phonetics 1.5.3 → 1.5.4

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 04d203a13c081a435c9d94accd6af2dc517d7992349cda641f415e35c5cc787d
-  data.tar.gz: b8c3376c2db3878a0d3c409814abf0bb70d6fc403ce3866a11d65e8bcb01c5c8
+  metadata.gz: e8f26a3fc9c4a6729a0ddd04beee049efc4d1c0e8e2b887cfd14974de71d989a
+  data.tar.gz: 4449ecff87444214065a0f964c2cd7a0ba2ce93f52b3cbeae14e4777afcba3df
 SHA512:
-  metadata.gz: 5dfc217801d09dd1d835debd6440886e56c790c8886291147302b4a7b16f4aafeda51943cc56657d743e9dfab62cd663fae3adb3ea48656e0a6b5ce0cd1130ab
-  data.tar.gz: 1e41e3f77efe3af5a8095a0be58767c3a6ceb3c51ef7add5bebc40db94e14c988e80de221943ef010b7c0818d34a63b84af2290c57c90c6c23a4e9a69f8f9eff
+  metadata.gz: 4309942350149685324fa56a9c642f9456c7abe1993ab7f0801fbf56883afc4e31750e969e6d49e3da0ecd7b16a8d93b9212f5469ddc1dcd4c86699d0f48cd46
+  data.tar.gz: 8007ae8ef77e2d9e5f055fc597c4c5ffbd9ee00007d9d31ed4d06ae96e9980bc2dfc92149c96bd94098faa7d0670c5d7cbd13a30f45acd8d8c6000853692f341

data/VERSION

@@ -1 +1 @@
-1.5.3
+1.5.4

data/ext/c_levenshtein/levenshtein.c

@@ -55,9 +55,9 @@ VALUE method_internal_phonetic_distance(VALUE self, VALUE _string1, VALUE _strin
     string1[i] = NUM2INT(string1_ruby[i]);
     debug("string1[%d] = %d\n", i, string1[i]);
   }
-  for (i = 0; i < string2_length; i++) {
-    string2[i] = NUM2INT(string2_ruby[i]);
-    debug("string2[%d] = %d\n", i, string2[i]);
+  for (j = 0; j < string2_length; j++) {
+    string2[j] = NUM2INT(string2_ruby[j]);
+    debug("string2[%d] = %d\n", i, string2[j]);
   }
 
   // one-dimensional representation of 2 dimentional array len(string1)+1 *
@@ -81,8 +81,8 @@ VALUE method_internal_phonetic_distance(VALUE self, VALUE _string1, VALUE _strin
   // phonetic edit distance for that matrix cell.
   // (Skipping i=0 and j=0 because set_initial filled in all cells where i
   // or j are zero-valued)
-  for(j = 1; j <= string2_length; j++){
-    for(i = 1; i <= string1_length; i++){
+  for (j = 1; j <= string2_length; j++){
+    for (i = 1; i <= string1_length; i++){
 
       // The cost of deletion or addition is the Levenshtein distance
       // calculation (the value in the cell to the left, upper-left, or above)
@@ -141,24 +141,22 @@ VALUE method_internal_phonetic_distance(VALUE self, VALUE _string1, VALUE _strin
 // "aek" -> [0.0, 1.0, 1.61, 2.61]
 void set_initial(double *d, int *string1, int string1_length, int *string2, int string2_length, bool verbose) {
 
-  double distance_between_first_phonemes;
+  double initial_distance;
   int i, j;
 
   if (string1_length == 0 || string2_length == 0) {
-    distance_between_first_phonemes = 0.0;
-  } else if (string1[0] == string2[0]) {
-    distance_between_first_phonemes = 0.0;
+    initial_distance = 0.0;
   } else {
-    distance_between_first_phonemes = phonetic_cost(string1[0], string2[0]);
+    initial_distance = 1.0;
   }
 
+  // The top-left is 0, the cell to the right and down are each 1 to start
   d[0] = (double) 0.0;
-  // Set the first value of string1's sequential phonetic calculation (maps to
-  // cell x=1, y=0)
-  d[1] = distance_between_first_phonemes;
-  // And of string2 (maps to cell x=0, y=1)
+  if (string1_length > 0) {
+    d[1] = initial_distance;
+  }
   if (string2_length > 0) {
-    d[string1_length+1] = distance_between_first_phonemes;
+    d[string1_length+1] = initial_distance;
   }
 
   debug("string1 length: %d\n", string1_length);
@@ -172,6 +170,14 @@ void set_initial(double *d, int *string1, int string1_length, int *string2, int
     // The same exact pattern down the left side of the matrix
     d[j * (string1_length+1)] = d[(j - 1) * (string1_length+1)] + phonetic_cost(string2[j-2], string2[j-1]);
   }
+
+  // And zero out the rest. If you're reading this please edit this to be
+  // faster.
+  for (j=1; j <= string2_length; j++) {
+    for (i=1; i <= string1_length; i++) {
+      d[j * (string1_length+1) + i] = (double) 0.0;
+    }
+  }
 }
 
 // A handy visualization for developers

data/lib/phonetics/ruby_levenshtein.rb

@@ -121,18 +121,13 @@ module Phonetics
     # Set the minimum scores equal to the distance between each phoneme,
     # sequentially.
     #
-    # The first value is always zero.
-    # The second value is always the phonetic distance between the first
-    # phonemes of each string.
+    # The first value is always zero, the second is always 1.
     # Subsequent values are the cumulative phonetic distance between each
     # phoneme within the same string.
     # "aek" -> [0, 1, 1.61, 2.61]
     def initial_distances(str1, str2)
-      starting_distance = if len1 == 0 || len2 == 0
-                            0
-                          else
-                            Phonetics.distance(str1[0], str2[0])
-                          end
+      starting_distance = 1
+      starting_distance = 0 if len1 == 0 || len2 == 0
 
       distances1 = (1..(str1.length - 1)).reduce([0, starting_distance]) do |acc, i|
         acc << acc.last + Phonetics.distance(str1[i - 1], str1[i])

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: phonetics
 version: !ruby/object:Gem::Version
-  version: 1.5.3
+  version: 1.5.4
 platform: ruby
 authors:
 - Jack Danger