chenx221/Textractor
1
0
Fork 0
forked from Public-Mirror/Textractor
Code Pull Requests Activity

improve repetition detection

Browse Source
This commit is contained in:
Akash Mozumdar 2018-09-13 02:50:03 -04:00
parent 20956f6412
commit 329d56c969
1 changed files with 18 additions and 17 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

35
extensions/removerepeat.cpp
View File

@ -1,7 +1,6 @@
#include "extensions.h" #include "extensions.h"
#include <set> #include <set>
#include <mutex> #include <mutex>
#include <algorithm>
bool RemoveRepeatedChars(std::wstring& sentence) bool RemoveRepeatedChars(std::wstring& sentence)
{ {
@ -18,44 +17,46 @@ bool RemoveRepeatedChars(std::wstring& sentence)
if ((j - i) % repeatNumber != 0) return false; if ((j - i) % repeatNumber != 0) return false;
else break; else break;
// Removes every repeatNumber'th character. std::wstring newSentence = L"";
sentence.erase(std::remove_if(sentence.begin(), sentence.end(), [&](const wchar_t& c) {return (&c - &*sentence.begin()) % repeatNumber != 0; }), sentence.end()); for (int i = 0; i < sentence.size(); ++i) if (i % repeatNumber == 0) newSentence.push_back(sentence[i]);
sentence = newSentence;
return true; return true;
} }
bool RemoveCyclicRepeats(std::wstring& sentence) bool RemoveCyclicRepeats(std::wstring& sentence)
{ {
unsigned int realLength = 6; // If the first 6 characters appear later on, there's probably a repetition issue. unsigned int junkLength = 0;
if (sentence.size() < realLength) return false; wchar_t junk[2000] = {};
wchar_t realSentence[2000] = {}; while (wcsstr(sentence.c_str() + junkLength, junk))
memcpy(realSentence, sentence.c_str(), realLength * sizeof(wchar_t));
while (wcsstr(sentence.c_str() + realLength, realSentence))
{ {
realSentence[realLength] = sentence[realLength]; junk[junkLength] = sentence[junkLength];
if (++realLength >= 2000) return false; if (++junkLength >= 2000) return false;
} }
if (realLength > 7) if (--junkLength >= 5) // If the first 5 characters appear later on, there's probably a repetition issue.
{ {
sentence = std::wstring(realSentence); sentence = std::wstring(sentence.c_str() + junkLength);
sentence.pop_back();
RemoveCyclicRepeats(sentence); RemoveCyclicRepeats(sentence);
return true; return true;
} }
return false; return false;
} }
bool RemoveRepeatedSentences(std::wstring& sentence, int threadHandle) bool RemoveRepeatedSentences(std::wstring& sentence, int handle)
{ {
static std::set<std::pair<int, std::wstring>> seenSentences; static std::set<std::pair<int, std::wstring>> seenSentences;
static std::mutex m; static std::mutex m;
std::lock_guard<std::mutex> l(m); std::lock_guard<std::mutex> l(m);
if (seenSentences.count({ threadHandle, sentence }) != 0) throw std::exception(); if (seenSentences.count({ handle, sentence }) != 0) throw std::exception();
seenSentences.insert({ threadHandle, sentence }); seenSentences.insert({ handle, sentence });
return false; return false;
} }
bool ProcessSentence(std::wstring& sentence, const InfoForExtension* miscInfo) bool ProcessSentence(std::wstring& sentence, const InfoForExtension* miscInfo)
{ {
if (GetProperty("hook address", miscInfo) == -1) return false; if (GetProperty("hook address", miscInfo) == -1) return false;
return RemoveRepeatedChars(sentence) | RemoveCyclicRepeats(sentence) | RemoveRepeatedSentences(sentence, GetProperty("text handle", miscInfo)); bool ret = false;
ret |= RemoveRepeatedChars(sentence);
ret |= RemoveCyclicRepeats(sentence);
ret |= RemoveRepeatedSentences(sentence, GetProperty("text handle", miscInfo));
return ret;
} }