Textractor_test/extensions/removerepeat.cpp

91 lines
2.6 KiB
C++
Raw Normal View History

2019-01-24 03:48:09 +08:00
#include "extension.h"
void RemoveRepeatedChars(std::wstring& sentence)
{
std::vector<int> repeatNumbers(sentence.size() + 1, 0);
int repeatNumber = 1;
wchar_t prevChar = L'\0';
for (auto nextChar : sentence)
{
if (nextChar == prevChar)
{
repeatNumber += 1;
}
else
{
prevChar = nextChar;
2019-02-21 11:12:26 +08:00
repeatNumbers.at(repeatNumber) += 1;
repeatNumber = 1;
}
}
if ((repeatNumber = std::distance(repeatNumbers.begin(), std::max_element(repeatNumbers.begin(), repeatNumbers.end()))) == 1) return;
std::wstring newSentence;
for (int i = 0; i < sentence.size();)
{
newSentence.push_back(sentence.at(i));
for (int j = i; j <= sentence.size(); ++j)
{
if (j == sentence.size() || sentence.at(i) != sentence.at(j))
{
i += (j - i) % repeatNumber == 0 ? repeatNumber : 1;
break;
}
}
}
2018-09-13 14:50:03 +08:00
sentence = newSentence;
}
void RemoveCyclicRepeats(std::wstring& sentence)
{
if (sentence.size() > 15000) return; // this algorithm is O(N^3) so if N > 15000 it's extremely slow
2019-02-21 11:12:26 +08:00
auto data = std::make_unique<wchar_t[]>(sentence.size() + 1);
wcscpy_s(data.get(), sentence.size() + 1, sentence.c_str());
wchar_t* dataEnd = data.get() + sentence.size();
int skip = 0, count = 0;
for (wchar_t* end = dataEnd; end - data.get() > skip; --end)
{
std::swap(*end, *dataEnd);
int junkLength = end - data.get() - skip;
auto junkFound = wcsstr(sentence.c_str() + skip + junkLength, data.get() + skip);
std::swap(*end, *dataEnd);
if (junkFound)
2018-11-15 13:06:49 +08:00
{
if (count && junkLength < min(skip / count, 4)) break;
2019-02-21 11:12:26 +08:00
skip += junkLength;
count += 1;
end = dataEnd;
2018-11-15 13:06:49 +08:00
}
2019-02-21 11:12:26 +08:00
}
if (count && skip / count >= 3) sentence = data.get() + skip;
}
2018-09-23 03:08:31 +08:00
bool ProcessSentence(std::wstring& sentence, SentenceInfo sentenceInfo)
{
2019-02-18 08:14:49 +08:00
if (sentenceInfo["text number"] == 0) return false;
RemoveRepeatedChars(sentence);
RemoveCyclicRepeats(sentence);
return true;
}
TEST(
{
std::wstring repeatedChars = L"aaaaaaaaaaaabbbbbbcccdddaabbbcccddd";
RemoveRepeatedChars(repeatedChars);
assert(repeatedChars.find(L"aaaabbcd") == 0);
std::wstring cyclicRepeats = L"_abcde_abcdef_abcdefg_abcdefg_abcdefg_abcdefg_abcdefg";
std::wstring buildupRepeats = L"__a_ab_abc_abcd_abcde_abcdef_abcdefg";
RemoveCyclicRepeats(cyclicRepeats);
2019-02-21 11:12:26 +08:00
RemoveCyclicRepeats(buildupRepeats);
assert(cyclicRepeats == L"_abcdefg");
assert(buildupRepeats == L"_abcdefg");
2019-01-24 03:48:09 +08:00
std::wstring empty = L"", one = L" ", normal = L"This is a normal sentence. はい";
2019-02-13 12:54:33 +08:00
ProcessSentence(empty, { SentenceInfo::DUMMY });
ProcessSentence(one, { SentenceInfo::DUMMY });
ProcessSentence(normal, { SentenceInfo::DUMMY });
2019-01-24 05:11:14 +08:00
assert(empty == L"" && one == L" " && normal == L"This is a normal sentence. はい");
}
);