merge.c

harvest是一个下载html网页得机器人

                                           &i_mode);
                /* is next input item the same as current except
                   for the delete flag? */
                cmp = (*is->method->compare_item)(i_item, f_item);
                if (!cmp && i_mode)   /* delete/insert nop? */
                {
                    /* yes! insert as if it was an insert only */
                    memcpy (r_item, i_item, i_item_ptr - i_item);
                    i_item_ptr = i_item;
                    i_more = (*data->read_item)(data->clientData, &i_item_ptr,
                                                &i_mode);
                }
                else
                {
                    /* no! delete the item */
                    r_item = NULL;
		    last_dirty = 1;
                    mb[ptr].dirty = 2;
                }
            }
            else
            {
                memcpy (r_item, f_item, f_item_ptr - f_item);

                /* move i */
                i_item_ptr = i_item;
                i_more = (*data->read_item)(data->clientData, &i_item_ptr,
                                           &i_mode);
            }
            /* move f */
            f_item_ptr = f_item;
            f_more = isc_read_item (pp, &f_item_ptr);
        }
        else if (cmp > 0)               /* insert f */
        {
            memcpy (r_item, f_item, f_item_ptr - f_item);
            /* move f */
            f_item_ptr = f_item;
            f_more = isc_read_item (pp, &f_item_ptr);
        }
        else                            /* insert i */
        {
            if (!i_mode)                /* delete item which isn't there? */
            {
                logf (LOG_FATAL, "Inconsistent register at offset %d",
                                 r_offset);
                abort ();
            }
            memcpy (r_item, i_item, i_item_ptr - i_item);
            mb[ptr].dirty = 2;
	    last_dirty = 1;
            /* move i */
            i_item_ptr = i_item;
            i_more = (*data->read_item)(data->clientData, &i_item_ptr,
                                        &i_mode);
        }
        if (r_item)  /* insert resulting item? */
        {
            char *r_out_ptr = r_buf + r_offset;
            int new_offset;

            (*is->method->code_item)(ISAMC_ENCODE, r_clientData,
                                     &r_out_ptr, &r_item);
            new_offset = r_out_ptr - r_buf; 

            numKeys++;

            if (border < new_offset && border >= r_offset)
            {
                if (debug > 2)
                    logf (LOG_LOG, "isc: border %d %d", ptr, border);
                /* Max size of current block category reached ...
                   make new virtual block entry */
                mb[++ptr].block = 0;
                mb[ptr].dirty = 1;
                mb[ptr].offset = r_offset;
                if (cat == is->max_cat && ptr >= is->method->max_blocks_mem)
                {
                    /* We are dealing with block(s) of max size. Block(s)
                       except one will be flushed. Note: the block(s) are
                       surely not the last one(s).
                     */
                    if (debug > 2)
                        logf (LOG_LOG, "isc: flush B %d sections", ptr-1);
                    flush_blocks (is, mb, ptr-1, r_buf, &firstpos, cat,
                                  0, &pp->numKeys);
                    mb[0].block = mb[ptr-1].block;
                    mb[0].dirty = mb[ptr-1].dirty;
                    memcpy (r_buf, r_buf + mb[ptr-1].offset,
                            mb[ptr].offset - mb[ptr-1].offset);
                    mb[0].offset = 0;

                    mb[1].block = mb[ptr].block;
                    mb[1].dirty = mb[0].dirty;
                    mb[1].offset = mb[ptr].offset - mb[ptr-1].offset;
                    memcpy (r_buf + mb[1].offset, r_buf + r_offset,
                            new_offset - r_offset);
                    new_offset = (new_offset - r_offset) + mb[1].offset;
                    ptr = 1;
                }
                border = get_border (is, mb, ptr, cat, firstpos);
            }
            r_offset = new_offset;
        }
        if (cat < is->max_cat && ptr >= is->method->filecat[cat].mblocks)
        {
            /* Max number blocks in current category reached ->
               must switch to next category (with larger block size) 
            */
            int j = 0;

            (is->files[cat].no_remap)++;
            /* delete all original block(s) read so far */
            for (i = 0; i < ptr; i++)
                if (mb[i].block)
                    isc_release_block (is, pp->cat, mb[i].block);
            /* also delete all block to be read in the future */
            pp->deleteFlag = 1;

            /* remap block offsets */
            assert (mb[j].offset == 0);
            cat++;
            mb[j].dirty = 1;
            mb[j].block = 0;
	    mb[ptr].offset = r_offset;
            for (i = 1; i < ptr; i++)
            {
                int border = is->method->filecat[cat].ifill -
                         ISAMC_BLOCK_OFFSET_1 + mb[j].offset;
                if (debug > 3)
                    logf (LOG_LOG, "isc: remap %d border=%d", i, border);
                if (mb[i+1].offset > border && mb[i].offset <= border)
                {
                    if (debug > 3)
                        logf (LOG_LOG, "isc:  to %d %d", j, mb[i].offset);
                    mb[++j].dirty = 1;
                    mb[j].block = 0;
                    mb[j].offset = mb[i].offset;
                }
            }
            if (debug > 2)
                logf (LOG_LOG, "isc: remap from %d to %d sections to cat %d",
                      ptr, j, cat);
            ptr = j;
            border = get_border (is, mb, ptr, cat, firstpos);
	    if (debug > 3)
		logf (LOG_LOG, "isc: border=%d r_offset=%d", border, r_offset);
        }
    }
    if (mb[ptr].offset < r_offset)
    {   /* make the final boundary offset */
        mb[++ptr].dirty = 1; 
        mb[ptr].block = 0; 
        mb[ptr].offset = r_offset;
    }
    else
    {   /* empty output. Release last block if any */
        if (cat == pp->cat && mb[ptr].block)
        {
            if (debug > 3)
                logf (LOG_LOG, "isc: release C");
            isc_release_block (is, pp->cat, mb[ptr].block);
            mb[ptr].block = 0;
	    if (ptr > 0)
		mb[ptr-1].dirty = 1;
        }
    }

    if (debug > 2)
        logf (LOG_LOG, "isc: flush C, %d sections", ptr);

    if (firstpos)
    {
        /* we have to patch initial block with num keys if that
           has changed */
        if (numKeys != isc_pp_num (pp))
        {
            if (debug > 2)
                logf (LOG_LOG, "isc: patch num keys firstpos=%d num=%d",
                                firstpos, numKeys);
            bf_write (is->files[cat].bf, firstpos, ISAMC_BLOCK_OFFSET_N,
                      sizeof(numKeys), &numKeys);
        }
    }
    else if (ptr > 0)
    {   /* we haven't flushed initial block yet and there surely are some
           blocks to flush. Make first block dirty if numKeys differ */
        if (numKeys != isc_pp_num (pp))
            mb[0].dirty = 1;
    }
    /* flush rest of block(s) in r_buf */
    flush_blocks (is, mb, ptr, r_buf, &firstpos, cat, 1, &numKeys);

    (*is->method->code_stop)(ISAMC_ENCODE, r_clientData);
    if (!firstpos)
        cat = 0;
    if (debug > 1)
        logf (LOG_LOG, "isc: isc_merge return %d %d", cat, firstpos);
    isc_pp_close (pp);
    return cat + firstpos * 8;
}

/*
 * $Log: merge.c,v $
 * Revision 1.23  2003/06/23 15:36:11  adam
 * Implemented isamb_unlink.
 *
 * Revision 1.22  2003/03/05 16:41:10  adam
 * Fix GCC warnings
 *
 * Revision 1.21  2002/08/02 19:26:56  adam
 * Towards GPL
 *
 * Revision 1.20  1999/11/30 13:48:04  adam
 * Improved installation. Updated for inclusion of YAZ header files.
 *
 * Revision 1.19  1999/07/14 12:12:07  heikki
 * Large-block isam-h  (may not work too well... Abandoning for isam-d)
 *
 * Revision 1.17  1999/07/13 14:22:17  heikki
 * Better allocation strategy in isamh_merge
 *
 * Revision 1.16  1999/07/08 14:23:27  heikki
 * Fixed a bug in isamh_pp_read and cleaned up a bit
 *
 * Revision 1.15  1999/07/07 09:36:04  heikki
 * Fixed an assertion in isamh
 *
 * Revision 1.13  1999/07/06 09:37:05  heikki
 * Working on isamh - not ready yet.
 *
 * Revision 1.12  1999/06/30 15:03:55  heikki
 * first take on isamh, the append-only isam structure
 *
 * Revision 1.11  1999/05/26 07:49:14  adam
 * C++ compilation.
 *
 * Revision 1.10  1998/03/19 12:22:09  adam
 * Minor change.
 *
 * Revision 1.9  1998/03/19 10:04:38  adam
 * Minor changes.
 *
 * Revision 1.8  1998/03/18 09:23:55  adam
 * Blocks are stored in chunks on free list - up to factor 2 in speed.
 * Fixed bug that could occur in block category rearrangemen.
 *
 * Revision 1.7  1998/03/11 11:18:18  adam
 * Changed the isc_merge to take into account the mfill (minimum-fill).
 *
 * Revision 1.6  1998/03/06 13:54:03  adam
 * Fixed two nasty bugs in isc_merge.
 *
 * Revision 1.5  1997/02/12 20:42:43  adam
 * Bug fix: during isc_merge operations, some pages weren't marked dirty
 * even though they should be. At this point the merge operation marks
 * a page dirty if the previous page changed at all. A better approach is
 * to mark it dirty if the last key written changed in previous page.
 *
 * Revision 1.4  1996/11/08 11:15:31  adam
 * Number of keys in chain are stored in first block and the function
 * to retrieve this information, isc_pp_num is implemented.
 *
 * Revision 1.3  1996/11/04 14:08:59  adam
 * Optimized free block usage.
 *
 * Revision 1.2  1996/11/01 13:36:46  adam
 * New element, max_blocks_mem, that control how many blocks of max size
 * to store in memory during isc_merge.
 * Function isc_merge now ignores delete/update of identical keys and
 * the proper blocks are then non-dirty and not written in flush_blocks.
 *
 * Revision 1.1  1996/11/01  08:59:15  adam
 * First version of isc_merge that supports update/delete.
 *
 */