isamc.c

harvest是一个下载html网页得机器人

	assert (is->files[cat].alloc_entries_num > 0);
	is->files[cat].alloc_entries_num--;
	if (!is->files[cat].alloc_entries_num)  /* last one in block? */
	{
	    memcpy (&is->files[cat].head.freelist, abuf + sizeof(int),
		    sizeof(int));
	    is->files[cat].head_is_dirty = 1;

	    if (is->files[cat].head.freelist)
	    {
		bf_read (is->files[cat].bf, is->files[cat].head.freelist,
			 0, 0, abuf);
		memcpy (&is->files[cat].alloc_entries_num, abuf,
			sizeof(is->files[cat].alloc_entries_num));
		assert (is->files[cat].alloc_entries_num);
	    }
	}
	else
	    memcpy (&block, abuf + sizeof(int) + sizeof(int) *
		    is->files[cat].alloc_entries_num, sizeof(int));
    }
    return block;
}

static void release_block (ISAMC is, int cat, int pos)
{
    char *abuf = is->files[cat].alloc_buf;
    int block = is->files[cat].head.freelist;

    (is->files[cat].no_released)++;

    if (block && !is->files[cat].alloc_entries_num) /* must read block */
    {
	bf_read (is->files[cat].bf, block, 0, 0, abuf);
	memcpy (&is->files[cat].alloc_entries_num, abuf,
		sizeof(is->files[cat].alloc_entries_num));
	assert (is->files[cat].alloc_entries_num > 0);
    }
    assert (is->files[cat].alloc_entries_num <= is->files[cat].alloc_entries_max);
    if (is->files[cat].alloc_entries_num == is->files[cat].alloc_entries_max)
    {
	assert (block);
	memcpy (abuf, &is->files[cat].alloc_entries_num, sizeof(int));
	bf_write (is->files[cat].bf, block, 0, 0, abuf);
	is->files[cat].alloc_entries_num = 0;
    }
    if (!is->files[cat].alloc_entries_num) /* make new buffer? */
    {
	memcpy (abuf + sizeof(int), &block, sizeof(int));
	is->files[cat].head.freelist = pos;
	is->files[cat].head_is_dirty = 1; 
    }
    else
    {
	memcpy (abuf + sizeof(int) +
		is->files[cat].alloc_entries_num*sizeof(int),
		&pos, sizeof(int));
    }
    is->files[cat].alloc_entries_num++;
}
#else
static void flush_block (ISAMC is, int cat)
{
    char *abuf = is->files[cat].alloc_buf;
    xfree (abuf);
}

static int alloc_block (ISAMC is, int cat)
{
    int block;
    char buf[sizeof(int)];

    is->files[cat].head_is_dirty = 1;
    (is->files[cat].no_allocated)++;
    if ((block = is->files[cat].head.freelist))
    {
        bf_read (is->files[cat].bf, block, 0, sizeof(int), buf);
        memcpy (&is->files[cat].head.freelist, buf, sizeof(int));
    }
    else
        block = (is->files[cat].head.lastblock)++;
    return block;
}

static void release_block (ISAMC is, int cat, int pos)
{
    char buf[sizeof(int)];
   
    (is->files[cat].no_released)++;
    is->files[cat].head_is_dirty = 1; 
    memcpy (buf, &is->files[cat].head.freelist, sizeof(int));
    is->files[cat].head.freelist = pos;
    bf_write (is->files[cat].bf, pos, 0, sizeof(int), buf);
}
#endif

int isc_alloc_block (ISAMC is, int cat)
{
    int block = 0;

    if (is->files[cat].fc_list)
    {
        int j, nb;
        for (j = 0; j < is->files[cat].fc_max; j++)
            if ((nb = is->files[cat].fc_list[j]) && (!block || nb < block))
            {
                is->files[cat].fc_list[j] = 0;
		block = nb;
                break;
            }
    }
    if (!block)
        block = alloc_block (is, cat);
    if (is->method->debug > 3)
        logf (LOG_LOG, "isc: alloc_block in cat %d: %d", cat, block);
    return block;
}

void isc_release_block (ISAMC is, int cat, int pos)
{
    if (is->method->debug > 3)
        logf (LOG_LOG, "isc: release_block in cat %d: %d", cat, pos);
    if (is->files[cat].fc_list)
    {
        int j;
        for (j = 0; j<is->files[cat].fc_max; j++)
            if (!is->files[cat].fc_list[j])
            {
                is->files[cat].fc_list[j] = pos;
                return;
            }
    }
    release_block (is, cat, pos);
}

static void init_fc (ISAMC is, int cat)
{
    int j = 100;
        
    is->files[cat].fc_max = j;
    is->files[cat].fc_list = (int *)
	xmalloc (sizeof(*is->files[0].fc_list) * j);
    while (--j >= 0)
        is->files[cat].fc_list[j] = 0;
}

static void release_fc (ISAMC is, int cat)
{
    int b, j = is->files[cat].fc_max;

    while (--j >= 0)
        if ((b = is->files[cat].fc_list[j]))
        {
            release_block (is, cat, b);
            is->files[cat].fc_list[j] = 0;
        }
}

void isc_pp_close (ISAMC_PP pp)
{
    ISAMC is = pp->is;

    (*is->method->code_stop)(ISAMC_DECODE, pp->decodeClientData);
    xfree (pp->buf);
    xfree (pp);
}

ISAMC_PP isc_pp_open (ISAMC is, ISAMC_P ipos)
{
    ISAMC_PP pp = (ISAMC_PP) xmalloc (sizeof(*pp));
    char *src;
   
    pp->cat = isc_type(ipos);
    pp->pos = isc_block(ipos); 

    src = pp->buf = (char *) xmalloc (is->method->filecat[pp->cat].bsize);

    pp->next = 0;
    pp->size = 0;
    pp->offset = 0;
    pp->is = is;
    pp->decodeClientData = (*is->method->code_start)(ISAMC_DECODE);
    pp->deleteFlag = 0;
    pp->numKeys = 0;

    if (pp->pos)
    {
        src = pp->buf;
        isc_read_block (is, pp->cat, pp->pos, src);
        memcpy (&pp->next, src, sizeof(pp->next));
        src += sizeof(pp->next);
        memcpy (&pp->size, src, sizeof(pp->size));
        src += sizeof(pp->size);
        memcpy (&pp->numKeys, src, sizeof(pp->numKeys));
        src += sizeof(pp->numKeys);
        assert (pp->next != pp->pos);
        pp->offset = src - pp->buf; 
        assert (pp->offset == ISAMC_BLOCK_OFFSET_1);
        if (is->method->debug > 2)
            logf (LOG_LOG, "isc: read_block size=%d %d %d next=%d",
                 pp->size, pp->cat, pp->pos, pp->next);
    }
    return pp;
}

/* returns non-zero if item could be read; 0 otherwise */
int isc_pp_read (ISAMC_PP pp, void *buf)
{
    return isc_read_item (pp, (char **) &buf);
}

/* read one item from file - decode and store it in *dst.
   Returns
     0 if end-of-file
     1 if item could be read ok and NO boundary
     2 if item could be read ok and boundary */
int isc_read_item (ISAMC_PP pp, char **dst)
{
    ISAMC is = pp->is;
    char *src = pp->buf + pp->offset;

    if (pp->offset >= pp->size)
    {
	if (!pp->next)
	{
	    pp->pos = 0;
	    return 0; /* end of file */
	}
	if (pp->next > pp->pos)
	{
	    if (pp->next == pp->pos + 1)
		is->files[pp->cat].no_next++;
	    else
	    {
		is->files[pp->cat].no_forward++;
		is->files[pp->cat].sum_forward += pp->next - pp->pos;
	    }
	}
	else
	{
	    if (pp->next + 1 == pp->pos)
		is->files[pp->cat].no_prev++;
	    else
	    {
		is->files[pp->cat].no_backward++;
		is->files[pp->cat].sum_backward += pp->pos - pp->next;
	    }
	}
	/* out new block position */
        pp->pos = pp->next;
        src = pp->buf;
	/* read block and save 'next' and 'size' entry */
        isc_read_block (is, pp->cat, pp->pos, src);
        memcpy (&pp->next, src, sizeof(pp->next));
        src += sizeof(pp->next);
        memcpy (&pp->size, src, sizeof(pp->size));
        src += sizeof(pp->size);
        /* assume block is non-empty */
        assert (src - pp->buf == ISAMC_BLOCK_OFFSET_N);
        assert (pp->next != pp->pos);
        if (pp->deleteFlag)
            isc_release_block (is, pp->cat, pp->pos);
        (*is->method->code_item)(ISAMC_DECODE, pp->decodeClientData, dst, &src);
        pp->offset = src - pp->buf; 
        if (is->method->debug > 2)
            logf (LOG_LOG, "isc: read_block size=%d %d %d next=%d",
                 pp->size, pp->cat, pp->pos, pp->next);
        return 2;
    }
    (*is->method->code_item)(ISAMC_DECODE, pp->decodeClientData, dst, &src);
    pp->offset = src - pp->buf; 
    return 1;
}

int isc_pp_num (ISAMC_PP pp)
{
    return pp->numKeys;
}