Add Expression.ANY constant that accepts, contains, and matches with everything other than an empty Or. This will be used as the default value for Alignment arguments.

PiperOrigin-RevId: 630344671

Author: isingoo

nisaba/scripts/natural_translit/utils/expression.py

@@ -112,6 +112,19 @@ def _symbols_of(
     else:
       return [[other]]
 
+  def is_any(self) -> bool:
+    if isinstance(self, sym.Symbol):
+      return False
+    if len(self) == 1:
+      return self.item(0).is_any()
+    return self is Expression.ANY
+
+  def is_eps(self) -> bool:
+    return isinstance(self, sym.Symbol) and self.symbol.is_eps()
+
+  def is_nor(self) -> bool:
+    return isinstance(self, sym.Symbol) and self.symbol.is_nor()
+
   def accepts(
       self, other: 'Expression.OR_SYMBOL', equivalent: bool = False
   ) -> bool:
@@ -124,6 +137,10 @@ def accepts(
     Returns:
       bool
     """
+    if self.is_any() or other.is_any():
+      if equivalent:
+        return self.is_any() and other.is_any()
+      return True
     self_symbols, other_symbols = self.symbols(), self._symbols_of(other)
     self_len, other_len = len(self_symbols), len(other_symbols)
     if (
@@ -198,6 +215,8 @@ def contains_symbol_list(
       return True
     if search_for == [sym.Symbol.CTRL.nor]:
       return False
+    if self.is_any():
+      return True
     # Loop over symbol lists, eg: [[a, b, c, d], [e, f, g]]
     for symbol_list in self.symbols():
       while symbol_list:
@@ -251,15 +270,19 @@ def contains(
       a.contains(b, head=True): [a, b, c, d] starts with [a, b]
       a.contains(b, tail=True): False
     """
+    if self.is_any() or other.is_any():
+      return not self.is_nor() and not other.is_nor()
     for sym_list in self._symbols_of(other):
       if self.contains_symbol_list(sym_list, head, tail):
         return True
     return False
 
   def _symbol_contains(self, other: sym.Symbol) -> bool:
+    if other.is_any():
+      return True
     self_symbols = self.symbols()
     return [sym.Symbol.CTRL.eps] in self_symbols or (
-        other != sym.Symbol.CTRL.nor and [other] in self_symbols
+        not other.is_nor() and [other] in self_symbols
     )
 
   def is_contained(
@@ -278,13 +301,13 @@ def matches(self, other: 'Expression.OR_SYMBOL') -> bool:
     return self.contains(other, head=True, tail=True)
 
   # head_matches and tail_matches require at least one symbol match unless
-  # both expressions are empty Cats. For example, if a rule requires a vowel as
-  # following context but there is no following context, the rule shouldn't
-  # apply.
+  # both expressions are empty Cats or one of the expressions is Expression.ANY
+  # For example, if a rule requires a vowel as following context but there is no
+  # following context, the rule shouldn't apply.
 
   def head_matches(self, other: 'Expression.OR_SYMBOL') -> bool:
     if self and not other:
-      return False
+      return other.is_any()
     return self.contains(other, head=True)
 
   def is_prefix(self, other: 'Expression.OR_SYMBOL') -> bool:
@@ -294,7 +317,7 @@ def is_prefix(self, other: 'Expression.OR_SYMBOL') -> bool:
 
   def tail_matches(self, other: 'Expression.OR_SYMBOL') -> bool:
     if self and not other:
-      return False
+      return other.is_any()
     return self.contains(other, tail=True)
 
   def is_suffix(self, other: 'Expression.OR_SYMBOL') -> bool:
@@ -316,6 +339,9 @@ def repeat(self, n: int = 2) -> 'Cat':
     return Cat(*([self] * n))
 
 
+Expression.ANY = Expression('any_expression')
+
+
 class Atomic(Expression, sym.Symbol):
   """An instance of a single symbol."""
 
@@ -392,7 +418,10 @@ def __str__(self):
 
   def add(self, *items: Expression) -> 'Cat':
     for item in items:
-      self._add_item(item)
+      if item.is_any():
+        self._items.append(item)
+      else:
+        self._add_item(item)
     return self
 
   def symbols(self) -> list[list[sym.Symbol]]:
@@ -462,6 +491,13 @@ def add(self, *items: Expression) -> 'Or':
       self
     """
     for item in items:
+      # If the Expression.ANY is in Or, don't add any items.
+      if Expression.ANY in self:
+        break
+      # If the item is any, the other items are irrelevant.
+      if item.is_any():
+        self._items = [item]
+        break
       if self.accepts(item):
         if item.accepts(self) and item.state_count() < self.state_count():
           self._update(item)

nisaba/scripts/natural_translit/utils/expression_test.py

@@ -51,6 +51,8 @@ def test_atomic_read(self):
 
   def test_control(self):
     self.assertTrue(exp.Atomic.CTRL.unk.is_control())
+    self.assertTrue(exp.Atomic.CTRL.eps.is_eps())
+    self.assertTrue(exp.Atomic.CTRL.nor.is_nor())
 
   def test_symbol_inventory_lookup(self):
     self.assertEqual(_ATM.lookup(_ATM.a, 'atm_sym'), _SYM.a)
@@ -98,6 +100,7 @@ def test_cat_items(self):
     self.AssertStrEqual(cat, '(a b a)')
     self.assertIsNot(cat.item(0), cat.item(2))
     self.AssertEquivalent(cat.item(0), (cat.item(2)))
+    self.assertTrue(exp.Cat(exp.Expression.ANY).is_any())
 
   def test_cat_nested(self):
     cat1 = exp.Cat(_ATM.a, _ATM.b)
@@ -132,10 +135,13 @@ def test_or_items(self):
     or2 = exp.Or(_ATM.b, _ATM.c)
     or3 = exp.Or(or1, or2)
     or4 = exp.Or(_ATM.a, _ATM.b, _ATM.a)
+    or5 = or1.copy().add(exp.Expression.ANY)
     self.AssertStrEqual(or2, '(b | c)')
     self.AssertStrEqual(or3, '(a | b | c)')
     self.AssertStrEqual(or4, '(a | b)')
     self.assertLen(or4, 2)
+    self.assertNotIn(_ATM.a, or5)
+    self.AssertAccepts(or5, _ATM.a)
 
   def test_or_nested(self):
     cat1 = exp.Cat(_ATM.a)
@@ -214,6 +220,8 @@ def test_state_count(self):
   def test_equivalent(self):
     or0 = exp.Or()
     self.AssertEquivalent(exp.Atomic.CTRL.eps, sym.Symbol.CTRL.eps)
+    self.AssertEquivalent(exp.Expression.ANY, exp.Expression.ANY)
+    self.AssertNotEquivalent(exp.Expression.ANY, exp.Atomic.CTRL.eps)
     self.AssertEquivalent(exp.Cat(), exp.Atomic.CTRL.eps)
     self.AssertEquivalent(exp.Cat(), exp.Cat())
     self.AssertNotEquivalent(or0, exp.Atomic.CTRL.nor)
@@ -229,13 +237,21 @@ def test_equivalent(self):
   def test_contains_controls(self):
     eps = exp.Atomic.CTRL.eps
     nor = exp.Atomic.CTRL.nor
+    any_exp = exp.Expression.ANY
     self.AssertContains(eps, eps)
     self.AssertContains(nor, eps)
+    self.AssertNotContains(nor, any_exp)
     self.AssertNotContains(eps, nor)
     self.AssertNotContains(nor, nor)
+    self.AssertNotContains(any_exp, nor)
 
   def test_contains_expressions(self):
     cat_abc = _ATM.a + _ATM.b + _ATM.c
+    any_exp = exp.Expression.ANY
+    self.AssertContains(any_exp, _ATM.a)
+    self.AssertContains(_ATM.a, any_exp)
+    self.AssertContains(any_exp, cat_abc)
+    self.AssertContains(cat_abc, any_exp)
     self.AssertContains(cat_abc, exp.Cat())
     self.AssertNotContains(cat_abc, exp.Or())
     self.AssertContains(cat_abc, exp.Or(exp.Cat()))
@@ -250,8 +266,15 @@ def test_contains_expressions(self):
   def test_matches(self):
     abc_or_cd = (_ATM.a + _ATM.b + _ATM.c) | (_ATM.c + _ATM.d)
     a_or_c_b_or_d = (_ATM.a | _ATM.c) + (_ATM.b | _ATM.d)
+    any_exp = exp.Expression.ANY
+    self.AssertMatches(any_exp, a_or_c_b_or_d)
+    self.AssertMatches(a_or_c_b_or_d, any_exp)
     self.AssertMatches(abc_or_cd, a_or_c_b_or_d)
     self.AssertNotMatches(abc_or_cd, _ATM.a + _ATM.b + _ATM.d)
+    self.assertTrue(any_exp.is_prefix(abc_or_cd))
+    self.assertTrue(any_exp.is_suffix(abc_or_cd))
+    self.assertTrue(abc_or_cd.is_prefix(any_exp))
+    self.assertTrue(abc_or_cd.is_suffix(any_exp))
     self.assertTrue(exp.Cat().is_prefix(exp.Cat()))
     self.assertTrue(exp.Cat().is_suffix(exp.Cat()))
     self.assertFalse(exp.Cat().is_prefix(abc_or_cd))

nisaba/scripts/natural_translit/utils/symbol.py

@@ -106,6 +106,15 @@ def __str__(self) -> str:
   def is_control(self) -> bool:
     return self in Symbol.CTRL
 
+  def is_any(self) -> bool:
+    return False
+
+  def is_eps(self) -> bool:
+    return self is Symbol.CTRL.eps
+
+  def is_nor(self) -> bool:
+    return self is Symbol.CTRL.nor
+
   def description(self, show_features: bool = False) -> str:
     """A string that describes the symbol."""
     text = 'alias: %s  index: %s' % (self.alias, self.index)