fixed #17 which reappeared when I fixed #46

markvanderloo · Sep 9, 2016 · 10b1be2 · 10b1be2
1 parent 9978434
commit 10b1be2
Show file tree

Hide file tree

Showing 5 changed files with 28 additions and 9 deletions.
diff --git a/pkg/DESCRIPTION b/pkg/DESCRIPTION
@@ -21,7 +21,7 @@ Imports:
     parallel
 URL: https://github.com/markvanderloo/stringdist
 BugReports: https://github.com/markvanderloo/stringdist/issues
-Date: 2015-12-30
+Date: 2016-09-09
 Suggests:
     testthat
 RoxygenNote: 5.0.1
diff --git a/pkg/NEWS b/pkg/NEWS
@@ -3,6 +3,8 @@ version 0.9.4.2
   (thanks to Max Fritsche)
 - bugfix in stringdistmatrix(a): Would segfault on q-gram w/input > ~7k strings
   and q>1 (thanks to Connor McKay)
+- bugfix in jaccard distance: distance not always correct when passing multiple
+  strings (thanks to Robert Carlson)
 
 version 0.9.4.1
 - stringdistmatrix(a) now outputs long vectors (issue #45, thanks to Wouter

diff --git a/pkg/src/Rstringdist.c b/pkg/src/Rstringdist.c
@@ -205,9 +205,13 @@ SEXP R_amatch(SEXP x, SEXP table, SEXP method
         len_T = T->str_len[j];
         if (len_X != NA_INTEGER && len_T != NA_INTEGER ){        // both are char (usual case)
           d = stringdist(sd, str, len_X, *tab, len_T);
+//Rprintf("d = %8.4f ",d);
           if ( d <= maxDist && d < d1){ 
             index = j + 1;
-            if ( ABS(d) < 1e-14 ) break; // exact match
+            if ( fabs(d) < 1e-14 ){ 
+ //               Rprintf(" helleu!\n");
+                  break; // exact match
+            }
             d1 = d;
           }
         } else if ( len_X == NA_INTEGER && len_T == NA_INTEGER ) {  // both are NA

diff --git a/pkg/src/qgram.c b/pkg/src/qgram.c
@@ -305,15 +305,28 @@ static void getjaccard(qtree *Q, double *d){
     ++d[0];
   } 
   // denominator: |x V y|
-  ++d[1];
+  if ( Q->n[0] > 0 || Q->n[1] > 0){
+    ++d[1];
+  }
   // clean up and continue
   Q->n[0] = 0;
   Q->n[1] = 0;
   getjaccard(Q->left,d);
   getjaccard(Q->right,d);
 }
 
-
+/* for testing purposes only
+static void print_qtree(qtree *Q, int q){
+  if (Q==NULL) return;
+  Rprintf("q=%d ",q);
+  Rprintf("qgram = {");
+  for(int i = 0; i < q; i++)
+    Rprintf("%03d ",Q->qgram[i]);
+  Rprintf("}");
+  Rprintf("n = [%2.0f %2.0f]\n", Q->n[0], Q->n[1]);
+  print_qtree(Q->left,q);
+  print_qtree(Q->right,q);
+}*/
 
 /*Get qgram distances 
  * Input
@@ -356,7 +369,8 @@ double qgram_dist(
   *Qp = push_string(t, y, q, *Qp, 1, 2);
   if (*Qp == NULL) return -2.0;
 
- qtree *Q = *Qp;
+
+  qtree *Q = *Qp;
   switch ( distance ){
     case 0:
       getdist(Q,dist);
@@ -373,7 +387,7 @@ double qgram_dist(
       }
       break;
     case 2:
-      getjaccard(Q,dist);
+      getjaccard(*Qp,dist);
       dist[0] = 1.0 - dist[0]/dist[1];
       break;
     default:

diff --git a/pkg/src/stringdist.c b/pkg/src/stringdist.c
@@ -124,12 +124,11 @@ double stringdist(Stringdist *S, unsigned int *str_a, int len_a, unsigned int *s
     case lcs :
       return lcs_dist(str_a, len_a, str_b, len_b, S->work); 
     case qgram :
-      return qgram_dist(str_a, len_a, str_b, len_b, S->q, &(S->tree), 0L);
+      return qgram_dist(str_a, len_a, str_b, len_b, S->q, &S->tree, 0L);
     case cosine :
       return qgram_dist(str_a, len_a, str_b, len_b, S->q, &S->tree, 1L);
     case jaccard :
-      d = qgram_dist(str_a, len_a, str_b, len_b, S->q, &S->tree, 2L);
-      break;
+      return qgram_dist(str_a, len_a, str_b, len_b, S->q, &S->tree, 2L);
     case jw :
       return jaro_winkler_dist(str_a, len_a, str_b, len_b, S->p, S->weight, S->work);
     case soundex :