 
+[test:python_version < "3.11"]
+tomli
+
 [transphone]
 scriptconv[phonemizers]
 transphone

scriptconv-0.0.4a22/tests/test_he_extra_declares_phonikud.py

@@ -0,0 +1,50 @@
+"""Pins the ``he`` extra to actually provide the modules the Hebrew
+phonemizer/diacritizer import.
+
+``scriptconv/phonemizers/he.py`` does ``from phonikud import phonemize`` --
+that is the ``phonikud`` PyPI distribution, a separate package from
+``phonikud-onnx`` (which ``scriptconv/diacritics.py`` uses for niqqud
+restoration via ``from phonikud_onnx import Phonikud``). Both backends are
+exercised by the Hebrew pipeline, so ``pip install scriptconv[he]`` must
+pull in both distributions -- declaring only one leaves the other import
+failing at runtime with a message that just repeats what was already
+installed.
+"""
+import re
+from pathlib import Path
+
+try:
+    import tomllib
+except ModuleNotFoundError:  # tomllib is stdlib only from Python 3.11
+    import tomli as tomllib
+
+PYPROJECT = Path(__file__).resolve().parent.parent / "pyproject.toml"
+
+
+def _he_extra_requirement_names():
+    data = tomllib.loads(PYPROJECT.read_text())
+    he = data["project"]["optional-dependencies"]["he"]
+    names = set()
+    for req in he:
+        match = re.match(r"^[A-Za-z0-9_.-]+", req)
+        if match:
+            names.add(match.group(0).lower())
+    return names
+
+
+def test_he_extra_declares_both_phonikud_distributions():
+    names = _he_extra_requirement_names()
+    # phonikud-onnx: used by diacritics.py for niqqud restoration
+    assert "phonikud-onnx" in names
+    # phonikud: used by phonemizers/he.py for `from phonikud import phonemize`
+    assert "phonikud" in names
+
+
+def test_he_py_imports_from_declared_distribution():
+    he_py = (PYPROJECT.parent / "scriptconv" / "phonemizers" / "he.py").read_text()
+    assert "from phonikud import" in he_py
+    names = _he_extra_requirement_names()
+    assert "phonikud" in names, (
+        "he.py imports the 'phonikud' package but the [he] extra does not "
+        "declare it"
+    )

{scriptconv-0.0.4a20 → scriptconv-0.0.4a22}/tests/test_phonemizers_base.py

@@ -5,6 +5,7 @@ from scriptconv.phonemizers import (
     BasePhonemizer,
     GraphemePhonemizer,
     LANG_DEFAULTS,
+    MissingLanguageError,
     PHONEMIZER_REGISTRY,
     Phonemizer,
     UnicodeCodepointPhonemizer,
@@ -81,6 +82,22 @@ class TestBaseContract(unittest.TestCase):
             BasePhonemizer.match_lang("zz-XX", ["en", "pt"])
         self.assertEqual(BasePhonemizer.match_lang("en-US", ["en", "pt"]), "en")
 
+    def test_match_lang_unsupported_tag_keeps_plain_valueerror(self):
+        # a real, present-but-unsupported tag (e.g. Tetun) is a phonemizer
+        # capability gap, not a missing-language config defect
+        with self.assertRaises(ValueError) as cm:
+            BasePhonemizer.match_lang("tet", ["en", "pt"])
+        self.assertNotIsInstance(cm.exception, MissingLanguageError)
+        self.assertIn("unsupported language code", str(cm.exception))
+
+    def test_match_lang_missing_language_raises_distinct_error(self):
+        for missing in (None, "", "und"):
+            with self.assertRaises(MissingLanguageError) as cm:
+                BasePhonemizer.match_lang(missing, ["eu"])
+            # still a ValueError so existing `except ValueError` callers work
+            self.assertIsInstance(cm.exception, ValueError)
+            self.assertNotIn("unsupported language code", str(cm.exception))
+
     def test_unicode_codepoint_nfd(self):
         u = UnicodeCodepointPhonemizer()
         self.assertEqual(len(u.phonemize_string("ã", "pt")), 2)  # a + combining
@@ -205,3 +222,29 @@ class TestModelForwarding(unittest.TestCase):
         self.assertIsInstance(g, GraphemePhonemizer)
 
 
+class TestEspeakGetLangChinese(unittest.TestCase):
+    """espeak-ng ships no bare "zh"/region-tagged Chinese voice, only "cmn"
+    and "yue" — get_lang must map BCP-47 Chinese tags onto those directly."""
+
+    def _get_lang(self):
+        from scriptconv.phonemizers.mul import EspeakPhonemizer
+        return EspeakPhonemizer.get_lang
+
+    def test_zh_region_tags_resolve_to_supported_voice(self):
+        from scriptconv.phonemizers.mul import EspeakPhonemizer
+        get_lang = self._get_lang()
+        for code in ("zh", "zh-CN", "zh-TW", "zh-HK", "zh-Hant-TW"):
+            resolved = get_lang(code)
+            self.assertIn(resolved, EspeakPhonemizer.ESPEAK_LANGS,
+                           f"{code} -> {resolved}")
+
+    def test_en_gb_special_case_unchanged(self):
+        get_lang = self._get_lang()
+        self.assertEqual(get_lang("en-gb"), "en-gb-x-rp")
+
+    def test_unsupported_code_still_raises(self):
+        get_lang = self._get_lang()
+        with self.assertRaises(ValueError):
+            get_lang("xx-not-a-real-lang")
+
+