X-Git-Url: https://wannabe.guru.org/gitweb/?a=blobdiff_plain;f=unscrambler.py;h=8fcd65d4e3ef732bc529cd15f3d3de1fa13dede2;hb=4c8cd8d958afa46eafceda7a7295e31bfd884a8c;hp=3abb6d817a633147382d0ac82fea37cf4450e816;hpb=36fea7f15ed17150691b5b3ead75450e575229ef;p=python_utils.git

diff --git a/unscrambler.py b/unscrambler.py
index 3abb6d8..8fcd65d 100644
--- a/unscrambler.py
+++ b/unscrambler.py
@@ -121,13 +121,9 @@ class Unscrambler(object):
 
     # 52 bits
     @staticmethod
-    def _compute_word_fingerprint(
-        word: str, population: Mapping[str, int]
-    ) -> int:
+    def _compute_word_fingerprint(word: str, population: Mapping[str, int]) -> int:
         fp = 0
-        for pair in sorted(
-            population.items(), key=lambda x: x[1], reverse=True
-        ):
+        for pair in sorted(population.items(), key=lambda x: x[1], reverse=True):
             letter = pair[0]
             if letter in fprint_feature_bit:
                 count = pair[1]
@@ -146,9 +142,7 @@ class Unscrambler(object):
         population: Mapping[str, int],
     ) -> int:
         sig = 0
-        for pair in sorted(
-            population.items(), key=lambda x: x[1], reverse=True
-        ):
+        for pair in sorted(population.items(), key=lambda x: x[1], reverse=True):
             letter = pair[0]
             if letter not in letter_sigs:
                 continue
@@ -189,9 +183,7 @@ class Unscrambler(object):
         """
         population = list_utils.population_counts(word)
         fprint = Unscrambler._compute_word_fingerprint(word, population)
-        letter_sig = Unscrambler._compute_word_letter_sig(
-            letter_sigs, word, population
-        )
+        letter_sig = Unscrambler._compute_word_letter_sig(letter_sigs, word, population)
         assert fprint & letter_sig == 0
         sig = fprint | letter_sig
         return sig
@@ -226,25 +218,19 @@ class Unscrambler(object):
                 word = words_by_sigs[sig]
                 print(f'0x{sig:x}+{word}', file=f)
 
-    def lookup(
-        self, word: str, *, include_fuzzy_matches: bool = False
-    ) -> Dict[str, bool]:
+    def lookup(self, word: str, *, window_size: int = 5) -> Dict[str, bool]:
         """Looks up a potentially scrambled word optionally including near
         "fuzzy" matches.
 
         >>> u = Unscrambler()
-        >>> u.lookup('eanycleocipd', include_fuzzy_matches=False)
+        >>> u.lookup('eanycleocipd', window_size=0)
         {'encyclopedia': True}
 
         """
         sig = Unscrambler.compute_word_sig(word)
-        return self.lookup_by_sig(
-            sig, include_fuzzy_matches=include_fuzzy_matches
-        )
+        return self.lookup_by_sig(sig, window_size=window_size)
 
-    def lookup_by_sig(
-        self, sig: int, *, include_fuzzy_matches: bool = False
-    ) -> Dict[str, bool]:
+    def lookup_by_sig(self, sig: int, *, window_size: int = 5) -> Dict[str, bool]:
         """Looks up a word that has already been translated into a signature by
         a previous call to Unscrambler.compute_word_sig.  Optionally returns
         near "fuzzy" matches.
@@ -254,23 +240,23 @@ class Unscrambler(object):
         18491949645300288339
 
         >>> u = Unscrambler()
-        >>> u.lookup_by_sig(sig, include_fuzzy_matches=True)
+        >>> u.lookup_by_sig(sig)
         {'pupigerous': False, 'pupigenous': False, 'unpurposing': False, 'superpurgation': False, 'unsupporting': False, 'superseptuaginarian': True, 'purpurogallin': False, 'scuppaug': False, 'purpurigenous': False, 'purpurogenous': False, 'proppage': False}
 
         """
         ret = {}
         (exact, location) = list_utils.binary_search(self.sigs, sig)
-        start = location - 5
+        start = location - window_size
         if start < 0:
             start = 0
-        end = location + 6
+        end = location + 1 + window_size
         if end > len(self.words):
             end = len(self.words)
 
         for x in range(start, end):
             word = self.words[x]
             fsig = self.sigs[x]
-            if include_fuzzy_matches is True or (fsig == sig):
+            if window_size > 0 or (fsig == sig):
                 ret[word] = fsig == sig
         return ret