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

repetition filters bail after 30 seconds

Browse Source
This commit is contained in:
Akash Mozumdar 2019-10-07 01:00:54 -04:00
parent 9e86ecb987
commit 6a6c208c20
2 changed files with 4 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
extensions/removerepeatphrase.cpp
View File

@ -46,8 +46,9 @@ bool ProcessSentence(std::wstring& sentence, SentenceInfo sentenceInfo)
// This algorithm looks for repeating substrings (in other words, common prefixes among the set of suffixes) of the sentence with length > 6
// It then looks for any regions of characters at least twice as long as the substring made up only of characters in the substring, and erases them
// If this results in the substring being completely erased from the string, the substring is copied to the last location where it was located in the original string
auto timeout = GetTickCount64() + 30'000; // give up if taking over 30 seconds
std::vector<int> suffixArray = GenerateSuffixArray(sentence);
for (int i = 0; i + 1 < sentence.size(); ++i)
for (int i = 0; i + 1 < sentence.size() && GetTickCount64() < timeout; ++i)
{
int commonPrefixLength = 0;
for (int j = suffixArray[i], k = suffixArray[i + 1]; j < sentence.size() && k < sentence.size(); ++j, ++k)

5
extensions/removerepeatphrase2.cpp
View File

@ -5,13 +5,12 @@ bool ProcessSentence(std::wstring& sentence, SentenceInfo sentenceInfo)
if (sentenceInfo["text number"] == 0) return false;
// This algorithm looks at all the prefixes of the sentence: if a prefix is found later in the sentence, it is removed from the beginning and the process is repeated
// Complexity O(N^3) so executing for N > 10,000 dangerous
if (sentence.size() > 10000) return false;
auto timeout = GetTickCount64() + 30'000; // give up if taking over 30 seconds
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)
for (wchar_t* end = dataEnd; end - data.get() > skip && GetTickCount64() < timeout; --end)
{
std::swap(*end, *dataEnd);
int junkLength = end - data.get() - skip;