zipup.c

给出了 zip 压缩算法的完整实现过程。

/*
  Copyright (c) 1990-2006 Info-ZIP.  All rights reserved.

  See the accompanying file LICENSE, version 2005-Feb-10 or later
  (the contents of which are also included in zip.h) for terms of use.
  If, for some reason, all these files are missing, the Info-ZIP license
  also may be found at:  ftp://ftp.info-zip.org/pub/infozip/license.html
*/
/*
 *  zipup.c by Mark Adler and Jean-loup Gailly.
 */
#define __ZIPUP_C

#include "zip.h"
#include <ctype.h>

#ifndef UTIL            /* This module contains no code for Zip Utilities */

#include "revision.h"
#include "crypt.h"
#ifdef USE_ZLIB
#  include "zlib.h"
#endif

#ifdef OS2
#  include "os2/os2zip.h"
#endif

#if defined(MMAP)
#  include <sys/mman.h>
#  ifndef PAGESIZE   /* used to be SYSV, what about pagesize on SVR3 ? */
#    define PAGESIZE getpagesize()
#  endif
#  if defined(NO_VALLOC) && !defined(valloc)
#    define valloc malloc
#  endif
#endif

/* Use the raw functions for MSDOS and Unix to save on buffer space.
   They're not used for VMS since it doesn't work (raw is weird on VMS).
 */

#ifdef AMIGA
#  include "amiga/zipup.h"
#endif /* AMIGA */

#ifdef AOSVS
#  include "aosvs/zipup.h"
#endif /* AOSVS */

#ifdef ATARI
#  include "atari/zipup.h"
#endif

#ifdef __BEOS__
#  include "beos/zipup.h"
#endif

#ifdef __ATHEOS__
#  include "atheos/zipup.h"
#endif /* __ATHEOS__ */

#ifdef __human68k__
#  include "human68k/zipup.h"
#endif /* __human68k__ */

#ifdef MACOS
#  include "macos/zipup.h"
#endif

#ifdef DOS
#  include "msdos/zipup.h"
#endif /* DOS */

#ifdef OS2
#  include "os2/zipup.h"
#endif /* OS2 */

#ifdef RISCOS
#  include "acorn/zipup.h"
#endif

#ifdef TOPS20
#  include "tops20/zipup.h"
#endif

#ifdef UNIX
#  include "unix/zipup.h"
#endif

#ifdef CMS_MVS
#  include "zipup.h"
#endif /* CMS_MVS */

#ifdef TANDEM
#  include "zipup.h"
#endif /* TANDEM */

#ifdef VMS
#  include "vms/zipup.h"
#endif /* VMS */

#ifdef QDOS
#  include "qdos/zipup.h"
#endif /* QDOS */

#ifdef WIN32
#  include "win32/zipup.h"
#endif

#ifdef THEOS
#  include "theos/zipup.h"
#endif

/* Local functions */
#ifndef RISCOS
   local int suffixes OF((char *, char *));
#else
   local int filetypes OF((char *, char *));
#endif
local unsigned file_read OF((char *buf, unsigned size));
#ifdef USE_ZLIB
  local int zl_deflate_init OF((int pack_level));
#else /* !USE_ZLIB */
# ifdef ZP_NEED_MEMCOMPR
    local unsigned mem_read OF((char *buf, unsigned size));
# endif
#endif /* ?USE_ZLIB */
local ulg filecompress OF((struct zlist far *z_entry, FILE *zipfile,
                           int *cmpr_method));

/* Deflate "internal" global data (currently not in zip.h) */
#if defined(MMAP) || defined(BIG_MEM)
# ifdef USE_ZLIB
    local uch *window = NULL;   /* Used to read all input file at once */
    local ulg window_size;      /* size of said window */
# else /* !USE_ZLIB */
    extern uch *window;         /* Used to read all input file at once */
#endif /* ?USE_ZLIB */
#endif /* MMAP || BIG_MEM */
#ifndef USE_ZLIB
  extern ulg window_size;       /* size of said window */

  unsigned (*read_buf) OF((char *buf, unsigned size)) = file_read;
  /* Current input function. Set to mem_read for in-memory compression */
#endif /* !USE_ZLIB */


/* Local data */
local ulg crc;          /* crc on uncompressed file data */
local ftype ifile;      /* file to compress */
#if defined(MMAP) || defined(BIG_MEM)
  local ulg remain;
  /* window bytes not yet processed.
   *  special value "(ulg)-1L" reserved to signal normal reads.
   */
#endif /* MMAP || BIG_MEM */
#ifdef USE_ZLIB
  local int deflInit;           /* flag: zlib deflate is initialized */
  local z_stream zstrm;         /* zlib's data interface structure */
  local char *f_ibuf = NULL;
  local char *f_obuf = NULL;
#else /* !USE_ZLIB */
  local FILE *zfile;            /* output zip file */
  local char file_outbuf[1024]; /* output buffer for compression to file */

# ifdef ZP_NEED_MEMCOMPR
    local char *in_buf;
    /* Current input buffer, in_buf is used only for in-memory compression. */
    local unsigned in_offset;
    /* Current offset in input buffer. in_offset is used only for in-memory
     * compression. On 16 bit machines, the buffer is limited to 64K.
     */
    local unsigned in_size;
    /* size of current input buffer */
# endif /* ZP_NEED_MEMCOMPR */
#endif /* ?USE_ZLIB */

#ifdef DEBUG
  ulg isize;           /* input file size. global only for debugging */
#else /* !DEBUG */
  local ulg isize;     /* input file size. */
#endif /* ?DEBUG */


int percent(n, m)
ulg n;
ulg m;               /* n is the original size, m is the new size */
/* Return the percentage compression from n to m using only integer
   operations */
{
  if (n > 0xffffffL)            /* If n >= 16M */
  {                             /*  then divide n and m by 256 */
    n += 0x80;  n >>= 8;
    m += 0x80;  m >>= 8;
  }
  return n > m ? (int)(1 + (200 * (n - m)/n)) / 2 : 0;
}

#ifndef RISCOS

local int suffixes(a, s)
char *a;                /* name to check suffix of */
char *s;                /* list of suffixes separated by : or ; */
/* Return true if a ends in any of the suffixes in the list s. */
{
  int m;                /* true if suffix matches so far */
  char *p;              /* pointer into special */
  char *q;              /* pointer into name a */

#ifdef QDOS
  short dlen = devlen(a);
  a = a + dlen;
#endif

  m = 1;
#ifdef VMS
  if( (q = strrchr(a,';')) != NULL )    /* Cut out VMS file version */
    --q;
  else
    q = a + strlen(a) - 1;
#else /* !VMS */
  q = a + strlen(a) - 1;
#endif /* ?VMS */
  for (p = s + strlen(s) - 1; p >= s; p--)
    if (*p == ':' || *p == ';')
    {
      if (m)
        return 1;
      else
      {
        m = 1;
#ifdef VMS
        if( (q = strrchr(a,';')) != NULL )      /* Cut out VMS file version */
          --q;
        else
          q = a + strlen(a) - 1;
#else /* !VMS */
        q = a + strlen(a) - 1;
#endif /* ?VMS */
      }
    }
    else
    {
      m = m && q >= a && case_map(*p) == case_map(*q);
      q--;
    }
  return m;
}

#else /* RISCOS */

local int filetypes(a, s)
char *a;                /* extra field of file to check filetype of */
char *s;                /* list of filetypes separated by : or ; */
/* Return true if a is any of the filetypes in the list s. */
{
 char *p;              /* pointer into special */
 char typestr[4];     /* filetype hex string taken from a */

 if ((((unsigned*)a)[2] & 0xFFF00000) != 0xFFF00000) {
 /* The file is not filestamped, always try to compress it */
   return 0;
 }

 sprintf(typestr,"%.3X",(((unsigned*)a)[2] & 0x000FFF00) >> 8);

 for (p=s;p<=s+strlen(s)-3;p+=3) { /* p+=3 to skip 3 hex type */
   while (*p==':' || *p==';')
     p++;

   if (typestr[0]==toupper(p[0]) && typestr[1]==toupper(p[1]) && typestr[2]==toupper(p[2]))
     return 1;
 }
 return 0;
}
#endif /* ?RISCOS */



/* 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 zipup(z, y)
struct zlist far *z;    /* zip entry to compress */
FILE *y;                /* output file */
/* Compress the file z->name into the zip entry described by *z and write
   it to the file *y. Encrypt if requested.  Return an error code in the
   ZE_ class.  Also, update tempzn by the number of bytes written. */
{
  iztimes f_utim;       /* UNIX GMT timestamps, filled by filetime() */
  ulg tim;              /* time returned by filetime() */
  ulg a = 0L;           /* attributes returned by filetime() */
  char *b;              /* malloc'ed file buffer */
  extent k = 0;         /* result of zread */
  int l = 0;            /* true if this file is a symbolic link */
  int m;                /* method for this entry */
  ulg o = 0, p;         /* offsets in zip file */
  long q = -3L;         /* size returned by filetime */
  int r;                /* temporary variable */
  ulg s = 0L;           /* size of compressed data */
  int isdir;            /* set for a directory name */
  int set_type = 0;     /* set if file type (ascii/binary) unknown */
  ulg last_o;           /* used to check if we wrapped beyond what fseek can handle */

  z->nam = strlen(z->iname);
  isdir = z->iname[z->nam-1] == (char)0x2f; /* ascii[(unsigned)('/')] */

  if ((tim = filetime(z->name, &a, &q, &f_utim)) == 0 || q == -3L)
    return ZE_OPEN;

  /* q is set to -1 if the input file is a device, -2 for a volume label */
  if (q == -2L) {
     isdir = 1;
     q = 0;
  } else if (isdir != ((a & MSDOS_DIR_ATTR) != 0)) {
     /* don't overwrite a directory with a file and vice-versa */
     return ZE_MISS;
  }
  z->att = (ush)UNKNOWN; /* will be changed later */
  z->atx = 0; /* may be changed by set_extra_field() */

  /* Free the old extra fields which are probably obsolete */
  if (z->ext) {
    free((zvoid *)(z->extra));
  }
  if (z->cext && z->extra != z->cextra) {
    free((zvoid *)(z->cextra));
  }
  z->extra = z->cextra = NULL;
  z->ext = z->cext = 0;

#if defined(MMAP) || defined(BIG_MEM)
  remain = (ulg)-1L; /* changed only for MMAP or BIG_MEM */
#endif /* MMAP || BIG_MEM */
#if (!defined(USE_ZLIB) || defined(MMAP) || defined(BIG_MEM))
  window_size = 0L;
#endif /* !USE_ZLIB || MMAP || BIG_MEM */

  /* Select method based on the suffix and the global method */
#ifndef RISCOS
  m = special != NULL && suffixes(z->name, special) ? STORE : method;
#else /* RISCOS  must set m after setting extra field */
  m = method;
#endif /* ?RISCOS */

  /* Open file to zip up unless it is stdin */
  if (strcmp(z->name, "-") == 0)
  {
    ifile = (ftype)zstdin;
#if defined(MSDOS) || defined(__human68k__)
    if (isatty(zstdin) == 0)  /* keep default mode if stdin is a terminal */
      setmode(zstdin, O_BINARY);
#endif
    z->tim = tim;
  }
  else
  {
#if !(defined(VMS) && defined(VMS_PK_EXTRA))
    if (extra_fields) {
      /* create extra field and change z->att and z->atx if desired */
      set_extra_field(z, &f_utim);
#ifdef QLZIP
      if(qlflag)
          a |= (S_IXUSR) << 16;   /* Cross compilers don't set this */
#endif
#ifdef RISCOS
      m = special != NULL && filetypes(z->extra, special) ? STORE : method;
#endif /* RISCOS */
    }
#endif /* !(VMS && VMS_PK_EXTRA) */
    l = issymlnk(a);
    if (l) {
      ifile = fbad;
      m = STORE;
    }
    else if (isdir) { /* directory */
      ifile = fbad;
      m = STORE;
      q = 0;
    }
#ifdef THEOS
    else if (((a >> 16) & S_IFMT) == S_IFLIB) {   /* library */
      ifile = fbad;
      m = STORE;
      q = 0;
    }
#endif
    else {
#ifdef CMS_MVS
      if (bflag) {
        if ((ifile = zopen(z->name, fhowb)) == fbad)
           return ZE_OPEN;
      }
      else
#endif /* CMS_MVS */
      if ((ifile = zopen(z->name, fhow)) == fbad)
         return ZE_OPEN;