scriptconv 0.0.4a28__tar.gz → 0.0.4a30__tar.gz

{scriptconv-0.0.4a28/scriptconv.egg-info → scriptconv-0.0.4a30}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: scriptconv
-Version: 0.0.4a28
+Version: 0.0.4a30
 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
@@ -63,6 +63,9 @@ Requires-Dist: g2p_barranquenho; extra == "pt-phonemizers"
 Provides-Extra: mwl
 Requires-Dist: scriptconv[phonemizers]; extra == "mwl"
 Requires-Dist: mwl_phonemizer; extra == "mwl"
+Provides-Extra: ghana
+Requires-Dist: scriptconv[phonemizers]; extra == "ghana"
+Requires-Dist: ghana-g2p>=0.1.1; extra == "ghana"
 Provides-Extra: gl
 Requires-Dist: scriptconv[phonemizers]; extra == "gl"
 Requires-Dist: pycotovia; extra == "gl"
@@ -126,6 +129,7 @@ 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: tugaphone>=1.2.0a1; extra == "test"
+Requires-Dist: ghana-g2p>=0.1.1; extra == "test"
 Requires-Dist: tomli; python_version < "3.11" and extra == "test"
 Dynamic: license-file
 

{scriptconv-0.0.4a28 → scriptconv-0.0.4a30}/pyproject.toml

@@ -48,6 +48,7 @@ o2i = ["scriptconv[phonemizers]", "orthography2ipa"]
 eu = ["scriptconv[phonemizers]", "euskaphone", "ahotts_g2p"]
 pt-phonemizers = ["scriptconv[phonemizers]", "tugaphone>=1.2.0a1", "g2p_barranquenho"]
 mwl = ["scriptconv[phonemizers]", "mwl_phonemizer"]
+ghana = ["scriptconv[phonemizers]", "ghana-g2p>=0.1.1"]
 gl = ["scriptconv[phonemizers]", "pycotovia"]
 he = ["scriptconv[phonemizers]", "phonikud-onnx", "phonikud"]
 fa = ["scriptconv[phonemizers]", "persian_phonemizer"]
@@ -61,7 +62,7 @@ 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", "tugaphone>=1.2.0a1", "tomli; python_version < '3.11'"]
+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", "tugaphone>=1.2.0a1", "ghana-g2p>=0.1.1", "tomli; python_version < '3.11'"]
 
 [project.urls]
 Homepage = "https://github.com/TigreGotico/scriptconv"

{scriptconv-0.0.4a28 → scriptconv-0.0.4a30}/scriptconv/phonemizers/enums.py

@@ -84,3 +84,4 @@ class Phonemizer(str, Enum):
     BARRANQUENHO = "barranquenho"  # barranquenho (pt/es contact variety; o2i lattice)
     ORTHOGRAPHY2IPA = "orthography2ipa"  # multilingual data-driven IPA (o2i lattice)
     AFRICA_G2P = "africa_g2p"  # rule-based G2P for 400+ African languages (native units or IPA)
+    GHANA_G2P = "ghana_g2p"  # africa-g2p rules patched for 42 Ghanaian languages (native units or IPA)

scriptconv-0.0.4a30/scriptconv/phonemizers/ghana.py

@@ -0,0 +1,87 @@
+"""ghana-g2p-backed phonemizer.
+
+Wraps ``ghana-g2p`` (GhanaNLP, https://github.com/GhanaNLP/ghana-g2p,
+Apache-2.0), a grapheme-to-phoneme library for 42 Ghanaian languages. It is
+built on africa-g2p's rule tables and adds three things for Ghanaian text:
+Unicode codepoint normalisation (for example ``ʊ`` U+028A written for ``ʋ``
+U+028B), a patch table of missing or corrected letters, and donor rule sets
+for languages africa-g2p has no table for. Every result records whether its
+rules are the language's own, an equivalent code, or a related donor.
+
+Like :class:`~scriptconv.phonemizers.africa.AfricaG2PPhonemizer`, it emits
+IPA (:attr:`Alphabet.IPA`) or native-orthography phoneme units
+(:attr:`Alphabet.AFRICA_G2P`).
+
+``ghana-g2p`` depends on the PyPI ``africa-g2p`` package and loads THAT
+installed copy, not scriptconv's vendored ``africa_g2p``. The two can be
+different versions, so this backend and ``AfricaG2PPhonemizer`` can give
+different output for the same language.
+
+Optional extra: ``pip install scriptconv[ghana]``.
+"""
+from typing import Dict, List
+
+from scriptconv.phonemizers.base import BasePhonemizer, _check_alphabet, _primary_subtag
+from scriptconv.phonemizers.enums import Alphabet
+
+__all__ = ["GhanaG2PPhonemizer"]
+
+
+def _ghana_g2p():
+    """Import ``ghana_g2p`` lazily, naming the extra when it is missing."""
+    try:
+        import ghana_g2p
+    except ImportError as e:
+        raise ImportError(
+            "ghana-g2p is required for the Ghanaian phonemizer. Install it "
+            "with 'pip install ghana-g2p' (or 'pip install scriptconv[ghana]')."
+        ) from e
+    return ghana_g2p
+
+
+class GhanaG2PPhonemizer(BasePhonemizer):
+    """
+    Rule-based G2P for 42 Ghanaian languages, backed by ``ghana-g2p``.
+
+    Languages are ghana-g2p's own ISO 639-3 codes (``twi``, ``ewe``, ``dag``,
+    ``gaa``, ``hau``, ...), read at runtime from ``ghana_g2p.languages()``.
+    One engine per language is created on first use and cached.
+    """
+
+    def __init__(self, alphabet: Alphabet = Alphabet.IPA):
+        _check_alphabet(self, alphabet, [Alphabet.IPA, Alphabet.AFRICA_G2P])
+        super().__init__(alphabet=alphabet)
+        self._cache: Dict[str, object] = {}
+
+    def _engine(self, resolved_lang: str):
+        """Return, lazily creating and caching, the engine for *resolved_lang*."""
+        if resolved_lang not in self._cache:
+            self._cache[resolved_lang] = _ghana_g2p().GhanaG2P(resolved_lang)
+        return self._cache[resolved_lang]
+
+    @classmethod
+    def supported_langs(cls) -> List[str]:
+        """Return every ISO 639-3 code ghana-g2p lists."""
+        return [entry["code"] for entry in _ghana_g2p().languages()]
+
+    @classmethod
+    def get_lang(cls, target_lang: str) -> str:
+        """
+        Resolve *target_lang* to a ghana-g2p ISO 639-3 code.
+
+        Only the primary subtag is matched, exactly (``twi-GH`` -> ``twi``).
+        ISO 639-1 tags are not mapped (``ee`` is not ``ewe``).
+
+        Raises:
+            ValueError: If ghana-g2p has no such language.
+        """
+        key = _primary_subtag(target_lang)
+        if key in cls.supported_langs():
+            return key
+        raise ValueError(f"ghana-g2p: unsupported language {target_lang!r}")
+
+    def phonemize_string(self, text: str, lang: str) -> str:
+        engine = self._engine(self.get_lang(lang))
+        if self.alphabet == Alphabet.IPA:
+            return engine.ipa(text, sep=" ")
+        return engine.grapheme(text, sep=" ")

{scriptconv-0.0.4a28 → scriptconv-0.0.4a30}/scriptconv/phonemizers/registry.py

@@ -58,6 +58,7 @@ PHONEMIZER_REGISTRY: Dict[Phonemizer, Tuple[str, str, Optional[str]]] = {
     _P.ORTHOGRAPHY2IPA: (f"{_BASE}.o2ipa", "Orthography2IPAPhonemizer", "o2i"),
     # vendored (not an optional extra) -- see _vendored/africa_g2p
     _P.AFRICA_G2P: (f"{_BASE}.africa", "AfricaG2PPhonemizer", None),
+    _P.GHANA_G2P: (f"{_BASE}.ghana", "GhanaG2PPhonemizer", "ghana"),
     # ported in the CJK/AR stage; registered here so no member is silently
     # unmapped — resolution raises ImportError naming the pending extra
     _P.OPENJTALK: (f"{_BASE}.ja", "OpenJTaklPhonemizer", "ja-phonemizers"),

{scriptconv-0.0.4a28 → scriptconv-0.0.4a30}/scriptconv/version.py

@@ -2,7 +2,7 @@
 VERSION_MAJOR = 0
 VERSION_MINOR = 0
 VERSION_BUILD = 4
-VERSION_ALPHA = 28
+VERSION_ALPHA = 30
 # END_VERSION_BLOCK
 
 VERSION_STR = f"{VERSION_MAJOR}.{VERSION_MINOR}.{VERSION_BUILD}"

{scriptconv-0.0.4a28 → scriptconv-0.0.4a30/scriptconv.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: scriptconv
-Version: 0.0.4a28
+Version: 0.0.4a30
 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
@@ -63,6 +63,9 @@ Requires-Dist: g2p_barranquenho; extra == "pt-phonemizers"
 Provides-Extra: mwl
 Requires-Dist: scriptconv[phonemizers]; extra == "mwl"
 Requires-Dist: mwl_phonemizer; extra == "mwl"
+Provides-Extra: ghana
+Requires-Dist: scriptconv[phonemizers]; extra == "ghana"
+Requires-Dist: ghana-g2p>=0.1.1; extra == "ghana"
 Provides-Extra: gl
 Requires-Dist: scriptconv[phonemizers]; extra == "gl"
 Requires-Dist: pycotovia; extra == "gl"
@@ -126,6 +129,7 @@ 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: tugaphone>=1.2.0a1; extra == "test"
+Requires-Dist: ghana-g2p>=0.1.1; extra == "test"
 Requires-Dist: tomli; python_version < "3.11" and extra == "test"
 Dynamic: license-file
 

{scriptconv-0.0.4a28 → scriptconv-0.0.4a30}/scriptconv.egg-info/SOURCES.txt

@@ -29,6 +29,7 @@ scriptconv/phonemizers/en.py
 scriptconv/phonemizers/enums.py
 scriptconv/phonemizers/eu.py
 scriptconv/phonemizers/fa.py
+scriptconv/phonemizers/ghana.py
 scriptconv/phonemizers/gl.py
 scriptconv/phonemizers/he.py
 scriptconv/phonemizers/ja.py
@@ -508,6 +509,7 @@ tests/test_phonemizers_africa.py
 tests/test_phonemizers_base.py
 tests/test_phonemizers_cjk_ar.py
 tests/test_phonemizers_friendly_import_errors.py
+tests/test_phonemizers_ghana.py
 tests/test_phonemizers_halabi_iqra.py
 tests/test_phonemizers_pt_tugaphone.py
 tests/test_phonemizers_ru.py

{scriptconv-0.0.4a28 → scriptconv-0.0.4a30}/scriptconv.egg-info/requires.txt

@@ -33,6 +33,10 @@ ahotts_g2p
 scriptconv[phonemizers]
 persian_phonemizer
 
+[ghana]
+scriptconv[phonemizers]
+ghana-g2p>=0.1.1
+
 [gl]
 scriptconv[phonemizers]
 pycotovia
@@ -113,6 +117,7 @@ ovos-number-parser
 bifonia>=0.1.1
 stressonnx>=0.0.3a2
 tugaphone>=1.2.0a1
+ghana-g2p>=0.1.1
 
 [test:python_version < "3.11"]
 tomli

{scriptconv-0.0.4a28 → scriptconv-0.0.4a30}/tests/test_arpa_stress.py

@@ -48,9 +48,11 @@ class TestRoundTrip(unittest.TestCase):
         spelling must itself be IPA-stable.  The overwhelming majority
         round-trip byte-exactly."""
         random.seed(20260721)
-        cons = [c for c in _ARPA_BASE
-                if c not in _ARPA_VOWELS and c not in ("AX", "AXR")]
-        vows = [v for v in _ARPA_VOWELS if v not in ("AX", "AXR")]
+        # sorted: _ARPA_VOWELS is a set, and set order follows
+        # PYTHONHASHSEED, so without sorting the seeded draw differs per run
+        cons = sorted(c for c in _ARPA_BASE
+                      if c not in _ARPA_VOWELS and c not in ("AX", "AXR"))
+        vows = sorted(v for v in _ARPA_VOWELS if v not in ("AX", "AXR"))
         exact = fused = 0
         for _ in range(400):
             seq = []
@@ -66,8 +68,10 @@ class TestRoundTrip(unittest.TestCase):
                 exact += 1
             else:
                 fused += 1
-                # fusion residue must be IPA-stable
-                self.assertEqual(arpa_to_ipa(rt, stress=True), ipa, s)
+                # fusion residue must be IPA-stable, up to the one lossy
+                # alias: British GOAT əʊ is read as OW, which writes oʊ
+                self.assertEqual(arpa_to_ipa(rt, stress=True),
+                                 ipa.replace("əʊ", "oʊ"), s)
         self.assertGreater(exact, 350)
         self.assertLess(fused, 50)
 
@@ -82,6 +86,11 @@ class TestRoundTrip(unittest.TestCase):
         self.assertEqual(ipa_to_arpa(arpa_to_ipa("T SH")), "CH")
         self.assertEqual(arpa_to_ipa(fused, stress=True),
                          arpa_to_ipa("AH0 R", stress=True))
+        # schwa + UH is contiguous əʊ, the British GOAT alias for OW. This
+        # residue is lossy: OW writes back as oʊ, not əʊ.
+        goat = ipa_to_arpa(arpa_to_ipa("AH0 UH0", stress=True), stress=True)
+        self.assertEqual(goat, "OW0")
+        self.assertEqual(arpa_to_ipa(goat, stress=True), "oʊ")
 
 
 if __name__ == "__main__":

{scriptconv-0.0.4a28 → scriptconv-0.0.4a30}/tests/test_phonemizers_base.py

@@ -25,7 +25,8 @@ class TestEnums(unittest.TestCase):
         self.assertEqual(Phonemizer.VOSK.value, "vosk")
         self.assertEqual(Alphabet.XSAMPA.value, "x-sampa")
         self.assertEqual(Alphabet.VOSK.value, "vosk")
-        self.assertEqual(len(list(Phonemizer)), 45)  # +HALABI, +IQRA
+        self.assertEqual(Phonemizer.GHANA_G2P.value, "ghana_g2p")
+        self.assertEqual(len(list(Phonemizer)), 46)  # +HALABI, +IQRA, +GHANA_G2P
         self.assertEqual(len(list(Alphabet)), 23)  # incl. HALABI, VOSK, AFRICA_G2P
 
 

scriptconv-0.0.4a30/tests/test_phonemizers_ghana.py

@@ -0,0 +1,56 @@
+"""Tests for the ghana-g2p-backed phonemizer.
+
+ghana-g2p is in the ``test`` extra, so these tests call the real library.
+The expected strings are ghana-g2p 0.1.1 output. ``Akwaaba`` / ``a kʷ a a b a``
+is also the example in ghana-g2p's own README.
+"""
+import unittest
+
+from scriptconv.phonemizers import Alphabet, Phonemizer, get_phonemizer_class
+from scriptconv.phonemizers.ghana import GhanaG2PPhonemizer
+
+
+class TestGhanaG2PPhonemizer(unittest.TestCase):
+    def test_twi_ipa(self):
+        p = GhanaG2PPhonemizer(alphabet=Alphabet.IPA)
+        self.assertEqual(p.phonemize_string("Akwaaba", "twi"), "a kʷ a a b a")
+
+    def test_twi_native_units(self):
+        p = GhanaG2PPhonemizer(alphabet=Alphabet.AFRICA_G2P)
+        self.assertEqual(p.phonemize_string("Akwaaba", "twi"), "a kw a a b a")
+
+    def test_ewe_y_is_the_palatal_glide(self):
+        # ghana-g2p corrects africa-g2p's Ewe <y>, once IPA y, to /j/.
+        p = GhanaG2PPhonemizer()
+        self.assertEqual(p.phonemize_string("yaa", "ewe"), "j a a")
+
+    def test_dagbani_long_vowel(self):
+        p = GhanaG2PPhonemizer()
+        self.assertEqual(p.phonemize_string("Naawuni", "dag"), "n aː w u n i")
+
+    def test_supported_langs_are_the_42_ghanaian_codes(self):
+        langs = GhanaG2PPhonemizer.supported_langs()
+        self.assertEqual(len(langs), 42)
+        for code in ("twi", "ewe", "gaa", "dag", "hau"):
+            self.assertIn(code, langs)
+
+    def test_get_lang_strips_region_only(self):
+        self.assertEqual(GhanaG2PPhonemizer.get_lang("twi-GH"), "twi")
+        with self.assertRaises(ValueError):
+            GhanaG2PPhonemizer.get_lang("ee")
+
+    def test_engine_is_cached_per_language(self):
+        p = GhanaG2PPhonemizer()
+        first = p._engine("twi")
+        self.assertIs(p._engine("twi"), first)
+
+    def test_rejects_unsupported_alphabet(self):
+        with self.assertRaises(ValueError):
+            GhanaG2PPhonemizer(alphabet=Alphabet.VOSK)
+
+    def test_registry_resolves_the_member(self):
+        self.assertIs(get_phonemizer_class(Phonemizer.GHANA_G2P), GhanaG2PPhonemizer)
+
+
+if __name__ == "__main__":
+    unittest.main()