ftpdirectoryparser.cpp

linux下开源浏览器WebKit的源码,市面上的很多商用浏览器都是移植自WebKit

      */
      if (numtoks >= 4 && toklen[0] < 13 && 
          ((toklen[1] == 5 && *tokens[1] == '<') || isASCIIDigit(*tokens[1])) )
      {
        if (numtoks == 4
         && (toklen[2] == 8 || toklen[2] == 9)
         && (((tokens[2][2]) == '/' && (tokens[2][5]) == '/') ||
             ((tokens[2][2]) == '-' && (tokens[2][5]) == '-'))
         && (toklen[3] == 4 || toklen[3] == 5)
         && (tokens[3][toklen[3]-3]) == ':'
         && isASCIIDigit(tokens[2][0]) && isASCIIDigit(tokens[2][1])
         && isASCIIDigit(tokens[2][3]) && isASCIIDigit(tokens[2][4])
         && isASCIIDigit(tokens[2][6]) && isASCIIDigit(tokens[2][7])
         && (toklen[2] < 9 || isASCIIDigit(tokens[2][8]))
         && isASCIIDigit(tokens[3][toklen[3]-1]) && isASCIIDigit(tokens[3][toklen[3]-2])
         && isASCIIDigit(tokens[3][toklen[3]-4]) && isASCIIDigit(*tokens[3]) 
         )
        {
          lstyle = 'w';
        }
        else if ((numtoks == 6 || numtoks == 7)
         && toklen[2] == 3 && toklen[3] == 2
         && toklen[4] == 4 && toklen[5] == 5
         && (tokens[5][2]) == ':'
         && isASCIIAlpha(tokens[2][0]) && isASCIIAlpha(tokens[2][1])
         &&                          isASCIIAlpha(tokens[2][2])
         && isASCIIDigit(tokens[3][0]) && isASCIIDigit(tokens[3][1])
         && isASCIIDigit(tokens[4][0]) && isASCIIDigit(tokens[4][1])
         && isASCIIDigit(tokens[4][2]) && isASCIIDigit(tokens[4][3])
         && isASCIIDigit(tokens[5][0]) && isASCIIDigit(tokens[5][1])
         && isASCIIDigit(tokens[5][3]) && isASCIIDigit(tokens[5][4])
         /* could also check that (&(tokens[5][5]) - tokens[2]) == 17 */
        )
        {
          lstyle = 'w';
        }
        if (lstyle && state.listStyle != lstyle) /* first time */
        {
          p = tokens[1];   
          if (toklen[1] != 5 || p[0] != '<' || p[1] != 'D' || 
                 p[2] != 'I' || p[3] != 'R' || p[4] != '>')
          {
            for (pos = 0; lstyle && pos < toklen[1]; pos++)
            {
              if (!isASCIIDigit(*p++))
                lstyle = 0;
            }
          } /* not <DIR> */
        } /* if (first time) */
      } /* if (numtoks == ...) */

      if (lstyle == 'w')
      {
        state.parsedOne = true;
        state.listStyle = lstyle;

        result.caseSensitive = true;
        result.filename = tokens[0];
        result.filenameLength = toklen[0];
        result.type = FTPDirectoryEntry;

        p = tokens[1];
        if (isASCIIDigit(*p))
        {
          result.type = FTPFileEntry;
          pos = toklen[1];
          result.fileSize = String(p, pos);
        }

        p = tokens[2];
        if (toklen[2] == 3) /* Chameleon */
        {
          tbuf[0] = toASCIIUpper(p[0]);
          tbuf[1] = toASCIILower(p[1]);
          tbuf[2] = toASCIILower(p[2]);
          for (pos = 0; pos < (12*3); pos+=3)
          {
            if (tbuf[0] == month_names[pos+0] &&
                tbuf[1] == month_names[pos+1] && 
                tbuf[2] == month_names[pos+2])
            {
              result.modifiedTime.tm_mon = pos/3;
              result.modifiedTime.tm_mday = atoi(tokens[3]);
              result.modifiedTime.tm_year = atoi(tokens[4]) - 1900;
              break;
            }
          }          
          pos = 5; /* Chameleon toknum of date field */
        }
        else
        {
          result.modifiedTime.tm_mon = atoi(p+0)-1;
          result.modifiedTime.tm_mday = atoi(p+3);
          result.modifiedTime.tm_year = atoi(p+6);
          if (result.modifiedTime.tm_year < 80) /* SuperTCP */
            result.modifiedTime.tm_year += 100;

          pos = 3; /* SuperTCP toknum of date field */
        }

        result.modifiedTime.tm_hour = atoi(tokens[pos]);
        result.modifiedTime.tm_min = atoi(&(tokens[pos][toklen[pos]-2]));

        /* the caller should do this (if dropping "." and ".." is desired)
        if (result.type == FTPDirectoryEntry && result.filename[0] == '.' &&
            (result.filenameLength == 1 || (result.filenameLength == 2 &&
                                      result.filename[1] == '.')))
          return FTPJunkEntry;
        */

        return result.type;
      } /* (lstyle == 'w') */

    } /* if (!lstyle && (!state.listStyle || state.listStyle == 'w'))  */
#endif

    /* +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ */

#if defined(SUPPORT_DLS) /* dls -dtR */
    if (!lstyle && 
       (state.listStyle == 'D' || (!state.listStyle && state.numLines == 1)))
       /* /bin/dls lines have to be immediately recognizable (first line) */
    {
      /* I haven't seen an FTP server that delivers a /bin/dls listing,
       * but can infer the format from the lynx and mirror.pl projects.
       * Both formats are supported.
       *
       * Lynx says:
       * README              763  Information about this server\0
       * bin/                  -  \0
       * etc/                  =  \0
       * ls-lR                 0  \0
       * ls-lR.Z               3  \0
       * pub/                  =  Public area\0
       * usr/                  -  \0
       * morgan               14  -> ../real/morgan\0
       * TIMIT.mostlikely.Z\0
       *                   79215  \0
       *
       * mirror.pl says:
       * filename:  ^(\S*)\s+
       * size:      (\-|\=|\d+)\s+
       * month/day: ((\w\w\w\s+\d+|\d+\s+\w\w\w)\s+
       * time/year: (\d+:\d+|\d\d\d\d))\s+
       * rest:      (.+) 
       *
       * README              763  Jul 11 21:05  Information about this server
       * bin/                  -  Apr 28  1994
       * etc/                  =  11 Jul 21:04
       * ls-lR                 0   6 Aug 17:14
       * ls-lR.Z               3  05 Sep 1994
       * pub/                  =  Jul 11 21:04  Public area
       * usr/                  -  Sep  7 09:39
       * morgan               14  Apr 18 09:39  -> ../real/morgan
       * TIMIT.mostlikely.Z
       *                   79215  Jul 11 21:04
      */
      if (!state.listStyle && line[linelen-1] == ':' && 
          linelen >= 2 && toklen[numtoks-1] != 1)
      { 
        /* code in mirror.pl suggests that a listing may be preceded
         * by a PWD line in the form "/some/dir/names/here:"
         * but does not necessarily begin with '/'. *sigh*
        */
        pos = 0;
        p = line;
        while (pos < (linelen-1))
        {
          /* illegal (or extremely unusual) chars in a dirspec */
          if (*p == '<' || *p == '|' || *p == '>' ||
              *p == '?' || *p == '*' || *p == '\\')
            break;
          if (*p == '/' && pos < (linelen-2) && p[1] == '/')
            break;
          pos++;
          p++;
        }
        if (pos == (linelen-1))
        {
          state.listStyle = 'D';
          return FTPJunkEntry;
        }
      }

      if (!lstyle && numtoks >= 2)
      {
        pos = 22; /* pos of (\d+|-|=) if this is not part of a multiline */
        if (state.listStyle && carry_buf_len) /* first is from previous line */
          pos = toklen[1]-1; /* and is 'as-is' (may contain whitespace) */

        if (linelen > pos)
        {
          p = &line[pos];
          if ((*p == '-' || *p == '=' || isASCIIDigit(*p)) &&
              ((linelen == (pos+1)) || 
               (linelen >= (pos+3) && p[1] == ' ' && p[2] == ' ')) )
          {
            tokmarker = 1;
            if (!carry_buf_len)
            {
              pos = 1;
              while (pos < numtoks && (tokens[pos]+toklen[pos]) < (&line[23]))
                pos++;
              tokmarker = 0;
              if ((tokens[pos]+toklen[pos]) == (&line[23]))
                tokmarker = pos;
            }
            if (tokmarker)  
            {
              lstyle = 'D';
              if (*tokens[tokmarker] == '-' || *tokens[tokmarker] == '=')
              {
                if (toklen[tokmarker] != 1 ||
                   (tokens[tokmarker-1][toklen[tokmarker-1]-1]) != '/')
                  lstyle = 0;
              }              
              else
              {
                for (pos = 0; lstyle && pos < toklen[tokmarker]; pos++) 
                {
                  if (!isASCIIDigit(tokens[tokmarker][pos]))
                    lstyle = 0; 
                }
              }
              if (lstyle && !state.listStyle) /* first time */
              {
                /* scan for illegal (or incredibly unusual) chars in fname */
                for (p = tokens[0]; lstyle &&
                     p < &(tokens[tokmarker-1][toklen[tokmarker-1]]); p++)
                {
                  if (*p == '<' || *p == '|' || *p == '>' || 
                      *p == '?' || *p == '*' || *p == '/' || *p == '\\')
                    lstyle = 0;
                }
              }

            } /* size token found */
          } /* expected chars behind expected size token */
        } /* if (linelen > pos) */
      } /* if (!lstyle && numtoks >= 2) */

      if (!lstyle && state.listStyle == 'D' && !carry_buf_len)
      {
        /* the filename of a multi-line entry can be identified
         * correctly only if dls format had been previously established.
         * This should always be true because there should be entries
         * for '.' and/or '..' and/or CWD that precede the rest of the
         * listing.
        */
        pos = linelen;
        if (pos > (sizeof(state.carryBuffer)-1))
          pos = sizeof(state.carryBuffer)-1;
        memcpy( state.carryBuffer, line, pos );
        state.carryBufferLength = pos;
        return FTPJunkEntry;
      }

      if (lstyle == 'D')
      {
        state.parsedOne = true;
        state.listStyle = lstyle;

        p = &(tokens[tokmarker-1][toklen[tokmarker-1]]);
        result.filename = tokens[0];
        result.filenameLength = p - tokens[0];
        result.type  = FTPFileEntry;

        if (result.filename[result.filenameLength-1] == '/')
        {
          if (result.linknameLength == 1)
            result.type = FTPJunkEntry;
          else
          {
            result.filenameLength--;
            result.type  = FTPDirectoryEntry;
          }
        }
        else if (isASCIIDigit(*tokens[tokmarker]))
        {
          pos = toklen[tokmarker];
          result.fileSize = String(tokens[tokmarker], pos);
        }

        if ((tokmarker+3) < numtoks && 
              (&(tokens[numtoks-1][toklen[numtoks-1]]) - 
               tokens[tokmarker+1]) >= (1+1+3+1+4) )
        {
          pos = (tokmarker+3);
          p = tokens[pos];
          pos = toklen[pos];

          if ((pos == 4 || pos == 5)
          &&  isASCIIDigit(*p) && isASCIIDigit(p[pos-1]) && isASCIIDigit(p[pos-2])
          &&  ((pos == 5 && p[2] == ':') ||  
               (pos == 4 && (isASCIIDigit(p[1]) || p[1] == ':')))
             )
          {
            month_num = tokmarker+1; /* assumed position of month field */
            pos = tokmarker+2;       /* assumed position of mday field */
            if (isASCIIDigit(*tokens[month_num])) /* positions are reversed */
            {
              month_num++;
              pos--;
            }
            p = tokens[month_num];
            if (isASCIIDigit(*tokens[pos]) 
            && (toklen[pos] == 1 || 
                  (toklen[pos] == 2 && isASCIIDigit(tokens[pos][1])))
            && toklen[month_num] == 3
            && isASCIIAlpha(*p) && isASCIIAlpha(p[1]) && isASCIIAlpha(p[2])  )
            {
              pos = atoi(tokens[pos]);
              if (pos > 0 && pos <= 31)
              {
                result.modifiedTime.tm_mday = pos;
                month_num = 1;
                for (pos = 0; pos < (12*3); pos+=3)
                {
                  if (p[0] == month_names[pos+0] &&
                      p[1] == month_names[pos+1] &&
                      p[2] == month_names[pos+2])
                    break;
                  month_num++;
                }
                if (month_num > 12)
                  result.modifiedTime.tm_mday = 0;
                else
                  result.modifiedTime.tm_mon = month_num - 1;
              }
            }
            if (result.modifiedTime.tm_mday)
            {
              tokmarker += 3; /* skip mday/mon/yrtime (to find " -> ") */
              p = tokens[tokmarker];

              pos = atoi(p);
              if (pos > 24)
                result.modifiedTime.tm_year = pos-1900;
              else
              {
                if (p[1] == ':')
                  p--;
                result.modifiedTime.tm_hour = pos;
                result.modifiedTime.tm_min = atoi(p+3);
                if (!state.now)
                {
                  time_t now = time(NULL);
                  state.now = now * 1000000.0;
                  
                  // FIXME: This code has the year 2038 bug
                  gmtime_r(&now, &state.nowFTPTime);
                  state.nowFTPTime.tm_year += 1900;
                }
                result.modifiedTime.tm_year = state.nowFTPTime.tm_year;
                if ( (( state.nowFTPTime.tm_mon  << 4) + state.nowFTPTime.tm_mday) <
                     ((result.modifiedTime.tm_mon << 4) + result.modifiedTime.tm_mday) )
                  result.modifiedTime.tm_year--;
              } /* got year or time */
            } /* got month/mday */
          } /* may have year or time */
        } /* enough remaining to possibly have date/time */

        if (numtoks > (tokmarker+2))
        {
          pos = tokmarker+1;
          p = tokens[pos];
          if (toklen[pos] == 2 && *p == '-' && p[1] == '>')
          {
            p = &(tokens[numtoks-1][toklen[numtoks-1]]);
            result.type  = FTPLinkEntry;
            result.linkname = tokens[pos+1];
            result.linknameLength = p - result.linkname;
            if (result.linknameLength > 1 &&
                result.linkname[result.linknameLength-1] == '/')
              result.linknameLength--;
          }
        } /* if (numtoks > (tokmarker+2)) */

        /* the caller should do this (if dropping "." and ".." is desired)
        if (result.type == FTPDirectoryEntry && result.filename[0] == '.' &&
            (result.filenameLength == 1 || (result.filenameLength == 2 &&
                                      result.filename[1] == '.')))
          return FTPJunkEntry;
        */

        return result.type;

      } /* if (lstyle == 'D') */
    } /* if (!lstyle && (!state.listStyle || state.listStyle == 'D')) */
#endif

    /* +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ */

  } /* if (linelen > 0) */

  if (state.parsedOne || state.listStyle) /* junk if we fail to parse */
    return FTPJunkEntry;      /* this time but had previously parsed sucessfully */
  return FTPMiscEntry;        /* its part of a comment or error message */
}

} // namespace WebCore

#endif // ENABLE(FTPDIR)