Move language attribute to PhonologicalSymbol.Inventory and update th…

…e language of Graphemes from their inventory.

PiperOrigin-RevId: 730459942

nisaba/scripts/natural_translit/language_params/en.py

@@ -19,11 +19,14 @@
 from nisaba.scripts.natural_translit.script.inventories import latn as l
 
 
+LANGUAGE = g.Grapheme.LANGUAGE.en
+
+
 def _latn_inventory() -> g.Grapheme.Inventory:
   """Builds a grapheme inventory for English."""
   latn = l.LATN
   ph = x_uni.PHONEMES
-  gr = g.Grapheme.Inventory(g.Grapheme.GR_FEATURES.script.latn, 'en')
+  gr = g.Grapheme.Inventory(g.Grapheme.GR_FEATURES.script.latn, LANGUAGE)
   lowercase = [
       latn.a,
       latn.b,

nisaba/scripts/natural_translit/phonology/phonological_symbol.py

@@ -70,8 +70,14 @@ def has_feature(self, value: ft.Feature.Aspect.VALUES) -> bool:
   class Inventory(sym.Symbol.Inventory):
     """Phonological symbol inventory."""
 
-    def __init__(self, alias: str, typed: ty.TypeOrNothing = ty.UNSPECIFIED):
+    def __init__(
+        self,
+        alias,
+        language: lang.Language.OR_NOTHING = ty.UNSPECIFIED,
+        typed: ty.TypeOrNothing = ty.UNSPECIFIED,
+    ):
       super().__init__(alias, typed=ty.type_check(typed, PhonologicalSymbol))
+      self.language = ty.type_check(language, Phon.LANGUAGE.x_uni)
       self.atomics = i.Inventory()
 
     def _add_symbol_and_atomic(self, symbol: PhonologicalSymbol) -> bool:
@@ -166,8 +172,7 @@ class Inventory(PhonologicalSymbol.Inventory):
 
     def __init__(self, language: lang.Language.OR_NOTHING = ty.UNSPECIFIED):
       language = ty.type_check(language, Phon.LANGUAGE.x_uni)
-      super().__init__(language.alias, typed=Phon)
-      self.language = language
+      super().__init__(language.alias, language, Phon)
 
     def _add_phoneme(self, phoneme: Phon) -> bool:
       """Adds a phoneme to the inventory."""

nisaba/scripts/natural_translit/script/BUILD.bazel

@@ -45,8 +45,8 @@ py_library(
     srcs = ["grapheme.py"],
     deps = [
         "//nisaba/scripts/natural_translit/phonology:phonological_symbol",
+        "//nisaba/scripts/natural_translit/phonology/features:language",
         "//nisaba/scripts/natural_translit/utils:feature",
-        "//nisaba/scripts/natural_translit/utils:inventory",
         "//nisaba/scripts/natural_translit/utils:type_op",
         requirement("pycountry"),
     ],

nisaba/scripts/natural_translit/script/grapheme.py

@@ -21,8 +21,8 @@
 import pycountry
 
 from nisaba.scripts.natural_translit.phonology import phonological_symbol as ps
+from nisaba.scripts.natural_translit.phonology.features import language as lang
 from nisaba.scripts.natural_translit.utils import feature as ft
-from nisaba.scripts.natural_translit.utils import inventory as i
 from nisaba.scripts.natural_translit.utils import type_op as ty
 
 
@@ -168,13 +168,15 @@ def copy(
   class Inventory(ps.PhonologicalSymbol.Inventory):
     """Grapheme inventory."""
 
-    def __init__(self, script: Script, language: str = ''):
-      if language:
-        language += '_'
-      super().__init__(alias=language + script.alias, typed=Grapheme)
+    def __init__(
+        self,
+        script: Script,
+        language: lang.Language.OR_NOTHING = ty.UNSPECIFIED,
+    ):
+      alias_prefix = language.alias + '_' if language else ''
+      super().__init__(alias_prefix + script.alias, language, Grapheme)
       self.script = script
       self.prefix = self._prefix()
-      self.atomics = i.Inventory()
 
     def _prefix(self) -> int:
       return (
@@ -188,6 +190,7 @@ def add_graphemes(
       grs = []
       for gr in graphemes:
         if self._add_symbol_and_atomic(gr):
+          gr.language = self.language
           grs.append(gr)
           gr.inventory = self
       if list_alias:

nisaba/scripts/natural_translit/script/grapheme_test.py

@@ -16,20 +16,21 @@
 from nisaba.scripts.natural_translit.script import grapheme as g
 from nisaba.scripts.natural_translit.utils import test_op
 
-_UND_GRAPHEMES = g.Grapheme.Inventory(g.Grapheme.GR_FEATURES.script.und)
+_G = g.Grapheme
+_UND_GRAPHEMES = _G.Inventory(_G.GR_FEATURES.script.und)
 
 
-def _test_inventory() -> g.Grapheme.Inventory:
+def _test_inventory() -> _G.Inventory:
   """Test grapheme inventory."""
-  gr_inv = g.Grapheme.Inventory(g.Grapheme.GR_FEATURES.script.latn)
+  gr_inv = _G.Inventory(_G.GR_FEATURES.script.latn)
   gr_inv.add_graphemes(
       # Raw
-      g.Grapheme.from_char('a', 'a'),
+      _G.from_char('a', 'a'),
       # Abstract with custom text
-      g.Grapheme('nasal', '~'),
+      _G('nasal', '~'),
       # Abstract with no text
       # Eg. Graphemes dynamically generated by aligner for partial match.
-      g.Grapheme('ch_1'),
+      _G('ch_1'),
   )
   return gr_inv
 
@@ -40,102 +41,103 @@ class GraphemeTest(test_op.TestCase):
 
   def test_script_iso(self):
     self.AssertStrEqual(
-        g.Grapheme.GR_FEATURES.script.latn,
+        _G.GR_FEATURES.script.latn,
         'alias: latn text: Latin numeric: 215',
     )
 
   def test_script_custom(self):
     self.AssertStrEqual(
-        g.Grapheme.GR_FEATURES.script.br,
+        _G.GR_FEATURES.script.br,
         'alias: br text: Brahmic Parent numeric: 801',
     )
 
   def test_from_char_index(self):
     self.assertEqual(
-        g.Grapheme.from_char('a').index,
-        g.Grapheme.ReservedIndex.GRAPHEME_PREFIX + 97,
+        _G.from_char('a').index,
+        _G.ReservedIndex.GRAPHEME_PREFIX + 97,
     )
 
   def test_from_char_name(self):
-    self.assertEqual(g.Grapheme.from_char('🐱').name, 'CAT FACE U+1F431')
+    self.assertEqual(_G.from_char('🐱').name, 'CAT FACE U+1F431')
 
   def test_from_char_name_error(self):
     self.assertEqual(
-        g.Grapheme.from_char(chr(0xE027)).name,  # Unassignable PUA character
+        _G.from_char(chr(0xE027)).name,  # Unassignable PUA character
         'GRAPHEME U+E027'
     )
 
   def test_from_char_alias_explicit(self):
-    self.assertEqual(g.Grapheme.from_char('ç', 'c_ced').alias, 'c_ced')
+    self.assertEqual(_G.from_char('ç', 'c_ced').alias, 'c_ced')
 
   def test_from_char_alias_default(self):
-    self.assertEqual(g.Grapheme.from_char('ç').alias, 'u_0xe7')
+    self.assertEqual(_G.from_char('ç').alias, 'u_0xe7')
 
   def test_from_char_has_feature(self):
     self.AssertHasFeature(
-        g.Grapheme.from_char('a'), g.Grapheme.SYM_FEATURES.type.raw
+        _G.from_char('a'), _G.SYM_FEATURES.type.raw
     )
 
   def test_from_char_add_feature(self):
     self.AssertHasFeature(
-        g.Grapheme.from_char(
-            'n', features=g.Grapheme.DESCRIPTIVE_FEATURES.ph_class.consonant
+        _G.from_char(
+            'n', features=_G.DESCRIPTIVE_FEATURES.ph_class.consonant
         ),
-        g.Grapheme.DESCRIPTIVE_FEATURES.ph_class.consonant,
+        _G.DESCRIPTIVE_FEATURES.ph_class.consonant,
     )
 
   def test_control_index(self):
     self.assertEqual(
-        _UND_GRAPHEMES.CTRL.eps.index, g.Grapheme.ReservedIndex.CONTROL_PREFIX
+        _UND_GRAPHEMES.CTRL.eps.index, _G.ReservedIndex.CONTROL_PREFIX
     )
 
   def test_control_in_text_dict(self):
-    self.assertIn(g.Grapheme.CTRL.unk.text, _UND_GRAPHEMES.text_dict)
+    self.assertIn(_G.CTRL.unk.text, _UND_GRAPHEMES.text_dict)
 
   def test_control_not_in_raw_dict(self):
-    self.assertNotIn(g.Grapheme.CTRL.unk.text, _UND_GRAPHEMES.raw_dict)
+    self.assertNotIn(_G.CTRL.unk.text, _UND_GRAPHEMES.raw_dict)
 
   def test_control_in_index_dict(self):
-    self.assertIn(g.Grapheme.CTRL.oos.index, _UND_GRAPHEMES.index_dict)
+    self.assertIn(_G.CTRL.oos.index, _UND_GRAPHEMES.index_dict)
 
   def test_inventory(self):
     self.assertEqual(_UND_GRAPHEMES.alias, 'und')
     self.assertEqual(_UND_GRAPHEMES.prefix, 2_800_000)
+    self.assertEqual(_UND_GRAPHEMES.language, _G.LANGUAGE.x_uni)
 
   def test_add_grapheme_in_dicts(self):
     char = 'ß'
-    _UND_GRAPHEMES.add_graphemes(g.Grapheme.from_char(char, alias='ss'))
+    _UND_GRAPHEMES.add_graphemes(_G.from_char(char, alias='ss'))
     self.assertEqual(_UND_GRAPHEMES.raw_lookup(char), _UND_GRAPHEMES.ss)
     self.assertEqual(_UND_GRAPHEMES.text_lookup(char), _UND_GRAPHEMES.ss)
     self.assertEqual(
         _UND_GRAPHEMES.index_lookup(
-            g.Grapheme.ReservedIndex.GRAPHEME_PREFIX + ord(char)
+            _G.ReservedIndex.GRAPHEME_PREFIX + ord(char)
         ),
         _UND_GRAPHEMES.ss,
     )
     self.assertNotEqual(_UND_GRAPHEMES.atomics.ss, _UND_GRAPHEMES.ss)
     self.AssertEquivalent(_UND_GRAPHEMES.atomics.ss, _UND_GRAPHEMES.ss)
 
   def test_add_grapheme_recurring_alias(self):
-    _UND_GRAPHEMES.add_graphemes(g.Grapheme.from_char('œ', alias='oe'))
+    _UND_GRAPHEMES.add_graphemes(_G.from_char('œ', alias='oe'))
     self.assertFalse(
-        _UND_GRAPHEMES.add_graphemes(g.Grapheme.from_char('Œ', alias='oe'))
+        _UND_GRAPHEMES.add_graphemes(_G.from_char('Œ', alias='oe'))
     )
 
   def test_add_grapheme_wrong_type(self):
-    self.assertFalse(_UND_GRAPHEMES._add_symbol(g.Grapheme.CTRL.eps))
+    self.assertFalse(_UND_GRAPHEMES._add_symbol(_G.CTRL.eps))
 
   def test_add_graphemes(self):
     _UND_GRAPHEMES.add_graphemes(
-        g.Grapheme.from_char('(', 'prl'),
-        g.Grapheme.from_char(')', 'prr'),
+        _G.from_char('(', 'prl'),
+        _G.from_char(')', 'prr'),
         list_alias='parentheses',
     )
     self.assertIn(_UND_GRAPHEMES.prl, _UND_GRAPHEMES)
     self.assertIn(_UND_GRAPHEMES.prl, _UND_GRAPHEMES.parentheses)
 
   def test_get_grapheme_in_inventory(self):
-    _UND_GRAPHEMES.add_graphemes(g.Grapheme.from_char('æ', alias='ae'))
+    _UND_GRAPHEMES.add_graphemes(_G.from_char('æ', alias='ae'))
     self.assertEqual(_UND_GRAPHEMES.text_lookup('æ'), _UND_GRAPHEMES.ae)
 
   def test_get_grapheme_out_of_inventory_char(self):
@@ -149,7 +151,7 @@ def test_parse(self):
         _UND_GRAPHEMES.parse('ş'), [_UND_GRAPHEMES.get('u_' + hex(ord('ş')))]
     )
     self.assertIn(
-        g.Grapheme.ReservedIndex.GRAPHEME_PREFIX + ord('ş'),
+        _G.ReservedIndex.GRAPHEME_PREFIX + ord('ş'),
         _UND_GRAPHEMES.index_dict,
     )
 
@@ -168,7 +170,9 @@ def test_grapheme_description(self):
     )
 
   def test_import_graphemes(self):
-    new_inv = g.Grapheme.Inventory(g.Grapheme.GR_FEATURES.script.latn)
+    new_inv = _G.Inventory(
+        _G.GR_FEATURES.script.latn, _G.LANGUAGE.en
+    )
     new_inv.import_graphemes(
         _TEST_INVENTORY.a, _TEST_INVENTORY.nasal, list_alias='from_test'
     )
@@ -177,6 +181,10 @@ def test_import_graphemes(self):
         new_inv.a.description(show_features=True),
         _TEST_INVENTORY.a.description(show_features=True),
     )
+    self.assertEqual(new_inv.language, _G.LANGUAGE.en)
+    self.assertEqual(new_inv.a.language, _G.LANGUAGE.en)
+    self.AssertHasFeature(new_inv.a, _G.LANGUAGE.en)
+    self.AssertHasFeature(new_inv.a, _G.LANGUAGE.indo_european)
 
 
 if __name__ == '__main__':