void MangoGross::parseQuery (Scanner &scanner)
{
scanner.skipSpace();
char c = scanner.readChar();
if (c == '>')
{
if (scanner.readChar() != '=')
throw Error("expecting '=' after '>'");
_sign = 1;
}
else if (c == '<')
{
if (scanner.readChar() != '=')
throw Error("expecting '=' after '<'");
_sign = -1;
}
else if (c == '=')
_sign = 0;
else
throw Error("expected one of '<= >= =', got %c", c);
MoleculeGrossFormula::fromString(scanner, _query_gross);
ArrayOutput out(_conditions);
bool first = true;
if (_sign == 0)
{
QS_DEF(Array<char>, query_gross_str);
MoleculeGrossFormula::toString(_query_gross, query_gross_str);
out.printf("gross = '%s'", query_gross_str.ptr());
}
else for (int i = 0; i < NELEM(MangoIndex::counted_elements); i++)
{
int elem = MangoIndex::counted_elements[i];
if (_query_gross[elem] <= 0 && _sign == 1)
continue;
if (!first)
out.printf(" AND ");
first = false;
if (_sign == 1)
out.printf("cnt_%s >= %d", Element::toString(elem), _query_gross[elem]);
else // _sign == -1
out.printf("cnt_%s <= %d", Element::toString(elem), _query_gross[elem]);
}
out.writeChar(0);
}
int Element::read (Scanner &scanner)
{
char str[3] = {0, 0, 0};
str[0] = scanner.readChar();
if (islower(scanner.lookNext()))
str[1] = scanner.readChar();
return fromString(str);
}
bool MoleculeAutoLoader::tryMDLCT (Scanner &scanner, Array<char> &outbuf)
{
int pos = scanner.tell();
bool endmark = false;
QS_DEF(Array<char>, curline);
outbuf.clear();
while (!scanner.isEOF())
{
int len = scanner.readByte();
if (len > 90) // Molfiles and Rxnfiles actually have 80 characters limit
{
scanner.seek(pos, SEEK_SET);
// Garbage after endmark means end of data.
// (See the note below about endmarks)
if (endmark)
return true;
return false;
}
curline.clear();
while (len-- > 0)
{
if (scanner.isEOF())
{
scanner.seek(pos, SEEK_SET);
return false;
}
int c = scanner.readChar();
curline.push(c);
}
curline.push(0);
if (endmark)
{
// We can not properly read the data to the end as there
// is often garbage after the actual MDLCT data.
// Instead, we are doing this lousy check:
// "M END" or "$END MOL" can be followed only
// by "$END CTAB" (in Markush queries), or
// by "$MOL" (in Rxnfiles). Otherwise, this
// is actually the end of data.
if (strcmp(curline.ptr(), "$END CTAB") != 0 &&
strcmp(curline.ptr(), "$MOL") != 0)
{
scanner.seek(pos, SEEK_SET);
return true;
}
}
if (strcmp(curline.ptr(), "M END") == 0)
endmark = true;
else if (strcmp(curline.ptr(), "$END MOL") == 0)
endmark = true;
else
endmark = false;
outbuf.appendString(curline.ptr(), false);
outbuf.push('\n');
}
scanner.seek(pos, SEEK_SET);
// It happened once that a valid Molfile had successfully
// made its way through the above while() cycle, and thus
// falsely recognized as MDLCT. To fight this case, we include
// here a check that the last line was actually an endmark
return endmark;
}
