jdhuff.c

这是在PCA下的基于IPP库示例代码例子,在网上下了IPP的库之后,设置相关参数就可以编译该代码.

/*
 * jdhuff.c
 *
 * Copyright (C) 1991-1997, Thomas G. Lane.
 * This file is part of the Independent JPEG Group's software.
 * For conditions of distribution and use, see the accompanying README file.
 *
 * This file contains Huffman entropy decoding routines.
 *
 * Much of the complexity here has to do with supporting input suspension.
 * If the data source module demands suspension, we want to be able to back
 * up to the start of the current MCU.  To do this, we copy state variables
 * into local working storage, and update them back to the permanent
 * storage only upon successful completion of an MCU.
 */

#define JPEG_INTERNALS
#include "jinclude.h"
#include "jpeglib.h"
#include "jdhuff.h"   /* Declarations shared with jdphuff.c */
#ifdef USE_IPP
#include "jpegipp.h"
#endif

/*
 * Expanded entropy decoder object for Huffman decoding.
 *
 * The savable_state subrecord contains fields that change within an MCU,
 * but must not be updated permanently until we complete the MCU.
 */

typedef struct {
  int last_dc_val[MAX_COMPS_IN_SCAN]; /* last DC coef for each component */
#ifdef IPPJ_HUFF
  IppiDecodeHuffmanState* pDecHuffState;
#endif
} savable_state;


/* This macro is to work around compilers with missing or broken
 * structure assignment.  You'll need to fix this code if you have
 * such a compiler and you change MAX_COMPS_IN_SCAN.
 */

#ifndef IPPJ_HUFF
#ifndef NO_STRUCT_ASSIGN
#define ASSIGN_STATE(dest,src)  ((dest) = (src))
#else
#if MAX_COMPS_IN_SCAN == 4
#define ASSIGN_STATE(dest,src) \
  ((dest).last_dc_val[0] = (src).last_dc_val[0], \
   (dest).last_dc_val[1] = (src).last_dc_val[1], \
   (dest).last_dc_val[2] = (src).last_dc_val[2], \
   (dest).last_dc_val[3] = (src).last_dc_val[3])
#endif
#endif
#else
#ifndef NO_STRUCT_ASSIGN
#define ASSIGN_STATE(dest,src)  ((dest) = (src))
#else
#if MAX_COMPS_IN_SCAN == 4
#define ASSIGN_STATE(dest,src) \
  ((dest).last_dc_val[0] = (src).last_dc_val[0], \
   (dest).last_dc_val[1] = (src).last_dc_val[1], \
   (dest).last_dc_val[2] = (src).last_dc_val[2], \
   (dest).last_dc_val[3] = (src).last_dc_val[3], \
   (dest).pDecHuffState  = (src).pDecHuffState)
#endif
#endif
#endif


typedef struct {
  struct jpeg_entropy_decoder pub; /* public fields */

  /* These fields are loaded into local variables at start of each MCU.
   * In case of suspension, we exit WITHOUT updating them.
   */
  bitread_perm_state bitstate;  /* Bit buffer at start of MCU */
  savable_state saved;    /* Other state at start of MCU */

  /* These fields are NOT loaded into local working state. */
  unsigned int restarts_to_go;  /* MCUs left in this restart interval */

  /* Pointers to derived tables (these workspaces have image lifespan) */
  d_derived_tbl * dc_derived_tbls[NUM_HUFF_TBLS];
  d_derived_tbl * ac_derived_tbls[NUM_HUFF_TBLS];

  /* Precalculated info set up by start_pass for use in decode_mcu: */

  /* Pointers to derived tables to be used for each block within an MCU */
  d_derived_tbl * dc_cur_tbls[D_MAX_BLOCKS_IN_MCU];
  d_derived_tbl * ac_cur_tbls[D_MAX_BLOCKS_IN_MCU];
  /* Whether we care about the DC and AC coefficient values for each block */
  boolean dc_needed[D_MAX_BLOCKS_IN_MCU];
  boolean ac_needed[D_MAX_BLOCKS_IN_MCU];
} huff_entropy_decoder;

typedef huff_entropy_decoder * huff_entropy_ptr;


/*
 * Initialize for a Huffman-compressed scan.
 */

METHODDEF(void)
start_pass_huff_decoder (j_decompress_ptr cinfo)
{
  huff_entropy_ptr entropy = (huff_entropy_ptr) cinfo->entropy;
  int ci, blkn, dctbl, actbl;
  jpeg_component_info * compptr;

  /* Check that the scan parameters Ss, Se, Ah/Al are OK for sequential JPEG.
   * This ought to be an error condition, but we make it a warning because
   * there are some baseline files out there with all zeroes in these bytes.
   */
  if (cinfo->Ss != 0 || cinfo->Se != DCTSIZE2-1 ||
      cinfo->Ah != 0 || cinfo->Al != 0)
    WARNMS(cinfo, JWRN_NOT_SEQUENTIAL);

  for (ci = 0; ci < cinfo->comps_in_scan; ci++) {
    compptr = cinfo->cur_comp_info[ci];
    dctbl = compptr->dc_tbl_no;
    actbl = compptr->ac_tbl_no;
    /* Compute derived values for Huffman tables */
    /* We may do this more than once for a table, but it's not expensive */
#ifndef IPPJ_HUFF
    jpeg_make_d_derived_tbl(cinfo, TRUE, dctbl,
          & entropy->dc_derived_tbls[dctbl]);
    jpeg_make_d_derived_tbl(cinfo, FALSE, actbl,
          & entropy->ac_derived_tbls[actbl]);
#else
    if(cinfo->UseIPP) {
      jpeg_make_d_derived_tbl_intellib(cinfo, TRUE, dctbl,
            & entropy->dc_derived_tbls[dctbl]);
      jpeg_make_d_derived_tbl_intellib(cinfo, FALSE, actbl,
            & entropy->ac_derived_tbls[actbl]);
    } else {
      jpeg_make_d_derived_tbl(cinfo, TRUE, dctbl,
            & entropy->dc_derived_tbls[dctbl]);
      jpeg_make_d_derived_tbl(cinfo, FALSE, actbl,
            & entropy->ac_derived_tbls[actbl]);
    }
#endif
    /* Initialize DC predictions to 0 */
    entropy->saved.last_dc_val[ci] = 0;
  }

  /* Precalculate decoding info for each block in an MCU of this scan */
  for (blkn = 0; blkn < cinfo->blocks_in_MCU; blkn++) {
    ci = cinfo->MCU_membership[blkn];
    compptr = cinfo->cur_comp_info[ci];
    /* Precalculate which table to use for each block */
    entropy->dc_cur_tbls[blkn] = entropy->dc_derived_tbls[compptr->dc_tbl_no];
    entropy->ac_cur_tbls[blkn] = entropy->ac_derived_tbls[compptr->ac_tbl_no];
    /* Decide whether we really care about the coefficient values */
    if (compptr->component_needed) {
      entropy->dc_needed[blkn] = TRUE;
      /* we don't need the ACs if producing a 1/8th-size image */
      entropy->ac_needed[blkn] = (compptr->DCT_scaled_size > 1);
    } else {
      entropy->dc_needed[blkn] = entropy->ac_needed[blkn] = FALSE;
    }
  }

  /* Initialize bitread state variables */
  entropy->bitstate.bits_left = 0;
  entropy->bitstate.get_buffer = 0; /* unnecessary, but keeps Purify quiet */
  entropy->pub.insufficient_data = FALSE;

#ifdef IPPJ_HUFF
  if(cinfo->UseIPP) {
    int size = 0;
    entropy->saved.pDecHuffState = NULL;

    ippiDecodeHuffmanStateGetBufSize_JPEG_8u(&size);

    entropy->saved.pDecHuffState = (IppiDecodeHuffmanState*)
      (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE, size);

    ippiDecodeHuffmanStateInit_JPEG_8u(entropy->saved.pDecHuffState);
  }
#endif

  /* Initialize restart counter */
  entropy->restarts_to_go = cinfo->restart_interval;
}


/*
 * Compute the derived values for a Huffman table.
 * This routine also performs some validation checks on the table.
 *
 * Note this is also used by jdphuff.c.
 */

GLOBAL(void)
jpeg_make_d_derived_tbl (j_decompress_ptr cinfo, boolean isDC, int tblno,
       d_derived_tbl ** pdtbl)
{
  JHUFF_TBL *htbl;
  d_derived_tbl *dtbl;
  int p, i, l, si, numsymbols;
  int lookbits, ctr;
  char huffsize[257];
  unsigned int huffcode[257];
  unsigned int code;

  /* Note that huffsize[] and huffcode[] are filled in code-length order,
   * paralleling the order of the symbols themselves in htbl->huffval[].
   */

  /* Find the input Huffman table */
  if (tblno < 0 || tblno >= NUM_HUFF_TBLS)
    ERREXIT1(cinfo, JERR_NO_HUFF_TABLE, tblno);
  htbl =
    isDC ? cinfo->dc_huff_tbl_ptrs[tblno] : cinfo->ac_huff_tbl_ptrs[tblno];
  if (htbl == NULL)
    ERREXIT1(cinfo, JERR_NO_HUFF_TABLE, tblno);

  /* Allocate a workspace if we haven't already done so. */
  if (*pdtbl == NULL)
    *pdtbl = (d_derived_tbl *)
      (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
          SIZEOF(d_derived_tbl));
  dtbl = *pdtbl;
  dtbl->pub = htbl;   /* fill in back link */
  
  /* Figure C.1: make table of Huffman code length for each symbol */

  p = 0;
  for (l = 1; l <= 16; l++) {
    i = (int) htbl->bits[l];
    if (i < 0 || p + i > 256) /* protect against table overrun */
      ERREXIT(cinfo, JERR_BAD_HUFF_TABLE);
    while (i--)
      huffsize[p++] = (char) l;
  }
  huffsize[p] = 0;
  numsymbols = p;
  
  /* Figure C.2: generate the codes themselves */
  /* We also validate that the counts represent a legal Huffman code tree. */
  
  code = 0;
  si = huffsize[0];
  p = 0;
  while (huffsize[p]) {
    while (((int) huffsize[p]) == si) {
      huffcode[p++] = code;
      code++;
    }
    /* code is now 1 more than the last code used for codelength si; but
     * it must still fit in si bits, since no code is allowed to be all ones.
     */
    if (((INT32) code) >= (((INT32) 1) << si))
      ERREXIT(cinfo, JERR_BAD_HUFF_TABLE);
    code <<= 1;
    si++;
  }

  /* Figure F.15: generate decoding tables for bit-sequential decoding */

  p = 0;
  for (l = 1; l <= 16; l++) {
    if (htbl->bits[l]) {
      /* valoffset[l] = huffval[] index of 1st symbol of code length l,
       * minus the minimum code of length l
       */
      dtbl->valoffset[l] = (INT32) p - (INT32) huffcode[p];
      p += htbl->bits[l];
      dtbl->maxcode[l] = huffcode[p-1]; /* maximum code of length l */
    } else {
      dtbl->maxcode[l] = -1;  /* -1 if no codes of this length */
    }
  }
  dtbl->maxcode[17] = 0xFFFFFL; /* ensures jpeg_huff_decode terminates */

  /* Compute lookahead tables to speed up decoding.
   * First we set all the table entries to 0, indicating "too long";
   * then we iterate through the Huffman codes that are short enough and
   * fill in all the entries that correspond to bit sequences starting
   * with that code.
   */

  MEMZERO(dtbl->look_nbits, SIZEOF(dtbl->look_nbits));

  p = 0;
  for (l = 1; l <= HUFF_LOOKAHEAD; l++) {
    for (i = 1; i <= (int) htbl->bits[l]; i++, p++) {
      /* l = current code's length, p = its index in huffcode[] & huffval[]. */
      /* Generate left-justified code followed by all possible bit sequences */
      lookbits = huffcode[p] << (HUFF_LOOKAHEAD-l);
      for (ctr = 1 << (HUFF_LOOKAHEAD-l); ctr > 0; ctr--) {
        dtbl->look_nbits[lookbits] = l;
        dtbl->look_sym[lookbits] = htbl->huffval[p];
        lookbits++;
      }
    }
  }

  /* Validate symbols as being reasonable.
   * For AC tables, we make no check, but accept all byte values 0..255.
   * For DC tables, we require the symbols to be in range 0..15.
   * (Tighter bounds could be applied depending on the data depth and mode,
   * but this is sufficient to ensure safe decoding.)
   */
  if (isDC) {
    for (i = 0; i < numsymbols; i++) {
      int sym = htbl->huffval[i];
      if (sym < 0 || sym > 15)
        ERREXIT(cinfo, JERR_BAD_HUFF_TABLE);
    }
  }
}

#ifdef IPPJ_HUFF
GLOBAL(void)
jpeg_make_d_derived_tbl_intellib(
  j_decompress_ptr cinfo,
  boolean          isDC,
  int              tblno,
  d_derived_tbl**  pdtbl)
{
  int            size;
  JHUFF_TBL*     htbl;
  d_derived_tbl* dtbl;
  IppStatus      status;

  /* Note that huffsize[] and huffcode[] are filled in code-length order,
   * paralleling the order of the symbols themselves in htbl->huffval[].
   */

  /* Find the input Huffman table */
  if(tblno < 0 || tblno >= NUM_HUFF_TBLS)
  {
    ERREXIT1(cinfo,JERR_NO_HUFF_TABLE,tblno);
  }

  htbl =
    isDC ? cinfo->dc_huff_tbl_ptrs[tblno] : cinfo->ac_huff_tbl_ptrs[tblno];

  if(htbl == NULL)
  {
    ERREXIT1(cinfo,JERR_NO_HUFF_TABLE,tblno);
  }

  /* Allocate a workspace if we haven't already done so. */
  if(*pdtbl == NULL)
  {
    *pdtbl = (d_derived_tbl *)
      (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
          SIZEOF(d_derived_tbl));

    ippiDecodeHuffmanSpecGetBufSize_JPEG_8u(&size);

    (*pdtbl)->pHuffTbl = (IppiDecodeHuffmanSpec*)
      (*cinfo->mem->alloc_small)((j_common_ptr)cinfo,JPOOL_IMAGE,size);
  }

  status = ippiDecodeHuffmanSpecInit_JPEG_8u(&htbl->bits[1],htbl->huffval,(*pdtbl)->pHuffTbl);
  if(ippStsNoErr != status)
  {
    ERREXIT(cinfo,JERR_BAD_HUFF_TABLE);
  }

  dtbl = *pdtbl;

  return;
} /* jpeg_make_d_derived_tbl_intellib() */
#endif

/*
 * Out-of-line code for bit fetching (shared with jdphuff.c).
 * See jdhuff.h for info about usage.
 * Note: current values of get_buffer and bits_left are passed as parameters,
 * but are returned in the corresponding fields of the state struct.
 *
 * On most machines MIN_GET_BITS should be 25 to allow the full 32-bit width
 * of get_buffer to be used.  (On machines with wider words, an even larger
 * buffer could be used.)  However, on some machines 32-bit shifts are
 * quite slow and take time proportional to the number of places shifted.
 * (This is true with most PC compilers, for instance.)  In this case it may
 * be a win to set MIN_GET_BITS to the minimum value of 15.  This reduces the
 * average shift distance at the cost of more calls to jpeg_fill_bit_buffer.
 */

#ifdef SLOW_SHIFT_32
#define MIN_GET_BITS  15  /* minimum allowable value */
#else
#define MIN_GET_BITS  (BIT_BUF_SIZE-7)
#endif

GLOBAL(boolean)
jpeg_fill_bit_buffer (bitread_working_state * state,
          register bit_buf_type get_buffer, register int bits_left,
          int nbits)
/* Load up the bit buffer to a depth of at least nbits */
{
  /* Copy heavily used state fields into locals (hopefully registers) */
  register const JOCTET * next_input_byte = state->next_input_byte;
  register size_t bytes_in_buffer = state->bytes_in_buffer;
  j_decompress_ptr cinfo = state->cinfo;

  /* Attempt to load at least MIN_GET_BITS bits into get_buffer. */
  /* (It is assumed that no request will be for more than that many bits.) */
  /* We fail to do so only if we hit a marker or are forced to suspend. */

  if (cinfo->unread_marker == 0) {  /* cannot advance past a marker */
    while (bits_left < MIN_GET_BITS) {
      register int c;

      /* Attempt to read a byte */
      if (bytes_in_buffer == 0) {
        if (! (*cinfo->src->fill_input_buffer) (cinfo))
          return FALSE;
        next_input_byte = cinfo->src->next_input_byte;
        bytes_in_buffer = cinfo->src->bytes_in_buffer;
      }
      bytes_in_buffer--;
      c = GETJOCTET(*next_input_byte++);

      /* If it's 0xFF, check and discard stuffed zero byte */
      if (c == 0xFF) {
        /* Loop here to discard any padding FF's on terminating marker,