chenx221/Textractor_test
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

add cyclic repetition detection in textthread

Browse Source
This commit is contained in:
Akash Mozumdar 2018-10-17 00:42:51 -04:00
parent abba2d77e3
commit 3ad311293f
2 changed files with 9 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
GUI/host/textthread.cc
View File

@ -6,6 +6,7 @@
#include "host.h"
#include "const.h"
#include <regex>
#include <algorithm>
TextThread::TextThread(ThreadParam tp, DWORD status) : handle(threadCounter++), name(Host::GetHookName(tp.pid, tp.hook)), tp(tp), status(status) {}
@ -31,6 +32,11 @@ void TextThread::Flush()
if (buffer.size() < maxBufferSize && GetTickCount() - timestamp < flushDelay) return;
sentence = buffer;
buffer.clear();
bool hasRepetition = false;
for (std::wsmatch results; std::regex_search(sentence, results, std::wregex(L"([^\\x00]{6,})\\1\\1")); hasRepetition = true) sentence = results[1];
if (hasRepetition) repeatingChars = std::unordered_set<wchar_t>(sentence.begin(), sentence.end());
else repeatingChars.clear();
}
AddSentence(sentence);
}
@ -51,6 +57,7 @@ void TextThread::AddText(const BYTE* data, int len)
buffer += status & USING_UNICODE
? std::wstring((wchar_t*)data, len / 2)
: StringToWideString(std::string((char*)data, len), status & USING_UTF8 ? CP_UTF8 : SHIFT_JIS);
if (std::all_of(buffer.begin(), buffer.end(), [&](wchar_t c) { return repeatingChars.count(c) > 0; })) buffer.clear();
timestamp = GetTickCount();
}

2
GUI/host/textthread.h
View File

@ -6,6 +6,7 @@
#include "common.h"
#include "types.h"
#include <unordered_set>
class TextThread
{
@ -34,6 +35,7 @@ private:
std::wstring buffer;
std::wstring storage;
std::unordered_set<wchar_t> repeatingChars;
std::recursive_mutex ttMutex;
DWORD status;