forked from Public-Mirror/Textractor
refactor replacer
This commit is contained in:
parent
c074bbf506
commit
5d83c9736b
@ -5,15 +5,18 @@
|
|||||||
#include <process.h>
|
#include <process.h>
|
||||||
|
|
||||||
extern const wchar_t* REPLACER_INSTRUCTIONS;
|
extern const wchar_t* REPLACER_INSTRUCTIONS;
|
||||||
|
|
||||||
constexpr auto REPLACE_SAVE_FILE = u8"SavedReplacements.txt";
|
constexpr auto REPLACE_SAVE_FILE = u8"SavedReplacements.txt";
|
||||||
|
|
||||||
std::atomic<std::filesystem::file_time_type> replaceFileLastWrite;
|
std::atomic<std::filesystem::file_time_type> replaceFileLastWrite = {};
|
||||||
std::shared_mutex m;
|
std::shared_mutex m;
|
||||||
|
|
||||||
struct
|
class Trie
|
||||||
{
|
{
|
||||||
public:
|
public:
|
||||||
void Put(std::wstring original, std::wstring replacement)
|
Trie(const std::unordered_map<std::wstring, std::wstring>& replacements)
|
||||||
|
{
|
||||||
|
for (const auto& [original, replacement] : replacements)
|
||||||
{
|
{
|
||||||
Node* current = &root;
|
Node* current = &root;
|
||||||
for (auto ch : original)
|
for (auto ch : original)
|
||||||
@ -22,27 +25,32 @@ public:
|
|||||||
else current = (next = std::make_unique<Node>()).get();
|
else current = (next = std::make_unique<Node>()).get();
|
||||||
if (current != &root) current->value = replacement;
|
if (current != &root) current->value = replacement;
|
||||||
}
|
}
|
||||||
|
}
|
||||||
|
|
||||||
std::pair<int, std::optional<std::wstring>> Lookup(const std::wstring& text) const
|
std::wstring Replace(const std::wstring& sentence) const
|
||||||
{
|
{
|
||||||
std::pair<int, std::optional<std::wstring>> result = {};
|
std::wstring result;
|
||||||
int length = 0;
|
for (int i = 0; i < sentence.size();)
|
||||||
|
{
|
||||||
|
std::wstring replacement(1, sentence[i]);
|
||||||
|
int originalLength = 1;
|
||||||
|
|
||||||
const Node* current = &root;
|
const Node* current = &root;
|
||||||
for (auto ch : text)
|
for (int j = i; j < sentence.size() + 1; ++j)
|
||||||
{
|
{
|
||||||
if (Ignore(ch))
|
if (current->value)
|
||||||
{
|
{
|
||||||
length += 1;
|
replacement = current->value.value();
|
||||||
}
|
originalLength = j - i;
|
||||||
else if (current->next.count(ch) != 0)
|
|
||||||
{
|
|
||||||
auto& next = current->next.at(ch);
|
|
||||||
length += 1;
|
|
||||||
current = next.get();
|
|
||||||
if (current->value) result = { length, current->value };
|
|
||||||
}
|
}
|
||||||
|
if (current->next.count(sentence[j]) > 0) current = current->next.at(sentence[j]).get();
|
||||||
|
else if (Ignore(sentence[j]));
|
||||||
else break;
|
else break;
|
||||||
}
|
}
|
||||||
|
|
||||||
|
result += replacement;
|
||||||
|
i += originalLength;
|
||||||
|
}
|
||||||
return result;
|
return result;
|
||||||
}
|
}
|
||||||
|
|
||||||
@ -57,55 +65,34 @@ private:
|
|||||||
std::unordered_map<wchar_t, std::unique_ptr<Node>> next;
|
std::unordered_map<wchar_t, std::unique_ptr<Node>> next;
|
||||||
std::optional<std::wstring> value;
|
std::optional<std::wstring> value;
|
||||||
} root;
|
} root;
|
||||||
} replacementTrie;
|
} trie = { {} };
|
||||||
|
|
||||||
int Parse(const std::wstring& file)
|
std::unordered_map<std::wstring, std::wstring> Parse(const std::wstring& replacementScript)
|
||||||
{
|
{
|
||||||
std::lock_guard l(m);
|
std::unordered_map<std::wstring, std::wstring> replacements;
|
||||||
replacementTrie = {};
|
|
||||||
int count = 0;
|
|
||||||
size_t end = 0;
|
size_t end = 0;
|
||||||
while (true)
|
while (true)
|
||||||
{
|
{
|
||||||
size_t original = file.find(L"|ORIG|", end);
|
size_t original = replacementScript.find(L"|ORIG|", end);
|
||||||
size_t becomes = file.find(L"|BECOMES|", original);
|
size_t becomes = replacementScript.find(L"|BECOMES|", original);
|
||||||
if ((end = file.find(L"|END|", becomes)) == std::wstring::npos) break;
|
if ((end = replacementScript.find(L"|END|", becomes)) == std::wstring::npos) break;
|
||||||
replacementTrie.Put(file.substr(original + 6, becomes - original - 6), file.substr(becomes + 9, end - becomes - 9));
|
replacements[replacementScript.substr(original + 6, becomes - original - 6)] = replacementScript.substr(becomes + 9, end - becomes - 9);
|
||||||
count += 1;
|
|
||||||
}
|
}
|
||||||
return count;
|
return replacements;
|
||||||
}
|
|
||||||
|
|
||||||
bool Replace(std::wstring& sentence)
|
|
||||||
{
|
|
||||||
std::shared_lock l(m);
|
|
||||||
for (int i = 0; i < sentence.size(); ++i)
|
|
||||||
{
|
|
||||||
auto [length, replacement] = replacementTrie.Lookup(sentence.substr(i));
|
|
||||||
if (replacement)
|
|
||||||
{
|
|
||||||
sentence.replace(i, length, replacement.value());
|
|
||||||
i += replacement.value().size() - 1; // iterate to end of replacement
|
|
||||||
}
|
|
||||||
}
|
|
||||||
return true;
|
|
||||||
}
|
}
|
||||||
|
|
||||||
BOOL WINAPI DllMain(HMODULE hModule, DWORD ul_reason_for_call, LPVOID lpReserved)
|
BOOL WINAPI DllMain(HMODULE hModule, DWORD ul_reason_for_call, LPVOID lpReserved)
|
||||||
{
|
{
|
||||||
// not actually used to read/write, just to ensure it exists
|
|
||||||
static AutoHandle<> replacementFile = CreateFileA(REPLACE_SAVE_FILE, FILE_GENERIC_READ, FILE_SHARE_READ | FILE_SHARE_WRITE, nullptr, OPEN_ALWAYS, FILE_ATTRIBUTE_NORMAL, NULL);
|
|
||||||
switch (ul_reason_for_call)
|
switch (ul_reason_for_call)
|
||||||
{
|
{
|
||||||
case DLL_PROCESS_ATTACH:
|
case DLL_PROCESS_ATTACH:
|
||||||
{
|
{
|
||||||
std::vector<BYTE> file(std::istreambuf_iterator<char>(std::ifstream(REPLACE_SAVE_FILE, std::ios::in | std::ios::binary)), {});
|
std::vector<BYTE> file(std::istreambuf_iterator<char>(std::ifstream(REPLACE_SAVE_FILE, std::ios::in | std::ios::binary)), {});
|
||||||
if (Parse(std::wstring((wchar_t*)file.data(), file.size() / sizeof(wchar_t))) == 0)
|
if (Parse(std::wstring((wchar_t*)file.data(), file.size() / sizeof(wchar_t))).empty())
|
||||||
{
|
{
|
||||||
std::ofstream(REPLACE_SAVE_FILE, std::ios::out | std::ios::binary | std::ios::trunc).write((char*)REPLACER_INSTRUCTIONS, wcslen(REPLACER_INSTRUCTIONS) * sizeof(wchar_t));
|
std::ofstream(REPLACE_SAVE_FILE, std::ios::out | std::ios::binary | std::ios::trunc).write((char*)REPLACER_INSTRUCTIONS, wcslen(REPLACER_INSTRUCTIONS) * sizeof(wchar_t));
|
||||||
_spawnlp(_P_DETACH, "notepad", "notepad", REPLACE_SAVE_FILE, NULL); // show file to user
|
_spawnlp(_P_DETACH, "notepad", "notepad", REPLACE_SAVE_FILE, NULL); // show file to user
|
||||||
}
|
}
|
||||||
replaceFileLastWrite = std::filesystem::last_write_time(REPLACE_SAVE_FILE);
|
|
||||||
}
|
}
|
||||||
break;
|
break;
|
||||||
case DLL_PROCESS_DETACH:
|
case DLL_PROCESS_DETACH:
|
||||||
@ -117,31 +104,35 @@ BOOL WINAPI DllMain(HMODULE hModule, DWORD ul_reason_for_call, LPVOID lpReserved
|
|||||||
}
|
}
|
||||||
|
|
||||||
bool ProcessSentence(std::wstring& sentence, SentenceInfo)
|
bool ProcessSentence(std::wstring& sentence, SentenceInfo)
|
||||||
|
{
|
||||||
|
try
|
||||||
{
|
{
|
||||||
static_assert(std::has_unique_object_representations_v<decltype(replaceFileLastWrite)::value_type>);
|
static_assert(std::has_unique_object_representations_v<decltype(replaceFileLastWrite)::value_type>);
|
||||||
if (!replaceFileLastWrite.compare_exchange_strong(std::filesystem::last_write_time(REPLACE_SAVE_FILE), std::filesystem::last_write_time(REPLACE_SAVE_FILE)))
|
if (replaceFileLastWrite.exchange(std::filesystem::last_write_time(REPLACE_SAVE_FILE)) != std::filesystem::last_write_time(REPLACE_SAVE_FILE))
|
||||||
{
|
{
|
||||||
|
std::scoped_lock l(m);
|
||||||
std::vector<BYTE> file(std::istreambuf_iterator<char>(std::ifstream(REPLACE_SAVE_FILE, std::ios::in | std::ios::binary)), {});
|
std::vector<BYTE> file(std::istreambuf_iterator<char>(std::ifstream(REPLACE_SAVE_FILE, std::ios::in | std::ios::binary)), {});
|
||||||
Parse(std::wstring((wchar_t*)file.data(), file.size() / sizeof(wchar_t)));
|
trie = Trie(Parse(std::wstring((wchar_t*)file.data(), file.size() / sizeof(wchar_t))));
|
||||||
|
}
|
||||||
|
}
|
||||||
|
catch (std::filesystem::filesystem_error) {}
|
||||||
|
|
||||||
|
std::shared_lock l(m);
|
||||||
|
sentence = trie.Replace(sentence);
|
||||||
|
return true;
|
||||||
}
|
}
|
||||||
|
|
||||||
return Replace(sentence);
|
TEST(
|
||||||
}
|
|
||||||
|
|
||||||
TEST_SYNC(
|
|
||||||
{
|
{
|
||||||
assert(Parse(LR"(|ORIG| さよなら|BECOMES|goodbye|END|
|
auto replacements = Parse(LR"(
|
||||||
|
|ORIG|さよなら|BECOMES|goodbye |END|Ignore this text
|
||||||
|
And this text ツ
|
||||||
|ORIG|バカ|BECOMES|idiot|END|
|
|ORIG|バカ|BECOMES|idiot|END|
|
||||||
|ORIG|こんにちは |BECOMES|hello|END|)") == 3);
|
|ORIG|こんにちは |BECOMES| hello|END||ORIG|delete this|BECOMES||END|)");
|
||||||
std::wstring replaced = LR"(blahblah
|
assert(replacements.size() == 4);
|
||||||
さよなら バカ こんにちは)";
|
std::wstring original = LR"(Don't replace this
|
||||||
Replace(replaced);
|
さよなら バカ こんにちは delete this)";
|
||||||
assert(replaced.find(L"さよなら") == std::wstring::npos &&
|
std::wstring replaced = Trie(replacements).Replace(original);
|
||||||
replaced.find(L"バカ") == std::wstring::npos &&
|
assert(replaced == L"Don't replace thisgoodbye idiot hello");
|
||||||
replaced.find(L"こんにちは") == std::wstring::npos &&
|
|
||||||
replaced.find(L"goodbye") != std::wstring::npos &&
|
|
||||||
replaced.find(L"idiot") != std::wstring::npos &&
|
|
||||||
replaced.find(L"hello") != std::wstring::npos
|
|
||||||
);
|
|
||||||
}
|
}
|
||||||
);
|
);
|
||||||
|
Loading…
x
Reference in New Issue
Block a user