LIB_EXPORT uint32_t CC AgrepFindBest( const AgrepParams *self, int32_t threshold, const char *buf, int32_t len, AgrepMatch *match )
{
if( self != NULL && buf != NULL && match != NULL ) {
AgrepCallArgs args;
args.self = self;
args.threshold = threshold;
args.buf = buf;
args.buflen = len;
args.cb = AgrepFindBestCallback;
args.cbinfo = match;
match->score = -1;
AgrepFindAll(&args);
if (match->score != -1) {
return 1;
}
}
return 0;
}
LIB_EXPORT int32_t CC agrep ( AgrepFlags alg, char const* pattern, FILE* fp, int32_t print,
int32_t printcontext, int32_t k, int32_t findall, int32_t simulate )
{
AgrepParams *self = NULL;
int32_t count;
size_t len;
AgrepFlags mode = alg;
AgrepMatch match;
agrepinfo info;
char szBuf[BUF_SIZE];
AgrepCallArgs args;
int32_t linenum = 0;
char const* na = "ACGTN";
size_t const nPatternSize = strlen(pattern);
size_t const nBufSize = BUF_SIZE;
size_t const nMaxTextLen = nBufSize-nPatternSize-1;
size_t nExtendBufSize;
int32_t i;
AgrepMake(&self, mode, pattern);
count = 0;
while (!feof(fp) && fgets(szBuf, nMaxTextLen, fp))
{
++linenum;
len = strlen(szBuf);
if (szBuf[len-1] == '\n')
{
szBuf[len-1] = '\0';
--len;
}
/* Make szBuf somewhat circular to be able to find pattern cut in between */
nExtendBufSize = nPatternSize <= len ? nPatternSize : len;
memmove(szBuf+len, szBuf, nExtendBufSize - 1);
szBuf[len + nExtendBufSize - 1] = '\0';
if (simulate)
{
for (i = 0; szBuf[i]; ++i)
{
if (NULL != index(na, szBuf[i]))
{
szBuf[i] = index(na, szBuf[i]) - na;
}
}
}
if (findall)
{
AgrepBestMatch best_match = {9999, 0, 0}; /* TODO: it's not C89*/
info.buf = szBuf;
info.linenum = linenum;
info.original_length = len;
info.pBestMatch = &best_match;
args.self = self;
args.threshold = k;
args.buf = szBuf;
args.buflen = len + nPatternSize;
args.cb = simulate ? agrep_2na_callback : agrep_callback_find_best;
args.cbinfo = (void *)&info;
AgrepFindAll(&args);
if (!simulate) /* output best match */
{
printf("line=%d, pos=%d, len=%d, end=%d, score=%d: %.*s%s%.*s\n",
linenum, best_match.indexStart, best_match.nMatchLength,
(best_match.indexStart + best_match.nMatchLength) % (int32_t)len,
best_match.nScore,
(int)(min(best_match.nMatchLength, (int32_t)len - best_match.indexStart)), szBuf + best_match.indexStart,
(best_match.indexStart + best_match.nMatchLength) > (int32_t)len ? "|" : "",
(int)((best_match.indexStart + best_match.nMatchLength) > (int32_t)len ? best_match.nMatchLength - ((int32_t)len - best_match.indexStart) : 0), szBuf);
}
}
else if (AgrepFindFirst(self, k, szBuf, len + nPatternSize, &match))
{
++count;
if (print)
{
printf("%s\n", szBuf);
}
if (printcontext)
{
printf("%.*s\n", match.length, szBuf + match.position);
}
}
}
return count;
}
