|
| 1 | +#include "ruby.h" |
| 2 | + |
| 3 | +int min(int a, int b, int c) { |
| 4 | + int min = a; |
| 5 | + if (b < min) |
| 6 | + min = b; |
| 7 | + if( c < min) |
| 8 | + min = c; |
| 9 | + return min; |
| 10 | +} |
| 11 | + |
| 12 | +int levenshtein_distance(VALUE str1, VALUE str2) { |
| 13 | + int i, j, s1_len, s2_len, *d; |
| 14 | + char * s = RSTRING(str1)->ptr; |
| 15 | + char * t = RSTRING(str2)->ptr; |
| 16 | + s1_len = RSTRING(str1)->len; |
| 17 | + s2_len = RSTRING(str2)->len; |
| 18 | + |
| 19 | + if (s1_len == 0) { |
| 20 | + return s2_len; |
| 21 | + } else if (s2_len == 0) { |
| 22 | + return s1_len; |
| 23 | + } |
| 24 | + |
| 25 | + // We need one extra col and row for the matrix for starting values |
| 26 | + s1_len++; |
| 27 | + s2_len++; |
| 28 | + |
| 29 | + d = malloc(sizeof(int) * (s1_len) * (s2_len)); |
| 30 | + |
| 31 | + for (i = 0; i < s1_len; i++) { |
| 32 | + d[i] = i; // d[i, 0] = i |
| 33 | + } |
| 34 | + for (j = 0; j < s2_len; j++) { |
| 35 | + d[j*s1_len] = j; // d[0, j] = j |
| 36 | + } |
| 37 | + |
| 38 | + for (i = 1; i < s1_len; i++) { |
| 39 | + for (j = 1; j < s2_len; j++) { |
| 40 | + if (s[i-1] == t[j-1]) { |
| 41 | + d[j * s1_len + i] = d[(j-1) * s1_len + (i-1)]; |
| 42 | + } else { |
| 43 | + d[j * s1_len + i] = min( |
| 44 | + d[j * s1_len + (i-1)], |
| 45 | + d[(j-1) * s1_len + i], |
| 46 | + d[(j-1) * s1_len + (i-1)] |
| 47 | + ) + 1; |
| 48 | + } |
| 49 | + } |
| 50 | + } |
| 51 | + i = d[s1_len * s2_len -1]; |
| 52 | + free(d); |
| 53 | + return i; |
| 54 | +} |
| 55 | + |
| 56 | +static VALUE lev_dist(VALUE self, VALUE str1, VALUE str2) { |
| 57 | + return INT2FIX(levenshtein_distance( str1, str2 )); |
| 58 | +} |
| 59 | + |
| 60 | +static VALUE mAlgorithms; |
| 61 | +static VALUE mString; |
| 62 | + |
| 63 | +void Init_CString() { |
| 64 | + mAlgorithms = rb_define_module("Algorithms"); |
| 65 | + mString = rb_define_module_under(mAlgorithms, "String"); |
| 66 | + rb_define_singleton_method(mString, "levenshtein_dist", lev_dist, 2); |
| 67 | +} |
| 68 | + |
0 commit comments