unsigned long hash(char * str)
{
unsigned long hash = HASH_INIT;
unsigned int i;
for(i = 0; str[i] && (i<MAXLEN); i++)
{
HASH_STEP(hash,str[i]);
}
return hash;
}
uint64_t hash(char * str)
{
uint64_t hash = HASH_INIT;
unsigned int i;
for(i = 0; str[i] ; i++)
{
HASH_STEP(hash,str[i]);
}
return hash;
}
int main(int argc, char ** argv)
{
FILE * dictionnary;
FILE * pwds;
char line[LINELEN];
unsigned int i,j;
avl_tree_t * sroot;
struct rlimit rlim;
BLOOM_TYPE * bloom;
BLOOM_TYPE * bloom_cityhash;
unsigned long curhashes[MAXLEN];
unsigned int curstart[MAXLEN];
unsigned int maxlen;
unsigned int maxstart;
unsigned char ln2[LINELEN];
unsigned int len;
unsigned int minmatch;
rlim.rlim_cur = MAXMEM;
rlim.rlim_max = MAXMEM;
if(setrlimit(RLIMIT_AS, &rlim))
{
perror("setrlimit");
return 3;
}
if( (argc != 4) && (argc != 5) )
usage();
nblines = 0;
if(argc == 5)
{
minmatch = atoi(argv[4]);
if(minmatch == 0)
{
fprintf(stderr, "can't parse %s as int\n", argv[4]);
return 6;
}
}
else
minmatch = 1;
/* arbre contenant le dictionnaire pour éviter les FP */
sroot = avl_alloc_tree((avl_compare_t)strcmp, (avl_freeitem_t)free);
/* bloom table */
bloom = xmalloc(BLOOM_STORAGE);
memset(bloom, 0, BLOOM_STORAGE);
bloom_cityhash = xmalloc(BLOOM_STORAGE);
memset(bloom_cityhash, 0, BLOOM_STORAGE);
fprintf(stderr, "%lld bytes have been allocated for bloom filters\n", BLOOM_STORAGE*2);
if(strcmp(argv[1],"-")==0)
dictionnary = stdin;
else
dictionnary = fopen(argv[1], "r");
if(dictionnary == NULL)
{
perror(argv[1]);
return 2;
}
pwds = fopen(argv[2], "r");
if(pwds == NULL)
{
perror(argv[2]);
return 3;
}
while(fgets(line, LINELEN-1, dictionnary))
{
len = strlen(line);
if(len>MAXLEN)
continue;
if(len<minmatch)
continue;
nblines++;
if(nblines % 1000000 == 0)
fprintf(stderr, "%ldM dictionnary lines integrated\n", nblines/1000000);
line[len-1]=0; // trim !
SETBIT(bloom, hash(line));
SETBIT(bloom_cityhash, cityhash(line));
avl_insert(sroot, strdup(line));
}
printf("%ld lines\n", nblines);
fclose(dictionnary);
nbpass = 0;
nbfp = 0;
nbmatch = 0;
while(fgets((char*) line, LINELEN-1, pwds))
{
nbpass++;
if(nbpass % 500000 == 0)
fprintf(stderr, "%.1fM passwords analyzed [%s]\n", ((float)nbpass)/1000000.0, argv[3]);
maxlen = 0;
maxstart = 0;
memset(curstart, 0, sizeof(curstart));
for(i=0;i<MAXLEN;i++)
curhashes[i]=HASH_INIT;
for(i=0;line[i] && line[i]!='\n' && (i<MAXLEN*2);i++)
{
curhashes[i%MAXLEN]=HASH_INIT;
curstart[i%MAXLEN] = i;
for(j=0;j<MAXLEN;j++)
{
HASH_STEP(curhashes[j],line[i]);
}
for(j=0;j<MAXLEN;j++)
{
/* si on ne bat pas le maxlen courant ça ne sert à rien */
if( i-curstart[j]+1 < maxlen )
continue;
/* check bloom filter */
if(GETBIT(bloom, curhashes[j]))
{
memset(ln2, 0, sizeof(ln2));
memcpy(ln2, line+curstart[j], i-curstart[j]+1);
if(GETBIT(bloom_cityhash, cityhash((char*)ln2)))
{
if(avl_search(sroot, ln2)) /* evitons les FP */
{
maxlen = i-curstart[j]+1;
maxstart = curstart[j];
}
else
{
nbfp++;
}
}
}
}
}
if(maxlen<minmatch)
continue;
line[strlen((char*)line)-1] = 0;
printf("%s\t%s\t", argv[3], line);
if(maxstart>0)
{
memset(ln2, 0, sizeof(ln2));
strncpy((char*)ln2, (char*)line, maxstart);
printf("%s\t", ln2);
}
else
printf("\t");
if(maxstart+maxlen<strlen((char*)line))
printf("%s\t", line + maxstart + maxlen);
else
printf("\t");
line[maxstart+maxlen] = 0;
printf("%s\t%d\n", line+maxstart, nbpass);
nbmatch++;
}
fprintf(stderr, "%d passwords analyzed, %d matches, %d false positives\n", nbpass, nbmatch, nbfp);
return 0;
}