uuscan.c

来自「UUDeview是一个编码解码器」· C语言 代码 · 共 2,492 行 · 第 1/5 页

      *errcode = UURET_NOMEM;
      _FP_free (result);
      return NULL;
    }

    /* ignore empty lines at the beginning of a file */
    preheaders = ftell (datei);
    while (!feof (datei)) {
      if (UUBUSYPOLL(ftell(datei),progress.fsize)) SPCANCEL();
      if (_FP_fgets (line, 255, datei) == NULL)
	break;
      line[255] = '\0';
      if (!IsLineEmpty (line)) {
	fseek (datei, preheaders, SEEK_SET);
	break;
      }
      preheaders = ftell (datei);
    }
  }

  if (ferror(datei) || feof(datei)) {
    _FP_free (result);
    return NULL;
  }

  /*
   * If we are confident that this is a mail folder and are at the
   * beginning of the file, expecting to read some headers, scan
   * the envelope.
   */

  if (sstate.isfolder && sstate.mimestate == MS_HEADERS) {
    hcount = 0;
    lcount = 0;
    UUkillheaders (&sstate.envelope);

    /*
     * clean up leftovers from invalid messages
     */

    while (mssdepth) {
      mssdepth--;
      UUkillheaders (&(multistack[mssdepth].envelope));
      _FP_free (multistack[mssdepth].source);
    }

    prevpos = ftell (datei);
    if (_FP_fgets (line, 255, datei) == NULL) {
      _FP_free (result);
      return NULL;
    }
    line[255] = '\0';

    /*
     * Special handling for AOL folder files, which start off with a boundary.
     * We recognize them by a valid boundary line as the first line of a file.
     * Note that the rest of the scanning code becomes suspicious if a boun-
     * dary does never appear in a file -- this should save us from grave
     * false detection errors
     */

    if (!feof (datei) && line[0] == '-' && line[1] == '-' && line[2]) {
      while (line[strlen(line)-1] == '\012' ||
	     line[strlen(line)-1] == '\015') {
	line[strlen(line)-1] = '\0';
      }

      sstate.ismime            = 1;
      sstate.envelope.mimevers = _FP_strdup ("1.0");
      sstate.envelope.boundary = _FP_strdup (line+2);
      sstate.envelope.ctype    = _FP_strdup ("multipart/mixed");
      sstate.mimestate         = MS_SUBPART;

      *errcode = UURET_CONT;
      _FP_free (result);
      return NULL;
    }

    /*
     * Normal behavior: look for a RFC 822 header
     */

    while (!feof (datei) && !IsLineEmpty (line)) {
      if (IsKnownHeader (line))
	hcount++;
      if (UUBUSYPOLL(ftell(datei),progress.fsize)) SPCANCEL();
      if (IsHeaderLine (line)) {
	ptr1 = ScanHeaderLine (datei, line);
	if (ParseHeader (&sstate.envelope, ptr1) == NULL) {
	  *errcode = UURET_NOMEM;
	  _FP_free (result);
	  return NULL;
	}
      }
      /*
       * if we've read too many lines without finding headers, then
       * this probably isn't a mail folder after all
       */
      lcount++;
      if (lcount > WAITHEADER && hcount < hlcount.afternl) {
	fseek (datei, prevpos, SEEK_SET);
	line[0] = '\0';
	break;
      }

      if (_FP_fgets (line, 255, datei) == NULL)
	break;
      line[255] = '\0';
    }

    /* skip empty lines */
    prevpos = ftell (datei);
    if (IsLineEmpty (line)) {
      while (!feof (datei)) {
	if (_FP_fgets (line, 255, datei) == NULL)
	  break;
	if (UUBUSYPOLL(ftell(datei),progress.fsize)) SPCANCEL();
	if (!IsLineEmpty (line)) {
	  fseek (datei, prevpos, SEEK_SET);
	  line[255] = '\0';
	  break;
	}
	prevpos = ftell (datei);
      }
    }

    /*
     * If we don't have all valid MIME headers yet, but the following
     * line is a MIME header, accept it anyway.
     */

    if (!uu_more_mime &&
	sstate.envelope.mimevers == NULL &&
	sstate.envelope.ctype    == NULL &&
	sstate.envelope.ctenc    == NULL &&
	IsKnownHeader (line)) {
      /*
       * see above
       */
      if (_FP_fgets (line, 255, datei) == NULL) {
	line[0] = '\012';
	line[1] = '\0';
      }
      line[255] = '\0';

      while (!feof (datei) && !IsLineEmpty (line)) {
	if (IsKnownHeader (line))
	  hcount++;
	if (UUBUSYPOLL(ftell(datei),progress.fsize)) SPCANCEL();
	ptr1 = ScanHeaderLine (datei, line);
	if (ParseHeader (&sstate.envelope, ptr1) == NULL) {
	  *errcode = UURET_NOMEM;
	  _FP_free (result);
	  return NULL;
	}
	if (_FP_fgets (line, 255, datei) == NULL)
	  break;
	line[255] = '\0';
      }
      /* skip empty lines */
      prevpos = ftell (datei);
      while (!feof (datei)) {
	if (_FP_fgets (line, 255, datei) == NULL)
	  break;
	if (UUBUSYPOLL(ftell(datei),progress.fsize)) SPCANCEL();
	if (!IsLineEmpty (line)) {
	  fseek (datei, prevpos, SEEK_SET);
	  line[255] = '\0';
	  break;
	}
	prevpos = ftell (datei);
      }
    }

    /*
     * A partial multipart message probably has only a Content-Type
     * header but nothing else. In this case, at least simulate a
     * MIME message
     * if mimevers is not set but there are other well-known MIME
     * headers, don't be too picky about it.
     */
    if (sstate.envelope.ctype && sstate.envelope.mimevers==NULL  &&
	_FP_stristr (sstate.envelope.ctype, "multipart") != NULL &&
	sstate.envelope.boundary != NULL) {
      sstate.envelope.mimevers = _FP_strdup ("1.0");
      hcount = hlcount.afternl;
    }
    else if (sstate.envelope.mimevers==NULL && sstate.envelope.ctype &&
	     sstate.envelope.fname && sstate.envelope.ctenc) {
      sstate.envelope.mimevers = _FP_strdup ("1.0");
      hcount = hlcount.afternl;
    }

    if (sstate.envelope.mimevers != NULL) {
      /* this is a MIME file. check the Content-Type */
      sstate.ismime = 1;
      if (_FP_stristr (sstate.envelope.ctype, "multipart") != NULL) {
	if (sstate.envelope.boundary == NULL) {
	  UUMessage (uuscan_id, __LINE__, UUMSG_WARNING,
		     uustring (S_MIME_NO_BOUNDARY));
	  sstate.mimestate = MS_BODY;
	  _FP_free (sstate.envelope.ctype);
	  sstate.envelope.ctype = _FP_strdup ("text/plain");
	}
	else {
	  sstate.mimestate = MS_PREAMBLE;
	}
      }
      else {
	sstate.mimestate = MS_BODY;	/* just a `simple' message */
      }
    }
    else {
      /* not a folder after all */
      fseek (datei, prevpos, SEEK_SET);
      sstate.isfolder = 0;
      sstate.ismime   = 0;
    }
  }

  if (feof (datei) || ferror (datei)) { /* oops */
    _FP_free (result);
    return NULL;
  }

  /*
   * Handle MIME stuff
   */

  /*
   * Read Preamble. This must be ended by a sstate.envelope.boundary line.
   * If uu_usepreamble is set, we produce a result from this one
   */

  if (sstate.ismime && sstate.mimestate == MS_PREAMBLE) {
    result->startpos = ftell (datei);	/* remember start of preamble */
    prevpos          = ftell (datei);
    preheaders       = ftell (datei);

    blen   = strlen (sstate.envelope.boundary);
    lcount = 0;
    
    while (!feof (datei)) {
      if (_FP_fgets (line, 255, datei) == NULL) {
	line[0] = '\0';
	break;
      }
      if (UUBUSYPOLL(ftell(datei),progress.fsize)) SPCANCEL();
      if (line[0] == '-' && line[1] == '-' &&
	  strncmp (line+2, sstate.envelope.boundary, blen) == 0)
	break;
      if (!IsLineEmpty (line))
	lcount++;

      prevpos = ftell (datei);
    }
    if (feof (datei) || ferror (datei)) {
      UUMessage (uuscan_id, __LINE__, UUMSG_WARNING,
		 uustring (S_MIME_B_NOT_FOUND));
      /*
       * restart and try again; don't restart if uu_fast_scanning
       */
      sstate.isfolder  = 0;
      sstate.ismime    = 0;
      sstate.mimestate = MS_BODY;

      if (!uu_fast_scanning) {
	*errcode = UURET_CONT;
	fseek (datei, preheaders, SEEK_SET);
      }
      _FP_free (result);
      return NULL;
    }
    if (line[0] == '-' && line[1] == '-' &&
	strncmp (line+2, sstate.envelope.boundary, blen) == 0) {
      ptr1 = line + 2 + blen;
      if (*ptr1 == '-' && *(ptr1+1) == '-') {
	/* Empty Multipart Message. Duh. */
	sstate.mimestate = MS_EPILOGUE;
      }
      else {
	sstate.mimestate = MS_SUBPART;
      }
    }
    else { /* shouldn't happen */
      UUMessage (uuscan_id, __LINE__, UUMSG_WARNING,
		 uustring (S_MIME_B_NOT_FOUND));
      /*
       * restart and try again; don't restart if uu_fast_scanning
       */
      sstate.isfolder  = 0;
      sstate.ismime    = 0;
      sstate.mimestate = MS_BODY;

      if (!uu_fast_scanning) {
	*errcode = UURET_CONT;
	fseek (datei, preheaders, SEEK_SET);
      }
      _FP_free (result);
      return NULL;
    }
    /* produce result if uu_usepreamble is set */
    if (uu_usepreamble && lcount) {
      sprintf (line, "%04d.txt", ++mimseqno);
      result->subject  = _FP_strdup (sstate.envelope.subject);
      result->filename = _FP_strdup (line);
      result->origin   = _FP_strdup (sstate.envelope.from);
      result->mimeid   = _FP_strdup (sstate.envelope.mimeid);
      result->mimetype = _FP_strdup ("text/plain");
      result->mode     = 0644;
      result->uudet    = PT_ENCODED;	/* plain text */
      result->sfname   = _FP_strdup (fname);
      result->flags    = FL_SINGLE | FL_PROPER;
      /* result->startpos set from above */
      result->length   = prevpos - result->startpos;
      result->partno   = 1;

      /* MIME message, let's continue */
      *errcode = UURET_CONT;

      if ((sstate.envelope.subject != NULL && result->subject == NULL) ||
	  result->filename == NULL || result->sfname == NULL) {
	*errcode = UURET_NOMEM;
      }

      return result;
    }
    /* MIME message, let's continue */
    if (*errcode == UURET_OK)
      *errcode = UURET_CONT;

    /* otherwise, just return NULL */
    _FP_free (result);
    return NULL;
  }

  /*
   * Read Epilogue, the plain text after the last boundary.
   * This can either end with new headers from the next message of a
   * mail folder or with a `parent' boundary if we are inside an
   * encapsulated Multipart message. Oh yes, and of course the file
   * may also simply end :-)
   * Another possibility is that we might find plain encoded data
   * without encapsulating message. We're not _too_ flexible here,
   * we won't detect Base64, and require a proper `begin' line for
   * uuencoding and xxencoding
   * If uu_usepreamble is set, we produce a result from this one
   */

  if (sstate.ismime && sstate.mimestate == MS_EPILOGUE) {
    result->startpos = ftell (datei);	/* remember start of epilogue */
    prevpos          = ftell (datei);
    preheaders       = ftell (datei);
    preenc           = ftell (datei);
    hcount = lcount  = 0;
    ecount = bhflag  = 0;
    begflag = vflag  = 0;
    res = 0;

    /*
     * If we are in the outermost message and uu_fast_scanning, we
     * know (or assume) that no more messages will follow, so there's
     * no need to scan the rest.
     */
    if (uu_fast_scanning && mssdepth == 0) {
      /*
       * check if the epilogue is empty
       */
      while (!feof (datei) && !ferror (datei) && lcount<10 && res==0) {
	if (_FP_fgets (line, 255, datei) == NULL)
	  break;
	if (!IsLineEmpty (line))
	  res++;
	lcount++;
      }
      if (uu_usepreamble && res) {
	sprintf (line, "%04d.txt", ++mimseqno);
	result->subject  = _FP_strdup (sstate.envelope.subject);
	result->filename = _FP_strdup (line);
	result->origin   = _FP_strdup (sstate.envelope.from);
	result->mimeid   = _FP_strdup (sstate.envelope.mimeid);
	result->mimetype = _FP_strdup ("text/plain");
	result->mode     = 0644;
	result->uudet    = PT_ENCODED;	/* plain text */
	result->sfname   = _FP_strdup (fname);
	result->flags    = FL_SINGLE | FL_PROPER | FL_TOEND;
	result->partno   = 1;
	/* result->startpos set from above */
	result->length   = progress.fsize - result->startpos;

	if ((sstate.envelope.subject != NULL && result->subject == NULL) ||
	    result->filename == NULL || result->sfname == NULL) {
	  *errcode = UURET_NOMEM;
	}

	return result;
      }
      _FP_free (result);
      return NULL;
    }

    if (mssdepth > 0)
      blen = strlen (multistack[mssdepth-1].envelope.boundary);

    while (!feof (datei)) {
      if (_FP_fgets (line, 255, datei) == NULL) {
	line[0] = '\0';
	break;
      }
      if (UUBUSYPOLL(ftell(datei),progress.fsize)) SPCANCEL();
      line[255] = '\0';
      /* check for parent boundary */
      if (mssdepth > 0 && line[0] == '-' && line[1] == '-' &&
	  strncmp (line+2,
		   multistack[mssdepth-1].envelope.boundary, blen) == 0)
	break;

      /* check for next headers only at outermost level */
      if (mssdepth == 0 && IsKnownHeader (line)) {
	(void) ScanHeaderLine (datei, line);
	if (hcount == 0) {
	  preheaders = prevpos;
	  lcount     = 0;
	}
	hcount++; 
	lcount++;

	if (hcount >= hlcount.restart) {
	  /* okay, new headers */
	  break;
	}
      }
      else if (lcount > WAITHEADER) {
	hcount = 0;
	lcount = 0;
      }
      else if (hcount) {
	lcount++;
      }
      else {
	hcount = lcount = 0;
      }

      /* check for begin and encoded data only at outermost level */
      if (mssdepth == 0 && !uu_more_mime) {
	if (strncmp      (line, "begin ",       6) == 0 ||
	    _FP_strnicmp (line, "<pre>begin ", 11) == 0) {
	  preenc  = prevpos;
	  begflag = 1;
	}
	else if (strncmp (line, "end", 3) == 0 && begflag) {
	  ecount = ELC_COUNT;
	  break;
	}
	else if ((vflag = UUValidData (line, 0, &bhflag)) != 0) {
	  if (vflag == BH_ENCODED && bhflag == 0) {
	    /* very short BinHex file follows */
	    preenc = prevpos;
	    break;
	  }
	  /* remember that XX can easily be mistaken as Base64 */
	  if ((vflag == UU_ENCODED || vflag == XX_ENCODED ||
	       vflag == B64ENCODED) && begflag) {
	    if (++ecount >= ELC_COUNT)
	      break;
	  }
	  else {
	    begflag = 0;
	    ecount  = 0;
	  }
	}
	else {
	  begflag = 0;
	  ecount  = 0;
	}
      }

      if (!IsLineEmpty (line))
	res++;

      prevpos = ftell (datei);
    }

    if (mssdepth > 0 &&	line[0] == '-' && line[1] == '-' &&
	strncmp (line+2,
		 multistack[mssdepth-1].envelope.boundary, blen) == 0) {
      /* restore previous state */
      mssdepth--;
      UUkillheaders (&sstate.envelope);
      _FP_free  (sstate.source);
      memcpy (&sstate, &(multistack[mssdepth]), sizeof (scanstate));

      ptr1 = line + 2 + strlen (sstate.envelope.boundary);

      if (*ptr1 == '-' && *(ptr1+1) == '-') {
	sstate.mimestate = MS_EPILOGUE;
      }
      else {
	sstate.mimestate = MS_SUBPART;