int main(int argc,char *argv[])
{
int error=0, subkeys;
uint keylen, keylen2=0, inx, doc_cnt=0;
float weight= 1.0;
double gws, min_gws=0, avg_gws=0;
MI_INFO *info;
char buf[MAX_LEN], buf2[MAX_LEN], buf_maxlen[MAX_LEN], buf_min_gws[MAX_LEN];
ulong total=0, maxlen=0, uniq=0, max_doc_cnt=0;
struct { MI_INFO *info; } aio0, *aio=&aio0; /* for GWS_IN_USE */
MY_INIT(argv[0]);
memset(&main_thread_keycache_var, 0, sizeof(st_keycache_thread_var));
mysql_cond_init(PSI_NOT_INSTRUMENTED,
&main_thread_keycache_var.suspend);
if ((error= handle_options(&argc, &argv, my_long_options, get_one_option)))
exit(error);
if (count || dump)
verbose=0;
if (!count && !dump && !lstats && !query)
stats=1;
if (verbose)
setbuf(stdout,NULL);
if (argc < 2)
usage();
{
char *end;
inx= (uint) my_strtoll(argv[1], &end, 10);
if (*end)
usage();
}
init_key_cache(dflt_key_cache,MI_KEY_BLOCK_LENGTH,USE_BUFFER_INIT, 0, 0);
if (!(info=mi_open(argv[0], O_RDONLY,
HA_OPEN_ABORT_IF_LOCKED|HA_OPEN_FROM_SQL_LAYER)))
{
error=my_errno();
goto err;
}
*buf2=0;
aio->info=info;
if ((inx >= info->s->base.keys) ||
!(info->s->keyinfo[inx].flag & HA_FULLTEXT))
{
printf("Key %d in table %s is not a FULLTEXT key\n", inx, info->filename);
goto err;
}
mi_lock_database(info, F_EXTRA_LCK);
info->lastpos= HA_OFFSET_ERROR;
info->update|= HA_STATE_PREV_FOUND;
while (!(error=mi_rnext(info,NULL,inx)))
{
keylen=*(info->lastkey);
subkeys=ft_sintXkorr(info->lastkey+keylen+1);
if (subkeys >= 0)
ft_floatXget(weight, info->lastkey+keylen+1);
my_snprintf(buf,MAX_LEN,"%.*s",(int) keylen,info->lastkey+1);
my_casedn_str(default_charset_info,buf);
total++;
lengths[keylen]++;
if (count || stats)
{
if (strcmp(buf, buf2))
{
if (*buf2)
{
uniq++;
avg_gws+=gws=GWS_IN_USE;
if (count)
printf("%9u %20.7f %s\n",doc_cnt,gws,buf2);
if (maxlen<keylen2)
{
maxlen=keylen2;
my_stpcpy(buf_maxlen, buf2);
}
if (max_doc_cnt < doc_cnt)
{
max_doc_cnt=doc_cnt;
my_stpcpy(buf_min_gws, buf2);
min_gws=gws;
}
}
my_stpcpy(buf2, buf);
keylen2=keylen;
doc_cnt=0;
}
doc_cnt+= (subkeys >= 0 ? 1 : -subkeys);
}
if (dump)
{
if (subkeys>=0)
printf("%9lx %20.7f %s\n", (long) info->lastpos,weight,buf);
else
printf("%9lx => %17d %s\n",(long) info->lastpos,-subkeys,buf);
}
if (verbose && (total%HOW_OFTEN_TO_WRITE)==0)
printf("%10ld\r",total);
}
mi_lock_database(info, F_UNLCK);
if (count || stats)
{
if (*buf2)
{
uniq++;
avg_gws+=gws=GWS_IN_USE;
if (count)
printf("%9u %20.7f %s\n",doc_cnt,gws,buf2);
if (maxlen<keylen2)
{
maxlen=keylen2;
my_stpcpy(buf_maxlen, buf2);
}
if (max_doc_cnt < doc_cnt)
{
max_doc_cnt=doc_cnt;
my_stpcpy(buf_min_gws, buf2);
min_gws=gws;
}
}
}
if (stats)
{
count=0;
for (inx=0;inx<256;inx++)
{
count+=lengths[inx];
if ((ulong) count >= total/2)
break;
}
printf("Total rows: %lu\nTotal words: %lu\n"
"Unique words: %lu\nLongest word: %lu chars (%s)\n"
"Median length: %u\n"
"Average global weight: %f\n"
"Most common word: %lu times, weight: %f (%s)\n",
(long) info->state->records, total, uniq, maxlen, buf_maxlen,
inx, avg_gws/uniq, max_doc_cnt, min_gws, buf_min_gws);
}
if (lstats)
{
count=0;
for (inx=0; inx<256; inx++)
{
count+=lengths[inx];
if (count && lengths[inx])
printf("%3u: %10lu %5.2f%% %20lu %4.1f%%\n", inx,
(ulong) lengths[inx],100.0*lengths[inx]/total,(ulong) count,
100.0*count/total);
}
}
err:
if (error && error != HA_ERR_END_OF_FILE)
printf("got error %d\n",my_errno());
if (info)
mi_close(info);
mysql_cond_destroy(&main_thread_keycache_var.suspend);
return 0;
}
int main(int argc,char *argv[])
{
int error=0;
uint keylen, keylen2=0, inx, doc_cnt=0;
float weight= 1.0;
double gws, min_gws=0, avg_gws=0;
MARIA_HA *info;
char buf[MAX_LEN], buf2[MAX_LEN], buf_maxlen[MAX_LEN], buf_min_gws[MAX_LEN];
ulong total=0, maxlen=0, uniq=0, max_doc_cnt=0;
struct { MARIA_HA *info; } aio0, *aio=&aio0; /* for GWS_IN_USE */
MY_INIT(argv[0]);
if ((error= handle_options(&argc, &argv, my_long_options, get_one_option)))
exit(error);
maria_init();
if (count || dump)
verbose=0;
if (!count && !dump && !lstats && !query)
stats=1;
if (verbose)
setbuf(stdout,NULL);
if (argc < 2)
usage();
{
char *end;
inx= (uint) strtoll(argv[1], &end, 10);
if (*end)
usage();
}
init_pagecache(maria_pagecache, PAGE_BUFFER_INIT, 0, 0,
MARIA_KEY_BLOCK_LENGTH, 0, MY_WME);
if (!(info=maria_open(argv[0], O_RDONLY,
HA_OPEN_ABORT_IF_LOCKED|HA_OPEN_FROM_SQL_LAYER)))
{
error=my_errno;
goto err;
}
*buf2=0;
aio->info=info;
if ((inx >= info->s->base.keys) ||
!(info->s->keyinfo[inx].flag & HA_FULLTEXT))
{
printf("Key %d in table %s is not a FULLTEXT key\n", inx,
info->s->open_file_name.str);
goto err;
}
maria_lock_database(info, F_EXTRA_LCK);
info->cur_row.lastpos= HA_OFFSET_ERROR;
info->update|= HA_STATE_PREV_FOUND;
while (!(error=maria_rnext(info,NULL,inx)))
{
FT_WEIGTH subkeys;
keylen=*(info->lastkey_buff);
subkeys.i= ft_sintXkorr(info->lastkey_buff + keylen + 1);
if (subkeys.i >= 0)
weight= subkeys.f;
#ifdef HAVE_SNPRINTF
snprintf(buf,MAX_LEN,"%.*s",(int) keylen,info->lastkey_buff+1);
#else
sprintf(buf,"%.*s",(int) keylen,info->lastkey_buff+1);
#endif
my_casedn_str(default_charset_info,buf);
total++;
lengths[keylen]++;
if (count || stats)
{
if (strcmp(buf, buf2))
{
if (*buf2)
{
uniq++;
avg_gws+=gws=GWS_IN_USE;
if (count)
printf("%9u %20.7f %s\n",doc_cnt,gws,buf2);
if (maxlen<keylen2)
{
maxlen=keylen2;
strmov(buf_maxlen, buf2);
}
if (max_doc_cnt < doc_cnt)
{
max_doc_cnt=doc_cnt;
strmov(buf_min_gws, buf2);
min_gws=gws;
}
}
strmov(buf2, buf);
keylen2=keylen;
doc_cnt=0;
}
doc_cnt+= (subkeys.i >= 0 ? 1 : -subkeys.i);
}
if (dump)
{
if (subkeys.i >= 0)
printf("%9lx %20.7f %s\n", (long) info->cur_row.lastpos,weight,buf);
else
printf("%9lx => %17d %s\n",(long) info->cur_row.lastpos,-subkeys.i,
buf);
}
if (verbose && (total%HOW_OFTEN_TO_WRITE)==0)
printf("%10ld\r",total);
}
maria_lock_database(info, F_UNLCK);
if (count || stats)
{
if (*buf2)
{
uniq++;
avg_gws+=gws=GWS_IN_USE;
if (count)
printf("%9u %20.7f %s\n",doc_cnt,gws,buf2);
if (maxlen<keylen2)
{
maxlen=keylen2;
strmov(buf_maxlen, buf2);
}
if (max_doc_cnt < doc_cnt)
{
max_doc_cnt=doc_cnt;
strmov(buf_min_gws, buf2);
min_gws=gws;
}
}
}
if (stats)
{
count=0;
for (inx=0;inx<256;inx++)
{
count+=lengths[inx];
if ((ulong) count >= total/2)
break;
}
printf("Total rows: %lu\nTotal words: %lu\n"
"Unique words: %lu\nLongest word: %lu chars (%s)\n"
"Median length: %u\n"
"Average global weight: %f\n"
"Most common word: %lu times, weight: %f (%s)\n",
(long) info->state->records, total, uniq, maxlen, buf_maxlen,
inx, avg_gws/uniq, max_doc_cnt, min_gws, buf_min_gws);
}
if (lstats)
{
count=0;
for (inx=0; inx<256; inx++)
{
count+=lengths[inx];
if (count && lengths[inx])
printf("%3u: %10lu %5.2f%% %20lu %4.1f%%\n", inx,
(ulong) lengths[inx],100.0*lengths[inx]/total,(ulong) count,
100.0*count/total);
}
}
err:
if (error && error != HA_ERR_END_OF_FILE)
printf("got error %d\n",my_errno);
if (info)
maria_close(info);
maria_end();
return 0;
}
