Aligned v3 behavior to the fixed version

digitalwave · Aug 14, 2020 · 1052599 · 1052599
1 parent 1b7f23c
commit 1052599
Show file tree

Hide file tree

Showing 4 changed files with 80 additions and 22 deletions.
diff --git a/CHANGES b/CHANGES
@@ -2,6 +2,7 @@ v0.3 - YYYY-MM-DD
 -----------------
   * refactoring regex.cc for pcre4msc3
   * add `-f` argument for pcre4msc3
+  * aligned v3 behavior to the fix (#2348)
 
 v0.2 - 2020-04-02
 -----------------

diff --git a/src/msc3.cc b/src/msc3.cc
@@ -12,7 +12,7 @@ void showhelp(char * name) {
     std::cout << "OPTIONS:" << std::endl;
     std::cout << "\t-h\tThis help" << std::endl;
     std::cout << "\t-n N\titerate pcre_regex as Nth times. Default value is 1." << std::endl;
-    std::cout << "\t-f\tForce to use modified regex matching method." << std::endl;
+    std::cout << "\t-f\tForce to use old v3 regex matching method." << std::endl;
     std::cout << "\t-t T\tExpects a float value; if the (last) pcre_exec time is greather than this," << std::endl;
     std::cout << "\t    \tthe exit status of program will non-zero." << std::endl;
     std::cout << "\t-d  \tShow debug information." << std::endl;
@@ -25,7 +25,7 @@ int main(int argc, char ** argv) {
     char * patternfile = NULL, * subjectfile = NULL;
     char c;
     int icnt = 1, rc = 0;
-    bool use_fixed = false;
+    bool use_old = false;
     float time_limit = 0.0;
     double m_sub = 0.0;
     int debuglevel = 0;  // may be later we can use different level...
@@ -41,7 +41,7 @@ int main(int argc, char ** argv) {
                 showhelp(argv[0]);
                 return EXIT_SUCCESS;
             case 'f':
-                use_fixed = true;
+                use_old = true;
                 break;
             case 'n':
                 icnt = atoi(optarg);
@@ -119,18 +119,18 @@ int main(int argc, char ** argv) {
 
     re = new Regex(pattern, debuglevel);
     std::list<SMatch> retval;
+    std::vector<SMatchCapture> captures;
 
     for(int i = 0; i < icnt; i++) {
 
         re->m_retList.clear();
 
         clock_t m_start = clock();
-        if (use_fixed == false) {
-            retval = re->searchAll(subject);
+        if (use_old == false) {
+            re->searchOneMatch(subject, captures);
         }
         else {
-            rc = re->searchAll2(subject, ((debuglevel == 1) ? 10 : 0));
-            retval = re->m_retList;
+            retval = re->searchAll(subject);
         }
         clock_t m_end = clock();
         m_sub = (m_end - m_start) / double(CLOCKS_PER_SEC);
@@ -148,26 +148,51 @@ int main(int argc, char ** argv) {
     // show captured substrings if debug was set
     if (debuglevel == 1) {
         debugvalue(debuglevel, "CAPTURES", "");
-        retval.reverse();
-        for(auto s: retval) {
-            std::string subpatt = "";
-            if (s.offset() > 0) {
-                subpatt += subject.substr(0, s.offset());
+        if (use_old == false) {
+            for (const SMatchCapture& capture : captures) {
+                const std::string capture_substring(subject.substr(capture.m_offset, capture.m_length));
+                std::string subpatt = "";
+                if (capture.m_offset > 0) {
+                    subpatt += subject.substr(0, capture.m_offset);
+                }
+                subpatt += BOLDGREEN + capture_substring + RESET;
+                if (capture.m_offset + capture_substring.size() < subject.size()) {
+                    subpatt += subject.substr(capture.m_offset + capture_substring.size());
+                }
+                std::cout << subpatt << std::endl;
             }
-            subpatt += BOLDGREEN + s.str() + RESET;
-            if (s.offset() + s.str().size() < subject.size()) {
-                subpatt += subject.substr(s.offset() + s.str().size());
+
+            debugvalue(debuglevel, "OVECTOR", "");
+            std::cout << "[";
+            size_t si = 0;
+            for(auto capture: captures) {
+                const std::string capture_substring(subject.substr(capture.m_offset, capture.m_length));
+                std::cout << capture.m_offset << ", " << capture.m_offset + capture_substring.size() << ((si++ < captures.size()-1) ? ", " : "");
             }
-            std::cout << subpatt << std::endl;
+            std::cout << "]" << std::endl;
         }
+        else {
+            retval.reverse();
+            for(auto s: retval) {
+                std::string subpatt = "";
+                if (s.offset() > 0) {
+                    subpatt += subject.substr(0, s.offset());
+                }
+                subpatt += BOLDGREEN + s.str() + RESET;
+                if (s.offset() + s.str().size() < subject.size()) {
+                    subpatt += subject.substr(s.offset() + s.str().size());
+                }
+                std::cout << subpatt << std::endl;
+            }
 
-        debugvalue(debuglevel, "OVECTOR", "");
-        std::cout << "[";
-        size_t si = 0;
-        for(auto s: retval) {
-            std::cout << s.offset() << ", " << s.offset() + s.str().size() << ((si++ < retval.size()-1) ? ", " : "");
+            debugvalue(debuglevel, "OVECTOR", "");
+            std::cout << "[";
+            size_t si = 0;
+            for(auto s: retval) {
+                std::cout << s.offset() << ", " << s.offset() + s.str().size() << ((si++ < retval.size()-1) ? ", " : "");
+            }
+            std::cout << "]" << std::endl;
         }
-        std::cout << "]" << std::endl;
     }
     // end debug
 

diff --git a/src/regex.cc b/src/regex.cc
@@ -68,6 +68,25 @@ Regex::~Regex() {
     }
 }
 
+bool Regex::searchOneMatch(const std::string& s, std::vector<SMatchCapture>& captures) const {
+    const char *subject = s.c_str();
+    int ovector[OVECCOUNT];
+
+    int rc = pcre_exec(m_pc, m_pce, subject, s.size(), 0, 0, ovector, OVECCOUNT);
+
+    for (int i = 0; i < rc; i++) {
+        size_t start = ovector[2*i];
+        size_t end = ovector[2*i+1];
+        size_t len = end - start;
+        if (end > s.size()) {
+            continue;
+        }
+        SMatchCapture capture(i, start, len);
+        captures.push_back(capture);
+    }
+
+    return (rc > 0);
+}
 
 std::list<SMatch> Regex::searchAll(const std::string& s) {
     const char *subject = s.c_str();

diff --git a/src/regex.h b/src/regex.h
@@ -7,6 +7,7 @@
 #include <fstream>
 #include <string>
 #include <list>
+#include <vector>
 
 #include "regexutils.h"
 
@@ -36,6 +37,17 @@ class SMatch {
     size_t m_offset;
 };
 
+struct SMatchCapture {
+    SMatchCapture(size_t group, size_t offset, size_t length) :
+	m_group(group),
+	m_offset(offset),
+	m_length(length) { }
+
+    size_t m_group; // E.g. 0 = full match; 6 = capture group 6
+    size_t m_offset; // offset of match within the analyzed string
+    size_t m_length;
+};
+
 class Regex {
  public:
     explicit Regex(const std::string& pattern_, int debuglevel);
@@ -52,6 +64,7 @@ class Regex {
     std::list<SMatch> m_retList;
 
     std::list<SMatch> searchAll(const std::string& s);
+    bool searchOneMatch(const std::string& s, std::vector<SMatchCapture>& captures) const;
     int searchAll2(const std::string& s, size_t capturelen);
 };