feat: PCRE2 JIT

Author: wfjsw

src/operators/verify_cc.cc

@@ -36,6 +36,8 @@ namespace operators {
 VerifyCC::~VerifyCC() {
 #if WITH_PCRE2
     pcre2_code_free(m_pc);
+    pcre2_match_context_free(m_pmc);
+    pcre2_jit_stack_free(m_pcjs);
 #else
     if (m_pc != NULL) {
         pcre_free(m_pc);
@@ -105,6 +107,11 @@ bool VerifyCC::init(const std::string &param2, std::string *error) {
     if (m_pc == NULL) {
         return false;
     }
+
+    m_pcje = pcre2_jit_compile(m_pc, PCRE2_JIT_COMPLETE);
+    m_pmc = pcre2_match_context_create(NULL);
+    m_pcjs = pcre2_jit_stack_create(32*1024, 512*1024, NULL);
+    pcre2_jit_stack_assign(m_pmc, NULL, m_pcjs);
 #else
     const char *errptr = NULL;
     int erroffset = 0;
@@ -142,8 +149,14 @@ bool VerifyCC::evaluate(Transaction *t, RuleWithActions *rule,
     PCRE2_SPTR pcre2_i = reinterpret_cast<PCRE2_SPTR>(i.c_str());
     pcre2_match_data *match_data = pcre2_match_data_create_from_pattern(m_pc, NULL);
 
+    int ret;
     for (offset = 0; offset < target_length; offset++) {
-        int ret = pcre2_match(m_pc, pcre2_i, target_length, offset, 0, match_data, NULL);
+
+        if (m_pcje == 0) {
+            ret = pcre2_jit_match(m_pc, pcre2_i, target_length, offset, 0, match_data, m_pmc);
+        } else {
+            ret = pcre2_match(m_pc, pcre2_i, target_length, offset, 0, match_data, m_pmc);
+        }
 
         /* If there was no match, then we are done. */
         if (ret < 0) {

src/operators/verify_cc.h

@@ -53,6 +53,9 @@ class VerifyCC : public Operator {
  private:
 #if WITH_PCRE2
     pcre2_code *m_pc;
+    pcre2_match_context *m_pmc;
+    int m_pcje;
+    pcre2_jit_stack *m_pcjs;
 #else
     pcre *m_pc;
     pcre_extra *m_pce;

src/utils/regex.cc

@@ -73,6 +73,11 @@ Regex::Regex(const std::string& pattern_, bool ignoreCase)
     PCRE2_SIZE erroroffset = 0;
     m_pc = pcre2_compile(pcre2_pattern, PCRE2_ZERO_TERMINATED,
         pcre2_options, &errornumber, &erroroffset, NULL);
+
+    m_pcje = pcre2_jit_compile(m_pc, PCRE2_JIT_COMPLETE);
+    m_pmc = pcre2_match_context_create(NULL);
+    m_pcjs = pcre2_jit_stack_create(32*1024, 512*1024, NULL);
+    pcre2_jit_stack_assign(m_pmc, NULL, m_pcjs);
 #else
     const char *errptr = NULL;
     int erroffset;
@@ -92,6 +97,8 @@ Regex::Regex(const std::string& pattern_, bool ignoreCase)
 Regex::~Regex() {
 #if WITH_PCRE2
     pcre2_code_free(m_pc);
+    pcre2_match_context_free(m_pmc);
+    pcre2_jit_stack_free(m_pcjs);
 #else
     if (m_pc != NULL) {
         pcre_free(m_pc);
@@ -118,8 +125,13 @@ std::list<SMatch> Regex::searchAll(const std::string& s) const {
 
     pcre2_match_data *match_data = pcre2_match_data_create_from_pattern(m_pc, NULL);
     do {
-        rc = pcre2_match(m_pc, pcre2_s, s.length(),
-                         offset, 0, match_data, NULL);
+        if (m_pcje == 0) {
+            rc = pcre2_jit_match(m_pc, pcre2_s, s.length(),
+                            offset, 0, match_data, m_pmc);
+        } else {
+            rc = pcre2_match(m_pc, pcre2_s, s.length(),
+                            offset, 0, match_data, m_pmc);
+        }
         PCRE2_SIZE *ovector = pcre2_get_ovector_pointer(match_data);
 #else
     const char *subject = s.c_str();
@@ -159,7 +171,12 @@ bool Regex::searchOneMatch(const std::string& s, std::vector<SMatchCapture>& cap
 #ifdef WITH_PCRE2
     PCRE2_SPTR pcre2_s = reinterpret_cast<PCRE2_SPTR>(s.c_str());
     pcre2_match_data *match_data = pcre2_match_data_create_from_pattern(m_pc, NULL);
-    int rc = pcre2_match(m_pc, pcre2_s, s.length(), 0, 0, match_data, NULL);
+    int rc;
+    if (m_pcje == 0) {
+        rc = pcre2_jit_match(m_pc, pcre2_s, s.length(), 0, 0, match_data, m_pmc);
+    } else {
+        rc = pcre2_match(m_pc, pcre2_s, s.length(), 0, 0, match_data, m_pmc);
+    }
     PCRE2_SIZE *ovector = pcre2_get_ovector_pointer(match_data);
 #else
     const char *subject = s.c_str();
@@ -198,7 +215,7 @@ bool Regex::searchGlobal(const std::string& s, std::vector<SMatchCapture>& captu
             pcre2_options = PCRE2_NOTEMPTY_ATSTART | PCRE2_ANCHORED;
         }
         int rc = pcre2_match(m_pc, pcre2_s, s.length(),
-                         startOffset, pcre2_options, match_data, NULL);
+                            startOffset, pcre2_options, match_data, m_pmc);
         PCRE2_SIZE *ovector = pcre2_get_ovector_pointer(match_data);
 
 #else
@@ -270,9 +287,14 @@ int Regex::search(const std::string& s, SMatch *match) const {
 #ifdef WITH_PCRE2
     PCRE2_SPTR pcre2_s = reinterpret_cast<PCRE2_SPTR>(s.c_str());
     pcre2_match_data *match_data = pcre2_match_data_create_from_pattern(m_pc, NULL);
-    int ret = pcre2_match(m_pc, pcre2_s, s.length(),
-        0, 0, match_data, NULL) > 0;
-
+    int ret;
+    if (m_pcje == 0) {
+        ret = pcre2_match(m_pc, pcre2_s, s.length(),
+            0, 0, match_data, m_pmc) > 0;
+    } else {
+        ret = pcre2_match(m_pc, pcre2_s, s.length(),
+            0, 0, match_data, m_pmc) > 0;
+    }
     if (ret > 0) { // match
         PCRE2_SIZE *ovector = pcre2_get_ovector_pointer(match_data);
 #else
@@ -297,7 +319,12 @@ int Regex::search(const std::string& s) const {
 #ifdef WITH_PCRE2
     PCRE2_SPTR pcre2_s = reinterpret_cast<PCRE2_SPTR>(s.c_str());
     pcre2_match_data *match_data = pcre2_match_data_create_from_pattern(m_pc, NULL);
-    int rc = pcre2_match(m_pc, pcre2_s, s.length(), 0, 0, match_data, NULL);
+    int rc;
+    if (m_pcje == 0) {
+        rc = pcre2_jit_match(m_pc, pcre2_s, s.length(), 0, 0, match_data, m_pmc);
+    } else {
+        rc = pcre2_match(m_pc, pcre2_s, s.length(), 0, 0, match_data, m_pmc);
+    }
     pcre2_match_data_free(match_data);
     if (rc > 0) {
         return 1; // match

src/utils/regex.h

@@ -85,6 +85,9 @@ class Regex {
  private:
 #if WITH_PCRE2
     pcre2_code *m_pc;
+    pcre2_match_context *m_pmc;
+    int m_pcje;
+    pcre2_jit_stack *m_pcjs;
 #else
     pcre *m_pc = NULL;
     pcre_extra *m_pce = NULL;