int main(int argc, char **argv) {
int32_t ch, index, excl_index;
const char *name;
bool load_default, exclusive;
load_default = true;
exclusive = false;
index = 0;
while (1) {
if ((ch = getopt_long_only(argc, argv, "", opt_field, &index)) < 0)
break;
switch (ch) {
case OPT_LOG:
log_load(optarg);
break;
case OPT_CONFIG:
config_load(optarg);
load_default = false;
break;
case OPT_EXCL:
if (exclusive)
error("Cannot use --%s with --%s",
opt_field[index].name,
opt_field[excl_index].name);
exclusive = true;
excl_index = index;
case OPT_NRML:
name = opt_field[index].name;
if (opt_field[index].has_arg)
config_set(name, optarg);
else
config_set(name, "true");
break;
case OPT_VERSION:
version();
return 1;
case OPT_HELP:
default:
usage();
return 1;
}
}
if (load_default)
config_default();
mt_init();
store_load();
bucket_load();
crypt_load();
volume_load();
return 0;
}
int merge(struct tmpindex *tmpidx,struct ehash *idx)
{
struct posting *p;
unsigned int len;
char *invertedlist;
char *merged;
int mergedlen;
struct bucket bkt;
struct page *page;
p = tmpidx->list;
while(p)
{
invertedlist = ehash_find(idx,p->c,&len);
if(invertedlist!=NULL && len==USHRT_MAX) /* invertedlist is bigger than a page! */
{
page = cache_pagein(idx->cache,*((int*)invertedlist),*((off_t*)(invertedlist+sizeof(int))));
if(page == NULL)
return -1;
bucket_load(&bkt,page);
if(merge_invertedlist_page(p,&bkt,idx->cache,idx->freemap) != 0)
return -1;
}
else
{
merged = merge_invertedlist(p,invertedlist,len,&mergedlen);
if(merged == NULL)
return -1;
if(ehash_insert(idx,p->c,merged,mergedlen) != 0)
return -1;
free(merged); /* merge_invertedlist's return value is malloced!!! */
}
merged = NULL;
tmpidx->list = p->next;
posting_free(p);
tmpidx->count--;
p = tmpidx->list;
}
tmpindex_clear(tmpidx);
return 0;
}
static int merge_invertedlist_page(struct posting *p,struct bucket *bkt, struct cache *cache,struct freemap *freemap)
{
unsigned int docnum;
unsigned int lastdoc;
unsigned int firstdoc;
int len_firstdoc;
int fd;
off_t offset;
unsigned int size;
unsigned short invertedlen;
unsigned int add_size;
struct vector *v;
struct bucket tmp;
struct page *page;
int page_data_size;
int available_size;
char *ptr;
assert(p);
v = p->invertedlist;
len_firstdoc = vbyte_decompress(v->vector,v->vector+v->len,&firstdoc);
if(len_firstdoc <= 0)
goto exit;
docnum = *((unsigned int*)bkt->record);
lastdoc = *((unsigned int*)((char*)bkt->record+sizeof(unsigned int)));
assert(firstdoc > lastdoc);
page_data_size = PAGE_SIZE - sizeof(struct HEAD);
invertedlen = bkt->head->num; /* special use for such kind of page,bkt->head->num
no longer store the number of records, it stores
invertedlen instead. */
add_size = v->len - len_firstdoc;
bkt->head->num += (v->len-len_firstdoc+vbyte_len(firstdoc-lastdoc));
*((unsigned int*)bkt->record) += p->count;
((unsigned int*)bkt->record)[1] = p->lastdoc;
/* switch to the last page */
tmp = *bkt;
while(tmp.head->next.fd != 0)
{
fd = tmp.head->next.fd;
offset = tmp.head->next.offset;
page = cache_pagein(cache,fd,offset);
if(page == NULL)
goto exit;
bucket_load(&tmp,page);
}
/* add invertedlist in the page */
if((char*)tmp.record+tmp.head->ptr + vbyte_len(firstdoc-lastdoc) < (char*)tmp.head + PAGE_SIZE)
tmp.head->ptr += vbyte_compress((char*)tmp.record+tmp.head->ptr,(char*)tmp.head+PAGE_SIZE,firstdoc-lastdoc);
else
{
size = PAGE_SIZE;
if(freemap_malloc(freemap,&size,&fd,&offset) != 0)
goto exit;
page = cache_newpage(cache,fd,offset);
if(page == NULL)
goto exit;
bucket_init(&tmp,page,0);
tmp.head->next.fd = fd;
tmp.head->next.offset = offset;
tmp.head->ptr = 0;
tmp.head->ptr += vbyte_compress((char*)tmp.record+tmp.head->ptr,(char*)tmp.head+PAGE_SIZE,firstdoc-lastdoc);
}
ptr = v->vector+len_firstdoc;
while(add_size > 0)
{
if((char*)tmp.record + tmp.head->ptr + add_size < (char*)tmp.head + PAGE_SIZE)
{
memcpy((char*)tmp.record+tmp.head->ptr,ptr,add_size);
tmp.head->ptr += add_size;
add_size = 0;
}
else
{
available_size = PAGE_SIZE-sizeof(struct HEAD)-tmp.head->ptr;
memcpy((char*)tmp.record+tmp.head->ptr,ptr,available_size);
ptr += available_size;
tmp.head->ptr += available_size;
add_size -= available_size;
size = PAGE_SIZE;
if(freemap_malloc(freemap,&size,&fd,&offset) != 0)
goto exit;
page = cache_newpage(cache,fd,offset);
if(page == NULL)
goto exit;
tmp.head->next.fd = fd;
tmp.head->next.offset = offset;
bucket_init(&tmp,page,0);
tmp.head->ptr = 0;
}
}
return 0;
exit:
return -1;
}
