levenshtein.c in phonetics-1.5.3

- old
+ new

@@ -1,35 +1,33 @@
 #include "ruby.h"
+#include <stdbool.h>
 #include "./phonetic_cost.h"
 
-#define NDEBUG true
-#ifdef NDEBUG
-#define debug(M, ...)
-#else
-#define debug(M, ...) printf(M, ##__VA_ARGS__)
-#endif
+#define debug(M, ...) if (verbose) printf(M, ##__VA_ARGS__)
 
 VALUE Binding = Qnil;
 
 /* Function declarations */
 
 void Init_c_levenshtein();
-void set_initial(double *d, int *string1, int string1_length, int *string2, int string2_length);
-void print_matrix(double *d, int *string1, int string1_length, int *string2, int string2_length);
-VALUE method_internal_phonetic_distance(VALUE self, VALUE _string1, VALUE _string2);
 
+void set_initial(double *d, int *string1, int string1_length, int *string2, int string2_length, bool verbose);
+void print_matrix(double *d, int *string1, int string1_length, int *string2, int string2_length, bool verbose);
+VALUE method_internal_phonetic_distance(VALUE self, VALUE _string1, VALUE _string2, VALUE _verbose);
+
 /* Function implemitations */
 
 void Init_c_levenshtein() {
 	Binding = rb_define_module("PhoneticsLevenshteinCBinding");
-	rb_define_method(Binding, "internal_phonetic_distance", method_internal_phonetic_distance, 2);
+	rb_define_method(Binding, "internal_phonetic_distance", method_internal_phonetic_distance, 3);
 }
 
-VALUE method_internal_phonetic_distance(VALUE self, VALUE _string1, VALUE _string2){
+VALUE method_internal_phonetic_distance(VALUE self, VALUE _string1, VALUE _string2, VALUE _verbose){
 
   VALUE *string1_ruby = RARRAY_PTR(_string1);
   VALUE *string2_ruby = RARRAY_PTR(_string2);
+  bool verbose = _verbose;
   int string1_length = (int) RARRAY_LEN(_string1);
   int string2_length = (int) RARRAY_LEN(_string2);
   // We name them as 'strings' but in C-land we're representing our strings as
   // arrays of `int`s, where each int represents a consistent (if unusual)
   // encoding of a grapheme cluster (a symbol for a phoneme).
@@ -45,11 +43,11 @@
          cost;
   int i, j;               // Frequently overwritten loop vars
 
   // Guard clause for two empty strings
   if (string1_length == 0 && string2_length == 0)
-    return DBL2NUM(0.0d);
+    return DBL2NUM(0.0);
 
   //
   // Intial data setup
   //
 
@@ -72,14 +70,14 @@
   // this case the last spot in the array)
   //
   
   // First, set the top row and left column of the matrix using the sequential
   // phonetic edit distance of string1 and string2, respectively
-  set_initial(d, string1, string1_length, string2, string2_length);
+  set_initial(d, string1, string1_length, string2, string2_length, verbose);
 
   debug("before:\n");
-  print_matrix(d, string1, string1_length, string2, string2_length);
+  print_matrix(d, string1, string1_length, string2, string2_length, verbose);
 
   // Then walk through the matrix and fill in each cell with the lowest-cost
   // 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)
@@ -115,11 +113,11 @@
         min = replace;
       }
 
       d[(j * (string1_length+1)) + i] = min + cost;
       debug("\n");
-      print_matrix(d, string1, string1_length, string2, string2_length);
+      print_matrix(d, string1, string1_length, string2, string2_length, verbose);
     }
   }
 
   // The final element in the `d` array is the value of the shortest path from
   // the top-left to the bottom-right of the matrix.
@@ -139,11 +137,11 @@
 // The second value is always the phonetic distance between the first
 // phonemes of each string.
 // Subsequent values are the cumulative phonetic distance between each
 // phoneme within the same string.
 // "aek" -> [0.0, 1.0, 1.61, 2.61]
-void set_initial(double *d, int *string1, int string1_length, int *string2, int string2_length) {
+void set_initial(double *d, int *string1, int string1_length, int *string2, int string2_length, bool verbose) {
 
   double distance_between_first_phonemes;
   int i, j;
 
   if (string1_length == 0 || string2_length == 0) {
@@ -152,10 +150,11 @@
     distance_between_first_phonemes = 0.0;
   } else {
     distance_between_first_phonemes = phonetic_cost(string1[0], string2[0]);
   }
 
+  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 (string2_length > 0) {
@@ -174,10 +173,10 @@
     d[j * (string1_length+1)] = d[(j - 1) * (string1_length+1)] + phonetic_cost(string2[j-2], string2[j-1]);
   }
 }
 
 // A handy visualization for developers
-void print_matrix(double *d, int *string1, int string1_length, int *string2, int string2_length) {
+void print_matrix(double *d, int *string1, int string1_length, int *string2, int string2_length, bool verbose) {
   int i, j;
   debug("              ");
   for (i=0; i < string1_length; i++) {
     debug("%8.d ", string1[i]);
   }