zipfile.c

给出了 zip 压缩算法的完整实现过程。

  PUTLG(z->tim, f);
  PUTLG(z->crc, f);
  PUTLG(z->siz, f);
  PUTLG(z->len, f);
  PUTSH(z->nam, f);
  PUTSH(z->cext, f);
  PUTSH(z->com, f);
  PUTSH(z->dsk, f);
  PUTSH(z->att, f);
  PUTLG(z->atx, f);
  PUTLG(z->off, f);
#ifdef EBCDIC
  if (z->com)
    memtoasc(z->comment, z->comment, z->com);
#endif /* EBCDIC */
  if (fwrite(z->iname, 1, z->nam, f) != z->nam ||
      (z->cext && fwrite(z->cextra, 1, z->cext, f) != z->cext) ||
      (z->com && fwrite(z->comment, 1, z->com, f) != z->com))
    return ZE_TEMP;
  return ZE_OK;
}


int putend(n, s, c, m, z, f)
unsigned n;             /* number of entries in central directory */
ulg s;                  /* size of central directory */
ulg c;                  /* offset of central directory */
extent m;               /* length of zip file comment (0 if none) */
char *z;                /* zip file comment if m != 0 */
FILE *f;                /* file to write to */
/* Write the end of central directory data to file *f.  Return an error code
   in the ZE_ class. */
{
  PUTLG(ENDSIG, f);
  PUTSH(0, f);
  PUTSH(0, f);
  PUTSH(n, f);
  PUTSH(n, f);
  PUTLG(s, f);
  PUTLG(c, f);
  PUTSH(m, f);
/* Write the comment, if any */
#ifdef EBCDIC
  memtoasc(z, z, m);
#endif
  if (m && fwrite(z, 1, m, f) != m)
    return ZE_TEMP;

#ifdef HANDLE_AMIGA_SFX
  if (amiga_sfx_offset && zipbeg /* -J zeroes this */) {
    s = ftell(f);
    while (s & 3) s++, putc(0, f);   /* final marker must be longword aligned */
    PUTLG(0xF2030000 /* 1010 in Motorola byte order */, f);
    c = (s - amiga_sfx_offset - 4) / 4;  /* size of archive part in longwords */
    if (fseek(f, amiga_sfx_offset, SEEK_SET) != 0)
      return ZE_TEMP;
    c = ((c >> 24) & 0xFF) | ((c >> 8) & 0xFF00)
         | ((c & 0xFF00) << 8) | ((c & 0xFF) << 24);     /* invert byte order */
    PUTLG(c, f);
    fseek(f, 0, SEEK_END);                                    /* just in case */
  }
#endif
  return ZE_OK;
}


/* Note: a zip "entry" includes a local header (which includes the file
   name), an encryption header if encrypting, the compressed data
   and possibly an extended local header. */

int zipcopy(z, x, y)
struct zlist far *z;    /* zip entry to copy */
FILE *x, *y;            /* source and destination files */
/* Copy the zip entry described by *z from file *x to file *y.  Return an
   error code in the ZE_ class.  Also update tempzn by the number of bytes
   copied. */
{
  ulg n;                /* holds local header offset */

  Trace((stderr, "zipcopy %s\n", z->zname));
  n = (ulg)(4 + LOCHEAD) + (ulg)z->nam + (ulg)z->ext;

  if (fix > 1) {
    if (fseek(x, z->off + n, SEEK_SET)) /* seek to compressed data */
      return ferror(x) ? ZE_READ : ZE_EOF;

    if (fix > 2) {
      /* Update length of entry's name, it may have been changed.  This is
         needed to support the ZipNote ability to rename archive entries. */
      z->nam = strlen(z->iname);
      n = (ulg)(4 + LOCHEAD) + (ulg)z->nam + (ulg)z->ext;
    }

    /* do not trust the old compressed size */
    if (putlocal(z, y) != ZE_OK)
      return ZE_TEMP;

    z->off = tempzn;
    tempzn += n;
    n = z->siz;
  } else {
    if (fseek(x, z->off, SEEK_SET))     /* seek to local header */
      return ferror(x) ? ZE_READ : ZE_EOF;

    z->off = tempzn;
    n += z->siz;
  }
  /* copy the compressed data and the extended local header if there is one */
  if (z->lflg & 8) n += 16;
  tempzn += n;
  return fcopy(x, y, n);
}


#ifndef UTIL

#ifdef USE_EF_UT_TIME

local int ef_scan_ut_time(ef_buf, ef_len, ef_is_cent, z_utim)
char *ef_buf;                   /* buffer containing extra field */
extent ef_len;                  /* total length of extra field */
int ef_is_cent;                 /* flag indicating "is central extra field" */
iztimes *z_utim;                /* return storage: atime, mtime, ctime */
/* This function scans the extra field for EF_TIME or EF_IZUNIX blocks
 * containing Unix style time_t (GMT) values for the entry's access, creation
 * and modification time.
 * If a valid block is found, all time stamps are copied to the iztimes
 * structure.
 * The presence of an EF_TIME or EF_IZUNIX2 block results in ignoring
 * all data from probably present obsolete EF_IZUNIX blocks.
 * If multiple blocks of the same type are found, only the information from
 * the last block is used.
 * The return value is the EF_TIME Flags field (simulated in case of an
 * EF_IZUNIX block) or 0 in case of failure.
 */
{
  int flags = 0;
  unsigned eb_id;
  extent eb_len;
  int have_new_type_eb = FALSE;

  if (ef_len == 0 || ef_buf == NULL)
    return 0;

  Trace((stderr,"\nef_scan_ut_time: scanning extra field of length %u\n",
         ef_len));
  while (ef_len >= EB_HEADSIZE) {
    eb_id = SH(EB_ID + ef_buf);
    eb_len = SH(EB_LEN + ef_buf);

    if (eb_len > (ef_len - EB_HEADSIZE)) {
      /* Discovered some extra field inconsistency! */
      Trace((stderr,"ef_scan_ut_time: block length %u > rest ef_size %u\n",
             eb_len, ef_len - EB_HEADSIZE));
      break;
    }

    switch (eb_id) {
      case EF_TIME:
        flags &= ~0x00ff;       /* ignore previous IZUNIX or EF_TIME fields */
        have_new_type_eb = TRUE;
        if ( eb_len >= EB_UT_MINLEN && z_utim != NULL) {
           unsigned eb_idx = EB_UT_TIME1;
           Trace((stderr,"ef_scan_ut_time: Found TIME extra field\n"));
           flags |= (ef_buf[EB_HEADSIZE+EB_UT_FLAGS] & 0x00ff);
           if ((flags & EB_UT_FL_MTIME)) {
              if ((eb_idx+4) <= eb_len) {
                 z_utim->mtime = LG((EB_HEADSIZE+eb_idx) + ef_buf);
                 eb_idx += 4;
                 Trace((stderr,"  Unix EF modtime = %ld\n", z_utim->mtime));
              } else {
                 flags &= ~EB_UT_FL_MTIME;
                 Trace((stderr,"  Unix EF truncated, no modtime\n"));
              }
           }
           if (ef_is_cent) {
              break;            /* central version of TIME field ends here */
           }
           if (flags & EB_UT_FL_ATIME) {
              if ((eb_idx+4) <= eb_len) {
                 z_utim->atime = LG((EB_HEADSIZE+eb_idx) + ef_buf);
                 eb_idx += 4;
                 Trace((stderr,"  Unix EF acctime = %ld\n", z_utim->atime));
              } else {
                 flags &= ~EB_UT_FL_ATIME;
              }
           }
           if (flags & EB_UT_FL_CTIME) {
              if ((eb_idx+4) <= eb_len) {
                 z_utim->ctime = LG((EB_HEADSIZE+eb_idx) + ef_buf);
                 /* eb_idx += 4; */  /* superfluous for now ... */
                 Trace((stderr,"  Unix EF cretime = %ld\n", z_utim->ctime));
              } else {
                 flags &= ~EB_UT_FL_CTIME;
              }
           }
        }
        break;

      case EF_IZUNIX2:
        if (!have_new_type_eb) {
           flags &= ~0x00ff;    /* ignore any previous IZUNIX field */
           have_new_type_eb = TRUE;
        }
        break;

      case EF_IZUNIX:
        if (eb_len >= EB_UX_MINLEN) {
           Trace((stderr,"ef_scan_ut_time: Found IZUNIX extra field\n"));
           if (have_new_type_eb) {
              break;            /* Ignore IZUNIX extra field block ! */
           }
           z_utim->atime = LG((EB_HEADSIZE+EB_UX_ATIME) + ef_buf);
           z_utim->mtime = LG((EB_HEADSIZE+EB_UX_MTIME) + ef_buf);
           Trace((stderr,"  Unix EF access time = %ld\n",z_utim->atime));
           Trace((stderr,"  Unix EF modif. time = %ld\n",z_utim->mtime));
           flags |= (EB_UT_FL_MTIME | EB_UT_FL_ATIME);  /* signal success */
        }
        break;

      case EF_THEOS:
/*      printf("Not implemented yet\n"); */
        break;

      default:
        break;
    }
    /* Skip this extra field block */
    ef_buf += (eb_len + EB_HEADSIZE);
    ef_len -= (eb_len + EB_HEADSIZE);
  }

  return flags;
}

int get_ef_ut_ztime(z, z_utim)
struct zlist far *z;
iztimes *z_utim;
{
  int r;

#ifdef IZ_CHECK_TZ
  if (!zp_tz_is_valid) return 0;
#endif

  /* First, scan local extra field. */
  r = ef_scan_ut_time(z->extra, z->ext, FALSE, z_utim);

  /* If this was not successful, try central extra field, but only if
     it is really different. */
  if (!r && z->cext > 0 && z->cextra != z->extra)
    r = ef_scan_ut_time(z->cextra, z->cext, TRUE, z_utim);

  return r;
}

#endif /* USE_EF_UT_TIME */


local void cutpath(p, delim)
char *p;                /* path string */
int delim;              /* path component separator char */
/* Cut the last path component off the name *p in place.
 * This should work on both internal and external names.
 */
{
  char *r;              /* pointer to last path delimiter */

#ifdef VMS                      /* change [w.x.y]z to [w.x]y.DIR */
  if ((r = MBSRCHR(p, ']')) != NULL)
  {
    *r = 0;
    if ((r = MBSRCHR(p, '.')) != NULL)
    {
      *r = ']';
      strcat(r, ".DIR;1");     /* this assumes a little padding--see PAD */
    } else {
      *p = 0;
    }
  } else {
    if ((r = MBSRCHR(p, delim)) != NULL)
      *r = 0;
    else
      *p = 0;
  }
#else /* !VMS */
  if ((r = MBSRCHR(p, delim)) != NULL)
    *r = 0;
  else
    *p = 0;
#endif /* ?VMS */
}

int trash()
/* Delete the compressed files and the directories that contained the deleted
   files, if empty.  Return an error code in the ZE_ class.  Failure of
   destroy() or deletedir() is ignored. */
{
  extent i;             /* counter on deleted names */
  extent n;             /* number of directories to delete */
  struct zlist far **s; /* table of zip entries to handle, sorted */
  struct zlist far *z;  /* current zip entry */

  /* Delete marked names and count directories */
  n = 0;
  for (z = zfiles; z != NULL; z = z->nxt)
    if (z->mark == 1 || z->trash)
    {
      z->mark = 1;
      if (z->iname[z->nam - 1] != (char)0x2f) { /* don't unlink directory */
        if (verbose)
          fprintf(mesg, "zip diagnostic: deleting file %s\n", z->name);
        if (destroy(z->name)) {
          zipwarn("error deleting ", z->name);
        }
        /* Try to delete all paths that lead up to marked names. This is
         * necessary only with the -D option.
         */
        if (!dirnames) {
          cutpath(z->name, '/');  /* XXX wrong ??? */
          cutpath(z->iname, 0x2f); /* 0x2f = ascii['/'] */
          z->nam = strlen(z->iname);
          if (z->nam > 0) {
            z->iname[z->nam - 1] = (char)0x2f;
            z->iname[z->nam++] = '\0';
          }
          if (z->nam > 0) n++;
        }
      } else {
        n++;
      }
    }

  /* Construct the list of all marked directories. Some may be duplicated
   * if -D was used.
   */
  if (n)
  {
    if ((s = (struct zlist far **)malloc(n*sizeof(struct zlist far *))) ==
        NULL)
      return ZE_MEM;
    n = 0;
    for (z = zfiles; z != NULL; z = z->nxt) {
      if (z->mark && z->nam > 0 && z->iname[z->nam - 1] == (char)0x2f /* '/' */
          && (n == 0 || strcmp(z->name, s[n-1]->name) != 0)) {
        s[n++] = z;
      }
    }
    /* Sort the files in reverse order to get subdirectories first.
     * To avoid problems with strange naming conventions as in VMS,
     * we sort on the internal names, so x/y/z will always be removed
     * before x/y. On VMS, x/y/z > x/y but [x.y.z] < [x.y]
     */
    qsort((char *)s, n, sizeof(struct zlist far *), rqcmp);

    for (i = 0; i < n; i++) {
      char *p = s[i]->name;
      if (*p == '\0') continue;
      if (p[strlen(p) - 1] == '/') { /* keep VMS [x.y]z.dir;1 intact */
        p[strlen(p) - 1] = '\0';
      }
      if (i == 0 || strcmp(s[i]->name, s[i-1]->name) != 0) {
        if (verbose) {
          fprintf(mesg, "deleting directory %s (if empty)                \n",
                  s[i]->name);
        }
        deletedir(s[i]->name);
      }
    }
    free((zvoid *)s);
  }
  return ZE_OK;
}

#endif /* !UTIL */