Merge pull request #3543 from tidyverse/feature-3335-wide-performance

Performance fixes for wide data frames #3335
tidyverse · May 3, 2018 · 5da3d0b · 5da3d0b
2 parents 71fdbde + 32dba91
commit 5da3d0b
Show file tree

Hide file tree

Showing 3 changed files with 10 additions and 3 deletions.
diff --git a/NEWS.md b/NEWS.md
@@ -123,6 +123,8 @@
 
 * `sample_n()` and `sample_frac()` on grouped data frame are now faster especially for those with large number of groups (#3193, @saurfang).
 
+* Improved performance for wide tibbles (#3335). 
+
 ## Internal
 
 * Compute variable names for joins in R (#3430).

diff --git a/inst/include/dplyr/Result/LazySubsets.h b/inst/include/dplyr/Result/LazySubsets.h
@@ -9,7 +9,7 @@ namespace dplyr {
 
 class LazySubsets : public ILazySubsets {
 public:
-  LazySubsets(const DataFrame& df) : nr(df.nrows()) {
+  LazySubsets(const DataFrame& df) : symbol_map(df.size(), df.names()), summary_map(), data(df.size()), nr(df.nrows()) {
     int nvars = df.size();
     if (nvars) {
       CharacterVector names = df.names();
@@ -18,8 +18,7 @@ class LazySubsets : public ILazySubsets {
         if (Rf_inherits(column, "matrix")) {
           stop("matrix as column is not supported");
         }
-        symbol_map.insert(names[i]);
-        data.push_back(df[i]);
+        data[i] = df[i];
       }
     }
   }

diff --git a/inst/include/tools/SymbolMap.h b/inst/include/tools/SymbolMap.h
@@ -25,6 +25,12 @@ class SymbolMap {
 public:
   SymbolMap(): lookup(), names() {}
 
+  SymbolMap(int n, const CharacterVector& names_): lookup(n), names((SEXP)names_) {
+    for (int i = 0; i < n; i++) {
+      lookup.insert(std::make_pair(names_[i], i));
+    }
+  }
+
   SymbolMap(const SymbolVector& names_): lookup(), names(names_) {}
 
   SymbolMapIndex insert(const SymbolString& name) {