scriptconv 0.0.3a13__tar.gz → 0.0.3a15__tar.gz

{scriptconv-0.0.3a13 → scriptconv-0.0.3a15}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: scriptconv
-Version: 0.0.3a13
+Version: 0.0.3a15
 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

{scriptconv-0.0.3a13 → scriptconv-0.0.3a15}/scriptconv/__main__.py

@@ -12,6 +12,7 @@ Usage
     python -m scriptconv restyle <convention> <style> <text>
     python -m scriptconv conventions [--script CODE]
     python -m scriptconv route <src> <dst>
+    python -m scriptconv phonemize <lang> <text> [--phonemizer ID] [--alphabet A]
 
 Examples
 --------
@@ -69,6 +70,12 @@ def main(argv: list[str] | None = None) -> int:
     rt.add_argument("src", help="Source representation id")
     rt.add_argument("dst", help="Target representation id")
 
+    ph = sub.add_parser("phonemize", help="Phonemize text (per-language default backend)")
+    ph.add_argument("lang", help="Language code")
+    ph.add_argument("text", help="Text to phonemize")
+    ph.add_argument("--phonemizer", default=None, help="Backend override (Phonemizer value)")
+    ph.add_argument("--alphabet", default="ipa", help="Output alphabet (default ipa)")
+
     args = p.parse_args(argv)
 
     if args.command is None:
@@ -86,6 +93,7 @@ def main(argv: list[str] | None = None) -> int:
         "restyle": lambda: _do_restyle(args),
         "conventions": lambda: _do_conventions(args),
         "route": lambda: _do_route(args),
+        "phonemize": lambda: _do_phonemize(args),
     }
     dispatch[args.command]()
     return 0
@@ -108,6 +116,16 @@ def _do_convert(args: argparse.Namespace) -> None:
                          f"graph nodes: {', '.join(DEFAULT_GRAPH.nodes)})")
 
 
+def _do_phonemize(args: argparse.Namespace) -> None:
+    from scriptconv.phonemizers import Alphabet, Phonemizer, phonemize
+    try:
+        override = Phonemizer(args.phonemizer) if args.phonemizer else None
+        print(phonemize(args.text, args.lang, Alphabet(args.alphabet),
+                        override=override))
+    except (ValueError, ImportError) as e:
+        raise SystemExit(f"error: {e}")
+
+
 def _do_route(args: argparse.Namespace) -> None:
     from scriptconv.graph import DEFAULT_GRAPH
     try:

{scriptconv-0.0.3a13 → scriptconv-0.0.3a15}/scriptconv/phonemizers/__init__.py

@@ -21,6 +21,8 @@ from scriptconv.phonemizers.registry import (
     get_phonemizer,
     get_phonemizer_class,
     phonemizer_for_lang,
+    phonemize,
+    register,
 )
 
 __all__ = [
@@ -37,4 +39,6 @@ __all__ = [
     "get_phonemizer",
     "get_phonemizer_class",
     "phonemizer_for_lang",
+    "phonemize",
+    "register",
 ]

{scriptconv-0.0.3a13 → scriptconv-0.0.3a15}/scriptconv/phonemizers/base.py

@@ -45,8 +45,9 @@ class BasePhonemizer(metaclass=abc.ABCMeta):
         # optional (text, lang) -> str hook run before chunking; see module
         # docstring — scriptconv performs no normalization of its own
         self.normalizer = normalizer
-        # local path to a phonikud ONNX model (Hebrew diacritization);
-        # scriptconv never downloads — the consumer resolves the file
+        # local path to a phonikud ONNX model (Hebrew diacritization), or a
+        # zero-arg callable resolving one lazily; scriptconv never downloads
+        # — the consumer resolves the file
         self.phonikud_model = phonikud_model
 
         # diacritizer model name, for languages that need one. Arabic uses
@@ -59,7 +60,9 @@ class BasePhonemizer(metaclass=abc.ABCMeta):
     @property
     def phonikud(self):
         if self._phonikud is None:
-            if not self.phonikud_model:
+            model = self.phonikud_model() if callable(self.phonikud_model) \
+                else self.phonikud_model
+            if not model:
                 raise ValueError(
                     "Hebrew diacritization needs a local phonikud ONNX model: "
                     "pass phonikud_model=<path> (scriptconv never downloads "
@@ -70,7 +73,7 @@ class BasePhonemizer(metaclass=abc.ABCMeta):
                 raise ImportError(
                     "Hebrew diacritization needs phonikud-onnx — install "
                     "with `pip install scriptconv[he]`") from None
-            self._phonikud = Phonikud(self.phonikud_model)
+            self._phonikud = Phonikud(model)
         return self._phonikud
 
     def tashkeel(self, model: Optional[str] = None):

{scriptconv-0.0.3a13 → scriptconv-0.0.3a15}/scriptconv/phonemizers/registry.py

@@ -158,3 +158,42 @@ def phonemizer_for_lang(lang: str, alphabet: Alphabet = Alphabet.IPA,
         if alphabet in _EMITS.get(candidate, (Alphabet.IPA,)):
             return get_phonemizer(candidate, alphabet, model, **kwargs)
     return get_phonemizer(_P.ESPEAK, alphabet, model, **kwargs)
+
+
+def phonemize(text: str, lang: str, alphabet: Alphabet = Alphabet.IPA,
+              override: Optional[Phonemizer] = None,
+              model: Optional[str] = None, **kwargs) -> str:
+    """One-call facade: phonemize *text* with the language's default backend.
+
+    Equivalent to ``phonemizer_for_lang(...).phonemize_string(text, lang)``.
+    Construct the phonemizer yourself when phonemizing repeatedly — backends
+    cache models/dictionaries per instance.
+    """
+    return phonemizer_for_lang(lang, alphabet, override, model,
+                               **kwargs).phonemize_string(text, lang)
+
+
+def register(graph) -> None:
+    """Opt-in graph integration: add the phonemization edge to *graph*.
+
+    Adds the ``"text"`` representation — meaningful only with ``lang=``
+    routing context, and present only in graphs that opted in (the
+    :data:`scriptconv.graph.DEFAULT_GRAPH` stays orthography-only by
+    design) — and one dispatching ``text -> ipa`` edge that resolves the
+    per-language default (honouring an ``override=`` context key).
+
+    Usage::
+
+        from scriptconv.graph import DEFAULT_GRAPH
+        from scriptconv import phonemizers
+        g = DEFAULT_GRAPH.extend(phonemizers.register)
+        g.convert("bom dia", "text", "ipa", lang="pt")
+    """
+    from scriptconv.graph import Edge
+
+    def _text_to_ipa(text: str, lang: str = "und",
+                     override: Optional[Phonemizer] = None, **_):
+        return phonemizer_for_lang(lang, Alphabet.IPA,
+                                   override).phonemize_string(text, lang)
+
+    graph.register(Edge("text", "ipa", _text_to_ipa, lossless=False))

{scriptconv-0.0.3a13 → scriptconv-0.0.3a15}/scriptconv/version.py

@@ -2,7 +2,7 @@
 VERSION_MAJOR = 0
 VERSION_MINOR = 0
 VERSION_BUILD = 3
-VERSION_ALPHA = 13
+VERSION_ALPHA = 15
 # END_VERSION_BLOCK
 
 VERSION_STR = f"{VERSION_MAJOR}.{VERSION_MINOR}.{VERSION_BUILD}"

{scriptconv-0.0.3a13 → scriptconv-0.0.3a15}/scriptconv.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: scriptconv
-Version: 0.0.3a13
+Version: 0.0.3a15
 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

{scriptconv-0.0.3a13 → scriptconv-0.0.3a15}/tests/test_phonemizers_base.py

@@ -130,3 +130,47 @@ class TestByT5NoNetwork(unittest.TestCase):
 
 if __name__ == "__main__":
     unittest.main()
+
+
+class TestFacadeAndGraph(unittest.TestCase):
+    def test_phonemize_facade(self):
+        from scriptconv.phonemizers import phonemize
+        out = phonemize("hola", "es", override=Phonemizer.GRAPHEMES)
+        self.assertEqual(out, "hola")
+
+    def test_register_is_opt_in_default_graph_untouched(self):
+        from scriptconv.graph import DEFAULT_GRAPH
+        from scriptconv import phonemizers
+        self.assertNotIn("text", DEFAULT_GRAPH.nodes)
+        g = DEFAULT_GRAPH.extend(phonemizers.register)
+        self.assertIn("text", g.nodes)
+        self.assertNotIn("text", DEFAULT_GRAPH.nodes)
+
+    def test_graph_edge_dispatches_with_override(self):
+        from scriptconv.graph import DEFAULT_GRAPH
+        from scriptconv import phonemizers
+        g = DEFAULT_GRAPH.extend(phonemizers.register)
+        out = g.convert("abc", "text", "ipa", lang="en",
+                        override=Phonemizer.GRAPHEMES)
+        self.assertEqual(out, "abc")
+
+    def test_graph_edge_chains_to_arpa(self):
+        from scriptconv.graph import DEFAULT_GRAPH
+        from scriptconv import phonemizers
+        g = DEFAULT_GRAPH.extend(phonemizers.register)
+        self.assertTrue(g.can_convert("text", "arpa"))
+
+
+class TestPhonikudModelResolver(unittest.TestCase):
+    def test_callable_resolver_invoked_lazily(self):
+        calls = []
+
+        def resolver():
+            calls.append(1)
+            return ""  # resolves to nothing -> still the explicit ValueError
+
+        g = GraphemePhonemizer(phonikud_model=resolver)
+        self.assertEqual(calls, [])  # not resolved at construction
+        with self.assertRaises(ValueError):
+            g.add_diacritics("שלום", "he")
+        self.assertEqual(calls, [1])