test: MRE for #283 shows swallowed edge after 1->2+2->1

roedoejet · Sep 1, 2023 · 91d46f6 · 91d46f6
1 parent b413089
commit 91d46f6
Showing 1 changed file with 48 additions and 5 deletions.
diff --git a/g2p/tests/test_mappings.py b/g2p/tests/test_mappings.py
@@ -166,19 +166,26 @@ def test_case_sensitive(self):
     def test_escape_special(self):
         mapping = Mapping([{"in": r"\d", "out": "digit"}])
         mapping_escaped = Mapping([{"in": r"\d", "out": "b"}], escape_special=True)
-        mapping_input_and_output_special_escaped = Mapping([{"in": "&", "out": "&"}], escape_special=True)
-        mapping_specific_from_fpcc = Mapping([{"in": r"^", "out": "A"}, {"in": "o", "out": r"."}], rule_ordering="apply-longest-first", escape_special=True)
+        mapping_input_and_output_special_escaped = Mapping(
+            [{"in": "&", "out": "&"}], escape_special=True
+        )
+        mapping_specific_from_fpcc = Mapping(
+            [{"in": r"^", "out": "A"}, {"in": "o", "out": r"."}],
+            rule_ordering="apply-longest-first",
+            escape_special=True,
+        )
         transducer = Transducer(mapping)
         transducer_escaped = Transducer(mapping_escaped)
-        transducer_escaped_input_output = Transducer(mapping_input_and_output_special_escaped)
+        transducer_escaped_input_output = Transducer(
+            mapping_input_and_output_special_escaped
+        )
         transducer_fpcc = Transducer(mapping_specific_from_fpcc)
         self.assertEqual(transducer("1").output_string, "digit")
         self.assertEqual(transducer(r"\d").output_string, r"\d")
         self.assertEqual(transducer_escaped("1").output_string, "1")
         self.assertEqual(transducer_escaped(r"\d").output_string, "b")
-        self.assertEqual(transducer_escaped_input_output('&').output_string, "&")
+        self.assertEqual(transducer_escaped_input_output("&").output_string, "&")
         self.assertEqual(transducer_fpcc("^o").output_string, "A.")
-
 
     def test_norm_form(self):
         mapping_nfc = Mapping([{"in": "a\u0301", "out": "a"}])  # Defaults to NFC
@@ -343,6 +350,42 @@ def test_g2p_studio_csv(self):
         )
         os.unlink(tf.name)
 
+    def test_vanishing_edge(self):
+        """1->2+2->1 should not eat the next edge"""
+        mapping = Mapping([{"in": "d{1}ef{2}", "out": "gh{1}i{2}"}])
+        transducer = Transducer(mapping)
+        # the rule itself works, with the indices doing what they're supposed to do
+        tg = transducer("def")
+        self.assertEqual(tg.output_string, "ghi")
+        self.assertEqual(tg.alignments(), [(0, 0), (0, 1), (1, 2), (2, 2)])
+        # but it swallows the next edge
+        tg = transducer("deft")
+        self.assertEqual(tg.output_string, "ghit")
+        # with vanishing edge bug, the following fails, getting only [(0, 0), (0, 1), (1, 2), (2, 2)]
+        self.assertEqual(tg.alignments(), [(0, 0), (0, 1), (1, 2), (2, 2), (3, 3)])
+
+        # make sure the fix works with "complex" cases too
+        tg = transducer("deftdefdef")
+        self.assertEqual(tg.output_string, "ghitghighi")
+        self.assertEqual(
+            tg.alignments(),
+            [
+                (0, 0),
+                (0, 1),
+                (1, 2),
+                (2, 2),
+                (3, 3),
+                (4, 4),
+                (4, 5),
+                (5, 6),
+                (6, 6),
+                (7, 7),
+                (7, 8),
+                (8, 9),
+                (9, 9),
+            ],
+        )
+
 
 if __name__ == "__main__":
     main()