Support unambiguous detection of language if only prefixes are supplied

o ceases search as soon as ambiguity is resolved

Simplify success exit criteria for detecting language

Correct language deduction if prefixes remain ambiguous

Author: pjkundert

src/mnemonic/mnemonic.py

@@ -93,15 +93,36 @@ def normalize_string(txt: t.AnyStr) -> str:
 
     @classmethod
     def detect_language(cls, code: str) -> str:
-        """Scan the Mnemonic until the language becomes unambiguous."""
+        """Scan the Mnemonic until the language becomes unambiguous, including as abbreviation prefixes.
+
+        Unfortunately, there are valid words that are ambiguous between languages, which are complete words
+        in one language and are prefixes in another:
+
+            english: abandon ... about
+            french:  abandon ... aboutir
+
+        If prefixes remain ambiguous, require exactly one language where word(s) match exactly.
+        """
         code = cls.normalize_string(code)
         possible = set(cls(lang) for lang in cls.list_languages())
-        for word in code.split():
-            possible = set(p for p in possible if word in p.wordlist)
+        words = set(code.split())
+        for word in words:
+            # possible languages have candidate(s) starting with the word/prefix
+            possible = set(
+                p for p in possible if any(c.startswith(word) for c in p.wordlist)
+            )
             if not possible:
                 raise ConfigurationError(f"Language unrecognized for {word!r}")
         if len(possible) == 1:
             return possible.pop().language
+        # Multiple languages match: A prefix in many, but an exact match in one determines language.
+        complete = set()
+        for word in words:
+            exact = set(p for p in possible if word in p.wordlist)
+            if len(exact) == 1:
+                complete.update(exact)
+        if len(complete) == 1:
+            return complete.pop().language
         raise ConfigurationError(
             f"Language ambiguous between {', '.join(p.language for p in possible)}"
         )

tests/test_mnemonic.py

@@ -57,6 +57,16 @@ def test_failed_checksum(self) -> None:
     def test_detection(self) -> None:
         self.assertEqual("english", Mnemonic.detect_language("security"))
 
+        self.assertEqual(
+            "english", Mnemonic.detect_language("fruit wave dwarf")
+        )  # ambiguous up to wave
+        self.assertEqual(
+            "english", Mnemonic.detect_language("fru wago dw")
+        )  # ambiguous french/english up to dwarf prefix
+        self.assertEqual(
+            "french", Mnemonic.detect_language("fru wago dur enje")
+        )  # ambiguous french/english up to enjeu prefix
+
         with self.assertRaises(Exception):
             Mnemonic.detect_language(
                 "jaguar xxxxxxx"
@@ -67,8 +77,20 @@ def test_detection(self) -> None:
                 "jaguar jaguar"
             )  # Ambiguous after examining all words
 
+        # Allowing word prefixes in language detection presents ambiguity issues.  Require exactly
+        # one language that matches all prefixes, or one language matching some word(s) exactly.
         self.assertEqual("english", Mnemonic.detect_language("jaguar security"))
         self.assertEqual("french", Mnemonic.detect_language("jaguar aboyer"))
+        self.assertEqual("english", Mnemonic.detect_language("abandon about"))
+        self.assertEqual("french", Mnemonic.detect_language("abandon aboutir"))
+        self.assertEqual("french", Mnemonic.detect_language("fav financer"))
+        self.assertEqual("czech", Mnemonic.detect_language("fav finance"))
+        with self.assertRaises(Exception):
+            Mnemonic.detect_language("favor finan")
+        self.assertEqual("czech", Mnemonic.detect_language("flanel"))
+        self.assertEqual("portuguese", Mnemonic.detect_language("flanela"))
+        with self.assertRaises(Exception):
+            Mnemonic.detect_language("flane")
 
     def test_utf8_nfkd(self) -> None:
         # The same sentence in various UTF-8 forms