/* ------------------ L2str ------------------- */
void __CDECL
L2str( const PLstr s )
{
if (LTYPE(*s)==LINTEGER_TY) {
#if defined(WCE) || defined(__BORLANDC__)
LTOA(LINT(*s),LSTR(*s),10);
#else
sprintf(LSTR(*s), "%ld", LINT(*s));
#endif
LLEN(*s) = STRLEN(LSTR(*s));
} else { /* LREAL_TY */
/* There is a problem with the Windows CE */
char str[50];
size_t len;
snprintf(str, sizeof(str), "%.*g", lNumericDigits, LREAL(*s));
/* --- remove the last dot from the number --- */
len = STRLEN(str);
#ifdef WCE
if (str[len-1] == '.') len--;
#endif
if (len>=LMAXLEN(*s)) Lfx(s,len);
MEMCPY(LSTR(*s),str,len);
LLEN(*s) = len;
}
LTYPE(*s) = LSTRING_TY;
} /* L2str */
/* we should change the string */
void __CDECL
_L2num( const PLstr s, const int type )
{
LASCIIZ(*s);
switch (type) {
case LINTEGER_TY:
/*////LINT(*s) = atol( LSTR(*s) ); */
LINT(*s) = (long)lLastScannedNumber;
LTYPE(*s) = LINTEGER_TY;
LLEN(*s) = sizeof(long);
break;
case LREAL_TY:
/*////LREAL(*s) = strtod( LSTR(*s), NULL ); */
LREAL(*s) = lLastScannedNumber;
if ((double)((long)LREAL(*s)) == LREAL(*s)) {
LINT(*s) = (long)LREAL(*s);
LTYPE(*s) = LINTEGER_TY;
LLEN(*s) = sizeof(long);
} else {
LTYPE(*s) = LREAL_TY;
LLEN(*s) = sizeof(double);
}
break;
default:
Lerror(ERR_BAD_ARITHMETIC,0);
}
} /* _L2num */
/* ------------------ L2int ------------------- */
void __CDECL
L2int( const PLstr s )
{
if (LTYPE(*s)==LREAL_TY) {
if ((double)((long)LREAL(*s)) == LREAL(*s))
LINT(*s) = (long)LREAL(*s);
else
Lerror(ERR_INVALID_INTEGER,0);
} else { /* LSTRING_TY */
LASCIIZ(*s);
switch (_Lisnum(s)) {
case LINTEGER_TY:
LINT(*s) = (long)lLastScannedNumber;
break;
case LREAL_TY:
LREAL(*s) = lLastScannedNumber;
if ((double)((long)LREAL(*s)) == LREAL(*s))
LINT(*s) = (long)LREAL(*s);
else
Lerror(ERR_INVALID_INTEGER,0);
break;
default:
Lerror(ERR_INVALID_INTEGER,0);
}
}
LTYPE(*s) = LINTEGER_TY;
LLEN(*s) = sizeof(long);
} /* L2int */
/* ---------------- Lstrcpy ----------------- */
void __CDECL
Lstrcpy( const PLstr to, const PLstr from )
{
if (LLEN(*from)==0) {
LLEN(*to) = 0;
LTYPE(*to) = LSTRING_TY;
} else {
if (LMAXLEN(*to)<=LLEN(*from)) Lfx(to,LLEN(*from));
switch ( LTYPE(*from) ) {
case LSTRING_TY:
MEMCPY( LSTR(*to), LSTR(*from), LLEN(*from) );
break;
case LINTEGER_TY:
LINT(*to) = LINT(*from);
break;
case LREAL_TY:
LREAL(*to) = LREAL(*from);
break;
}
LTYPE(*to) = LTYPE(*from);
LLEN(*to) = LLEN(*from);
}
} /* Lstrcpy */
/* ----------------- Lrdint ------------------ */
long __CDECL
Lrdint( const PLstr s )
{
if (LTYPE(*s)==LINTEGER_TY) return LINT(*s);
if (LTYPE(*s)==LREAL_TY) {
if ((double)((long)LREAL(*s)) == LREAL(*s))
return (long)LREAL(*s);
else
Lerror(ERR_INVALID_INTEGER,0);
} else { /* LSTRING_TY */
LASCIIZ(*s);
switch (_Lisnum(s)) {
case LINTEGER_TY:
/*///return atol( LSTR(*s) ); */
return (long)lLastScannedNumber;
case LREAL_TY:
/*///d = strtod( LSTR(*s), NULL );
//////if ((double)((long)d) == d)
////// return (long)d; */
if ((double)((long)lLastScannedNumber) == lLastScannedNumber)
return (long)lLastScannedNumber;
else
Lerror(ERR_INVALID_INTEGER,0);
break;
default:
Lerror(ERR_INVALID_INTEGER,0);
}
}
return 0; /* never gets here but keeps compiler happy */
} /* Lrdint */
/* ------------------ L2num ------------------- */
void __CDECL
L2num( const PLstr s )
{
switch (_Lisnum(s)) {
case LINTEGER_TY:
/*//LINT(*s) = atol( LSTR(*s) ); */
LINT(*s) = (long)lLastScannedNumber;
LTYPE(*s) = LINTEGER_TY;
LLEN(*s) = sizeof(long);
break;
case LREAL_TY:
/*///LREAL(*s) = strtod( LSTR(*s), NULL ); */
LREAL(*s) = lLastScannedNumber;
/*
//// Numbers like 2.0 should be treated as real and not as integer
//// because in cases like factorial while give an error result
//// if ((double)((long)LREAL(*s)) == LREAL(*s)) {
//// LINT(*s) = (long)LREAL(*s);
//// LTYPE(*s) = LINTEGER_TY;
//// LLEN(*s) = sizeof(long);
//// } else {
*/
LTYPE(*s) = LREAL_TY;
LLEN(*s) = sizeof(double);
/*
//// }
*/
break;
default:
Lerror(ERR_BAD_ARITHMETIC,0);
}
} /* L2num */
/* ---------------- Licpy ------------------ */
void __CDECL
Licpy( const PLstr to, const long from )
{
LLEN(*to) = sizeof(long);
LTYPE(*to) = LINTEGER_TY;
LINT(*to) = from;
} /* Licpy */
/* --------------------------------------------------------------- */
void R_sql( const int func )
{
int rc;
const char *tail;
if (ARGN!=1) Lerror(ERR_INCORRECT_CALL,0);
get_s(1);
if (!sqldb) Lerror(ERR_DATABASE,1);
if (sqlstmt) {
sqlite3_finalize(sqlstmt);
sqlstmt = NULL;
}
Licpy(ARGR, sqlite3_prepare_v2(sqldb, LSTR(*ARG1), LLEN(*ARG1), &sqlstmt, &tail));
if (LINT(*ARGR) == SQLITE_OK) {
if (sqlite3_bind_parameter_count(sqlstmt)==0) {
rc = sqlite3_step(sqlstmt);
if (rc == SQLITE_ROW)
Lscpy(ARGR, "ROW");
else
if (rc == SQLITE_DONE)
Licpy(ARGR, SQLITE_OK);
else
Licpy(ARGR, rc);
}
}
} /* R_sql */
/* ---------------- Lintdiv ---------------- */
void __CDECL
Lintdiv( const PLstr to, const PLstr A, const PLstr B )
{
double b;
b = Lrdreal(B);
if (b == 0) Lerror(ERR_ARITH_OVERFLOW,0);
LINT(*to) = (long) (Lrdreal(A) / b);
LTYPE(*to) = LINTEGER_TY;
LLEN(*to) = sizeof(long);
} /* Lintdiv */
/* ------------------ L2real ------------------- */
void __CDECL
L2real( const PLstr s )
{
if (LTYPE(*s)==LINTEGER_TY)
LREAL(*s) = (double)LINT(*s);
else { /* LSTRING_TY */
LASCIIZ(*s);
if (_Lisnum(s)!=LSTRING_TY)
/*/////LREAL(*s) = strtod( LSTR(*s), NULL ); */
LREAL(*s) = lLastScannedNumber;
else
Lerror(ERR_BAD_ARITHMETIC,0);
}
LTYPE(*s) = LREAL_TY;
LLEN(*s) = sizeof(double);
} /* L2real */
/* ----------------- Lrdreal ------------------ */
double __CDECL
Lrdreal( const PLstr s )
{
if (LTYPE(*s)==LREAL_TY) return LREAL(*s);
if (LTYPE(*s)==LINTEGER_TY)
return (double)LINT(*s);
else { /* LSTRING_TY */
LASCIIZ(*s);
if (_Lisnum(s)!=LSTRING_TY)
/*///// return strtod( LSTR(*s), NULL ); */
return lLastScannedNumber;
else
Lerror(ERR_BAD_ARITHMETIC,0);
}
return 0.0;
} /* Lrdreal */
StringSet::StringSet(LINT nitems, // number of items
PatternPar par, // npats, patlen, corr, conf & conf_var
Taxonomy *ptax, // taxonomy (optional)
FLOAT rept, // repetition-level
FLOAT rept_var // variation in repetition-level
)
: tax(ptax)
{
NormalDist conf(par.conf, par.conf_var);
ExpDist freq;
ExpDist corr_lvl;
PoissonDist len(par.patlen-1); // string length
NormalDist repeat(rept, rept_var);
UniformDist ud;
items = new ItemSet(nitems, tax); // associate probabilities with items
LINT i, j, num_same;
FLOAT tot;
npats = par.npats;
// last_pat = 0;
pat = new StringP [npats];
for (i = 0; i < npats; i++)
{
pat[i] = new String( 1+len() );
// fill correlated items
if (par.corr > 0 && i > 0) { // correlated patterns
// each pattern has some items same as the previous pattern
num_same = LINT( pat[i]->size() * par.corr * corr_lvl() + 0.5 );
if ( num_same > pat[i-1]->size() )
num_same = pat[i-1]->size();
if ( num_same > pat[i]->size() )
num_same = pat[i]->size();
// choose num_same items at random from previous pattern
Choose shuffle(pat[i-1]->size(), num_same);
for (j = 0; j < num_same; j++)
pat[i]->items[j] = pat[i-1]->item( shuffle.pos(j) );
// pat[i-1]->shuffle(num_same);
// for (j = 0; j < num_same; j++)
// pat[i]->items[j] = pat[i-1]->rand_item(j);
}
else { // no correlation
num_same = 0;
}
if (rept == 0) {
// fill remaining items at random
for (j = num_same; j < pat[i]->size(); j++)
pat[i]->items[j] = items->get_item();
// pat[i]->items[j] = LINT(1 + nitems * rand());
}
else {
// some items are repetitions
FLOAT rept_lvl = repeat();
for (j = num_same; j < pat[i]->size(); j++)
if ( j > 0 && ud() < rept_lvl ) // pick a previous item
pat[i]->items[j] = pat[i]->items[ LINT(j*ud()) ];
else // pick random item
pat[i]->items[j] = items->get_item();
}
pat[i]->prob = freq(); // prob. that this pattern will be picked
pat[i]->conf = conf(); // used in Transaction::add and CustSeq::add
// to decide how many items to drop from
// this pattern to corrupt it
}
if (tax) {
// weight probabilites with geometric mean of probabilities of items
for (i = 0; i < npats; i++)
{
DOUBLE weight = 1;
for (j = 0; j < pat[i]->size(); j++)
weight *= items->weight(pat[i]->items[j]);
// cerr << "WEIGHT = " << weight;
weight = pow(weight, DOUBLE(1)/pat[i]->size());
// cerr << " " << weight << endl;
pat[i]->prob *= weight;
}
}
// normalize probabilites (why -- see get_pat)
cum_prob = new FLOAT [npats];
tot = 0;
for (i = 0; i < npats; i++)
tot += pat[i]->prob;
for (i = 0; i < npats; i++)
pat[i]->prob /= tot;
// calulate cumulative probabilities
cum_prob[0] = pat[0]->prob;
for (i = 1; i < npats; i++)
cum_prob[i] = cum_prob[i-1] + pat[i]->prob;
// cerr << cum_prob[npats-1] << endl << flush;
// allocate space for answer
LINT maxlen = 0;
for (i = 1; i < npats; i++)
if (pat[i]->size() > maxlen)
maxlen = pat[i]->size();
answer = new String(maxlen);
}
void get_args(SeqPar &par, int argc, char **argv)
{
LINT arg_pos = 2;
strcpy(data_file, "data");
strcpy(pat_file, "pat");
strcpy(tax_file, "tax");
while (arg_pos < argc)
{
if (strcmp(argv[arg_pos], "-ncust") == 0) {
// g++ LINT LINT
par.ncust = LINT (1000 * atof(argv[++arg_pos]));
cat_fname(".ncust_", argv[arg_pos]);
arg_pos++;
if (par.ncust < 1) exit_fail_with_msg("ntrans must be >= 1");
continue;
}
else if (strcmp(argv[arg_pos], "-slen") == 0) {
par.slen = atof(argv[++arg_pos]);
cat_fname(".slen_", argv[arg_pos]);
arg_pos++;
if (par.slen < 1) exit_fail_with_msg("slen must be >= 1");
continue;
}
else if (strcmp(argv[arg_pos], "-tlen") == 0) {
par.tlen = atof(argv[++arg_pos]);
cat_fname(".tlen_", argv[arg_pos]);
arg_pos++;
if (par.tlen < 1) exit_fail_with_msg("tlen must be >= 1");
continue;
}
else if (strcmp(argv[arg_pos], "-nitems") == 0) {
// g++ LINT
par.nitems = LINT (1000 * atof(argv[++arg_pos]));
cat_fname(".nitems_", argv[arg_pos]);
arg_pos++;
if (par.nitems < 1) exit_fail_with_msg("nitems must be >= 1");
continue;
}
else if (strcmp(argv[arg_pos], "-rept") == 0) {
par.rept = atof(argv[++arg_pos]);
cat_fname(".rept_", argv[arg_pos]);
arg_pos++;
if (par.rept < 0 || par.rept > 1)
exit_fail_with_msg("repetition-level must be between 0 and 1");
continue;
}
else if (strcmp(argv[arg_pos], "-seq.npats") == 0) {
par.lseq.npats = atoi(argv[++arg_pos]);
cat_fname(".seq.npats_", argv[arg_pos]);
arg_pos++;
if (par.lseq.npats < 1) exit_fail_with_msg("npats must be >= 1");
continue;
}
else if (strcmp(argv[arg_pos], "-seq.patlen") == 0) {
par.lseq.patlen = atof(argv[++arg_pos]);
cat_fname(".seq.patlen_", argv[arg_pos]);
arg_pos++;
if (par.lseq.patlen <= 0) exit_fail_with_msg("patlen must be > 0");
continue;
}
else if (strcmp(argv[arg_pos], "-seq.corr") == 0) {
par.lseq.corr = atof(argv[++arg_pos]);
cat_fname(".seq.corr_", argv[arg_pos]);
arg_pos++;
continue;
}
else if (strcmp(argv[arg_pos], "-seq.conf") == 0) {
par.lseq.conf = atof(argv[++arg_pos]);
cat_fname(".seq.conf_", argv[arg_pos]);
arg_pos++;
if (par.lseq.conf > 1 || par.lseq.conf < 0)
exit_fail_with_msg("conf must be between 0 and 1");
continue;
}
else if (strcmp(argv[arg_pos], "-lit.npats") == 0) {
par.lits.npats = atoi(argv[++arg_pos]);
cat_fname(".lit.npats_", argv[arg_pos]);
arg_pos++;
if (par.lits.npats < 1) exit_fail_with_msg("npats must be >= 1");
continue;
}
else if (strcmp(argv[arg_pos], "-lit.patlen") == 0) {
par.lits.patlen = atof(argv[++arg_pos]);
cat_fname(".lit.patlen_", argv[arg_pos]);
arg_pos++;
if (par.lits.patlen <= 0) exit_fail_with_msg("patlen must be > 0");
continue;
}
else if (strcmp(argv[arg_pos], "-lit.corr") == 0) {
par.lits.corr = atof(argv[++arg_pos]);
cat_fname(".lit.corr_", argv[arg_pos]);
arg_pos++;
continue;
}
else if (strcmp(argv[arg_pos], "-lit.conf") == 0) {
par.lits.conf = atof(argv[++arg_pos]);
cat_fname(".lit.conf_", argv[arg_pos]);
arg_pos++;
if (par.lits.conf > 1 || par.lits.conf < 0)
exit_fail_with_msg("conf must be between 0 and 1");
continue;
}
else if (strcmp(argv[arg_pos], "-fname") == 0) {
strcpy(data_file, argv[++arg_pos]);
strcat(data_file, ".data");
strcpy(pat_file, argv[arg_pos]);
strcat(pat_file, ".pat");
strcpy(tax_file, argv[arg_pos++]);
strcat(tax_file, ".tax");
userfile = true;
continue;
}
else if (strcmp(argv[arg_pos], "-ascii") == 0) {
par.ascii = true;
cat_fname(".ascii", "");
arg_pos++;
continue;
}
else if (strcmp(argv[arg_pos], "-version") == 0) {
cout << VERSION << endl;
exit(EXIT_SUCCESS);
}
else {
command_line(par);
}
} // end while
}
void get_args(TaxPar &par, int argc, char **argv)
{
LINT arg_pos = 2;
strcpy(data_file, "data");
strcpy(pat_file, "pat");
strcpy(tax_file, "tax");
while (arg_pos < argc)
{
if (strcmp(argv[arg_pos], "-ntrans") == 0) {
// g++ LINT !!
par.ntrans = LINT (1000 * atof(argv[++arg_pos]));
cat_fname(".ntrans_", argv[arg_pos]);
arg_pos++;
if (par.ntrans < 1) exit_fail_with_msg("ntrans must be >= 1");
continue;
}
else if (strcmp(argv[arg_pos], "-tlen") == 0) {
par.tlen = atof(argv[++arg_pos]);
cat_fname(".tlen_", argv[arg_pos]);
arg_pos++;
if (par.tlen < 1) exit_fail_with_msg("tlen must be >= 1");
continue;
}
else if (strcmp(argv[arg_pos], "-nitems") == 0) {
// g++ the LINT worm is back again!
par.nitems = LINT (1000 * atof(argv[++arg_pos]));
cat_fname(".nitems_", argv[arg_pos]);
arg_pos++;
if (par.nitems < 1) exit_fail_with_msg("nitems must be >= 1");
continue;
}
else if (strcmp(argv[arg_pos], "-nroots") == 0) {
par.nroots = atoi(argv[++arg_pos]);
cat_fname(".nroots_", argv[arg_pos]);
arg_pos++;
if (par.nroots < 1) exit_fail_with_msg("nroots must be >= 1");
continue;
}
else if (strcmp(argv[arg_pos], "-nlevels") == 0) {
par.nlevels = atof(argv[++arg_pos]);
cat_fname(".nlevels_", argv[arg_pos]);
arg_pos++;
if (par.nlevels < 1) exit_fail_with_msg("nlevels must be >= 1");
continue;
}
else if (strcmp(argv[arg_pos], "-fanout") == 0) {
par.fanout = atof(argv[++arg_pos]);
cat_fname(".fanout_", argv[arg_pos]);
arg_pos++;
if (par.fanout < 1) exit_fail_with_msg("fanout must be >= 1");
continue;
}
else if (strcmp(argv[arg_pos], "-depth") == 0) {
par.depth_ratio = atof(argv[++arg_pos]);
cat_fname(".depth_", argv[arg_pos]);
arg_pos++;
if (par.depth_ratio <= 0) exit_fail_with_msg("fanout must be > 0");
continue;
}
else if (strcmp(argv[arg_pos], "-npats") == 0) {
par.lits.npats = atoi(argv[++arg_pos]);
cat_fname(".npats_", argv[arg_pos]);
arg_pos++;
if (par.lits.npats < 1) exit_fail_with_msg("npats must be >= 1");
continue;
}
else if (strcmp(argv[arg_pos], "-patlen") == 0) {
par.lits.patlen = atof(argv[++arg_pos]);
cat_fname(".patlen_", argv[arg_pos]);
arg_pos++;
if (par.lits.patlen <= 0) exit_fail_with_msg("patlen must be > 0");
continue;
}
else if (strcmp(argv[arg_pos], "-corr") == 0) {
par.lits.corr = atof(argv[++arg_pos]);
cat_fname(".corr_", argv[arg_pos]);
arg_pos++;
continue;
}
else if (strcmp(argv[arg_pos], "-conf") == 0) {
par.lits.conf = atof(argv[++arg_pos]);
cat_fname(".conf_", argv[arg_pos]);
arg_pos++;
if (par.lits.conf > 1 || par.lits.conf < 0)
exit_fail_with_msg("conf must be between 0 and 1");
continue;
}
else if (strcmp(argv[arg_pos], "-fname") == 0) {
strcpy(data_file, argv[++arg_pos]);
strcat(data_file, ".data");
strcpy(pat_file, argv[arg_pos]);
strcat(pat_file, ".pat");
strcpy(tax_file, argv[arg_pos++]);
strcat(tax_file, ".tax");
userfile = true;
continue;
}
else if (strcmp(argv[arg_pos], "-ascii") == 0) {
par.ascii = true;
cat_fname(".ascii", "");
arg_pos++;
continue;
}
else if (strcmp(argv[arg_pos], "-randseed") == 0) {
par.seed = atoi(argv[++arg_pos]);
arg_pos++;
if (par.seed >= 0)
exit_fail_with_msg("randseed must be negative");
continue;
}
else if (strcmp(argv[arg_pos], "-version") == 0) {
cout << VERSION << endl;
exit(EXIT_SUCCESS);
}
else {
command_line(par);
}
} // end while
par.calc_values();
}
