Use Abseil's character class functions.

They are locale-agnostic whereas the libc equivalents aren't...
and thus incur indirections due to checking the current locale.

Change-Id: I5e1350ff33c3090c345b4166afb6b80d92d4a0df
Reviewed-on: https://code-review.googlesource.com/c/re2/+/61890
Reviewed-by: Alex Chernyakhovsky <[email protected]>
Reviewed-by: Paul Wankadia <[email protected]>

re2/fuzzing/re2_fuzzer.cc

@@ -105,7 +105,7 @@ void TestOneInput(absl::string_view pattern, const RE2::Options& options,
   // counted repetition is involved - whereas the marginal benefit is zero.
   // Crudely limit the use of 'k', 'K', 's' and 'S' too because they become
   // three-element character classes when case-insensitive and using UTF-8.
-  // TODO(junyer): Handle [:isalnum:] et al. when they start to cause pain.
+  // TODO(junyer): Handle [[:alnum:]] et al. when they start to cause pain.
   int char_class = 0;
   int backslash_p = 0;  // very expensive, so handle specially
   for (size_t i = 0; i < pattern.size(); i++) {

re2/parse.cc

@@ -26,6 +26,7 @@
 #include <vector>
 
 #include "absl/base/macros.h"
+#include "absl/strings/ascii.h"
 #include "util/logging.h"
 #include "util/utf.h"
 #include "re2/pod_array.h"
@@ -1322,14 +1323,14 @@ bool Regexp::ParseState::MaybeConcatString(int r, ParseFlags flags) {
 // Parses a decimal integer, storing it in *np.
 // Sets *s to span the remainder of the string.
 static bool ParseInteger(absl::string_view* s, int* np) {
-  if (s->empty() || !isdigit((*s)[0] & 0xFF))
+  if (s->empty() || !absl::ascii_isdigit((*s)[0] & 0xFF))
     return false;
   // Disallow leading zeros.
-  if (s->size() >= 2 && (*s)[0] == '0' && isdigit((*s)[1] & 0xFF))
+  if (s->size() >= 2 && (*s)[0] == '0' && absl::ascii_isdigit((*s)[1] & 0xFF))
     return false;
   int n = 0;
   int c;
-  while (!s->empty() && isdigit(c = (*s)[0] & 0xFF)) {
+  while (!s->empty() && absl::ascii_isdigit(c = (*s)[0] & 0xFF)) {
     // Avoid overflow.
     if (n >= 100000000)
       return false;
@@ -1468,7 +1469,7 @@ static bool ParseEscape(absl::string_view* s, Rune* rp,
   int code;
   switch (c) {
     default:
-      if (c < Runeself && !isalpha(c) && !isdigit(c)) {
+      if (c < Runeself && !absl::ascii_isalnum(c)) {
         // Escaped non-word characters are always themselves.
         // PCRE is not quite so rigorous: it accepts things like
         // \q, but we don't.  We once rejected \_, but too many

re2/re2.cc

@@ -27,6 +27,7 @@
 
 #include "absl/base/macros.h"
 #include "absl/container/fixed_array.h"
+#include "absl/strings/ascii.h"
 #include "absl/strings/str_format.h"
 #include "util/logging.h"
 #include "util/strutil.h"
@@ -975,7 +976,7 @@ bool RE2::CheckRewriteString(absl::string_view rewrite,
     if (c == '\\') {
       continue;
     }
-    if (!isdigit(c)) {
+    if (!absl::ascii_isdigit(c)) {
       *error = "Rewrite schema error: "
                "'\\' must be followed by a digit or '\\'.";
       return false;
@@ -1005,7 +1006,7 @@ int RE2::MaxSubmatch(absl::string_view rewrite) {
     if (*s == '\\') {
       s++;
       int c = (s < end) ? *s : -1;
-      if (isdigit(c)) {
+      if (absl::ascii_isdigit(c)) {
         int n = (c - '0');
         if (n > max)
           max = n;
@@ -1029,7 +1030,7 @@ bool RE2::Rewrite(std::string* out,
     }
     s++;
     int c = (s < end) ? *s : -1;
-    if (isdigit(c)) {
+    if (absl::ascii_isdigit(c)) {
       int n = (c - '0');
       if (n >= veclen) {
         if (options_.log_errors()) {
@@ -1110,13 +1111,13 @@ static const char* TerminateNumber(char* buf, size_t nbuf, const char* str,
                                    size_t* np, bool accept_spaces) {
   size_t n = *np;
   if (n == 0) return "";
-  if (n > 0 && isspace(*str)) {
+  if (n > 0 && absl::ascii_isspace(*str)) {
     // We are less forgiving than the strtoxxx() routines and do not
     // allow leading spaces. We do allow leading spaces for floats.
     if (!accept_spaces) {
       return "";
     }
-    while (n > 0 && isspace(*str)) {
+    while (n > 0 && absl::ascii_isspace(*str)) {
       n--;
       str++;
     }