zipup.c

给出了 zip 压缩算法的完整实现过程。

    }

    z->tim = tim;

#if defined(VMS) && defined(VMS_PK_EXTRA)
    /* vms_get_attributes must be called after vms_open() */
    if (extra_fields) {
      /* create extra field and change z->att and z->atx if desired */
      vms_get_attributes(ifile, z, &f_utim);
    }
#endif /* VMS && VMS_PK_EXTRA */

#if defined(MMAP) || defined(BIG_MEM)
    /* Map ordinary files but not devices. This code should go in fileio.c */
    if (!translate_eol && m != STORE && q != -1L && (ulg)q > 0 &&
        (ulg)q + MIN_LOOKAHEAD > (ulg)q) {
# ifdef MMAP
      /* Map the whole input file in memory */
      if (window != NULL)
        free(window);  /* window can't be a mapped file here */
      window_size = (ulg)q + MIN_LOOKAHEAD;
      remain = window_size & (PAGESIZE-1);
      /* If we can't touch the page beyond the end of file, we must
       * allocate an extra page.
       */
      if (remain > MIN_LOOKAHEAD) {
        window = (uch*)mmap(0, window_size, PROT_READ, MAP_PRIVATE, ifile, 0);
      } else {
        window = (uch*)valloc(window_size - remain + PAGESIZE);
        if (window != NULL) {
          window = (uch*)mmap((char*)window, window_size - remain, PROT_READ,
                        MAP_PRIVATE | MAP_FIXED, ifile, 0);
        } else {
          window = (uch*)(-1);
        }
      }
      if (window == (uch*)(-1)) {
        Trace((mesg, " mmap failure on %s\n", z->name));
        window = NULL;
        window_size = 0L;
        remain = (ulg)-1L;
      } else {
        remain = (ulg)q;
      }
# else /* !MMAP, must be BIG_MEM */
      /* Read the whole input file at once */
      window_size = (ulg)q + MIN_LOOKAHEAD;
      window = window ? (uch*) realloc(window, (unsigned)window_size)
                      : (uch*) malloc((unsigned)window_size);
      /* Just use normal code if big malloc or realloc fails: */
      if (window != NULL) {
        remain = (ulg)zread(ifile, (char*)window, q+1);
        if (remain != (ulg)q) {
          fprintf(mesg, " q=%lu, remain=%lu ", (ulg)q, remain);
          error("can't read whole file at once");
        }
      } else {
        window_size = 0L;
      }
# endif /* ?MMAP */
    }
#endif /* MMAP || BIG_MEM */

  } /* strcmp(z->name, "-") == 0 */

  if (q == 0)
    m = STORE;
  if (m == BEST)
    m = DEFLATE;

  /* Do not create STORED files with extended local headers if the
   * input size is not known, because such files could not be extracted.
   * So if the zip file is not seekable and the input file is not
   * on disk, obey the -0 option by forcing deflation with stored block.
   * Note however that using "zip -0" as filter is not very useful...
   * ??? to be done.
   */

  /* Fill in header information and write local header to zip file.
   * This header will later be re-written since compressed length and
   * crc are not yet known.
   */

  /* (Assume ext, cext, com, and zname already filled in.) */
#if defined(OS2) || defined(WIN32)
  z->vem = (ush)(z->dosflag ? (dosify ? 20 : /* Made under MSDOS by PKZIP 2.0 */
                               (0 + Z_MAJORVER * 10 + Z_MINORVER))
                 : OS_CODE + Z_MAJORVER * 10 + Z_MINORVER);
  /* For a FAT file system, we cheat and pretend that the file
   * was not made on OS2/WIN32 but under DOS. unzip is confused otherwise.
   */
#else /* !(OS2 || WIN32) */
  z->vem = (ush)(dosify ? 20 : OS_CODE + Z_MAJORVER * 10 + Z_MINORVER);
#endif /* ?(OS2 || WIN32) */

  z->ver = (ush)(m == STORE ? 10 : 20); /* Need PKUNZIP 2.0 except for store */
  z->crc = 0;  /* to be updated later */
  /* Assume first that we will need an extended local header: */
  z->flg = 8;  /* to be updated later */
#if CRYPT
  if (key != NULL) {
    z->flg |= 1;
    /* Since we do not yet know the crc here, we pretend that the crc
     * is the modification time:
     */
    z->crc = z->tim << 16;
  }
#endif /* CRYPT */
  z->lflg = z->flg;
  z->how = (ush)m;                              /* may be changed later  */
  z->siz = (ulg)(m == STORE && q >= 0 ? q : 0); /* will be changed later */
  z->len = (ulg)(q != -1L ? q : 0);             /* may be changed later  */
  z->dsk = 0;
  if (z->att == (ush)UNKNOWN) {
      z->att = BINARY;                    /* set sensible value in header */
      set_type = 1;
  }
  /* Attributes from filetime(), flag bits from set_extra_field(): */
#if defined(DOS) || defined(OS2) || defined(WIN32)
  z->atx = z->dosflag ? a & 0xff : a | (z->atx & 0x0000ff00);
#else
  z->atx = dosify ? a & 0xff : a | (z->atx & 0x0000ff00);
#endif /* DOS || OS2 || WIN32 */
  z->off = tempzn;
  if ((r = putlocal(z, y)) != ZE_OK) {
    if (ifile != fbad)
      zclose(ifile);
    return r;
  }
  tempzn += 4 + LOCHEAD + z->nam + z->ext;

#if CRYPT
  if (key != NULL) {
    crypthead(key, z->crc, y);
    z->siz += RAND_HEAD_LEN;  /* to be updated later */
    tempzn += RAND_HEAD_LEN;
  }
#endif /* CRYPT */
  if (ferror(y)) {
    if (ifile != fbad)
      zclose(ifile);
    ZIPERR(ZE_WRITE, "unexpected error on zip file");
  }

  last_o = o;
  o = ftell(y); /* for debugging only, ftell can fail on pipes */
  if (ferror(y))
    clearerr(y);

  if (last_o > o) {
    /* could be wrap around */
    ZIPERR(ZE_BIG, "seek wrap - zip file too big to write");
  }

  /* Write stored or deflated file to zip file */
  isize = 0L;
  crc = CRCVAL_INITIAL;

  if (m == DEFLATE) {
    if (set_type) z->att = (ush)UNKNOWN; /* is finally set in filecompress() */
    s = filecompress(z, y, &m);
#ifndef PGP
    if (z->att == (ush)BINARY && translate_eol) {
        zipwarn("-l used on binary file", "");
    }
#endif
  }
  else if (!isdir)
  {
    if ((b = malloc(SBSZ)) == NULL)
       return ZE_MEM;

    if (l) {
      k = rdsymlnk(z->name, b, SBSZ);
/*
 * compute crc first because zfwrite will alter the buffer b points to !!
 */
      crc = crc32(crc, (uch *) b, k);
      if (zfwrite(b, 1, k, y) != k)
      {
        free((zvoid *)b);
        return ZE_TEMP;
      }
      isize = k;

#ifdef MINIX
      q = k;
#endif /* MINIX */
    }
    else
    {
      while ((k = file_read(b, SBSZ)) > 0 && k != (extent) EOF)
      {
        if (zfwrite(b, 1, k, y) != k)
        {
          if (ifile != fbad)
            zclose(ifile);
          free((zvoid *)b);
          return ZE_TEMP;
        }
#ifndef WINDLL
        if (verbose) putc('.', stderr);
        fflush(stderr);
#else
        if (verbose) fprintf(stdout,"%c",'.');
#endif
      }
    }
    free((zvoid *)b);
    s = isize;
  }
  if (ifile != fbad && zerr(ifile)) {
    perror("\nzip warning");
    zipwarn("could not read input file: ", z->zname);
  }
  if (ifile != fbad)
    zclose(ifile);
#ifdef MMAP
  if (remain != (ulg)-1L) {
    munmap((caddr_t) window, window_size);
    window = NULL;
  }
#endif /*MMAP */

  tempzn += s;
  p = tempzn; /* save for future fseek() */

#if (!defined(MSDOS) || defined(OS2))
#if !defined(VMS) && !defined(CMS_MVS) && !defined(__mpexl)
  /* Check input size (but not in VMS -- variable record lengths mess it up)
   * and not on MSDOS -- diet in TSR mode reports an incorrect file size)
   */
#ifndef TANDEM /* Tandem EOF does not match byte count unless Unstructured */
  if (!translate_eol && q != -1L && isize != (ulg)q)
  {
    Trace((mesg, " i=%lu, q=%lu ", isize, q));
    zipwarn(" file size changed while zipping ", z->name);
  }
#endif /* !TANDEM */
#endif /* !VMS && !CMS_MVS && !__mpexl */
#endif /* (!MSDOS || OS2) */

  /* Try to rewrite the local header with correct information */
  z->crc = crc;
  z->siz = s;
#if CRYPT
  if (key != NULL)
    z->siz += RAND_HEAD_LEN;
#endif /* CRYPT */
  z->len = isize;
#ifdef BROKEN_FSEEK
  if (!fseekable(y) || fseek(y, z->off, SEEK_SET))
#else
  if (fseek(y, z->off, SEEK_SET))
#endif
  {
    if (z->how != (ush) m)
       error("can't rewrite method");
    if (m == STORE && q < 0)
       ZIPERR(ZE_PARMS, "zip -0 not supported for I/O on pipes or devices");
    if ((r = putextended(z, y)) != ZE_OK)
      return r;
    tempzn += 16L;
    z->flg = z->lflg; /* if flg modified by inflate */
  } else {
     /* seek ok, ftell() should work, check compressed size */
#if !defined(VMS) && !defined(CMS_MVS)
    if (p - o != s) {
      fprintf(mesg, " s=%ld, actual=%ld ", s, p-o);
      error("incorrect compressed size");
    }
#endif /* !VMS && !CMS_MVS */
    z->how = (ush)m;
    z->ver = (ush)(m == STORE ? 10 : 20);  /* Need PKUNZIP 2.0 unless STORED */
    if ((z->flg & 1) == 0)
      z->flg &= ~8; /* clear the extended local header flag */
    z->lflg = z->flg;
    /* rewrite the local header: */
    if ((r = putlocal(z, y)) != ZE_OK)
      return r;
    if (fseek(y, p, SEEK_SET))
      return ZE_READ;
    if ((z->flg & 1) != 0) {
      /* encrypted file, extended header still required */
      if ((r = putextended(z, y)) != ZE_OK)
        return r;
      tempzn += 16L;
    }
  }
  /* Free the local extra field which is no longer needed */
  if (z->ext) {
    if (z->extra != z->cextra) {
      free((zvoid *)(z->extra));
      z->extra = NULL;
    }
    z->ext = 0;
  }

  /* Display statistics */
  if (noisy)
  {
    if (verbose)
      fprintf(mesg, "\t(in=%lu) (out=%lu)", isize, s);
    if (m == DEFLATE)
      fprintf(mesg, " (deflated %d%%)\n", percent(isize, s));
    else
      fprintf(mesg, " (stored 0%%)\n");
    fflush(mesg);
  }
#ifdef WINDLL
  if (lpZipUserFunctions->ServiceApplication != NULL)
     {
     if ((*lpZipUserFunctions->ServiceApplication)(z->zname, isize))
        {
        ZIPERR(ZE_ABORT, "User terminated operation");
        }
     }
#endif
  return ZE_OK;
}


local unsigned file_read(buf, size)
  char *buf;
  unsigned size;
/* Read a new buffer from the current input file, perform end-of-line
 * translation, and update the crc and input file size.
 * IN assertion: size >= 2 (for end-of-line translation)
 */
{
  unsigned len;
  char *b;

#if defined(MMAP) || defined(BIG_MEM)
  if (remain == 0L) {
    return 0;
  } else if (remain != (ulg)-1L) {
    /* The window data is already in place. We still compute the crc
     * by 32K blocks instead of once on whole file to keep a certain
     * locality of reference.
     */
    Assert(buf == (char*)window + isize, "are you lost?");
    if ((ulg)size > remain) size = (unsigned)remain;
    if (size > WSIZE) size = WSIZE; /* don't touch all pages at once */
    remain -= (ulg)size;
    len = size;
  } else
#endif /* MMAP || BIG_MEM */
  if (translate_eol == 0) {
    len = zread(ifile, buf, size);
    if (len == (unsigned)EOF || len == 0) return len;
#ifdef OS390
    b = buf;
    if (aflag == ASCII) {
      while (*b != '\0') {
        *b = (char)ascii[(uch)*b];
        b++;
      }
    }
#endif
  } else if (translate_eol == 1) {
    /* Transform LF to CR LF */
    size >>= 1;
    b = buf+size;
    size = len = zread(ifile, b, size);
    if (len == (unsigned)EOF || len == 0) return len;
#ifdef EBCDIC
    if (aflag == ASCII)
    {
       do {
          char c;

          if ((c = *b++) == '\n') {
             *buf++ = CR; *buf++ = LF; len++;
          } else {
            *buf++ = (char)ascii[(uch)c];
          }
       } while (--size != 0);
    }
    else
#endif /* EBCDIC */
    {
       do {
          if ((*buf++ = *b++) == '\n') *(buf-1) = CR, *buf++ = LF, len++;
       } while (--size != 0);
    }
    buf -= len;

  } else {
    /* Transform CR LF to LF and suppress final ^Z */
    b = buf;
    size = len = zread(ifile, buf, size-1);
    if (len == (unsigned)EOF || len == 0) return len;
    buf[len] = '\n'; /* I should check if next char is really a \n */
#ifdef EBCDIC
    if (aflag == ASCII)
    {
       do {
          char c;