use 'Rf_mkCharLenCE() as appropriate

RcppCore · Oct 25, 2018 · 8c26ace · 8c26ace
1 parent a669a19
commit 8c26ace
Show file tree

Hide file tree

Showing 5 changed files with 25 additions and 4 deletions.
diff --git a/ChangeLog b/ChangeLog
@@ -1,3 +1,9 @@
+2018-10-25  Kevin Ushey  <kevinushey@gmail.com>
+
+	* inst/include/Rcpp/String.h: Use Rf_mkCharLenCE() as appropriate
+	* inst/unitTests/cpp/String.cpp: Add unit tests
+	* inst/unitTests/runit.String.R: Add unit tests
+
 2018-10-12  Dirk Eddelbuettel  <edd@debian.org>
 
 	* DESCRIPTION (Date, Version): Roll minor version

diff --git a/inst/NEWS.Rd b/inst/NEWS.Rd
@@ -12,6 +12,8 @@
       \item The constructor for \code{NumericMatrix(not_init(n,k))} was
       corrected (Romain in \ghpr{904}, Dirk in \ghpr{905}, and also
       Romain in \ghpr{908} fixing \ghpr{907}).
+      \item Rcpp::String no longer silently drops embedded NUL bytes
+      in strings -- instead, an R error is signaled. (\ghit{916})
     }
     \item Changes in Rcpp Deployment:
     \itemize{

diff --git a/inst/include/Rcpp/String.h b/inst/include/Rcpp/String.h
@@ -366,7 +366,7 @@ namespace Rcpp {
 
         inline SEXP get_sexp() const {
             RCPP_STRING_DEBUG_1("String::get_sexp const (valid = %d) ", valid);
-            return valid ? data : Rf_mkCharCE(buffer.c_str(), enc);
+            return valid ? data : Rf_mkCharLenCE(buffer.c_str(), buffer.size(), enc);
         }
 
         inline SEXP get_sexp() {
@@ -395,9 +395,12 @@ namespace Rcpp {
             enc = encoding;
 
             if (valid) {
-                data = Rcpp_ReplaceObject(data, Rf_mkCharCE(Rf_translateCharUTF8(data), encoding));
+                const void* vmax = vmaxget();
+                const char* translated = Rf_translateCharUTF8(data);
+                data = Rcpp_ReplaceObject(data, Rf_mkCharCE(translated, encoding));
+                vmaxset(vmax);
             } else {
-                data = Rf_mkCharCE(buffer.c_str(), encoding);
+                data = Rf_mkCharLenCE(buffer.c_str(), buffer.size(), encoding);
                 Rcpp_PreserveObject(data);
                 valid = true;
             }
@@ -469,7 +472,7 @@ namespace Rcpp {
         inline void setData() {
             RCPP_STRING_DEBUG("setData");
             if (!valid) {
-                data = Rf_mkCharCE(buffer.c_str(), enc);
+                data = Rf_mkCharLenCE(buffer.c_str(), buffer.size(), enc);
                 Rcpp_PreserveObject(data);
                 valid = true;
             }

diff --git a/inst/unitTests/cpp/String.cpp b/inst/unitTests/cpp/String.cpp
@@ -114,3 +114,9 @@ String test_String_ctor_encoding2() {
     y.set_encoding(CE_UTF8);
     return y;
 }
+
+// [[Rcpp::export]]
+String test_String_embeddedNul() {
+    std::string bad("abc\0abc", 7);
+    return String(bad);
+}
diff --git a/inst/unitTests/runit.String.R b/inst/unitTests/runit.String.R
@@ -94,4 +94,8 @@ if (.runThisTest) {
         checkEquals(Encoding(test_String_ctor_encoding2()), "UTF-8")
     }
 
+    test.String.embeddedNul <- function() {
+        checkException(test_String_embeddedNul())
+    }
+
 }