void bingoGetName (Scanner &scanner, Array<char> &result)
{
QS_DEF(Array<char>, str);
bool single_line = Scanner::isSingleLine(scanner);
result.clear();
if (single_line)
{
scanner.readLine(str, true);
int idx = str.find('|');
int idx2 = 0;
if (idx >= 0)
{
int tmp = str.find(idx + 1, str.size(), '|');
if (tmp > 0)
idx2 = tmp + 1;
}
BufferScanner scan2(str.ptr() + idx2);
while (!scan2.isEOF())
{
if (isspace(scan2.readChar()))
break;
}
scan2.skipSpace();
if (!scan2.isEOF())
scan2.readLine(result, false);
}
else
{
scanner.readLine(result, false);
if (result.size() >= 4 && strncmp(result.ptr(), "$RXN", 4) == 0)
scanner.readLine(result, false);
}
}
/**
* Initialize Table
* ================================
*
* The following reads in the PEG in the passed filename and converts the
* grammar specification into a set of tokens for traversal when parsing other files
* according to said grammar.
*/
void Parser::initializeTable(Scanner& input)
{
// These regexes are used to read in a custom PEG grammar.
// Refer to /grammars/arithmetic.peg for a more thorough explanation
// on the grammar. Note all other terms can be manipulated just
// by reading in the remainder of a line or reading in words.
Regex arrowOperator = Regex::fromPool("pparser-arrow", "\\->");
Regex markedWord = Regex::fromPool("pparser-marked-word", "\\A+'?");
// On any given line, the first two terminals should be the nonterminal
// being defined and the arrow operator or we've encountered a comment.
while(input.peek() != EOF) {
if(input.peek() == PPARSER_COMMENT) {
input.readLine();
} else {
// First read in nonterminal and find start if possible
std::string nonterminal = input.next(markedWord);
if(nonterminal.back() == PPARSER_START) {
nonterminal.pop_back();
if(start.empty()) {
start = nonterminal;
} else {
throw InvalidGrammar("Multiple starting nonterminals", input.getCurrentState());
}
}
// Can now read in the arrow operator
input.next(arrowOperator);
// Rest of line is dedicated to definition
table[nonterminal] = std::make_shared<Choices>(input);
}
}
// Must have at least one starting nonterminal
if(start.empty()) {
throw InvalidGrammar("No starting nonterminal specified");
}
}
void _importSMILES (OracleEnv &env, const char *table, const char *smiles_col,
const char *id_col, const char *file_name)
{
FileScanner fscanner(file_name);
AutoPtr<GZipScanner> gzscanner;
Scanner *scanner;
int nwritten = 0;
QS_DEF(Array<char>, id);
QS_DEF(Array<char>, str);
env.dbgPrintfTS("importing into table %s\n", table);
// detect if input is gzipped
byte magic[2];
int pos = fscanner.tell();
fscanner.readCharsFix(2, (char *)magic);
fscanner.seek(pos, SEEK_SET);
if (magic[0] == 0x1f && magic[1] == 0x8b)
{
gzscanner.reset(new GZipScanner(fscanner));
scanner = gzscanner.get();
}
else
scanner = &fscanner;
while (!scanner->isEOF())
{
id.clear();
scanner->readLine(str, false);
BufferScanner strscan(str);
strscan.skipSpace();
while (!strscan.isEOF() && !isspace(strscan.readChar()))
;
strscan.skipSpace();
if (strscan.lookNext() == '|')
{
strscan.readChar();
while (!strscan.isEOF() && strscan.readChar() != '|')
;
strscan.skipSpace();
}
if (!strscan.isEOF() && id_col != 0)
strscan.readLine(id, true);
OracleStatement statement(env);
statement.append("INSERT INTO %s(%s", table, smiles_col);
if (id_col != 0)
statement.append(", %s", id_col);
statement.append(") VALUES(:smiles");
if (id_col != 0)
{
if (id.size() > 1)
statement.append(", :id");
else
statement.append(", NULL");
}
statement.append(")");
statement.prepare();
str.push(0);
statement.bindStringByName(":smiles", str.ptr(), str.size());
if (id.size() > 1)
statement.bindStringByName(":id", id.ptr(), id.size());
statement.execute();
nwritten++;
if (nwritten % 1000 == 0)
{
env.dbgPrintfTS("imported %d items, commiting\n", nwritten);
OracleStatement::executeSingle(env, "COMMIT");
}
}
if (nwritten % 1000 != 0)
{
env.dbgPrintfTS("imported %d items, commiting\n", nwritten);
OracleStatement::executeSingle(env, "COMMIT");
}
}