From 54c1b508d5b63a76a0da8a583ea593bcccad8edb Mon Sep 17 00:00:00 2001
From: Akash Mozumdar <akashmozumdar@gmail.com>
Date: Fri, 28 Dec 2018 11:57:06 -0500
Subject: [PATCH] upgrade repetition remover and add tests

---
 extensions/CMakeLists.txt      |  2 ++
 extensions/extensiontester.cpp | 11 +++++++
 extensions/removerepeat.cpp    | 59 ++++++++++++++++++++++++++--------
 3 files changed, 58 insertions(+), 14 deletions(-)
 create mode 100644 extensions/extensiontester.cpp

diff --git a/extensions/CMakeLists.txt b/extensions/CMakeLists.txt
index aacfaa8..a5c1276 100644
--- a/extensions/CMakeLists.txt
+++ b/extensions/CMakeLists.txt
@@ -12,6 +12,8 @@ add_library(Google\ Translate SHARED googletranslate.cpp extensionimpl.cpp)
 add_library(Regex\ Filter SHARED regexfilter.cpp extensionimpl.cpp)
 add_library(Remove\ Repetition SHARED removerepeat.cpp extensionimpl.cpp)
 
+add_executable(Extension_Tests extensiontester.cpp)
+
 target_link_libraries(Bing\ Translate winhttp Qt5::Widgets)
 target_link_libraries(Extra\ Window Qt5::Widgets)
 target_link_libraries(Google\ Translate winhttp Qt5::Widgets)
diff --git a/extensions/extensiontester.cpp b/extensions/extensiontester.cpp
new file mode 100644
index 0000000..c84f3a9
--- /dev/null
+++ b/extensions/extensiontester.cpp
@@ -0,0 +1,11 @@
+#include "common.h"
+#include <filesystem>
+
+int main()
+{
+	wchar_t path[MAX_PATH] = {};
+	GetModuleFileNameW(NULL, path, MAX_PATH);
+	*(wcsrchr(path, L'\\') + 1) = 0;
+	for (auto file : std::filesystem::directory_iterator(path))
+		if (file.path().extension() == L".dll") LoadLibraryW(file.path().c_str());
+}
diff --git a/extensions/removerepeat.cpp b/extensions/removerepeat.cpp
index 5544887..10fdd45 100644
--- a/extensions/removerepeat.cpp
+++ b/extensions/removerepeat.cpp
@@ -1,22 +1,34 @@
 #include "extension.h"
+#include "defs.h"
 
 void RemoveRepeatedChars(std::wstring& sentence)
 {
-	int repeatNumber = 0;
-	wchar_t prevChar = sentence[0];
-	for (auto c : sentence)
-		if (c == prevChar) repeatNumber++;
-		else break;
-	if (repeatNumber == 1) return;
+	std::vector<int> repeatNumbers(sentence.size() + 1, 0);
+	int repeatNumber = 1;
+	wchar_t prevChar = L'\0';
+	for (auto nextChar : sentence)
+		if (nextChar == prevChar) repeatNumber++;
+		else
+		{
+			prevChar = nextChar;
+			++repeatNumbers.at(repeatNumber);
+			repeatNumber = 1;
+		}
+	if ((repeatNumber = std::distance(repeatNumbers.begin(), std::max_element(repeatNumbers.begin(), repeatNumbers.end()))) == 1) return;
 
-	for (int i = 0; i < sentence.size(); i += repeatNumber)
-		for (int j = i; j < sentence.size(); ++j)
-			if (sentence[j] != sentence[i])
-				if ((j - i) % repeatNumber != 0) return;
-				else break;
-
-	std::wstring newSentence = L"";
-	for (int i = 0; i < sentence.size(); i += repeatNumber) newSentence.push_back(sentence[i]);
+	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;
+			}
+		}
+	}
 	sentence = newSentence;
 }
 
@@ -38,3 +50,22 @@ bool ProcessSentence(std::wstring& sentence, SentenceInfo sentenceInfo)
 	RemoveCyclicRepeats(sentence);
 	return true;
 }
+
+TEST(
+	{
+		std::wstring repeatedChars = L"aaaaaaaaaaaabbbbbbcccdddaabbbcccddd";
+		RemoveRepeatedChars(repeatedChars);
+		assert(repeatedChars.find(L"aaaabbcd") == 0);
+
+		std::wstring cyclicRepeats = L"abcdeabcdefabcdefgabcdefgabcdefgabcdefgabcdefg";
+		RemoveCyclicRepeats(cyclicRepeats);
+		assert(cyclicRepeats == L"abcdefg");
+
+		InfoForExtension tester{ "hook address", 0, nullptr };
+		std::wstring empty = L"", one = L" ", normal = L"This is a normal sentence. はい。";
+		ProcessSentence(empty, { &tester });
+		ProcessSentence(one, { &tester });
+		ProcessSentence(normal, { &tester });
+		assert(empty == L"" && one == L" " && normal == L"This is a normal sentence. はい。");
+	}
+);