Fix: re2_detect with NA #8

qinwf · May 8, 2016 · ca857bd · ca857bd
1 parent d5deeac
commit ca857bd
Show file tree

Hide file tree

Showing 4 changed files with 46 additions and 7 deletions.
diff --git a/inst/include/re2r.h b/inst/include/re2r.h
@@ -128,6 +128,6 @@ XPtr<RE2Obj> cpp_re2_compile(const char* pattern,
 
 SEXP toprotect_optstring_sexp(const optstring& input);
 SEXP toprotect_vec_string_sexp(const vector<string>& input);
-
+vector<tr2::optional<string>> as_vec_opt_string(CharacterVector& input);
 
 #endif
diff --git a/src/re2r_match.cpp b/src/re2r_match.cpp
@@ -329,7 +329,12 @@ SEXP cpp_detect(CharacterVector& input,
     LogicalVector res(input.size());
     auto resi  = res.begin();
     for(auto it = 0; it != input.size(); it++, resi ++){
-        auto r_char = R_CHAR(STRING_ELT(inputx, it));
+        auto rstr = STRING_ELT(inputx, it);
+        if (rstr == NA_STRING){
+            *resi = NA_LOGICAL;
+            continue;
+        }
+        auto r_char = R_CHAR(rstr);
         *resi = pattern->Match( r_char ,0, strlen(r_char),
                                 anchor_type, nullptr, 0);
     }
@@ -338,22 +343,25 @@ SEXP cpp_detect(CharacterVector& input,
 
 struct BoolP : public Worker
 {
-    vector<string>& input;
+    vector<tr2::optional<string>>& input;
     RVector<int> output;
     RE2& tt;
     RE2::Options& opt;
     const RE2::Anchor anchor_type;
 
-    BoolP (vector<string>&  input_,RVector<int> output_, RE2& tt_, RE2::Options& opt_, const RE2::Anchor&  anchor_type_)
+    BoolP (vector<tr2::optional<string>>&  input_,RVector<int> output_, RE2& tt_, RE2::Options& opt_, const RE2::Anchor&  anchor_type_)
         : input(input_), output(output_), tt(tt_), opt(opt_), anchor_type(anchor_type_){}
 
     void operator()(std::size_t begin, std::size_t end) {
         RE2 pattern(tt.pattern(),opt);
         std::transform(input.begin() + begin,
                        input.begin() + end,
                        output.begin() + begin,
-                       [this,&pattern](string& x)->int{
-                           return pattern.Match(x, 0, (int) x.length(),
+                       [this,&pattern](tr2::optional<string>& x)->int{
+                           if (!bool(x)){
+                                return NA_LOGICAL;
+                           }
+                           return pattern.Match(x.value(), 0, (int) x.value().length(),
                                                 anchor_type, nullptr, 0);
                        });
     }
@@ -366,7 +374,7 @@ SEXP cpp_detect_parallel(CharacterVector& input,
                 RE2::Anchor anchor_type){
     LogicalVector reso(input.size());
     RVector<int> res(reso);
-    vector<string> inputv = as<vector<string>>(input);
+    auto inputv = as_vec_opt_string(input);
     BoolP pobj(inputv, res, *pattern, opt, anchor_type);
     parallelFor(0, input.size(), pobj, 1000000);
     return wrap(reso);

diff --git a/src/re2r_util.cpp b/src/re2r_util.cpp
@@ -14,6 +14,23 @@ SEXP toprotect_vec_string_sexp(const vector<string>& input){
     return x;
 }
 
+vector<tr2::optional<string>> as_vec_opt_string(CharacterVector& input){
+    SEXP inputx = input;
+    vector<tr2::optional<string>> res;
+    res.reserve(input.size());
+
+    for(auto it = 0; it != input.size(); it++){
+        auto rstr = STRING_ELT(inputx, it);
+        if (rstr == NA_STRING){
+            res.push_back(tr2::nullopt);
+            continue;
+        }else{
+            res.push_back(tr2::make_optional(string(R_CHAR(rstr))));
+        }
+    }
+    return res;
+}
+
 SEXP toprotect_optstring_sexp(const optstring& input){
     SEXP x;
     PROTECT(x = Rf_allocVector(STRSXP, input.size()));

diff --git a/tests/testthat/test-match_group.R b/tests/testthat/test-match_group.R
@@ -117,3 +117,17 @@ test_that("tolist",{
         re2_pmatch_all(str,"(?P<testname>this)( is)")
         )
 })
+
+library(stringi)
+
+test_that("match NA",{
+    expect_identical(structure(c(NA, "sd"), .Dim = c(2L, 1L), .Dimnames = list(NULL, "?nocapture")), re2_match(c(NA,"sd"),"sd"))
+    expect_identical(re2_match(c(NA,"sd"),"sd"),re2_pmatch(c(NA,"sd"),"sd"))
+
+    expect_identical(c(NA,TRUE), re2_detect(c(NA,"sd"),"sd"))
+    expect_identical(c(NA, FALSE), re2_detect(c(NA,"sd"),"NA"))
+
+    expect_identical(re2_detect(c(NA,"sd"),"NA"),re2_pdetect(c(NA,"sd"),"NA"))
+    expect_identical(re2_detect(c(NA,"sd"),"sd"),re2_pdetect(c(NA,"sd"),"sd"))
+
+})