int state_load(memfile_t *file)
{
stateheader_t header;
block_t *block;
u32 size = 0;
int i;
readvar(header.ident,4);
readvar(header.version,2);
readvar(header.flags,2);
readvar(header.usize,4);
readvar(header.csize,4);
readvar(header.crc32,4);
log_printf("state_load: state header loaded. version %04X\n",header.version);
while(memfile_eof(file) == 0 && size < header.usize) {
if((block = block_load(file)) == 0)
break;
size += 8 + block->size;
log_printf("state_load: loaded block '%4s' (%08X) (%d bytes)\n",&block->type,block->type,block->size);
for(i=0;blockinfo[i].type;i++) {
if(blockinfo[i].type == block->type) {
blockinfo[i].func(STATE_LOAD,block->data);
break;
}
}
if(blockinfo[i].type == 0) {
log_printf("state_load: no handler for block type '%4s' (%d bytes)\n",&block->type,block->size);
}
block_destroy(block);
}
return(0);
}
entry_t *hfile_loup(hfile_t *fp, sds key)
{
int32_t fd;
meta_t *m;
entry_t *t;
sds firstkey;
if (!fp || !key)
return(NULL);
if (-1 == bloom_check(fp->bloom,key))
return(NULL);
firstkey = (index_head(fp->metas))->key;
if (sdscmp(key,firstkey) < 0)
return(NULL);
if (sdscmp(key,fp->fileinfo->lastkey) > 0)
return(NULL);
m = index_loup(fp->metas,key);
if (m->state != IN_CACHE) {
if ((fd = open(fp->name,O_RDONLY)) < 0)
/* error handle */
return(NULL);
fp->blocks[m->id] = block_load(fd,m->offset,m->blocksize);
m->state = IN_CACHE;
close(fd);
}
t = block_loup(fp->blocks[m->id],key);
return(t);
}
int block_load_article(wp_dump *d, char *name, int block, wp_article *a) {
char *text = xalloc(BZ_MAX_BLOCK);
int text_len = BZ_MAX_BLOCK, article_len;
char *start;
debug("opening %s from block %d (%llu)", name, block, (d->block_map)[block]);
block_load(d, block, text, &text_len);
if((article_len = article_bounds(name, text, &start, text_len)) < 0) {
if(article_len == BOUNDS_HIT_END) {
/* the start was in the block, but we hit the end of the block before finding
* the end of the article text. Load the next block and re-enter...
* We only consider the case of loading one extra block; no Wikipedia article will span
* more than two (it'd have to be >900kb for it to be possible).
*/
if(!d->block_map[block + 1])
fatal("trying to load another block, but don't have any blocks left. "
"This probably indicates a malformed dump.");
debug("Loading another block");
if(!(text = realloc(text, BZ_MAX_BLOCK * 2))) fatal("realloc");
int n_text_len = BZ_MAX_BLOCK;
block_load(d, block + 1, text + text_len, &n_text_len);
if((article_len = article_bounds(name, text, &start, n_text_len + text_len)) < 0)
fatal("additional block loading failure: %d", article_len);
} else
fatal("couldn't find %s in block %d", name, block);
}
debug("start: 0x%x, len: %d", start, article_len);
debug("a->text: 0x%x", a->text);
*(start + article_len) = '\0';
strncpy(a->text, start, article_len + 1);
a->block = block;
free(text);
return article_len;
}
