pgLab/pglablib/util.cpp

#include "util.h"
#include "CsvWriter.h"
#include "SqlLexer.h"
#include <QApplication>
#include <QTextStream>
#include <QClipboard>
#include <sstream>
#include <stdexcept>

// Supported range from microseconds to seconds
// min:sec to hours::min::sec
QString msfloatToHumanReadableString(float ms)
{
	QString unit;
	float val;
	int deci = 2;
	if (ms < 1.0f) {
		val = ms * 1000.f;
		//result = QString::asprintf("%0.3f", ms * 1000.0f);
		unit = u8"μs";
	}
	else if (ms >= 1000.0) {
		val = ms / 1000.0f;
		unit = "s";
		if (val >= 60.0) {
			int secs = val;
			int min = secs / 60.0;
			secs -= min * 60;
			if (min >= 60) {
				int hour = min / 60;
				min -= hour * 60;
				return QString::asprintf("%d:%02d:%02d", hour, min, secs);
			}
			else {
				return QString::asprintf("%02d:%02d", min, secs);
			}
		}
	}
	else {
		val = ms;
		unit = "ms";
	}

//		if (val >= 1000.f) {
//			deci = 0;
//		}
//		else
	if (val >= 100.f) {
		deci = 0;
	}
	else if (val >= 10.f) {
		deci = 1;
	}
	QString result = QString::asprintf("%0.*f", deci, val);
	return result + unit;
}

void exportTable(const QTableView *view, QTextStream &out)
{
	auto model = view->model();
	if (model) {
		CsvWriter csv(&out);
		csv.setSeperator('\t');
		csv.setQuote('"');

		auto cols = model->columnCount();
		auto rows = model->rowCount();

        // header
        for (int col = 0; col < cols; ++col) {
            auto display_text = model->headerData(col, Qt::Horizontal, Qt::DisplayRole).toString();
            csv.writeField(display_text);
        }
        csv.nextRow();

		for (auto row = 0; row < rows; ++row) {
			for (int col = 0; col < cols; ++col) {
				auto idx = model->index(row, col);
				auto display_text = idx.data(Qt::DisplayRole).toString();
				csv.writeField(display_text);
			}
			csv.nextRow();
		}
		out.flush();
	}
}

void copySelectionToClipboard(const QTableView *view)
{
	//QAbstractItemModel * model = resultModel; //view->model();
	QItemSelectionModel * selection = view->selectionModel();

	QModelIndexList selectedIndexes = selection->selectedIndexes();

	if (selectedIndexes.count() > 0) {
		QString clipboard_string;
        QTextStream out(&clipboard_string, QTextStream::WriteOnly);
		CsvWriter csv(&out);
		csv.setSeperator('\t');
		csv.setQuote('"');

		int previous_row = selectedIndexes[0].row();
	//		for (int i = 0; i < selectedIndexes.count(); ++i) {
	//			QModelIndex current = selectedIndexes[i];
		for (auto current : selectedIndexes) {
			if (current.row() > previous_row) {
				csv.nextRow();
				previous_row = current.row();
			}
			QString display_text = current.data(Qt::DisplayRole).toString();
			csv.writeField(display_text);
		}
		out.flush();
		QApplication::clipboard()->setText(clipboard_string);
	}
}

QString EscapeForCString(QString qi)
{
	int escape_count = 0;
	for (QChar c : qi)
		if (c == '"' || c == '\\') ++escape_count;

	QString out;
	out.reserve(qi.size() + escape_count);
	for (QChar c : qi)
		if (c == '"' || c == '\\') {
			out += "\\";
			out += c;
		}
		else
			out += c;
	return out;
}

QString ConvertToMultiLineCString(const QString &in_)
{
	// We need to atleast escape " and \ and also any multi byte utf8 char

	// remove empty lines at start
	int last_nl_idx = 0;
	for (int idx = 0; idx < in_.length(); ++idx) {
		QChar c = in_[idx];
		if (c == '\n') last_nl_idx = idx+1;
		if (!c.isSpace()) {
			break;
		}
	}
	QString in = in_.right(in_.length() - last_nl_idx);
	int idx;
	for (idx = in.length() - 1; idx >= 0 && in[idx].isSpace(); --idx) ;
	++idx;
	in.truncate(idx);

	SqlLexer lexer(in, LexerState::Null, true);
	QString out;
	QString line = "\"";
	QString comment;
	while (true) {
		SqlToken token = lexer.nextBasicToken();
		if (token.ok) {
			if (token.tokenType == BasicTokenType::Comment) {
				// save comment is seperate variable
                comment = "//" + token.out.right(token.out.length()-2);
				// Trim whitespace on right
				int idx;
				for (idx = comment.length() - 1; idx >= 0 && comment[idx].isSpace(); --idx) ;
				++idx;
				comment.truncate(idx);
			}
			else if (token.tokenType == BasicTokenType::End || token.tokenType == BasicTokenType::NewLine) {
				// trim right
				{
					int idx;
					for (idx = line.length() - 1; idx >= 0 && line[idx].isSpace(); --idx) ;
					++idx;
					if (!comment.isEmpty()) {
						// put the whitespace in front of the comment so it will be outside the contents of the string literal but alignment of comments is preserved
                        comment = line.right(line.length() - (idx)) + comment;
					}
					line.truncate(idx);
				}

				out += line;
				if (token.tokenType == BasicTokenType::End) {
					out += "\"";
					out += comment;
					break;
				}
				else {
					out += "\\n\"";
					out += comment;
					out += "\n";
					line = "\"";
				}
				comment.clear();
			}
			else {
				line += EscapeForCString(token.out);
			}
		}
		else {
			// error during lexical analysis, need to recover
			throw std::runtime_error("Unrecognized input");
		}
	}
	return out;
}


QString ConvertToMultiLineRawCppString(const QString &in)
{
    const QString delim = "__SQL__";
    
    QString out;
    out.append("R\"" + delim + "(\n");
    out.append(in.trimmed());
    out.append("\n)" + delim + "\"");
    return out;
}

namespace {

// index points at the first char of the token. On success index is placed after the opening quote and quote is set to '"' or '\''.
bool readOpeningQuote(const QString &in, int &index, QChar &quote)
{
	QChar c = in[index];
	if (c == '\'' || c == '\"') {
		quote = c;
		++index;
		return true;
	}
	if (c == 'L' || c == 'U') {
		if (in.length() > index + 1 && in[index + 1] == '\"') {
			quote = '\"';
			index += 2;
			return true;
		}
		return false;
	}
	if (c == 'u') {
		if (in.length() > index + 2 && in[index + 1] == '8' && in[index + 2] == '\"') {
			quote = '\"';
			index += 3;
			return true;
		}
		if (in.length() > index + 1 && in[index + 1] == '\"') {
			quote = '\"';
			index += 2;
			return true;
		}
	}
	return false;
}

int hexDigitValue(QChar c)
{
	if (c >= '0' && c <= '9') return c.unicode() - '0';
	if (c >= 'a' && c <= 'f') return c.unicode() - 'a' + 10;
	if (c >= 'A' && c <= 'F') return c.unicode() - 'A' + 10;
	return -1;
}

// Reads exactly `count` hex digits starting at `first`. Fails on out-of-range or non-hex input.
bool parseHexDigits(const QString &in, int first, int count, char32_t &value)
{
	if (in.length() < first + count) return false;
	value = 0;
	for (int i = first; i < first + count; ++i) {
		int digit = hexDigitValue(in[i]);
		if (digit < 0) return false;
		value = value * 16 + static_cast<char32_t>(digit);
	}
	return true;
}

// Appends a code point as UTF-16. Rejects > 0x10FFFF and surrogates D800-DFFF.
bool appendCodePoint(QString &out, char32_t cp)
{
	if (cp > 0x10FFFF || (cp >= 0xD800 && cp <= 0xDFFF)) return false;
	if (QChar::requiresSurrogates(cp)) {
		out += QChar(QChar::highSurrogate(cp));
		out += QChar(QChar::lowSurrogate(cp));
	}
	else {
		out += QChar(static_cast<char16_t>(cp));
	}
	return true;
}

// index points at the first octal digit. On success index points at the last consumed digit.
bool decodeOctalEscape(const QString &in, int &index, QString &out)
{
	if (in.length() <= index + 2) return false;
	int value = 0;
	for (int i = index; i <= index + 2; ++i) {
		QChar digit = in[i];
		if (digit < '0' || digit > '7') return false;
		value = value * 8 + (digit.unicode() - '0');
	}
	if (value >= 0x80) return false;
	out += static_cast<QChar>(static_cast<char>(value));
	index += 2;
	return true;
}

// index points at the char after the backslash. On success index points at the last consumed char.
bool decodeEscape(const QString &in, int &index, QString &out)
{
	switch (in[index].unicode()) {
	case 'a':
		out += '\x07';
		break;
	case 'b':
		out += '\x08';
		break;
	case 'f':
		out += '\x0c';
		break;
	case 'n':
		out += '\n';
		break;
	case 'r':
	case 'v':
		break;
	case '0': case '1': case '2': case '3':
	case '4': case '5': case '6': case '7':
		return decodeOctalEscape(in, index, out);
	case 'x': {
		char32_t value = 0;
		if (!parseHexDigits(in, index + 1, 2, value)) return false;
		out += static_cast<QChar>(static_cast<char>(value));
		index += 2;
		break;
	}
	case 'u': {
		char32_t value = 0;
		if (!parseHexDigits(in, index + 1, 4, value)) return false;
		out += QChar(static_cast<ushort>(value));
		index += 4;
		break;
	}
	case 'U': {
		char32_t value = 0;
		if (!parseHexDigits(in, index + 1, 8, value)) return false;
		if (!appendCodePoint(out, value)) return false;
		index += 8;
		break;
	}
	default:
		out += in[index];
		break;
	}
	return true;
}

// index points at the first char after the opening quote. Stops at the closing quote (consumed), at newline (consumed) or at end of input.
bool readStringBody(const QString &in, int &index, QChar quote, QString &out)
{
	while (index < in.length()) {
		QChar c = in[index];
		if (c == quote || c == '\n') {
			++index;
			return true;
		}
		if (c == '\\') {
			++index;
			if (index == in.length()) return true;
			if (!decodeEscape(in, index, out)) return false;
		}
		else {
			out += c;
		}
		++index;
	}
	return true;
}

}

QString ConvertLangToSqlString(const QString &in)
{
	// Assume mostly C++ for now but allow some other things like
	// - single quotes (php)
	// - concattenation operators . (php) and + (java)

	// Allow cpp prefixes L u8 u U

	// Parser flow, we start in whitespace state and search for prefix|opening quote
	// parse string and process escapes
	// if escape is \r strip it if \n go to new line
	// until we reach matching end quote
	// skip whitespace and + or .

	QString output;
	int index = 0;
	QChar quote = '\0';
	while (true) {
		// skip all whitespace untill we encounter something else
		// we also skip concatenation operators. Note this code is not trying to validate
		// for correct syntax so it will quite happily accept many incorrect constructs
		// that doesn't matter however as we are just trying to strip everything which is not SQL.
		while (index < in.length() && (in[index].isSpace() || in[index] == '+' || in[index] == '.')) ++index;
		if (index == in.length()) break;
		if (!readOpeningQuote(in, index, quote)) break;
		if (!readStringBody(in, index, quote, output)) break;
	}

	return output;
}