/*!
* l_dnaJoin()
*
* Input: dad (dest dma; add to this one)
* das (<optional> source dna; add from this one)
* istart (starting index in das)
* iend (ending index in das; use -1 to cat all)
* Return: 0 if OK, 1 on error
*
* Notes:
* (1) istart < 0 is taken to mean 'read from the start' (istart = 0)
* (2) iend < 0 means 'read to the end'
* (3) if das == NULL, this is a no-op
*/
l_int32
l_dnaJoin(L_DNA *dad,
L_DNA *das,
l_int32 istart,
l_int32 iend)
{
l_int32 n, i;
l_float64 val;
PROCNAME("l_dnaJoin");
if (!dad)
return ERROR_INT("dad not defined", procName, 1);
if (!das)
return 0;
if (istart < 0)
istart = 0;
n = l_dnaGetCount(das);
if (iend < 0 || iend >= n)
iend = n - 1;
if (istart > iend)
return ERROR_INT("istart > iend; nothing to add", procName, 1);
for (i = istart; i <= iend; i++) {
l_dnaGetDValue(das, i, &val);
l_dnaAddNumber(dad, val);
}
return 0;
}
/*!
* \brief l_dnaHashCreateFromDna()
*
* \param[in] da
* \return dahash if OK; 1 on error
*
* <pre>
* Notes:
* (1) The values stored in the %dahash are indices into %da;
* %dahash has no use without %da.
* </pre>
*/
L_DNAHASH *
l_dnaHashCreateFromDna(L_DNA *da)
{
l_int32 i, n;
l_uint32 nsize;
l_uint64 key;
l_float64 val;
L_DNAHASH *dahash;
PROCNAME("l_dnaHashCreateFromDna");
if (!da)
return (L_DNAHASH *)ERROR_PTR("da not defined", procName, NULL);
n = l_dnaGetCount(da);
findNextLargerPrime(n / 20, &nsize); /* buckets in hash table */
dahash = l_dnaHashCreate(nsize, 8);
for (i = 0; i < n; i++) {
l_dnaGetDValue(da, i, &val);
l_hashFloat64ToUint64(nsize, val, &key);
l_dnaHashAdd(dahash, key, (l_float64)i);
}
return dahash;
}
/*!
* \brief l_dnaRemoveDupsByAset()
*
* \param[in] das
* \return dad with duplicates removed, or NULL on error
*/
L_DNA *
l_dnaRemoveDupsByAset(L_DNA *das)
{
l_int32 i, n;
l_float64 val;
L_DNA *dad;
L_ASET *set;
RB_TYPE key;
PROCNAME("l_dnaRemoveDupsByAset");
if (!das)
return (L_DNA *)ERROR_PTR("das not defined", procName, NULL);
set = l_asetCreate(L_FLOAT_TYPE);
dad = l_dnaCreate(0);
n = l_dnaGetCount(das);
for (i = 0; i < n; i++) {
l_dnaGetDValue(das, i, &val);
key.ftype = val;
if (!l_asetFind(set, key)) {
l_dnaAddNumber(dad, val);
l_asetInsert(set, key);
}
}
l_asetDestroy(&set);
return dad;
}
/*!
* \brief l_dnaIntersectionByHash()
*
* \param[in] da1, da2
* \return dad intersection of the number arrays, or NULL on error
*
* <pre>
* Notes:
* (1) This uses the same method for building the intersection set
* as ptaIntersectionByHash() and sarrayIntersectionByHash().
* </pre>
*/
L_DNA *
l_dnaIntersectionByHash(L_DNA *da1,
L_DNA *da2)
{
l_int32 n1, n2, nsmall, nbuckets, i, index1, index2;
l_uint32 nsize2;
l_uint64 key;
l_float64 val;
L_DNAHASH *dahash1, *dahash2;
L_DNA *da_small, *da_big, *dad;
PROCNAME("l_dnaIntersectionByHash");
if (!da1)
return (L_DNA *)ERROR_PTR("da1 not defined", procName, NULL);
if (!da2)
return (L_DNA *)ERROR_PTR("da2 not defined", procName, NULL);
/* Put the elements of the biggest array into a dnahash */
n1 = l_dnaGetCount(da1);
n2 = l_dnaGetCount(da2);
da_small = (n1 < n2) ? da1 : da2; /* do not destroy da_small */
da_big = (n1 < n2) ? da2 : da1; /* do not destroy da_big */
dahash1 = l_dnaHashCreateFromDna(da_big);
/* Build up the intersection of numbers. Add to %dad
* if the number is in da_big (using dahash1) but hasn't
* yet been seen in the traversal of da_small (using dahash2). */
dad = l_dnaCreate(0);
nsmall = l_dnaGetCount(da_small);
findNextLargerPrime(nsmall / 20, &nsize2); /* buckets in hash table */
dahash2 = l_dnaHashCreate(nsize2, 0);
nbuckets = l_dnaHashGetCount(dahash2);
for (i = 0; i < nsmall; i++) {
l_dnaGetDValue(da_small, i, &val);
l_dnaFindValByHash(da_big, dahash1, val, &index1);
if (index1 >= 0) { /* found */
l_dnaFindValByHash(da_small, dahash2, val, &index2);
if (index2 == -1) { /* not found */
l_dnaAddNumber(dad, val);
l_hashFloat64ToUint64(nbuckets, val, &key);
l_dnaHashAdd(dahash2, key, (l_float64)i);
}
}
}
l_dnaHashDestroy(&dahash1);
l_dnaHashDestroy(&dahash2);
return dad;
}
/*!
* \brief l_dnaMakeHistoByHash()
*
* \param[in] das
* \param[out] pdahash hash map: val --> index
* \param[out] pdav array of values: index --> val
* \param[out] pdac histo array of counts: index --> count
* \return 0 if OK; 1 on error
*
* <pre>
* Notes:
* (1) Generates and returns a dna of occurrences (histogram),
* an aligned dna of values, and an associated hashmap.
* The hashmap takes %dav and a value, and points into the
* histogram in %dac.
* (2) The dna of values, %dav, is aligned with the histogram %dac,
* and is needed for fast lookup. It is a hash set, because
* the values are unique.
* (3) Lookup is simple:
* l_dnaFindValByHash(dav, dahash, val, &index);
* if (index >= 0)
* l_dnaGetIValue(dac, index, &icount);
* else
* icount = 0;
* </pre>
*/
l_ok
l_dnaMakeHistoByHash(L_DNA *das,
L_DNAHASH **pdahash,
L_DNA **pdav,
L_DNA **pdac)
{
l_int32 i, n, nitems, index, count;
l_uint32 nsize;
l_uint64 key;
l_float64 val;
L_DNA *dac, *dav;
L_DNAHASH *dahash;
PROCNAME("l_dnaMakeHistoByHash");
if (pdahash) *pdahash = NULL;
if (pdac) *pdac = NULL;
if (pdav) *pdav = NULL;
if (!pdahash || !pdac || !pdav)
return ERROR_INT("&dahash, &dac, &dav not all defined", procName, 1);
if (!das)
return ERROR_INT("das not defined", procName, 1);
if ((n = l_dnaGetCount(das)) == 0)
return ERROR_INT("no data in das", procName, 1);
findNextLargerPrime(n / 20, &nsize); /* buckets in hash table */
dahash = l_dnaHashCreate(nsize, 8);
dac = l_dnaCreate(n); /* histogram */
dav = l_dnaCreate(n); /* the values */
for (i = 0, nitems = 0; i < n; i++) {
l_dnaGetDValue(das, i, &val);
/* Is this value already stored in dav? */
l_dnaFindValByHash(dav, dahash, val, &index);
if (index >= 0) { /* found */
l_dnaGetIValue(dac, (l_float64)index, &count);
l_dnaSetValue(dac, (l_float64)index, count + 1);
} else { /* not found */
l_hashFloat64ToUint64(nsize, val, &key);
l_dnaHashAdd(dahash, key, (l_float64)nitems);
l_dnaAddNumber(dav, val);
l_dnaAddNumber(dac, 1);
nitems++;
}
}
*pdahash = dahash;
*pdac = dac;
*pdav = dav;
return 0;
}
/*!
* \brief l_dnaRemoveDupsByHash()
*
* \param[in] das
* \param[out] pdad hash set
* \param[out] pdahash [optional] dnahash used for lookup
* \return 0 if OK; 1 on error
*
* <pre>
* Notes:
* (1) Generates a dna with unique values.
* (2) The dnahash is built up with dad to assure uniqueness.
* It can be used to find if an element is in the set:
* l_dnaFindValByHash(dad, dahash, val, &index)
* </pre>
*/
l_ok
l_dnaRemoveDupsByHash(L_DNA *das,
L_DNA **pdad,
L_DNAHASH **pdahash)
{
l_int32 i, n, index, items;
l_uint32 nsize;
l_uint64 key;
l_float64 val;
L_DNA *dad;
L_DNAHASH *dahash;
PROCNAME("l_dnaRemoveDupsByHash");
if (pdahash) *pdahash = NULL;
if (!pdad)
return ERROR_INT("&dad not defined", procName, 1);
*pdad = NULL;
if (!das)
return ERROR_INT("das not defined", procName, 1);
n = l_dnaGetCount(das);
findNextLargerPrime(n / 20, &nsize); /* buckets in hash table */
dahash = l_dnaHashCreate(nsize, 8);
dad = l_dnaCreate(n);
*pdad = dad;
for (i = 0, items = 0; i < n; i++) {
l_dnaGetDValue(das, i, &val);
l_dnaFindValByHash(dad, dahash, val, &index);
if (index < 0) { /* not found */
l_hashFloat64ToUint64(nsize, val, &key);
l_dnaHashAdd(dahash, key, (l_float64)items);
l_dnaAddNumber(dad, val);
items++;
}
}
if (pdahash)
*pdahash = dahash;
else
l_dnaHashDestroy(&dahash);
return 0;
}
/*!
* l_dnaConvertToNuma()
*
* Input: da
* Return: na, or null on error
*/
NUMA *
l_dnaConvertToNuma(L_DNA *da)
{
l_int32 i, n;
l_float64 val;
NUMA *na;
PROCNAME("l_dnaConvertToNuma");
if (!da)
return (NUMA *)ERROR_PTR("da not defined", procName, NULL);
n = l_dnaGetCount(da);
na = numaCreate(n);
for (i = 0; i < n; i++) {
l_dnaGetDValue(da, i, &val);
numaAddNumber(na, val);
}
return na;
}
/*!
* \brief l_dnaIntersectionByAset()
*
* \param[in] da1, da2
* \return dad with the intersection of the two arrays, or NULL on error
*
* <pre>
* Notes:
* (1) See sarrayIntersection() for the approach.
* (2) Here, the key in building the sorted tree is the number itself.
* (3) Operations using an underlying tree are O(nlogn), which is
* typically less efficient than hashing, which is O(n).
* </pre>
*/
L_DNA *
l_dnaIntersectionByAset(L_DNA *da1,
L_DNA *da2)
{
l_int32 n1, n2, i, n;
l_float64 val;
L_ASET *set1, *set2;
RB_TYPE key;
L_DNA *da_small, *da_big, *dad;
PROCNAME("l_dnaIntersectionByAset");
if (!da1)
return (L_DNA *)ERROR_PTR("da1 not defined", procName, NULL);
if (!da2)
return (L_DNA *)ERROR_PTR("da2 not defined", procName, NULL);
/* Put the elements of the largest array into a set */
n1 = l_dnaGetCount(da1);
n2 = l_dnaGetCount(da2);
da_small = (n1 < n2) ? da1 : da2; /* do not destroy da_small */
da_big = (n1 < n2) ? da2 : da1; /* do not destroy da_big */
set1 = l_asetCreateFromDna(da_big);
/* Build up the intersection of floats */
dad = l_dnaCreate(0);
n = l_dnaGetCount(da_small);
set2 = l_asetCreate(L_FLOAT_TYPE);
for (i = 0; i < n; i++) {
l_dnaGetDValue(da_small, i, &val);
key.ftype = val;
if (l_asetFind(set1, key) && !l_asetFind(set2, key)) {
l_dnaAddNumber(dad, val);
l_asetInsert(set2, key);
}
}
l_asetDestroy(&set1);
l_asetDestroy(&set2);
return dad;
}
/*!
* \brief l_dnaFindValByHash()
*
* \param[in] da
* \param[in] dahash containing indices into %da
* \param[in] val searching for this number in %da
* \param[out] pindex index into da if found; -1 otherwise
* \return 0 if OK; 1 on error
*
* <pre>
* Notes:
* (1) Algo: hash %val into a key; hash the key to get the dna
* in %dahash (that holds indices into %da); traverse
* the dna of indices looking for %val in %da.
* </pre>
*/
l_ok
l_dnaFindValByHash(L_DNA *da,
L_DNAHASH *dahash,
l_float64 val,
l_int32 *pindex)
{
l_int32 i, nbuckets, nvals, indexval;
l_float64 vali;
l_uint64 key;
L_DNA *da1;
PROCNAME("l_dnaFindValByHash");
if (!pindex)
return ERROR_INT("&index not defined", procName, 1);
*pindex = -1;
if (!da)
return ERROR_INT("da not defined", procName, 1);
if (!dahash)
return ERROR_INT("dahash not defined", procName, 1);
nbuckets = l_dnaHashGetCount(dahash);
l_hashFloat64ToUint64(nbuckets, val, &key);
da1 = l_dnaHashGetDna(dahash, key, L_NOCOPY);
if (!da1) return 0;
/* Run through da1, looking for this %val */
nvals = l_dnaGetCount(da1);
for (i = 0; i < nvals; i++) {
l_dnaGetIValue(da1, i, &indexval);
l_dnaGetDValue(da, indexval, &vali);
if (val == vali) {
*pindex = indexval;
return 0;
}
}
return 0;
}
/*!
* \brief l_asetCreateFromDna()
*
* \param[in] da source dna
* \return set using the doubles in %da as keys
*/
L_ASET *
l_asetCreateFromDna(L_DNA *da)
{
l_int32 i, n;
l_float64 val;
L_ASET *set;
RB_TYPE key;
PROCNAME("l_asetCreateFromDna");
if (!da)
return (L_ASET *)ERROR_PTR("da not defined", procName, NULL);
set = l_asetCreate(L_FLOAT_TYPE);
n = l_dnaGetCount(da);
for (i = 0; i < n; i++) {
l_dnaGetDValue(da, i, &val);
key.ftype = val;
l_asetInsert(set, key);
}
return set;
}
