perf. improvement/rx: Only compute dynamic regex in case of macro

On #1528 was added the support for macro expansion on @rx operator.
The performance improvement suggested on the pull request was not
thread safe, therefore removed. This patch adds a performance
improvement on top of #1528. The benchmarks points to 10x faster
results on OWASP CRS.

src/macro_expansion.cc

@@ -48,6 +48,18 @@ std::string MacroExpansion::expand(const std::string& input,
 }
 
 
+bool MacroExpansion::containsMacro(std::string& input) {
+   size_t pos = input.find("%{");
+    size_t end = input.find("}");
+
+    if (pos == std::string::npos || end == std::string::npos) {
+        return false;
+    }
+
+    return true;
+}
+
+
 std::string MacroExpansion::expand(const std::string& input,
     modsecurity::Rule *rule, Transaction *transaction) {
     std::string res;

src/macro_expansion.h

@@ -47,6 +47,8 @@ class MacroExpansion {
             return toupper(aa) == bb;
         });
     }
+
+    static bool containsMacro(std::string& input);
 };
 
 

src/operators/rx.cc

@@ -28,6 +28,15 @@ namespace modsecurity {
 namespace operators {
 
 
+bool Rx::init(const std::string &arg, std::string *error) {
+    m_containsMacro = MacroExpansion::containsMacro(m_param);
+    if (m_containsMacro == false) {
+        m_re = new Regex(m_param);
+    }
+
+    return true;
+}
+
 
 bool Rx::evaluate(Transaction *transaction, Rule *rule,
     const std::string& input, std::shared_ptr<RuleMessage> ruleMessage) {
@@ -39,8 +48,12 @@ bool Rx::evaluate(Transaction *transaction, Rule *rule,
         return true;
     }
 
+    if (m_containsMacro) {
         std::string eparam = MacroExpansion::expand(m_param, transaction);
         re = new Regex(eparam);
+    } else {
+        re = m_re;
+    }
 
     matches = re->searchAll(input);
     if (rule && rule->getActionsByName("capture").size() > 0 && transaction) {
@@ -62,7 +75,9 @@ bool Rx::evaluate(Transaction *transaction, Rule *rule,
         logOffset(ruleMessage, i.m_offset, i.m_length);
     }
 
+    if (m_containsMacro) {
         delete re;
+    }
 
     if (matches.size() > 0) {
         return true;

src/operators/rx.h

@@ -36,17 +36,25 @@ class Rx : public Operator {
  public:
     /** @ingroup ModSecurity_Operator */
     Rx(std::string op, std::string param, bool negation)
-        : Operator(op, param, negation) {
+        : Operator(op, param, negation),
+        m_containsMacro(false) {
         }
     Rx(std::string name, std::string param)
-        : Operator(name, param) {
+        : Operator(name, param),
+        m_containsMacro(false) {
     }
     explicit Rx(std::string param)
-        : Operator("Rx", param) {
+        : Operator("Rx", param),
+        m_containsMacro(false) {
     }
 
     ~Rx() {
+        if (m_containsMacro == false && m_re != NULL) {
+            delete m_re;
+            m_re = NULL;
         }
+    }
+
     bool evaluate(Transaction *transaction, Rule *rule,
         const std::string &input) override {
         return evaluate(transaction, NULL, input, NULL);
@@ -58,6 +66,11 @@ class Rx : public Operator {
     bool evaluate(Transaction *transaction, Rule *rule,
         const std::string& input,
         std::shared_ptr<RuleMessage> ruleMessage) override;
+
+    bool init(const std::string &arg, std::string *error);
+ private:
+    bool m_containsMacro;
+    Regex *m_re;
 };
 
 

src/parser/seclang-parser.yy

@@ -1027,6 +1027,11 @@ op_before_init:
     | OPERATOR_RX FREE_TEXT
       {
         OPERATOR_CONTAINER($$, new operators::Rx($2));
+        std::string error;
+        if ($$->init(driver.ref.back(), &error) == false) {
+            driver.error(@0, error);
+            YYERROR;
+        }
       }
     | OPERATOR_STR_EQ FREE_TEXT
       {

test/benchmark/basic_rules.conf

@@ -3,5 +3,3 @@ Include "../../modsecurity.conf-recommended"
 
 Include "owasp-v3/crs-setup.conf.example"
 Include "owasp-v3/rules/*.conf"
-Include "owasp-v3/crs-setup.conf.example"
-Include "owasp-v3/rules/*.conf"

test/benchmark/benchmark.cc

@@ -24,31 +24,8 @@
 
 using modsecurity::Transaction;
 
-char request_header[] =  "" \
-    "GET /tutorials/other/top-20-mysql-best-practices/ HTTP/1.1\n\r" \
-    "Host: net.tutsplus.com\n\r" \
-    "User-Agent: Mozilla/5.0 (Windows; U; Windows NT 6.1; en-US; rv:1.9.1.5)" \
-    " Gecko/20091102 Firefox/3.5.5 (.NET CLR 3.5.30729)\n\r" \
-    "Accept: text/html,application/xhtml+xml,application/xml; " \
-    "q=0.9,*/*;q=0.8\n\r" \
-    "Accept-Language: en-us,en;q=0.5\n\r" \
-    "Accept-Encoding: gzip,deflate\n\r" \
-    "Accept-Charset: ISO-8859-1,utf-8;q=0.7,*;q=0.7\n\r" \
-    "Keep-Alive: 300\n\r" \
-    "Connection: keep-alive\n\r" \
-    "Cookie: PHPSESSID=r2t5uvjq435r4q7ib3vtdjq120\n\r" \
-    "Pragma: no-cache\n\r" \
-    "Cache-Control: no-cache\n\r";
-
 char request_uri[] = "/test.pl?param1=test&para2=test2";
 
-char request_body[] = "";
-
-char response_headers[] = "" \
-    "HTTP/1.1 200 OK\n\r" \
-    "Content-Type: text/xml; charset=utf-8\n\r" \
-    "Content-Length: length\n\r";
-
 unsigned char response_body[] = "" \
     "<?xml version=\"1.0\" encoding=\"utf-8\"?>\n\r" \
     "<soap:Envelope xmlns:xsi=\"http://www.w3.org/2001/XMLSchema-instance\" " \
@@ -69,7 +46,7 @@ const char* const help_message = "Usage: benchmark [num_iterations|-h|-?|--help]
 
 int main(int argc, char *argv[]) {
 
-    unsigned long long NUM_REQUESTS(10000);
+    unsigned long long NUM_REQUESTS(1000000);
 
     if (argc > 1) {
         if (0 == strcmp(argv[1], "-h") ||

test/benchmark/download-owasp-v3-rules.sh

@@ -3,7 +3,7 @@
 
 git clone https://github.com/SpiderLabs/owasp-modsecurity-crs.git owasp-v3
 cd owasp-v3
-git checkout v3.0.0 -b tag3.0.0
+git checkout v3.0.2 -b tag3.0.2
 cd -
 
 echo 'Include "owasp-v3/crs-setup.conf.example"' >> basic_rules.conf