zipfile.c

给出了 zip 压缩算法的完整实现过程。

      {
        sprintf(errbuf, "%lu", (ulg)zcount + 1);
        zipwarn("zero-length name for entry #", errbuf);
#ifndef DEBUG
        farfree((zvoid far *)z);
        return ZE_FORM;
#endif
      }
      if ((z->iname = malloc(z->nam+1)) ==  NULL ||
          (z->cext && (z->cextra = malloc(z->cext)) == NULL) ||
          (z->com && (z->comment = malloc(z->com)) == NULL))
        return ZE_MEM;
      if (fread(z->iname, z->nam, 1, f) != 1 ||
          (z->cext && fread(z->cextra, z->cext, 1, f) != 1) ||
          (z->com && fread(z->comment, z->com, 1, f) != 1))
        return ferror(f) ? ZE_READ : ZE_EOF;
      z->iname[z->nam] = '\0';                  /* terminate name */
#ifdef EBCDIC
      if (z->com)
         memtoebc(z->comment, z->comment, z->com);
#endif /* EBCDIC */
      /* Update zipbeg offset, prepare for next header */
      if (z->off < zipbeg)
         zipbeg = z->off;
      zcount++;
      /* Read next signature */
      if (fread(b, 4, 1, f) != 1)
          return ferror(f) ? ZE_READ : ZE_EOF;
    }

    /* Point to start of header list and read local headers */
    z = zfiles;
    while (z != NULL) {
      /* Read next signature */
      if (fseek(f, z->off, SEEK_SET) != 0 || fread(b, 4, 1, f) != 1)
        return ferror(f) ? ZE_READ : ZE_EOF;
      if (LG(b) == LOCSIG) {
        if (fread(b, LOCHEAD, 1, f) != 1)
            return ferror(f) ? ZE_READ : ZE_EOF;
        z->lflg = SH(LOCFLG + b);
        n = SH(LOCNAM + b);
        z->ext = SH(LOCEXT + b);

        /* Compare name and extra fields */
        if (n != z->nam)
        {
#ifdef EBCDIC
          strtoebc(z->iname, z->iname);
#endif
          zipwarn("name lengths in local and central differ for ", z->iname);
          return ZE_FORM;
        }
        if ((t = malloc(z->nam)) == NULL)
          return ZE_MEM;
        if (fread(t, z->nam, 1, f) != 1)
        {
          free((zvoid *)t);
          return ferror(f) ? ZE_READ : ZE_EOF;
        }
        if (memcmp(t, z->iname, z->nam))
        {
          free((zvoid *)t);
#ifdef EBCDIC
          strtoebc(z->iname, z->iname);
#endif
          zipwarn("names in local and central differ for ", z->iname);
          return ZE_FORM;
        }
        free((zvoid *)t);
        if (z->ext)
        {
          if ((z->extra = malloc(z->ext)) == NULL)
            return ZE_MEM;
          if (fread(z->extra, z->ext, 1, f) != 1)
          {
            free((zvoid *)(z->extra));
            return ferror(f) ? ZE_READ : ZE_EOF;
          }
          if (z->ext == z->cext && memcmp(z->extra, z->cextra, z->ext) == 0)
          {
            free((zvoid *)(z->extra));
            z->extra = z->cextra;
          }
        }

        /* Check extended local header if there is one */
        if ((z->lflg & 8) != 0)
        {
          char buf2[16];
          ulg s;                        /* size of compressed data */

          s = LG(LOCSIZ + b);
          if (s == 0)
            s = LG((CENSIZ-CENVER) + (char far *)(&(z->ver)));
          if (fseek(f, (z->off + (4+LOCHEAD) + z->nam + z->ext + s), SEEK_SET)
              || (fread(buf2, 16, 1, f) != 1))
            return ferror(f) ? ZE_READ : ZE_EOF;
          if (LG(buf2) != EXTLOCSIG)
          {
#ifdef EBCDIC
            strtoebc(z->iname, z->iname);
#endif
            zipwarn("extended local header not found for ", z->iname);
            return ZE_FORM;
          }
          /* overwrite the unknown values of the local header: */
          for (n = 0; n < 12; n++)
            b[LOCCRC+n] = buf2[4+n];
        }

        /* Compare local header with that part of central header (except
           for the reserved bits in the general purpose flags and except
           for the already checked entry name length */
        u = (char far *)(&(z->ver));
        flg = SH((CENFLG-CENVER) + u);          /* Save central flags word */
        u[CENFLG-CENVER+1] &= 0x1f;             /* Mask reserved flag bits */
        b[LOCFLG+1] &= 0x1f;
        for (m = 0, n = 0; n < LOCNAM; n++)
          if (b[n] != u[n])
          {
            if (!m)
            {
              zipwarn("local and central headers differ for ", z->zname);
              m = 1;
            }
            if (noisy)
            {
              sprintf(errbuf, " offset %u--local = %02x, central = %02x",
                      (unsigned)n, (uch)b[n], (uch)u[n]);
              zipwarn(errbuf, "");
            }
          }
        if (m && !adjust)
          return ZE_FORM;

        /* Complete the setup of the zlist entry by translating the remaining
         * central header fields in memory, starting with the fields with
         * highest offset. This order of the conversion commands takes into
         * account potential buffer overlaps caused by structure padding.
         */
        z->len = LG((CENLEN-CENVER) + u);
        z->siz = LG((CENSIZ-CENVER) + u);
        z->crc = LG((CENCRC-CENVER) + u);
        z->tim = LG((CENTIM-CENVER) + u);   /* time and date into one long */
        z->how = SH((CENHOW-CENVER) + u);
        z->flg = flg;                       /* may be different from z->lflg */
        z->ver = SH((CENVER-CENVER) + u);

        /* Clear actions */
        z->mark = 0;
        z->trash = 0;
#ifdef UTIL
/* We only need z->iname in the utils */
        z->name = z->iname;
#ifdef EBCDIC
/* z->zname is used for printing and must be coded in native charset */
        if ((z->zname = malloc(z->nam+1)) ==  NULL)
          return ZE_MEM;
        strtoebc(z->zname, z->iname);
#else
        z->zname = z->iname;
#endif
#else /* !UTIL */
        z->zname = in2ex(z->iname);       /* convert to external name */
        if (z->zname == NULL)
          return ZE_MEM;
        z->name = z->zname;
#endif /* ?UTIL */
      }
      else {
#ifdef EBCDIC
        strtoebc(z->iname, z->iname);
#endif
        zipwarn("local header not found for ", z->iname);
        return ZE_FORM;
      }
#ifndef UTIL
      if (verbose)
        zipoddities(z);
#endif
      z = z->nxt;
    }

    if (zipbeg && noisy)
      fprintf(mesg, "%s: %s a preamble of %lu bytes\n",
              zipfile, adjust ? "adjusting offsets for" : "found", zipbeg);

#ifdef HANDLE_AMIGA_SFX
    if (zipbeg < 12 || (zipbeg & 3) != 0 /* must be longword aligned */)
      amiga_sfx_offset = 0;
    else if (amiga_sfx_offset) {
      char buf2[16];
      if (!fseek(f, zipbeg - 12, SEEK_SET) && fread(buf2, 12, 1, f) == 1) {
        if (LG(buf2 + 4) == 0xF1030000 /* 1009 in Motorola byte order */)
          /* could also check if LG(buf2) == 0xF2030000... no for now */
          amiga_sfx_offset = zipbeg - 4;
        else
          amiga_sfx_offset = 0L;
      }
    }
#endif /* HANDLE_AMIGA_SFX */
    return ZE_OK;
}


/*
 * readzipfile initializes the global variables that hold the zipfile
 * directory info and opens the zipfile. For the actual zipfile scan,
 * the subroutine scanzipf_reg() or scanzipf_fix() is called,
 * depending on the mode of operation (regular processing, or zipfix mode).
 */
int readzipfile()
/*
   The name of the zip file is pointed to by the global "zipfile".
   The globals zipbeg, zfiles, zcount, and zcomlen are initialized.
   Return an error code in the ZE_ class.
*/
{
  FILE *f;              /* zip file */
  int retval;           /* return code */
  int readable;         /* 1 if zipfile exists and is readable */

  /* Initialize zip file info */
  zipbeg = 0;
  zfiles = NULL;                        /* Points to first header */
  zcount = 0;                           /* number of files */
  zcomlen = 0;                          /* zip file comment length */
  retval = ZE_OK;
  f = NULL;                             /* shut up some compilers */

  /* If zip file exists, read headers and check structure */
#ifdef VMS
  if (zipfile == NULL || !(*zipfile) || !strcmp(zipfile, "-"))
    return ZE_OK;
  {
    int rtype;

    if ((VMSmunch(zipfile, GET_RTYPE, (char *)&rtype) == RMS$_NORMAL) &&
        (rtype == FAT$C_VARIABLE)) {
      fprintf(stderr,
     "\n     Error:  zipfile is in variable-length record format.  Please\n\
     run \"bilf b %s\" to convert the zipfile to fixed-length\n\
     record format.\n\n", zipfile);
      return ZE_FORM;
    }
  }
  readable = ((f = fopen(zipfile, FOPR)) != NULL);
#else /* !VMS */
  readable = (zipfile != NULL && *zipfile && strcmp(zipfile, "-") &&
              (f = fopen(zipfile, FOPR)) != NULL);
#endif /* ?VMS */
#ifdef MVS
  /* Very nasty special case for MVS.  Just because the zipfile has been
   * opened for reading does not mean that we can actually read the data.
   * Typical JCL to create a zipfile is
   *
   * //ZIPFILE  DD  DISP=(NEW,CATLG),DSN=prefix.ZIP,
   * //             SPACE=(CYL,(10,10))
   *
   * That creates a VTOC entry with an end of file marker (DS1LSTAR) of zero.
   * Alas the VTOC end of file marker is only used when the file is opened in
   * append mode.  When a file is opened in read mode, the "other" end of file
   * marker is used, a zero length data block signals end of file when reading.
   * With a brand new file which has not been written to yet, it is undefined
   * what you read off the disk.  In fact you read whatever data was in the same
   * disk tracks before the zipfile was allocated.  You would be amazed at the
   * number of application programmers who still do not understand this.  Makes
   * for interesting and semi-random errors, GIGO.
   *
   * Newer versions of SMS will automatically write a zero length block when a
   * file is allocated.  However not all sites run SMS or they run older levels
   * so we cannot rely on that.  The only safe thing to do is close the file,
   * open in append mode (we already know that the file exists), close it again,
   * reopen in read mode and try to read a data block.  Opening and closing in
   * append mode will write a zero length block where DS1LSTAR points, making
   * sure that the VTOC and internal end of file markers are in sync.  Then it
   * is safe to read data.  If we cannot read one byte of data after all that,
   * it is a brand new zipfile and must not be read.
   */
  if (readable)
  {
    char c;
    fclose(f);
    /* append mode */
    if ((f = fopen(zipfile, "ab")) == NULL) {
      ZIPERR(ZE_OPEN, zipfile);
    }
    fclose(f);
    /* read mode again */
    if ((f = fopen(zipfile, FOPR)) == NULL) {
      ZIPERR(ZE_OPEN, zipfile);
    }
    if (fread(&c, 1, 1, f) != 1) {
      /* no actual data */
      readable = 0;
      fclose(f);
    }
    else{
      fseek(f, 0, SEEK_SET);  /* at least one byte in zipfile, back to the start */
    }
  }
#endif /* MVS */
  if (readable)
  {
#ifndef UTIL
    retval = (fix && !adjust) ? scanzipf_fix(f) : scanzipf_reg(f);
#else
    retval = scanzipf_reg(f);
#endif

    /* Done with zip file for now */
    fclose(f);

    /* If one or more files, sort by name */
    if (zcount)
    {
      struct zlist far * far *x;    /* pointer into zsort array */
      struct zlist far *z;          /* pointer into zfiles linked list */
      extent zl_size = zcount * sizeof(struct zlist far *);

      if (zl_size / sizeof(struct zlist far *) != zcount ||
          (x = zsort = (struct zlist far **)malloc(zl_size)) == NULL)
        return ZE_MEM;
      for (z = zfiles; z != NULL; z = z->nxt)
        *x++ = z;
      qsort((char *)zsort, zcount, sizeof(struct zlist far *), zqcmp);
    }
  }
  return retval;
}


int putlocal(z, f)
struct zlist far *z;    /* zip entry to write local header for */
FILE *f;                /* file to write to */
/* Write a local header described by *z to file *f.  Return an error code
   in the ZE_ class. */
{
  PUTLG(LOCSIG, f);
  PUTSH(z->ver, f);
  PUTSH(z->lflg, f);
  PUTSH(z->how, f);
  PUTLG(z->tim, f);
  PUTLG(z->crc, f);
  PUTLG(z->siz, f);
  PUTLG(z->len, f);
  PUTSH(z->nam, f);
  PUTSH(z->ext, f);
  if (fwrite(z->iname, 1, z->nam, f) != z->nam ||
      (z->ext && fwrite(z->extra, 1, z->ext, f) != z->ext))
    return ZE_TEMP;
  return ZE_OK;
}

int putextended(z, f)
struct zlist far *z;    /* zip entry to write local header for */
FILE *f;                /* file to write to */
/* Write an extended local header described by *z to file *f.
 * Return an error code in the ZE_ class. */
{
  PUTLG(EXTLOCSIG, f);
  PUTLG(z->crc, f);
  PUTLG(z->siz, f);
  PUTLG(z->len, f);
  return ZE_OK;
}

int putcentral(z, f)
struct zlist far *z;    /* zip entry to write central header for */
FILE *f;                /* file to write to */
/* Write a central header described by *z to file *f.  Return an error code
   in the ZE_ class. */
{
  PUTLG(CENSIG, f);
  PUTSH(z->vem, f);
  PUTSH(z->ver, f);
  PUTSH(z->flg, f);
  PUTSH(z->how, f);