Changeset 610 for CMT/HEAD/source/cmt_system.cxx

Timestamp:

Apr 16, 2012, 12:17:30 PM (12 years ago)

Author:

rybkin

Message:

See C.L. 485

File:

• CMT/HEAD/source/cmt_system.cxx (modified) (2 diffs)

CMT/HEAD/source/cmt_system.cxx

@@ -2231 +2231 @@
 void CmtSystem::split (const cmt_string& text,
                        const cmt_string& separators,
-                       cmt_string_vector& strings)
+                       cmt_string_vector& strings,
+                       const bool& unquote)
 {
   static char* buffer = 0;
   static int allocated = 0;
-
-  bool finished = false;
 
   strings.clear ();
@@ -2277 +2276 @@
   */
 
-  char* current_word = buffer;
-
-  while (*current_word != 0)
-    {
-      size_t prefix_length;
-      size_t word_length;
-
-      /*
-        while ((*current_word == ' ') ||
-        (*current_word == '\t'))
-        {
-        current_word++;
-        }
-      */
-
-      // first skip all starting separators.
-
-      prefix_length = strspn (current_word, separators.c_str ());
-      if (prefix_length > 0)
-        {
-          // Move to the first non-separator character
-
-          current_word += prefix_length;
-        }
-
-      /*
-        Parse the next word.
-
-        It may contain enclosures in quote characters or not.
-        Quotes must be identical on both sides of each enclosure.
-      */
-
-      char* running_char = current_word;
-
-      word_length = 0;
-
-      for (;;)
-        {
-          size_t unquoted_length;
-          size_t separator_offset;
-
-          for (int p = 0;;)
+  char * b = buffer;
+  char * begin = b;
+  char * e;
+  char * pq = 0;
+  char * pqm = 0;
+  char q, ev;
+  char * token = 0;
+  bool before_q = false;
+  bool after_q = false;
+  bool matched = false;
+
+//   cerr << "split: " << buffer << endl;
+//   cerr << "seps: `";
+//   for (int i = 0; i < separators.size (); i++)
+//     cerr << separators[i];
+//   cerr << "'" << endl;
+
+  // sep...ab..."cd"ef...sep...
+  while (pq = strpbrk (begin, "\"\'"))
+    {
+      if (begin < pq && *(pq - 1) == '\\')
+        {// quote considered escaped
+          begin = pq + 1;
+          continue;
+        }
+
+      if ((b < pq && NULL == strchr (separators.c_str (), *(pq - 1))) ||
+          (b == pq && after_q))
+        before_q = true;
+      else
+        before_q = false;
+
+      // save quote found
+      q = *pq;
+      // terminate string for standard string functions
+      *pq = 0;
+
+      // parse string - up to quote - into a sequence of tokens
+      if (token = strtok (b, separators.c_str ()))
+        {
+          if (after_q)
+            strings.back () += token;
+          else
+            strings.add () = token;
+
+//        cerr << (after_q ? "append: " : "add: ")
+//             << "[" << token << "]{" << strings.back () << "}";
+
+          while (token = strtok (NULL, separators.c_str ()))
             {
-              unquoted_length = strcspn (running_char + p, "\"\'") + p;
-              if ((unquoted_length > 0) && (running_char[unquoted_length-1] == '\\'))
-                {
-                  p = unquoted_length + 1;
-                }
-              else
-                {
-                  break;
-                }
+              strings.add () = token;
+              //strings.push_back (token);
+              //cerr << "[" << token << "]{" << strings.back () << "}";
+            } // while ( token = strtok (NULL, separators.c_str ()) )
+          //cerr << endl;
+        }
+
+      if (unquote)
+        {
+          b = pq + 1;
+        }
+      else
+        {
+          // restore quote found
+          *pq = q;
+          b = pq;
+        }
+      begin = pq + 1;
+
+      // look for matching quote
+      matched = false;
+      while (pqm = strchr (begin, q))
+        {
+          // commented out because of
+          // inconsistency of quoting rules
+          // and backward compatibility
+//        if (begin < pqm && *(pqm - 1) == '\\')
+//          {// quote considered escaped
+//            begin = pqm + 1;
+//            continue;
+//          }
+          matched = true;
+
+          if (*(pqm + 1) && NULL == strchr (separators.c_str (), *(pqm + 1)))
+            after_q = true;
+          else
+            after_q = false;
+
+          if (unquote)
+            {
+              e = pqm;
             }
-
-          separator_offset = strcspn (running_char, separators.c_str ());
-
-          if (separator_offset <= unquoted_length)
-            {
-              // no quote in this word -> we are finished for this one.
-              running_char += separator_offset;
-              break;
-            }
-
-          // We have found a quoted enclosure. Move to it.
-
-          running_char += unquoted_length;
-
-          char quote = running_char[0];
-
-          // Remove it.
-          {
-            char* p = running_char;
-            while (p[1] != 0)
-              {
-                *p = p[1];
-                p++;
-              }
-            *p = 0;
-          }
-
-          // Look for the next occurence of this quote.
-          {
-            char* p = strchr (running_char, quote);
-            if (p == 0)
-              {
-                // Unmatched quote : the rest of the line will be taken as a word...
-                running_char += strlen (running_char);
-                finished = true;
-                break;
-              }
-            else
-              {
-                running_char = p;
-              }
-          }
-
-          // Now we remove the ending quote from the word
-          // (by shifting all remaining characters by one place to the left)
-
-          {
-            char* p = running_char;
-            while (p[1] != 0)
-              {
-                *p = p[1];
-                p++;
-              }
-            *p = 0;
-          }
-        }
-
-      word_length = running_char - current_word;
-
-      if (current_word[word_length] == 0)
-        {
-          finished = true;
-        }
-      else
-        {
-          current_word[word_length] = 0;
-        }
-
-      /*
-        if ((t[0] == '"') ||
-        (t[0] == '\'') ||
-        (t[0] == ':'))
-        {
-        char* quote;
-
-        t++;
-        quote = strchr (t, sep);
-        if (quote != 0) *quote = 0;
-        else finished = true;
-        }
-        else
-        {
-        int offset;
-
-        offset = strcspn (t, " \t:");
-        if ((offset < 0) || (t[offset] == 0)) finished = true;
-        if (!finished)
-        {
-        space = t + offset;
-        *space = 0;
-        }
-        }
-      */
-
-      // Store the current word into the vector of strings
-
-      {
-        cmt_string& s = strings.add ();
-        s = current_word;
-      }
-
-      if (finished) break;
-
-      // Move to the next possible word.
-      current_word += word_length + 1;
-    }
+          else
+            {
+              e = pqm + 1;
+              ev = *e;
+            }
+          // terminate string for standard string functions
+          *e = 0;
+          
+          if (before_q)
+            strings.back () += b;
+          else
+            strings.add () = b;
+
+//        cerr << (before_q ? "append: " : "add: ")
+//             << "|" << b << "|{" << strings.back () << "}" << endl;
+          
+          // restore e value
+          if (!unquote)
+            {
+              *e = ev;
+            }
+          b = pqm + 1;
+          begin = b;
+          break;
+        } // while (pqm = strchr (begin, q))
+
+      if (!matched)
+        { // unmatched quote : the rest of the line will be taken as a token
+          if (before_q)
+            strings.back () += b;
+          else
+            strings.add () = b;
+
+          // append quote to match
+          if (!unquote)
+            strings.back () += q;
+          
+//        cerr << (before_q ? "append: " : "add: ")
+//             << "|" << b << "|{" << strings.back () << "}" << endl;
+
+          b = buffer + strlen(buffer);
+          begin = b;
+          break;
+        }
+
+    } // while (pq = strpbrk (begin, "\"\'"))
+
+      // parse string - up to end - into a sequence of tokens
+      if (token = strtok (b, separators.c_str ()))
+        {
+          if (after_q)
+            strings.back () += token;
+          else
+            strings.add () = token;
+
+//        cerr << (after_q ? "append: " : "add: ")
+//             << "<" << token << ">{" << strings.back () << "}";
+
+          while (token = strtok (NULL, separators.c_str ()))
+            {
+              strings.add () = token;
+              //cerr << "<" << token << ">{" << strings.back () << "}";
+            } // while ( token = strtok (NULL, separators.c_str ()) )
+          //cerr << endl;
+        }
+//       cerr << "strings:";
+//       for (int i = 0; i < strings.size (); i++)
+//      cerr << " {" << strings[i] << "}";
+//       //     cerr << " `" << strings[i] << "'";
+//       cerr << endl;
 }