Fix build/re2.m4 file lookup

There's no such include file as "re2/re2_parse.h",
and there was a typo in RE2_POSSIBLE_PATHS.

build/re2.m4

@@ -31,7 +31,7 @@ AC_ARG_WITH(
 )
 
 if test "x${with_re2}" == "xno"; then
-    AC_DEFINE(HAVE_GEOIP, 0, [Support for RE2 was disabled by the utilization of --without-re2 or --with-re2=no])
+    AC_DEFINE(HAVE_RE2, 0, [Support for RE2 was disabled by the utilization of --without-re2 or --with-re2=no])
     AC_MSG_NOTICE([Support for RE2 was disabled by the utilization of --without-re2 or --with-re2=no])
     RE2_DISABLED=yes
 else
@@ -48,7 +48,7 @@ else
 
 #    if test "x${with_re2}" != "xyes" or test "x${with_re2}" == "xyes"; then
         if test "x${with_re2}" == "x" || test "x${with_re2}" == "xyes"; then
-            # Nothing about GeoIP was informed, using the pkg-config to figure things out.
+            # Nothing about RE2 was informed, using the pkg-config to figure things out.
             if test -n "${PKG_CONFIG}"; then
                 RE2_PKG_NAME=""
                 for x in ${RE2_POSSIBLE_LIB_NAMES}; do
@@ -58,7 +58,7 @@ else
                     fi
                 done
             fi
-            AC_MSG_NOTICE([Nothing about GeoIP was informed during the configure phase. Trying to detect it on the platform...])
+            AC_MSG_NOTICE([Nothing about RE2 was informed during the configure phase. Trying to detect it on the platform...])
             if test -n "${RE2_PKG_NAME}"; then
                 # Package was found using the pkg-config scripts
                 RE2_VERSION="`${PKG_CONFIG} ${RE2_PKG_NAME} --modversion`"
@@ -68,7 +68,7 @@ else
                 RE2_DISPLAY="${RE2_LDADD}, ${RE2_CFLAGS}"
             else
                 # If pkg-config did not find anything useful, go over file lookup.
-                for x in ${RE2_POSSIBLE_LIB_NAMES}; do
+                for x in ${RE2_POSSIBLE_PATHS}; do
                     CHECK_FOR_RE2_AT(${x})
                     if test -n "${RE2_VERSION}"; then
                         break
@@ -149,11 +149,11 @@ AC_DEFUN([CHECK_FOR_RE2_AT], [
            break
        fi
     done
-    if test -e "${path}/include/re2_parse.h"; then
+    if test -e "${path}/include/re2.h"; then
         re2_inc_path="${path}/include"
-    elif test -e "${path}/re2_parse.h"; then
+    elif test -e "${path}/re2.h"; then
         re2_inc_path="${path}"
-    elif test -e "${path}/include/re2/re2_parse.h"; then
+    elif test -e "${path}/include/re2/re2.h"; then
         re2_inc_path="${path}/include"
     fi
 

src/regex/backend/re2.cc

@@ -44,55 +44,71 @@ Re2::Re2(const std::string& pattern)
 {
 }
 
-std::list<RegexMatch> Re2::searchAll(const std::string& s) const {
+static bool do_match(
-    std::list<RegexMatch> retList;
+        const RE2 &re,
+        const char *s,
+        size_t n,
+        RegexMatch *m,
+        ssize_t max_groups,
+        size_t offset)
+{
+    if (m == nullptr) {
+        max_groups = 0;
+    }
 
-    re2::StringPiece subject(s);
+    // "+1" is required for full match (aka group 0)
+    size_t ngroups = re.NumberOfCapturingGroups() + 1;
+    if (max_groups >= 0 && max_groups < ngroups) {
+        ngroups = max_groups;
+    }
+    re2::StringPiece submatches[ngroups];
 
+    if (re.Match(re2::StringPiece(s, n), offset, n, RE2::UNANCHORED,
+        &submatches[0], ngroups)) {
+        if (ngroups != 0) {
+            RegexMatch::MatchGroupContainer groups;
+            groups.reserve(ngroups);
+            for (size_t i = 0; i < ngroups; i++) {
+                size_t start = submatches[i].data() - s;
+                std::string group = submatches[i].as_string();
+                groups.push_back(MatchGroup{start, std::move(group)});
+            }
+            *m = RegexMatch(std::move(groups));
+        }
+        return true;
+    }
+    return false;
+}
+
+std::vector<RegexMatch> Re2::searchAll(const std::string& s, bool overlapping) const {
+    std::vector<RegexMatch> res;
     size_t offset = 0;
-    while (offset <= s.size()) {
-        int ngroups = re.NumberOfCapturingGroups() + 1;
-        re2::StringPiece submatches[ngroups];
 
-        if (!re.Match(subject, offset, s.size(), RE2::UNANCHORED,
+    while (1) {
-            &submatches[0], ngroups)) {
+        RegexMatch m;
-            break;
+        bool match = do_match(re, s.data(), s.size(), &m, -1, offset);
-        }
+        if (!match) break;
 
-        for (int i = 0; i < ngroups; i++) {
+        if (overlapping) {
-            // N.B. StringPiece::as_string returns value, not reference
+            // start just after the beginning of the last match
-            auto match_string =  submatches[i].as_string();
+            offset = m.group(0).offset + 1;
-            auto start = &submatches[i][0] - &subject[0];
+        } else {
-            retList.push_front(RegexMatch(std::move(match_string), start));
+            // start just at the end of the last match
-        }
+            offset = m.group(0).offset + m.group(0).string.size();
-
+            if (offset == m.group(0).offset) {
-        offset = (&submatches[0][0] - &subject[0]) + submatches[0].length();
+                // empty match - advance by one to not match empty string repeatedly
-        if (submatches[0].size() == 0) {
+                offset++;
-            offset++;
+            }
         }
+        res.push_back(std::move(m));
     }
-
+    return res;
-    return retList;
 }
 
-int Re2::search(const std::string& s, RegexMatch *match) const {
+bool Re2::search(const std::string &s, RegexMatch *m, ssize_t max_groups) const {
-    re2::StringPiece subject(s);
+    return do_match(re, s.data(), s.size(), m, max_groups, 0);
-    re2::StringPiece submatches[1];
-    if (re.Match(subject, 0, s.size(), RE2::UNANCHORED, &submatches[0], 1)) {
-        // N.B. StringPiece::as_string returns value, not reference
-        auto match_string = submatches[0].as_string();
-        auto start = &submatches[0][0] - &subject[0];
-        *match = RegexMatch(std::move(match_string), start);
-        return 1;
-    } else {
-        return 0;
-    }
 }
 
-int Re2::search(const std::string& s) const {
-    re2::StringPiece subject(s);
-    return re.Match(subject, 0, s.size(), RE2::UNANCHORED, NULL, 0);
-}
 #endif
 
 }  // namespace backend