use cpp11 std::unordered_map in place of tr1, issue #8

src/Makevars

@@ -0,0 +1 @@
+CXX_STD = CXX11

src/Makevars.win

@@ -0,0 +1 @@
+CXX_STD = CXX11

src/key_collision_merge_funcs.cpp

@@ -43,10 +43,7 @@ CharacterVector merge_KC_clusters_no_dict(CharacterVector clusters,
   // Create unordered_map, using clusters as keys, values will be the indices
   // of each cluster in keys_vect.
   std::vector<std::string> cl = as<std::vector<std::string> >(clusters);
-  unordered_map<std::string, std::vector<int> > keys_vect_map = create_map(
-    keys_vect,
-    cl
-  );
+  refinr_map keys_vect_map = create_map(keys_vect, cl);
 
   // Iterate over clusters, make mass edits to output.
   std::vector<std::string>::iterator clust_end = cl.end();
@@ -90,14 +87,8 @@ CharacterVector merge_KC_clusters_dict(CharacterVector clusters,
   // keys_vect_map will be the indices of each cluster in keys_vect. Values for
   // keys_dict_map will be the indices of each cluster in keys_dict.
   std::vector<std::string> cl = as<std::vector<std::string> >(clusters);
-  unordered_map<std::string, std::vector<int> > keys_vect_map = create_map(
-    keys_vect,
-    cl
-  );
-  unordered_map<std::string, std::vector<int> > keys_dict_map = create_map(
-    keys_dict,
-    cl
-  );
+  refinr_map keys_vect_map = create_map(keys_vect, cl);
+  refinr_map keys_dict_map = create_map(keys_dict, cl);
 
   // Iterate over clusters, make mass edits to output.
   std::vector<std::string>::iterator clust_end = cl.end();

src/n_gram_merge_funcs.cpp

@@ -2,27 +2,24 @@
 #include"refinr_header.h"
 using namespace Rcpp;
 
+
 // Iterate over all clusters, make mass edits related to each cluster.
 // [[Rcpp::export]]
 CharacterVector merge_ngram_clusters(List clusters,
                                      CharacterVector clust_unlist,
                                      CharacterVector n_gram_keys,
                                      CharacterVector univect,
                                      CharacterVector vect) {
-  //int vect_len = vect.size();
   CharacterVector output = clone(vect);
 
+  // Subset clust_unlist to only keep unique values.
+  clust_unlist = unique(clust_unlist);
+
   // Create maps
   std::vector<std::string> cl_ul = as<std::vector<std::string> >(clust_unlist);
-  unordered_map<std::string, std::vector<int> > ngram_map = create_map(
-    n_gram_keys,
-    cl_ul
-  );
   std::vector<std::string> uni = as<std::vector<std::string> >(univect);
-  unordered_map<std::string, std::vector<int> > univect_map = create_map(
-    vect,
-    uni
-  );
+  refinr_map ngram_map = create_map(n_gram_keys, cl_ul);
+  refinr_map univect_map = create_map(vect, uni);
 
   List::iterator clust_end = clusters.end();
   List::iterator iter;
@@ -95,10 +92,7 @@ List get_ngram_initial_clusters(CharacterVector ngram_keys,
   // Create unordered_map, using unigram_dups as keys, values will be the
   // indices of each dup in unigram_keys.
   std::vector<std::string> dups = as<std::vector<std::string> >(unigram_dups);
-  unordered_map<std::string, std::vector<int> > unigram_map = create_map(
-    unigram_keys,
-    dups
-  );
+  refinr_map unigram_map = create_map(unigram_keys, dups);
 
   // Iterate over unigram_dups, for each value get the corresponding indices
   // from unigram_map, Use those indices to subset ngram_keys, save the subset

src/refinr_header.h

@@ -1,9 +1,13 @@
 #include <Rcpp.h>
 using namespace Rcpp;
 
-#include <tr1/unordered_map>
-using std::tr1::unordered_map;
+// [[Rcpp::plugins(cpp11)]]
 
+// Define std::unordered_map using std::string as keys and std::vector<int>
+// as values.
+typedef std::unordered_map<std::string, std::vector<int> > refinr_map;
+
+// utils
 LogicalVector equality(CharacterVector lookupvect, String charstring);
 LogicalVector cpp_in(CharacterVector x, CharacterVector y);
 bool complete_intersect(CharacterVector a, CharacterVector b);
@@ -12,6 +16,8 @@ CharacterVector cpp_paste_list(List input, std::string collapse_str);
 CharacterVector cpp_get_key_dups(CharacterVector keys);
 List cpp_flatten_list(List list_obj);
 String most_freq_str(CharacterVector x);
-unordered_map <std::string, std::vector<int> > create_map(CharacterVector vect, std::vector<std::string> clusters);
+refinr_map create_map(CharacterVector vect, std::vector<std::string> clusters);
+
+// key_collision_merge_funcs
 CharacterVector merge_KC_clusters_no_dict(CharacterVector clusters, CharacterVector vect, CharacterVector keys_vect);
 CharacterVector merge_KC_clusters_dict(CharacterVector clusters, CharacterVector vect, CharacterVector keys_vect, CharacterVector dict, CharacterVector keys_dict);

src/utils.cpp

@@ -2,30 +2,32 @@
 #include"refinr_header.h"
 using namespace Rcpp;
 
+
 // Utility helper functions, some are used within R functions, some are used
 // within c++ functions, some are used in both.
 
 
 // Create unordered_map with strings as keys, and integer vectors as values.
-unordered_map <std::string, std::vector<int> > create_map(
-    CharacterVector vect,
-    std::vector<std::string> clusters
-) {
-  int clust_len = clusters.size();
-  int vect_len = vect.size();
+refinr_map create_map(CharacterVector terms,
+                      std::vector<std::string> keys) {
+  int keys_len = keys.size();
+  int terms_len = terms.size();
 
   // Initialize unordered_map.
-  unordered_map<std::string, std::vector<int> > clust_map;
-  for(int i = 0; i < clust_len; ++i) {
-    clust_map[clusters[i]];
+  refinr_map out;
+  for(int i = 0; i < keys_len; ++i) {
+    out[keys[i]];
   }
 
   // Fill values of the map.
-  for(int i = 0; i < vect_len; ++i) {
-    clust_map[as<std::string>(vect[i])].push_back(i);
+  for(int i = 0; i < terms_len; ++i) {
+    refinr_map::iterator val = out.find(as<std::string>(terms[i]));
+    if(val != out.end()) {
+      val->second.push_back(i);
+    }
   }
 
-  return(clust_map);
+  return(out);
 }
 
 // Given a CharacterVector, return the string that appears most frequently.