scriptconv 0.0.4a24__tar.gz → 0.0.4a26__tar.gz

{scriptconv-0.0.4a24/scriptconv.egg-info → scriptconv-0.0.4a26}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: scriptconv
-Version: 0.0.4a24
+Version: 0.0.4a26
 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
@@ -58,7 +58,7 @@ Requires-Dist: euskaphone; extra == "eu"
 Requires-Dist: ahotts_g2p; extra == "eu"
 Provides-Extra: pt-phonemizers
 Requires-Dist: scriptconv[phonemizers]; extra == "pt-phonemizers"
-Requires-Dist: tugaphone; extra == "pt-phonemizers"
+Requires-Dist: tugaphone>=1.2.0a1; extra == "pt-phonemizers"
 Requires-Dist: g2p_barranquenho; extra == "pt-phonemizers"
 Provides-Extra: mwl
 Requires-Dist: scriptconv[phonemizers]; extra == "mwl"
@@ -125,6 +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: tugaphone>=1.2.0a1; extra == "test"
 Requires-Dist: tomli; python_version < "3.11" and extra == "test"
 Dynamic: license-file
 

{scriptconv-0.0.4a24 → scriptconv-0.0.4a26}/pyproject.toml

@@ -46,7 +46,7 @@ misaki = ["scriptconv[phonemizers]", "misaki"]
 byt5 = ["scriptconv[phonemizers]", "onnxruntime", "numpy"]
 o2i = ["scriptconv[phonemizers]", "orthography2ipa"]
 eu = ["scriptconv[phonemizers]", "euskaphone", "ahotts_g2p"]
-pt-phonemizers = ["scriptconv[phonemizers]", "tugaphone", "g2p_barranquenho"]
+pt-phonemizers = ["scriptconv[phonemizers]", "tugaphone>=1.2.0a1", "g2p_barranquenho"]
 mwl = ["scriptconv[phonemizers]", "mwl_phonemizer"]
 gl = ["scriptconv[phonemizers]", "pycotovia"]
 he = ["scriptconv[phonemizers]", "phonikud-onnx", "phonikud"]
@@ -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", "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", "tomli; python_version < '3.11'"]
 
 [project.urls]
 Homepage = "https://github.com/TigreGotico/scriptconv"

{scriptconv-0.0.4a24 → scriptconv-0.0.4a26}/scriptconv/phonemizers/base.py

@@ -196,7 +196,11 @@ class BasePhonemizer(metaclass=abc.ABCMeta):
         non-ASCII punctuation such as Arabic ``،``/``؟`` and curly quotes.
 
         Apostrophes and hyphens sandwiched between letters (e.g. "don't", "well-known")
-        are preserved so contractions and compounds aren't broken apart.
+        are preserved so contractions and compounds aren't broken apart, and any
+        punctuation character sandwiched between two digits (e.g. the ":" in a clock
+        time "16:30", the "," in "10,4", the "." in "92.073", the "-" in "1139-1185")
+        is preserved so the per-language normalizers still see it as a single token,
+        matching the digit:digit exemption in ``chunk_text``.
         """
         out = []
         chars = list(text)
@@ -205,6 +209,10 @@ class BasePhonemizer(metaclass=abc.ABCMeta):
                     and chars[i - 1].isalpha() and chars[i + 1].isalpha():
                 out.append(c)
                 continue
+            if 0 < i < len(chars) - 1 \
+                    and chars[i - 1].isdigit() and chars[i + 1].isdigit():
+                out.append(c)
+                continue
             if unicodedata.category(c).startswith("P"):
                 continue
             out.append(c)

{scriptconv-0.0.4a24 → scriptconv-0.0.4a26}/scriptconv/phonemizers/pt.py

@@ -23,6 +23,13 @@ class TugaphonePhonemizer(BasePhonemizer):
         """
         Validates and returns the closest supported language code.
 
+        A full sub-regional lect tag tugaphone itself knows about (e.g.
+        ``pt-PT-x-lisbon``, ``pt-PT-x-porto``) is passed through unchanged,
+        so its output stays distinct from the generic country lect. A loose
+        tag (``pt``, ``pt-PT``) falls back to :meth:`match_lang` against the
+        country-level lects, since ``tugaphone.resolve_lect`` silently
+        defaults an unrecognized tag to ``pt-PT`` rather than raising.
+
         Args:
             target_lang (str): The language code to validate.
 
@@ -32,12 +39,15 @@ class TugaphonePhonemizer(BasePhonemizer):
         Raises:
             ValueError: If the language code is unsupported.
         """
+        import tugaphone
+        dialects = {d.lower(): d for d in tugaphone.list_dialects()}
+        if target_lang and target_lang.replace("_", "-").lower() in dialects:
+            return dialects[target_lang.replace("_", "-").lower()]
         # this check is here only to throw an exception if invalid language is provided
         return cls.match_lang(target_lang, ["pt-PT", "pt-BR", "pt-AO", "pt-MZ", "pt-TL"])
 
     def phonemize_string(self, text: str, lang: str) -> str:
         lang = self.get_lang(lang)
-        # TODO - support regional dialects
         return self.tuga.phonemize_sentence(text, lang)
 
 

{scriptconv-0.0.4a24 → scriptconv-0.0.4a26}/scriptconv/version.py

@@ -2,7 +2,7 @@
 VERSION_MAJOR = 0
 VERSION_MINOR = 0
 VERSION_BUILD = 4
-VERSION_ALPHA = 24
+VERSION_ALPHA = 26
 # END_VERSION_BLOCK
 
 VERSION_STR = f"{VERSION_MAJOR}.{VERSION_MINOR}.{VERSION_BUILD}"

{scriptconv-0.0.4a24 → scriptconv-0.0.4a26/scriptconv.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: scriptconv
-Version: 0.0.4a24
+Version: 0.0.4a26
 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
@@ -58,7 +58,7 @@ Requires-Dist: euskaphone; extra == "eu"
 Requires-Dist: ahotts_g2p; extra == "eu"
 Provides-Extra: pt-phonemizers
 Requires-Dist: scriptconv[phonemizers]; extra == "pt-phonemizers"
-Requires-Dist: tugaphone; extra == "pt-phonemizers"
+Requires-Dist: tugaphone>=1.2.0a1; extra == "pt-phonemizers"
 Requires-Dist: g2p_barranquenho; extra == "pt-phonemizers"
 Provides-Extra: mwl
 Requires-Dist: scriptconv[phonemizers]; extra == "mwl"
@@ -125,6 +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: tugaphone>=1.2.0a1; extra == "test"
 Requires-Dist: tomli; python_version < "3.11" and extra == "test"
 Dynamic: license-file
 

{scriptconv-0.0.4a24 → scriptconv-0.0.4a26}/scriptconv.egg-info/SOURCES.txt

@@ -505,6 +505,7 @@ tests/test_phonemizers_base.py
 tests/test_phonemizers_cjk_ar.py
 tests/test_phonemizers_friendly_import_errors.py
 tests/test_phonemizers_halabi_iqra.py
+tests/test_phonemizers_pt_tugaphone.py
 tests/test_phonemizers_ru.py
 tests/test_readings.py
 tests/test_readings_zh.py

{scriptconv-0.0.4a24 → scriptconv-0.0.4a26}/scriptconv.egg-info/requires.txt

@@ -86,7 +86,7 @@ bifonia>=0.1.1
 
 [pt-phonemizers]
 scriptconv[phonemizers]
-tugaphone
+tugaphone>=1.2.0a1
 g2p_barranquenho
 
 [shami]
@@ -112,6 +112,7 @@ regex
 ovos-number-parser
 bifonia>=0.1.1
 stressonnx>=0.0.3a2
+tugaphone>=1.2.0a1
 
 [test:python_version < "3.11"]
 tomli

{scriptconv-0.0.4a24 → scriptconv-0.0.4a26}/tests/test_phonemizers_base.py

@@ -292,3 +292,39 @@ class TestChunkTextDigitColon(unittest.TestCase):
             ("12", ":", False),
             ("pontos", ".", True),
         ])
+
+
+class TestRemovePunctuationKeepsDigitInternal(unittest.TestCase):
+    """remove_punctuation must keep a punctuation character sandwiched
+    between two digits (clock times, decimals, ranges) — otherwise
+    "16:30" reaches the phonemizer as "1630" and gets read as one large
+    integer instead of an hour and a minute, matching chunk_text's
+    digit:digit exemption above."""
+
+    def test_clock_time_colon_kept(self):
+        self.assertEqual(
+            BasePhonemizer.remove_punctuation(
+                "A reunião é às 16:30, não te esqueças."),
+            "A reunião é às 16:30 não te esqueças",
+        )
+
+    def test_decimal_comma_kept(self):
+        self.assertEqual(BasePhonemizer.remove_punctuation("10,4"), "10,4")
+
+    def test_decimal_dot_kept(self):
+        self.assertEqual(
+            BasePhonemizer.remove_punctuation("92.073"), "92.073")
+
+    def test_range_hyphen_kept(self):
+        self.assertEqual(
+            BasePhonemizer.remove_punctuation("1139-1185"), "1139-1185")
+
+    def test_word_colon_still_stripped(self):
+        self.assertEqual(
+            BasePhonemizer.remove_punctuation("Nota: fim."), "Nota fim")
+
+    def test_ordinal_mark_still_stripped(self):
+        # "3.º" — the dot follows a digit but precedes a letter (the
+        # masculine ordinal indicator), so it is not digit:digit and is
+        # stripped like any other punctuation, giving "3º".
+        self.assertEqual(BasePhonemizer.remove_punctuation("3.º"), "3º")

scriptconv-0.0.4a26/tests/test_phonemizers_pt_tugaphone.py

@@ -0,0 +1,44 @@
+"""Tests for TugaphonePhonemizer sub-regional lect resolution.
+
+A sub-regional lect tag such as ``pt-PT-x-lisbon`` or ``pt-PT-x-porto`` must
+resolve to itself, and must phonemize distinctly from the generic country
+lect and from every other sub-regional lect. A loose tag like ``pt`` must
+still resolve to a default supported lect, and a non-Portuguese tag must be
+rejected.
+"""
+import unittest
+
+from scriptconv.phonemizers.pt import TugaphonePhonemizer
+
+
+class TestTugaphoneSubregionalLects(unittest.TestCase):
+
+    def setUp(self):
+        self.pho = TugaphonePhonemizer()
+
+    def test_subregional_lects_pass_through(self):
+        self.assertEqual(TugaphonePhonemizer.get_lang("pt-PT-x-lisbon"), "pt-PT-x-lisbon")
+        self.assertEqual(TugaphonePhonemizer.get_lang("pt-PT-x-porto"), "pt-PT-x-porto")
+
+    def test_subregional_lects_produce_distinct_output(self):
+        text = "O vinho verde"
+        generic = self.pho.phonemize_string(text, "pt-PT")
+        lisbon = self.pho.phonemize_string(text, "pt-PT-x-lisbon")
+        porto = self.pho.phonemize_string(text, "pt-PT-x-porto")
+        self.assertNotEqual(generic, lisbon)
+        self.assertNotEqual(generic, porto)
+        self.assertNotEqual(lisbon, porto)
+
+    def test_loose_tag_still_resolves(self):
+        # "pt" has no country/lect of its own; it must still resolve to a
+        # default supported lect instead of raising.
+        resolved = TugaphonePhonemizer.get_lang("pt")
+        self.assertIn(resolved, ["pt-PT", "pt-BR", "pt-AO", "pt-MZ", "pt-TL"])
+
+    def test_non_portuguese_still_rejected(self):
+        with self.assertRaises(ValueError):
+            TugaphonePhonemizer.get_lang("es")
+
+
+if __name__ == "__main__":
+    unittest.main()