Imported Upstream version 20150601upstream/20150601

Change-Id: Idfa2fac0fcc3d292a47b4eda4d8b2cf0c527eb56
Signed-off-by: DongHun Kwak <dh0128.kwak@samsung.com>

10 files changed, 55 insertions, 17 deletions

diff --git a/Makefile b/Makefile
index f4d6b52..fd93441 100644
--- a/Makefile
+++ b/Makefile
@@ -245,18 +245,21 @@ shared-bigtest: $(STESTS) $(SBIGTESTS)
 benchmark: obj/test/regexp_benchmark
 
 install: obj/libre2.a obj/so/libre2.so
-	mkdir -p $(DESTDIR)$(includedir)/re2 $(DESTDIR)$(libdir)
+	mkdir -p $(DESTDIR)$(includedir)/re2 $(DESTDIR)$(libdir)/pkgconfig
 	$(INSTALL_DATA) $(INSTALL_HFILES) $(DESTDIR)$(includedir)/re2
 	$(INSTALL) obj/libre2.a $(DESTDIR)$(libdir)/libre2.a
 	$(INSTALL) obj/so/libre2.so $(DESTDIR)$(libdir)/libre2.so.$(SONAME).0.0
 	ln -sf libre2.so.$(SONAME).0.0 $(DESTDIR)$(libdir)/libre2.so.$(SONAME)
 	ln -sf libre2.so.$(SONAME).0.0 $(DESTDIR)$(libdir)/libre2.so
+	sed -e "s#@prefix@#${prefix}#" re2.pc >$(DESTDIR)$(libdir)/pkgconfig/re2.pc
 
 testinstall:
 	@mkdir -p obj
 	cp testinstall.cc obj
+ifneq ($(shell uname),Darwin)
 	(cd obj && $(CXX) -I$(DESTDIR)$(includedir) -L$(DESTDIR)$(libdir) testinstall.cc -lre2 -pthread -static -o testinstall)
 	obj/testinstall
+endif
 	(cd obj && $(CXX) -I$(DESTDIR)$(includedir) -L$(DESTDIR)$(libdir) testinstall.cc -lre2 -pthread -o testinstall)
 	LD_LIBRARY_PATH=$(DESTDIR)$(libdir) obj/testinstall
 
diff --git a/README b/README
index 3143fbe..3cba956 100644
--- a/README
+++ b/README
@@ -25,5 +25,6 @@ under the BSD-style license found in the LICENSE file.
 RE2's native language is C++.
 An Erlang wrapper is at https://github.com/tuncer/re2/.
 An Inferno wrapper is at https://github.com/powerman/inferno-re2/.
+A Perl wrapper is at https://github.com/dgl/re-engine-RE2/ and on CPAN. 
 A Python wrapper is at https://github.com/facebook/pyre2/.
 A Ruby wrapper is at https://github.com/axic/rre2/.
diff --git a/re2.pc b/re2.pc
new file mode 100644
index 0000000..9e90cda
--- /dev/null
+++ b/re2.pc
@@ -0,0 +1,10 @@
+prefix=@prefix@
+exec_prefix=${prefix}
+includedir=${prefix}/include
+libdir=${exec_prefix}/lib
+
+Name: re2
+Description: RE2 is a fast, safe, thread-friendly regular expression engine.
+Version: 0.0.0
+Cflags: -I${includedir}
+Libs: -L${libdir} -lre2 -pthread
diff --git a/re2/compile.cc b/re2/compile.cc
index 4d84ef3..6478def 100644
--- a/re2/compile.cc
+++ b/re2/compile.cc
@@ -815,7 +815,8 @@ Frag Compiler::PostVisit(Regexp* re, Frag, Frag, Frag* child_frags,
         // If this range contains all of A-Za-z or none of it,
         // the fold flag is unnecessary; don't bother.
         bool fold = foldascii;
-        if ((i->lo <= 'A' && 'z' <= i->hi) || i->hi < 'A' || 'z' < i->lo)
+        if ((i->lo <= 'A' && 'z' <= i->hi) || i->hi < 'A' || 'z' < i->lo ||
+            ('Z' < i->lo && i->hi < 'a'))
           fold = false;
 
         AddRuneRange(i->lo, i->hi, fold);
diff --git a/re2/parse.cc b/re2/parse.cc
index 4f45041..e6b27d2 100644
--- a/re2/parse.cc
+++ b/re2/parse.cc
@@ -216,6 +216,7 @@ bool Regexp::ParseState::PushRegexp(Regexp* re) {
   // analysis does better with fewer character classes.
   // Similarly, [Aa] can be rewritten as a literal A with ASCII case folding.
   if (re->op_ == kRegexpCharClass) {
+    re->ccb_->RemoveAbove(rune_max_);
     if (re->ccb_->size() == 1) {
       Rune r = re->ccb_->begin()->lo;
       re->Decref();
@@ -378,7 +379,6 @@ bool Regexp::ParseState::PushLiteral(Rune r) {
       }
       r = CycleFoldRune(r);
     } while (r != r1);
-    re->ccb_->RemoveAbove(rune_max_);
     return PushRegexp(re);
   }
 
@@ -1832,7 +1832,6 @@ bool Regexp::ParseState::ParseCharClass(StringPiece* s,
 
   if (negated)
     re->ccb_->Negate();
-  re->ccb_->RemoveAbove(rune_max_);
 
   *out_re = re;
   return true;
diff --git a/re2/re2.cc b/re2/re2.cc
index 3a2262e..4e8af90 100644
--- a/re2/re2.cc
+++ b/re2/re2.cc
@@ -1142,7 +1142,7 @@ bool RE2::Arg::parse_uint_radix(const char* str,
   return true;
 }
 
-#ifdef RE2_HAVE_LONGLONG
+#if RE2_HAVE_LONGLONG
 bool RE2::Arg::parse_longlong_radix(const char* str,
                                    int n,
                                    void* dest,
diff --git a/re2/re2.h b/re2/re2.h
index 6bf7ac6..bad75bb 100644
--- a/re2/re2.h
+++ b/re2/re2.h
@@ -686,7 +686,7 @@ class RE2 {
   static inline Arg CRadix(unsigned int* x);
   static inline Arg CRadix(long* x);
   static inline Arg CRadix(unsigned long* x);
-  #ifdef RE2_HAVE_LONGLONG
+  #if RE2_HAVE_LONGLONG
   static inline Arg CRadix(long long* x);
   static inline Arg CRadix(unsigned long long* x);
   #endif
@@ -697,7 +697,7 @@ class RE2 {
   static inline Arg Hex(unsigned int* x);
   static inline Arg Hex(long* x);
   static inline Arg Hex(unsigned long* x);
-  #ifdef RE2_HAVE_LONGLONG
+  #if RE2_HAVE_LONGLONG
   static inline Arg Hex(long long* x);
   static inline Arg Hex(unsigned long long* x);
   #endif
@@ -708,7 +708,7 @@ class RE2 {
   static inline Arg Octal(unsigned int* x);
   static inline Arg Octal(long* x);
   static inline Arg Octal(unsigned long* x);
-  #ifdef RE2_HAVE_LONGLONG
+  #if RE2_HAVE_LONGLONG
   static inline Arg Octal(long long* x);
   static inline Arg Octal(unsigned long long* x);
   #endif
@@ -791,7 +791,7 @@ class RE2::Arg {
   MAKE_PARSER(unsigned int,       parse_uint);
   MAKE_PARSER(long,               parse_long);
   MAKE_PARSER(unsigned long,      parse_ulong);
-  #ifdef RE2_HAVE_LONGLONG
+  #if RE2_HAVE_LONGLONG
   MAKE_PARSER(long long,          parse_longlong);
   MAKE_PARSER(unsigned long long, parse_ulonglong);
   #endif
@@ -839,7 +839,7 @@ class RE2::Arg {
   DECLARE_INTEGER_PARSER(uint);
   DECLARE_INTEGER_PARSER(long);
   DECLARE_INTEGER_PARSER(ulong);
-  #ifdef RE2_HAVE_LONGLONG
+  #if RE2_HAVE_LONGLONG
   DECLARE_INTEGER_PARSER(longlong);
   DECLARE_INTEGER_PARSER(ulonglong);
   #endif
@@ -869,7 +869,7 @@ MAKE_INTEGER_PARSER(int,                int)
 MAKE_INTEGER_PARSER(unsigned int,       uint)
 MAKE_INTEGER_PARSER(long,               long)
 MAKE_INTEGER_PARSER(unsigned long,      ulong)
-#ifdef RE2_HAVE_LONGLONG
+#if RE2_HAVE_LONGLONG
 MAKE_INTEGER_PARSER(long long,          longlong)
 MAKE_INTEGER_PARSER(unsigned long long, ulonglong)
 #endif
diff --git a/re2/testing/compile_test.cc b/re2/testing/compile_test.cc
index 8d92105..d438b19 100644
--- a/re2/testing/compile_test.cc
+++ b/re2/testing/compile_test.cc
@@ -99,6 +99,10 @@ static Test tests[] = {
   { "[Aa]",
     "1. byte/i [61-61] -> 2\n"
     "2. match! 0\n" },
+  // Issue 20992936
+  { "[[-`]",
+    "1. byte [5b-60] -> 2\n"
+    "2. match! 0\n" },
 };
 
 TEST(TestRegexpCompileToProg, Simple) {
diff --git a/re2/testing/re2_test.cc b/re2/testing/re2_test.cc
index 8505195..ad15b34 100644
--- a/re2/testing/re2_test.cc
+++ b/re2/testing/re2_test.cc
@@ -1511,4 +1511,15 @@ TEST(RE2, Bug18523943) {
   CHECK(!RE2::PartialMatch((const char*)a, re, &s1));
 }
 
+TEST(RE2, Bug21371806) {
+  // Bug in parser accepting Unicode groups in Latin-1 mode,
+  // causing compiler to fail in DCHECK in prog.cc.
+
+  RE2::Options opt;
+  opt.set_encoding(RE2::Options::EncodingLatin1);
+
+  RE2 re("g\\p{Zl}]", opt);
+  CHECK(re.ok());
+}
+
 }  // namespace re2
diff --git a/util/util.h b/util/util.h
index 250ee93..a7ccafe 100644
--- a/util/util.h
+++ b/util/util.h
@@ -9,14 +9,14 @@
 #include <stdio.h>
 #include <string.h>
 #include <stdint.h>
-#include <stddef.h>         // For size_t
+#include <stddef.h>     // For size_t
 #include <assert.h>
 #include <stdarg.h>
 #include <time.h>
-#include <ctype.h>	// For isdigit, isalpha.
+#include <ctype.h>      // For isdigit, isalpha
 
 #if !defined(WIN32)
-#include <sys/time.h>  // For gettimeofday
+#include <sys/time.h>   // For gettimeofday
 #endif
 
 // C++
@@ -45,7 +45,7 @@ using std::sort;
 using std::swap;
 using std::make_pair;
 
-#if defined(__GNUC__) && !defined(USE_CXX0X) && !defined(_LIBCPP_ABI_VERSION) && !defined(OS_ANDROID)
+#if defined(__GNUC__) && !defined(USE_CXX0X) && !defined(_LIBCPP_ABI_VERSION)
 
 #include <tr1/unordered_set>
 using std::tr1::unordered_set;
@@ -53,7 +53,7 @@ using std::tr1::unordered_set;
 #else
 
 #include <unordered_set>
-#if defined(WIN32) || defined(OS_ANDROID)
+#if defined(WIN32)
 using std::tr1::unordered_set;
 #else
 using std::unordered_set;
@@ -92,13 +92,22 @@ typedef unsigned long ulong;
 typedef unsigned int uint;
 typedef unsigned short ushort;
 
+// Prevent the compiler from complaining about or optimizing away variables
+// that appear unused.
+#undef ATTRIBUTE_UNUSED
+#if defined(__GNUC__)
+#define ATTRIBUTE_UNUSED __attribute__ ((unused))
+#else
+#define ATTRIBUTE_UNUSED
+#endif
+
 // COMPILE_ASSERT causes a compile error about msg if expr is not true.
 #if __cplusplus >= 201103L
 #define COMPILE_ASSERT(expr, msg) static_assert(expr, #msg)
 #else
 template<bool> struct CompileAssert {};
 #define COMPILE_ASSERT(expr, msg) \
-  typedef CompileAssert<(bool(expr))> msg[bool(expr) ? 1 : -1]
+  typedef CompileAssert<(bool(expr))> msg[bool(expr) ? 1 : -1] ATTRIBUTE_UNUSED
 #endif
 
 // DISALLOW_COPY_AND_ASSIGN disallows the copy and operator= functions.