re2: parse float/double args with locale-independent from_chars

re2/re2.cc

@@ -19,6 +19,7 @@
 #include <atomic>
 #include <map>
 #include <string>
+#include <system_error>
 #include <utility>
 #include <vector>
 
@@ -28,6 +29,7 @@
 #include "absl/log/absl_check.h"
 #include "absl/log/absl_log.h"
 #include "absl/strings/ascii.h"
+#include "absl/strings/charconv.h"
 #include "absl/strings/str_format.h"
 #include "absl/strings/string_view.h"
 #include "re2/prog.h"
@@ -1176,11 +1178,12 @@ bool Parse(const char* str, size_t n, float* dest) {
   static const int kMaxLength = 200;
   char buf[kMaxLength+1];
   str = TerminateNumber(buf, sizeof buf, str, &n, true);
-  char* end;
-  errno = 0;
-  float r = strtof(str, &end);
-  if (end != str + n) return false;   // Leftover junk
-  if (errno) return false;
+  if (str[0] == '\0') return false;
+  // Use absl::from_chars() instead of strtof() to avoid the radix character
+  // (and thus the accepted grammar) depending on the process's LC_NUMERIC.
+  float r;
+  absl::from_chars_result res = absl::from_chars(str, str + n, r);
+  if (res.ec != std::errc() || res.ptr != str + n) return false;  // Leftover junk
   if (dest == NULL) return true;
   *dest = r;
   return true;
@@ -1192,11 +1195,12 @@ bool Parse(const char* str, size_t n, double* dest) {
   static const int kMaxLength = 200;
   char buf[kMaxLength+1];
   str = TerminateNumber(buf, sizeof buf, str, &n, true);
-  char* end;
-  errno = 0;
-  double r = strtod(str, &end);
-  if (end != str + n) return false;   // Leftover junk
-  if (errno) return false;
+  if (str[0] == '\0') return false;
+  // Use absl::from_chars() instead of strtod() to avoid the radix character
+  // (and thus the accepted grammar) depending on the process's LC_NUMERIC.
+  double r;
+  absl::from_chars_result res = absl::from_chars(str, str + n, r);
+  if (res.ec != std::errc() || res.ptr != str + n) return false;  // Leftover junk
   if (dest == NULL) return true;
   *dest = r;
   return true;

re2/testing/re2_test.cc

@@ -8,6 +8,7 @@
 #include "re2/re2.h"
 
 #include <errno.h>
+#include <locale.h>
 #include <stddef.h>
 #include <stdint.h>
 #include <string.h>
@@ -949,6 +950,37 @@ TEST(RE2, FloatingPointFullMatchTypes) {
   }
 }
 
+TEST(RE2, FloatingPointFullMatchIsLocaleIndependent) {
+  // Parsing must not depend on LC_NUMERIC: '.' is always the radix character
+  // regardless of the process locale. Skip if no comma-radix locale is present.
+  const char* saved = setlocale(LC_NUMERIC, NULL);
+  std::string savedcopy = saved != NULL ? saved : "C";
+  static const char* const kCommaLocales[] = {
+      "de_DE.UTF-8", "de_DE.utf8", "de_DE", "fr_FR.UTF-8", "nl_NL.UTF-8",
+  };
+  bool have_locale = false;
+  for (const char* name : kCommaLocales) {
+    if (setlocale(LC_NUMERIC, name) != NULL) {
+      have_locale = true;
+      break;
+    }
+  }
+  if (!have_locale)
+    GTEST_SKIP() << "no comma-radix locale available";
+
+  double d = 0;
+  EXPECT_TRUE(RE2::FullMatch("1.5", "(.*)", &d));
+  EXPECT_EQ(d, 1.5);
+  EXPECT_FALSE(RE2::FullMatch("1,5", "(.*)", &d));
+
+  float f = 0;
+  EXPECT_TRUE(RE2::FullMatch("1.5", "(.*)", &f));
+  EXPECT_EQ(f, 1.5f);
+  EXPECT_FALSE(RE2::FullMatch("1,5", "(.*)", &f));
+
+  setlocale(LC_NUMERIC, savedcopy.c_str());
+}
+
 TEST(RE2, FullMatchAnchored) {
   int i;
   // Check that matching is fully anchored