// the caller is responsible for delete[]'ing this
boost::uint8_t *copyBytes(const FPBReader_impl *dp_impl, unsigned int which) {
PRECONDITION(dp_impl, "bad reader pointer");
boost::uint8_t *res;
res = new boost::uint8_t[dp_impl->numBytesStoredPerFingerprint];
if (!dp_impl->df_lazy) {
boost::uint8_t *fpData = NULL;
extractBytes(dp_impl, which, fpData);
memcpy(static_cast<void *>(res), fpData,
dp_impl->numBytesStoredPerFingerprint);
} else {
extractBytes(dp_impl, which, res);
}
return res;
};
/*
* ShockBurst packet format (length in bytes in parenthesis):
*
* +--------------+---------------+----------------------------+-----------+
* | Preamble (1) | Address (3-5) | Payload (1-32) | CRC (1-2) |
* +--------------+---------------+----------------------------+-----------+
*
* CRC is calculated over the "Address" and "Payload" fields, and it is 2
* bytes in case of ANT.
*/
bool decodePacket(int32_t sample)
{
int i;
_threshold = extractThreshold();
if (detectPreamble()) {
uint8_t tmp_buf[ADDRESS_LENGTH + PAYLOAD_LENGTH + 2];
extractBytes(8, tmp_buf, sizeof(tmp_buf));
if (isShockBurstPacket(tmp_buf)) {
packetData.clear();
// address
uint64_t address = 0;
for (int i = 0; i < ADDRESS_LENGTH; ++i)
address |= tmp_buf[i] << (ADDRESS_LENGTH - 1 - i) * 8;
packetData["address"] = Pothos::Object(address);
// CRC
uint16_t crc = tmp_buf[ADDRESS_LENGTH + PAYLOAD_LENGTH] << 8 |
tmp_buf[ADDRESS_LENGTH + PAYLOAD_LENGTH + 1];
packetData["crc"] = Pothos::Object(crc);
// paylod
packetData["payload"] = Pothos::Object(std::vector<uint8_t>(
tmp_buf + ADDRESS_LENGTH,
tmp_buf + ADDRESS_LENGTH + PAYLOAD_LENGTH));
return true;
}
}
return false;
}
void containingNeighbors(const FPBReader_impl *dp_impl,
const boost::uint8_t *bv,
std::vector<unsigned int> &res) {
PRECONDITION(dp_impl, "bad reader pointer");
PRECONDITION(bv, "bad bv");
res.clear();
boost::uint64_t probeCount =
CalcBitmapPopcount(bv, dp_impl->numBytesStoredPerFingerprint);
boost::uint64_t startScan = 0, endScan = dp_impl->len;
if (dp_impl->popCountOffsets.size() == dp_impl->nBits + 2) {
startScan = dp_impl->popCountOffsets[probeCount];
// std::cerr << " scan: " << startScan << "-" << endScan << std::endl;
}
boost::uint8_t *dbv;
if (dp_impl->df_lazy) {
dbv = new boost::uint8_t[dp_impl->numBytesStoredPerFingerprint];
}
for (boost::uint64_t i = startScan; i < endScan; ++i) {
extractBytes(dp_impl, i, dbv);
if (CalcBitmapAllProbeBitsMatch(bv, dbv,
dp_impl->numBytesStoredPerFingerprint)) {
res.push_back(i);
}
}
if (dp_impl->df_lazy) delete[] dbv;
}
// the caller is responsible for delete'ing this
ExplicitBitVect *extractFP(const FPBReader_impl *dp_impl, unsigned int which) {
PRECONDITION(dp_impl, "bad reader pointer");
boost::uint8_t *fpData;
if (dp_impl->df_lazy) {
fpData = new boost::uint8_t[dp_impl->numBytesStoredPerFingerprint];
}
extractBytes(dp_impl, which, fpData);
boost::dynamic_bitset<> *resDBS = bytesToBitset(fpData, dp_impl->nBits);
if (dp_impl->df_lazy) delete[] fpData;
return new ExplicitBitVect(resDBS);
};
void tanimotoNeighbors(const FPBReader_impl *dp_impl, const boost::uint8_t *bv,
double threshold,
std::vector<std::pair<double, unsigned int> > &res,
bool usePopcountScreen, unsigned int readCache = 1000) {
PRECONDITION(dp_impl, "bad reader pointer");
PRECONDITION(bv, "bad bv");
RANGE_CHECK(-1e-6, threshold, 1.0 + 1e-6);
PRECONDITION(readCache > 0, "bad cache size");
res.clear();
boost::uint64_t probeCount =
CalcBitmapPopcount(bv, dp_impl->numBytesStoredPerFingerprint);
boost::uint64_t startScan = 0, endScan = dp_impl->len;
if (usePopcountScreen &&
dp_impl->popCountOffsets.size() == dp_impl->nBits + 2) {
// figure out the bounds based on equation 24 from:
// 1. Swamidass, S. J. & Baldi, P. Bounds and Algorithms for Fast Exact
// Searches of Chemical Fingerprints in Linear and Sublinear Time. J. Chem.
// Inf. Model. 47, 302–317 (2007).
// http://pubs.acs.org/doi/abs/10.1021/ci600358f
boost::uint32_t minDbCount =
static_cast<boost::uint32_t>(floor(threshold * probeCount));
boost::uint32_t maxDbCount =
(threshold > 1e-6)
? static_cast<boost::uint32_t>(ceil(probeCount / threshold))
: dp_impl->numBytesStoredPerFingerprint;
// std::cerr << "probeCount: " << probeCount << " bounds: " << minDbCount
// << "-" << maxDbCount << std::endl;
startScan = dp_impl->popCountOffsets[minDbCount];
endScan = dp_impl->popCountOffsets[maxDbCount + 1];
// std::cerr << " scan: " << startScan << "-" << endScan << std::endl;
}
boost::uint8_t *dbv;
if (dp_impl->df_lazy) {
dbv = new boost::uint8_t[dp_impl->numBytesStoredPerFingerprint * readCache];
}
for (boost::uint64_t i = startScan; i < endScan; i += readCache) {
unsigned int toRead = readCache;
if (i + toRead >= endScan) {
toRead = endScan - i;
}
extractBytes(dp_impl, i, dbv, toRead);
for (unsigned int j = 0; j < toRead; ++j) {
double tani =
CalcBitmapTanimoto(dbv + j * dp_impl->numBytesStoredPerFingerprint,
bv, dp_impl->numBytesStoredPerFingerprint);
if (tani >= threshold) {
res.push_back(std::make_pair(tani, i + j));
}
}
}
if (dp_impl->df_lazy) delete[] dbv;
}
void tverskyNeighbors(const FPBReader_impl *dp_impl, const boost::uint8_t *bv,
double ca, double cb, double threshold,
std::vector<std::pair<double, unsigned int> > &res,
bool usePopcountScreen) {
PRECONDITION(dp_impl, "bad reader pointer");
PRECONDITION(bv, "bad bv");
RANGE_CHECK(-1e-6, threshold, 1.0 + 1e-6);
res.clear();
boost::uint64_t probeCount =
CalcBitmapPopcount(bv, dp_impl->numBytesStoredPerFingerprint);
boost::uint64_t startScan = 0, endScan = dp_impl->len;
if (usePopcountScreen &&
dp_impl->popCountOffsets.size() == dp_impl->nBits + 2) {
// figure out the bounds based on equation 25 from:
// 1. Swamidass, S. J. & Baldi, P. Bounds and Algorithms for Fast Exact
// Searches of Chemical Fingerprints in Linear and Sublinear Time. J. Chem.
// Inf. Model. 47, 302–317 (2007).
// http://pubs.acs.org/doi/abs/10.1021/ci600358f
boost::uint32_t minDbCount = static_cast<boost::uint32_t>(floor(
(threshold * probeCount * ca) / (1. - threshold + threshold * ca)));
boost::uint32_t maxDbCount =
((threshold * cb) > 1e-6)
? static_cast<boost::uint32_t>(
ceil(probeCount * (1 - threshold + threshold * cb) /
(threshold * cb)))
: dp_impl->numBytesStoredPerFingerprint;
// std::cerr << "probeCount: " << probeCount << " bounds: " << minDbCount
// << "-" << maxDbCount << std::endl;
startScan = dp_impl->popCountOffsets[minDbCount];
endScan = dp_impl->popCountOffsets[maxDbCount + 1];
// std::cerr << " scan: " << startScan << "-" << endScan << std::endl;
}
boost::uint8_t *dbv;
if (dp_impl->df_lazy) {
dbv = new boost::uint8_t[dp_impl->numBytesStoredPerFingerprint];
}
for (boost::uint64_t i = startScan; i < endScan; ++i) {
extractBytes(dp_impl, i, dbv);
double sim = CalcBitmapTversky(
dbv, bv, dp_impl->numBytesStoredPerFingerprint, ca, cb);
// std::cerr << " i:" << i << " " << tani << " ? " << threshold <<
// std::endl;
if (sim >= threshold) {
res.push_back(std::make_pair(sim, i));
}
}
if (dp_impl->df_lazy) delete[] dbv;
}
double tversky(const FPBReader_impl *dp_impl, unsigned int which,
const ::boost::uint8_t *bv, double ca, double cb) {
PRECONDITION(dp_impl, "bad reader pointer");
PRECONDITION(bv, "bad bv pointer");
if (which >= dp_impl->len) {
throw ValueErrorException("bad index");
}
boost::uint8_t *fpData;
if (dp_impl->df_lazy) {
fpData = new boost::uint8_t[dp_impl->numBytesStoredPerFingerprint];
}
extractBytes(dp_impl, which, fpData);
double res = CalcBitmapTversky(fpData, bv,
dp_impl->numBytesStoredPerFingerprint, ca, cb);
if (dp_impl->df_lazy) delete[] fpData;
return res;
};
void header_view(FILE *fp) {
printf("\033[22;32mHeader information:\n\n");
extractBytes(fp, 0, 8, " \033[22;37m[\033[01;37m+\033[22;37m] Magic: \033[22;31m", false, false);
extractBytes(fp, 8, 4, " \033[22;37m[\033[01;37m+\033[22;37m] Checksum: \033[22;31m", false, false);
extractBytes(fp, 12, 20, " \033[22;37m[\033[01;37m+\033[22;37m] Signature: \033[22;31m", false, false);
extractBytes(fp, 32, 4, " \033[22;37m[\033[01;37m+\033[22;37m] File Size: \033[22;31m", true, true);
extractBytes(fp, 36, 4, " \033[22;37m[\033[01;37m+\033[22;37m] Header Size: \033[22;31m", true, true);
extractBytes(fp, 40, 4, " \033[22;37m[\033[01;37m+\033[22;37m] Endian Tag: \033[22;31m", false, false);
extractBytes(fp, 44, 4, " \033[22;37m[\033[01;37m+\033[22;37m] Link Size: \033[22;31m", true, true);
extractBytes(fp, 48, 4, " \033[22;37m[\033[01;37m+\033[22;37m] Link Offset: \033[22;31m", false, true);
extractBytes(fp, 52, 4, " \033[22;37m[\033[01;37m+\033[22;37m] Map Offset: \033[22;31m", false, true);
extractBytes(fp, 56, 4, " \033[22;37m[\033[01;37m+\033[22;37m] String Ids Size: \033[22;31m", true, true);
extractBytes(fp, 60, 4, " \033[22;37m[\033[01;37m+\033[22;37m] String Ids Offset: \033[22;31m", false, true);
extractBytes(fp, 64, 4, " \033[22;37m[\033[01;37m+\033[22;37m] Type Ids Size: \033[22;31m", true, true);
extractBytes(fp, 68, 4, " \033[22;37m[\033[01;37m+\033[22;37m] Type Ids Offset: \033[22;31m", false, true);
extractBytes(fp, 72, 4, " \033[22;37m[\033[01;37m+\033[22;37m] Proto Ids Size: \033[22;31m", true, true);
extractBytes(fp, 76, 4, " \033[22;37m[\033[01;37m+\033[22;37m] Proto Ids Offset: \033[22;31m", false, true);
extractBytes(fp, 80, 4, " \033[22;37m[\033[01;37m+\033[22;37m] Field Ids Size: \033[22;31m", true, true);
extractBytes(fp, 84, 4, " \033[22;37m[\033[01;37m+\033[22;37m] Field Ids Offset: \033[22;31m", false, true);
extractBytes(fp, 88, 4, " \033[22;37m[\033[01;37m+\033[22;37m] Method Ids Size: \033[22;31m", true, true);
extractBytes(fp, 92, 4, " \033[22;37m[\033[01;37m+\033[22;37m] Method Ids Offset: \033[22;31m", false, true);
extractBytes(fp, 96, 4, " \033[22;37m[\033[01;37m+\033[22;37m] Class Defs Size: \033[22;31m", true, true);
extractBytes(fp, 100, 4, " \033[22;37m[\033[01;37m+\033[22;37m] Class Defs Offset: \033[22;31m", false, true);
extractBytes(fp, 104, 4, " \033[22;37m[\033[01;37m+\033[22;37m] Data Size: \033[22;31m", true, true);
extractBytes(fp, 108, 4, " \033[22;37m[\033[01;37m+\033[22;37m] Data Offset: \033[22;31m", false, true);
printf("\n");
}