regexconvert.c

nedit 是一款linux下的开发源码的功能强大的编辑器

   switch (*Reg_Parse++) {
      case '^':
         emit_convert_byte ('^');
         break;

      case '$':
         emit_convert_byte ('$');
         break;

      case '<':
         emit_convert_byte ('<');
         break;

      case '>':
         emit_convert_byte ('>');
         break;

      case '.':
         emit_convert_byte ('.');

         *flag_param |= (HAS_WIDTH | SIMPLE); break;

      case '(':
         emit_convert_byte ('(');

         ret_val = chunk (PAREN, &flags_local);

         if (ret_val == 0) return 0;  /* Something went wrong. */

         /* Add HAS_WIDTH flag if it was set by call to chunk. */

         *flag_param |= flags_local & HAS_WIDTH;

         break;

      case '\0':
      case '|':
      case ')':
         CONVERT_FAIL ("internal error #3, `atom\'");  /* Supposed to be  */
                                                       /* caught earlier. */
      case '?':
      case '+':
      case '*':
         sprintf (Error_Text, "%c follows nothing", *(Reg_Parse - 1));
         CONVERT_FAIL (Error_Text);

      case '{':
         emit_convert_byte ('\\'); /* Quote braces. */
         emit_convert_byte ('{');

         break;

      case '[':
         {
            register unsigned int  last_value;
                     unsigned char last_emit = 0;
                     unsigned char buffer [500];
                              int  head = 0;
                              int  negated = 0;
                              int  do_brackets  = 1;
                              int  a_z_flag     = 0;
                              int  A_Z_flag     = 0;
                              int  zero_nine    = 0;
                              int  u_score_flag = 0;

            buffer [0]  = '\0';

            /* Handle characters that can only occur at the start of a class. */

            if (*Reg_Parse == '^') { /* Complement of range. */
               negated = 1;

               Reg_Parse++;
            }

            if (*Reg_Parse == ']' || *Reg_Parse == '-') {
               /* If '-' or ']' is the first character in a class,
                  it is a literal character in the class. */

               last_emit = *Reg_Parse;

               if (head >= 498) {
                  CONVERT_FAIL ("too much data in [] to convert.");
               }

               buffer [head++] = '\\'; /* Escape `]' and '-' for clarity. */
               buffer [head++] = *Reg_Parse;

               Reg_Parse++;
            }

            /* Handle the rest of the class characters. */

            while (*Reg_Parse != '\0' && *Reg_Parse != ']') {
               if (*Reg_Parse == '-') { /* Process a range, e.g [a-z]. */
                  Reg_Parse++;

                  if (*Reg_Parse == ']' || *Reg_Parse == '\0') {
                     /* If '-' is the last character in a class it is a literal
                        character.  If `Reg_Parse' points to the end of the
                        regex string, an error will be generated later. */

                     last_emit = '-';

                     if (head >= 498) {
                        CONVERT_FAIL ("too much data in [] to convert.");
                     }

                     buffer [head++] = '\\'; /* Escape '-' for clarity. */
                     buffer [head++] = '-';

                  } else {
                     if (*Reg_Parse == '\\') {
                        /* Handle escaped characters within a class range. */

                        Reg_Parse++;

                        if ((test = literal_escape (*Reg_Parse, 0))) {

                           buffer [head++] = '-';

                           if (*Reg_Parse != '\"') {
                              emit_convert_byte ('\\');
                           }

                           buffer [head++] = *Reg_Parse;
                           last_value = (unsigned int) test;
                        } else {
                           sprintf (
                              Error_Text,
                              "\\%c is an invalid escape sequence(3)",
                              *Reg_Parse);

                           CONVERT_FAIL (Error_Text);
                        }
                     } else {
                        last_value = U_CHAR_AT (Reg_Parse);

                        if (last_emit == '0' && last_value == '9') {
                           zero_nine = 1;
                           head--;
                        } else if (last_emit == 'a' && last_value == 'z') {
                           a_z_flag  = 1;
                           head--;
                        } else if (last_emit == 'A' && last_value == 'Z') {
                           A_Z_flag = 1;
                           head--;
                        } else {
                           buffer [head++] = '-';

                           if ((test = literal_escape (*Reg_Parse, 1))) {
                              /* Ordinary character matches an escape sequence;
                                 convert it to the escape sequence. */

                              if (head >= 495) {
                                 CONVERT_FAIL (
                                    "too much data in [] to convert.");
                              }

                              buffer [head++] = '\\';

                              if (test == '0') { /* Make octal escape. */
                                 test = *Reg_Parse;
                                 buffer [head++] = '0';
                                 buffer [head++] = ('0' + (test / 64));
                                 test -= (test / 64) * 64;
                                 buffer [head++] = ('0' + (test / 8));
                                 test -= (test / 8) * 8;
                                 buffer [head++] = ('0' +  test);
                              } else {
                                 buffer [head++] = test;
                              }
                           } else {
                              buffer [head++] = last_value;
                           }
                        }
                     }

                     if (last_emit > last_value) {
                        CONVERT_FAIL ("invalid [] range");
                     }

                     last_emit = (unsigned char) last_value;

                     Reg_Parse++;

                  } /* End class character range code. */
               } else if (*Reg_Parse == '\\') {
                  Reg_Parse++;

                  if ((test = literal_escape (*Reg_Parse, 0)) != '\0') {
                     last_emit = test;

                     if (head >= 498) {
                        CONVERT_FAIL ("too much data in [] to convert.");
                     }

                     if (*Reg_Parse != '\"') {
                        buffer [head++] = '\\';
                     }

                     buffer [head++] = *Reg_Parse;

                  } else {
                     sprintf (Error_Text,
                              "\\%c is an invalid escape sequence(1)",
                              *Reg_Parse);

                     CONVERT_FAIL (Error_Text);
                  }

                  Reg_Parse++;

                  /* End of class escaped sequence code */
               } else {
                  last_emit = *Reg_Parse;

                  if (*Reg_Parse == '_') {
                     u_score_flag = 1; /* Emit later if we can't do `\w'. */

                  } else if ((test = literal_escape (*Reg_Parse, 1))) {
                     /* Ordinary character matches an escape sequence;
                        convert it to the escape sequence. */

                     if (head >= 495) {
                        CONVERT_FAIL ("too much data in [] to convert.");
                     }

                     buffer [head++] = '\\';

                     if (test == '0') {  /* Make octal escape. */
                        test = *Reg_Parse;
                        buffer [head++] = '0';
                        buffer [head++] = ('0' + (test / 64));
                        test -= (test / 64) * 64;
                        buffer [head++] = ('0' + (test / 8));
                        test -= (test / 8) * 8;
                        buffer [head++] = ('0' +  test);
                     } else {
                        if (head >= 499) {
                           CONVERT_FAIL ("too much data in [] to convert.");
                        }

                        buffer [head++] = test;
                     }
                  } else {
                     if (head >= 499) {
                        CONVERT_FAIL ("too much data in [] to convert.");
                     }

                     buffer [head++] = *Reg_Parse;
                  }

                  Reg_Parse++;
               }
            } /* End of while (*Reg_Parse != '\0' && *Reg_Parse != ']') */

            if (*Reg_Parse != ']') CONVERT_FAIL ("missing right \']\'");

            buffer [head] = '\0';

            /* NOTE: it is impossible to specify an empty class.  This is
               because [] would be interpreted as "begin character class"
               followed by a literal ']' character and no "end character class"
               delimiter (']').  Because of this, it is always safe to assume
               that a class HAS_WIDTH. */

            Reg_Parse++; *flag_param |= HAS_WIDTH | SIMPLE;

            if (head == 0) {
               if (( a_z_flag &&  A_Z_flag &&  zero_nine &&  u_score_flag) ||
                   ( a_z_flag &&  A_Z_flag && !zero_nine && !u_score_flag) ||
                   (!a_z_flag && !A_Z_flag &&  zero_nine && !u_score_flag)) {

                   do_brackets = 0;
               }
            }

            if (do_brackets) {
               emit_convert_byte ('[');
               if (negated) emit_convert_byte ('^');
            }

            /* Output any shortcut escapes if we can. */

            while (a_z_flag || A_Z_flag || zero_nine || u_score_flag) {
               if (a_z_flag && A_Z_flag && zero_nine && u_score_flag) {
                  emit_convert_byte ('\\');

                  if (negated && !do_brackets) {
                     emit_convert_byte ('W');
                  } else {
                     emit_convert_byte ('w');
                  }

                  a_z_flag = A_Z_flag = zero_nine = u_score_flag = 0;
               } else if (a_z_flag && A_Z_flag) {
                  emit_convert_byte ('\\');

                  if (negated && !do_brackets) {
                     emit_convert_byte ('L');
                  } else {
                     emit_convert_byte ('l');
                  }

                  a_z_flag = A_Z_flag = 0;
               } else if (zero_nine) {
                  emit_convert_byte ('\\');

                  if (negated && !do_brackets) {
                     emit_convert_byte ('D');
                  } else {
                     emit_convert_byte ('d');
                  }

                  zero_nine = 0;
               } else if (a_z_flag) {
                  emit_convert_byte ('a');
                  emit_convert_byte ('-');
                  emit_convert_byte ('z');