Speed up numeric sorting by 10x.

Convert strings to int, sort, then convert back to strings.
This commit is contained in:
Andreas Jönsson 2015-05-09 13:20:37 +02:00
parent bcbe48b13f
commit e258bcb3a7
3 changed files with 65 additions and 36 deletions

View File

@ -749,11 +749,11 @@ generic_string stringJoin(const std::vector<generic_string> &strings, const gene
return joined;
}
int stoi_CountEmptyLinesAsMinimum(const generic_string &input)
int stoiStrict(const generic_string &input)
{
if (input.empty())
{
return INT_MIN;
throw std::invalid_argument("Empty input.");
}
else
{
@ -778,14 +778,14 @@ int stoi_CountEmptyLinesAsMinimum(const generic_string &input)
}
}
bool allLinesAreNumericOrEmpty(const std::vector<generic_string> &lines)
bool allLinesAreNumeric(const std::vector<generic_string> &lines)
{
const auto endit = lines.end();
for (auto it = lines.begin(); it != endit; ++it)
{
try
{
stoi_CountEmptyLinesAsMinimum(*it);
stoiStrict(*it);
}
catch (std::invalid_argument&)
{
@ -798,3 +798,49 @@ bool allLinesAreNumericOrEmpty(const std::vector<generic_string> &lines)
}
return true;
}
std::vector<generic_string> lexicographicSort(std::vector<generic_string> input, bool isDescending)
{
std::sort(input.begin(), input.end(), [isDescending](generic_string a, generic_string b)
{
if (isDescending)
{
return a.compare(b) > 0;
}
else
{
return a.compare(b) < 0;
}
});
return input;
}
std::vector<generic_string> numericSort(std::vector<generic_string> input, bool isDescending)
{
// Pre-condition: all strings in "input" are convertible to int with stoiStrict.
std::vector<int> inputAsInts;
std::map<int, generic_string> intToString; // Cache for ints converted to strings.
for (auto it = input.begin(); it != input.end(); ++it)
{
int converted = stoiStrict(*it);
inputAsInts.push_back(converted);
intToString[converted] = *it;
}
std::sort(inputAsInts.begin(), inputAsInts.end(), [isDescending](int a, int b)
{
if (isDescending)
{
return a > b;
}
else
{
return a < b;
}
});
std::vector<generic_string> output;
for (auto it = inputAsInts.begin(); it != inputAsInts.end(); ++it)
{
output.push_back(intToString[*it]);
}
return output;
}

View File

@ -190,7 +190,10 @@ generic_string stringToUpper(generic_string strToConvert);
generic_string stringReplace(generic_string subject, const generic_string& search, const generic_string& replace);
std::vector<generic_string> stringSplit(const generic_string& input, const generic_string &delimiter);
generic_string stringJoin(const std::vector<generic_string> &strings, const generic_string &separator);
int stoi_CountEmptyLinesAsMinimum(const generic_string &input);
bool allLinesAreNumericOrEmpty(const std::vector<generic_string> &lines);
int stoiStrict(const generic_string &input);
bool allLinesAreNumeric(const std::vector<generic_string> &lines);
std::vector<generic_string> numericSort(std::vector<generic_string> input, bool isDescending);
std::vector<generic_string> lexicographicSort(std::vector<generic_string> input, bool isDescending);
#endif //M30_IDE_COMMUN_H

View File

@ -2979,37 +2979,17 @@ void ScintillaEditView::sortLines(size_t fromLine, size_t toLine, bool isDescend
}
}
assert(toLine - fromLine + 1 == splitText.size());
const bool isNumericSort = allLinesAreNumericOrEmpty(splitText);
std::sort(splitText.begin(), splitText.end(), [isDescending, isNumericSort](generic_string a, generic_string b)
const bool isNumericSort = allLinesAreNumeric(splitText);
std::vector<generic_string> sortedText;
if (isNumericSort)
{
if (isDescending)
{
if (isNumericSort)
{
int numA = stoi_CountEmptyLinesAsMinimum(a);
int numB = stoi_CountEmptyLinesAsMinimum(b);
return numA > numB;
}
else
{
return a.compare(b) > 0;
}
}
else
{
if (isNumericSort)
{
int numA = stoi_CountEmptyLinesAsMinimum(a);
int numB = stoi_CountEmptyLinesAsMinimum(b);
return numA < numB;
}
else
{
return a.compare(b) < 0;
}
}
});
const generic_string joined = stringJoin(splitText, getEOLString());
sortedText = numericSort(splitText, isDescending);
}
else
{
sortedText = lexicographicSort(splitText, isDescending);
}
const generic_string joined = stringJoin(sortedText, getEOLString());
if (sortAllLines)
{
replaceTarget(joined.c_str(), startPos, endPos);