chm_lib.cpp

chm文件阅读器.开发工具EVC 支持ARM处理器和 模拟器环境 可直接编译通过 基于PPC的 也可以编译为WINCE下的程序

/*
 * set a parameter on the file handle.
 * valid parameter types:
 *          CHM_PARAM_MAX_BLOCKS_CACHED:
 *                 how many decompressed blocks should be cached?  A simple
 *                 caching scheme is used, wherein the index of the block is
 *                 used as a hash value, and hash collision results in the
 *                 invalidation of the previously cached block.
 */
void chm_set_param(struct chmFile *h,
                   int paramType,
                   int paramVal)
{
    switch (paramType)
    {
        case CHM_PARAM_MAX_BLOCKS_CACHED:
            
            if (paramVal != h->cache_num_blocks)
            {
                UChar **newBlocks;
                UInt64 *newIndices;
                int     i;

                /* allocate new cached blocks */
                newBlocks = (UChar **)malloc(paramVal * sizeof (UChar *));
                newIndices = (UInt64 *)malloc(paramVal * sizeof (UInt64));
                for (i=0; i<paramVal; i++)
                {
                    newBlocks[i] = NULL;
                    newIndices[i] = 0;
                }

                /* re-distribute old cached blocks */
                if (h->cache_blocks)
                {
                    for (i=0; i<h->cache_num_blocks; i++)
                    {
                        int newSlot = (int)(h->cache_block_indices[i] % paramVal);

                        if (h->cache_blocks[i])
                        {
                            /* in case of collision, destroy newcomer */
                            if (newBlocks[newSlot])
                            {
                                free(h->cache_blocks[i]);
                                h->cache_blocks[i] = NULL;
                            }
                            else
                            {
                                newBlocks[newSlot] = h->cache_blocks[i];
                                newIndices[newSlot] =
                                            h->cache_block_indices[i];
                            }
                        }
                    }

                    free(h->cache_blocks);
                    free(h->cache_block_indices);
                }

                /* now, set new values */
                h->cache_blocks = newBlocks;
                h->cache_block_indices = (__int64*)newIndices;
                h->cache_num_blocks = paramVal;
            }            
            break;

        default:
            break;
    }
}

/*
 * helper methods for chm_resolve_object
 */

/* skip a compressed dword */
static void _chm_skip_cword(UChar **pEntry)
{
    while (*(*pEntry)++ >= 0x80)
        ;
}

/* skip the data from a PMGL entry */
static void _chm_skip_PMGL_entry_data(UChar **pEntry)
{
    _chm_skip_cword(pEntry);
    _chm_skip_cword(pEntry);
    _chm_skip_cword(pEntry);
}

/* parse a compressed dword */
static UInt64 _chm_parse_cword(UChar **pEntry)
{
    UInt64 accum = 0;
    UChar temp;
    while ((temp=*(*pEntry)++) >= 0x80)
    {
        accum <<= 7;
        accum += temp & 0x7f;
    }

    return (accum << 7) + temp;
}

/* parse a utf-8 string into an ASCII char buffer */
static int _chm_parse_UTF8(UChar **pEntry, UInt64 count, char *path)
{
    /* XXX: implement UTF-8 support, including a real mapping onto
     *      ISO-8859-1?  probably there is a library to do this?  As is
     *      immediately apparent from the below code, I'm only handling files
     *      in which none of the strings contain UTF-8 multi-byte characters.
     */
    while (count != 0)
    {
        if (*(*pEntry) > 0x7f)
            return 0;

        *path++ = (char)(*(*pEntry)++);
        --count;
    }

    *path = '\0';
    return 1;
}

/* parse a PMGL entry into a chmUnitInfo struct; return 1 on success. */
static int _chm_parse_PMGL_entry(UChar **pEntry, struct chmUnitInfo *ui)
{
    UInt64 strLen;

    /* parse str len */
    strLen = _chm_parse_cword(pEntry);
    if (strLen > CHM_MAX_PATHLEN)
        return 0;

    /* parse path */
    if (! _chm_parse_UTF8(pEntry, strLen, ui->path))
        return 0;

    /* parse info */
    ui->space  = (int)_chm_parse_cword(pEntry);
    ui->start  = _chm_parse_cword(pEntry);
    ui->length = _chm_parse_cword(pEntry);
    return 1;
}

/* find an exact entry in PMGL; return NULL if we fail */
static UChar *_chm_find_in_PMGL(UChar *page_buf,
                         UInt32 block_len,
                         const char *objPath)
{
    /* XXX: modify this to do a binary search using the nice index structure
     *      that is provided for us.
     */
    struct chmPmglHeader header;
    UInt32 hremain;
    UChar *end;
    UChar *cur;
    UChar *temp;
    UInt64 strLen;
    char buffer[CHM_MAX_PATHLEN+1];

    /* figure out where to start and end */
    cur = page_buf;
    hremain = _CHM_PMGL_LEN;
    if (! _unmarshal_pmgl_header(&cur, (unsigned long *)&hremain, &header))
        return NULL;
    end = page_buf + block_len - (header.free_space);

    /* now, scan progressively */
    while (cur < end)
    {
        /* grab the name */
        temp = cur;
        strLen = _chm_parse_cword(&cur);
        if (! _chm_parse_UTF8(&cur, strLen, buffer))
            return NULL;

        /* check if it is the right name */
		if (!_stricmp(buffer, objPath))
           return temp;

        _chm_skip_PMGL_entry_data(&cur);
    }

    return NULL;
}

/* find which block should be searched next for the entry; -1 if no block */
static Int32 _chm_find_in_PMGI(UChar *page_buf,
                        UInt32 block_len,
                        const char *objPath)
{
    /* XXX: modify this to do a binary search using the nice index structure
     *      that is provided for us
     */
    struct chmPmgiHeader header;
    UInt32 hremain;
    int page=-1;
    UChar *end;
    UChar *cur;
    UInt64 strLen;
    char buffer[CHM_MAX_PATHLEN+1];

    /* figure out where to start and end */
    cur = page_buf;
    hremain = _CHM_PMGI_LEN;
    if (! _unmarshal_pmgi_header(&cur, (unsigned long*)&hremain, &header))
        return -1;
    end = page_buf + block_len - (header.free_space);

    /* now, scan progressively */
    while (cur < end)
    {
        /* grab the name */
        strLen = _chm_parse_cword(&cur);
        if (! _chm_parse_UTF8(&cur, strLen, buffer))
            return -1;

        /* check if it is the right name */
        if (_stricmp(buffer, objPath) > 0)
            return page;

        /* load next value for path */
        page = (int)_chm_parse_cword(&cur);
    }

    return page;
}

/* resolve a particular object from the archive */
int chm_resolve_object(struct chmFile *h,
                       const char *objPath,
                       struct chmUnitInfo *ui)
{
    /*
     * XXX: implement caching scheme for dir pages
     */

    Int32 curPage;

    /* buffer to hold whatever page we're looking at */
        UChar *page_buf = (UChar*)alloca(h->block_len);

    /* starting page */
    curPage = h->index_root;

    /* until we have either returned or given up */
    while (curPage != -1)
    {

        /* try to fetch the index page */
        if (_chm_fetch_bytes(h, page_buf,
                             (UInt64)h->dir_offset + (UInt64)curPage*h->block_len,
                             h->block_len) != h->block_len)
            return CHM_RESOLVE_FAILURE;

        /* now, if it is a leaf node: */
        if (memcmp(page_buf, _chm_pmgl_marker, 4) == 0)
        {
            /* scan block */
            UChar *pEntry = _chm_find_in_PMGL(page_buf,
                                              h->block_len,
                                              objPath);
            if (pEntry == NULL)
                return CHM_RESOLVE_FAILURE;

            /* parse entry and return */
            _chm_parse_PMGL_entry(&pEntry, ui);
            return CHM_RESOLVE_SUCCESS;
        }

        /* else, if it is a branch node: */
        else if (memcmp(page_buf, _chm_pmgi_marker, 4) == 0)
            curPage = _chm_find_in_PMGI(page_buf, h->block_len, objPath);

        /* else, we are confused.  give up. */
        else
            return CHM_RESOLVE_FAILURE;
    }

        /* didn't find anything.  fail. */
        return CHM_RESOLVE_FAILURE;
}

/*
 * utility methods for dealing with compressed data
 */

/* get the bounds of a compressed block.  return 0 on failure */
static int _chm_get_cmpblock_bounds(struct chmFile *h,
                             UInt64 block,
                             UInt64 *start,
                             Int64 *len)
{
    UChar buffer[8], *dummy;
    UInt32 remain;

    /* for all but the last block, use the reset table */
    if (block < h->reset_table.block_count-1)
    {
        /* unpack the start address */
        dummy = buffer;
        remain = 8;
        if (_chm_fetch_bytes(h, buffer,
                             (UInt64)h->data_offset
                                + (UInt64)h->rt_unit.start
                                + (UInt64)h->reset_table.table_offset
                                + (UInt64)block*8,
                             remain) != remain                            ||
            !_unmarshal_uint64(&dummy, (unsigned long *)&remain, start))
            return 0;

        /* unpack the end address */
        dummy = buffer;
        remain = 8;
        if (_chm_fetch_bytes(h, buffer,
                         (UInt64)h->data_offset
                                + (UInt64)h->rt_unit.start
                                + (UInt64)h->reset_table.table_offset
                                + (UInt64)block*8 + 8,
                         remain) != remain                                ||
            !_unmarshal_int64(&dummy, (unsigned long*)&remain, len))
            return 0;
    }

    /* for the last block, use the span in addition to the reset table */
    else
    {
        /* unpack the start address */
        dummy = buffer;
        remain = 8;
        if (_chm_fetch_bytes(h, buffer,
                             (UInt64)h->data_offset
                                + (UInt64)h->rt_unit.start
                                + (UInt64)h->reset_table.table_offset
                                + (UInt64)block*8,
                             remain) != remain                            ||
            !_unmarshal_uint64(&dummy, (unsigned long*)&remain, start))
            return 0;

        *len = h->reset_table.compressed_len;
    }

    /* compute the length and absolute start address */
    *len -= *start;
    *start += h->data_offset + h->cn_unit.start;

    return 1;
}

/* decompress the block.  must have lzx_mutex. */
static Int64 _chm_decompress_block(struct chmFile *h,
                                   UInt64 block,
                                   UChar **ubuffer)
{
        UChar *cbuffer = (UChar*) alloca(((unsigned int)h->reset_table.block_len + 6144));
    UInt64 cmpStart;                                    /* compressed start  */
    Int64 cmpLen;                                       /* compressed len    */
    int indexSlot;                                      /* cache index slot  */
    UChar *lbuffer;                                     /* local buffer ptr  */
    UInt32 blockAlign = (UInt32)(block % h->reset_blkcount); /* reset intvl. aln. */
    UInt32 i;                                           /* local loop index  */

    /* check if we need previous blocks */
    if (blockAlign != 0)
    {
        /* fetch all required previous blocks since last reset */
        for (i = h->reset_blkcount - blockAlign; i > 0; i--)
        {

            /* check if we most recently decompressed the previous block */
            if (h->lzx_last_block != block-i)
            {
                indexSlot = (int)((block-i) % h->cache_num_blocks);
                h->cache_block_indices[indexSlot] = block-i;
                if (! h->cache_blocks[indexSlot])