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

Added global modifier option

Browse Source
This commit is contained in:
Blu3train 2021-10-29 09:25:54 +02:00 committed by Akash Mozumdar
parent 06b93e3def
commit 2316f59415
2 changed files with 38 additions and 11 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

32
extensions/regexreplacer.cpp
View File

@ -17,7 +17,7 @@ BOOL WINAPI DllMain(HMODULE hModule, DWORD ul_reason_for_call, LPVOID lpReserved
case DLL_PROCESS_ATTACH:
{
if (!std::ifstream(REGEX_REPLACEMENTS_SAVE_FILE).good())
{
{
auto file = std::ofstream(REGEX_REPLACEMENTS_SAVE_FILE, std::ios::binary) << "\xff\xfe";
for (auto ch : std::wstring_view(REGEX_REPLACER_INSTRUCTIONS))
file << (ch == L'\n' ? std::string_view("\r\0\n", 4) : std::string_view((char*)&ch, 2));
@ -36,16 +36,38 @@ BOOL WINAPI DllMain(HMODULE hModule, DWORD ul_reason_for_call, LPVOID lpReserved
bool ProcessSentence(std::wstring& sentence, SentenceInfo sentenceInfo)
{
if (!sentenceInfo["current select"] || sentenceInfo["text number"] == 0) return false;
std::regex::flag_type mod;
std::regex_constants::match_flag_type flag;
std::ifstream stream(REGEX_REPLACEMENTS_SAVE_FILE, std::ios::binary);
BlockMarkupIterator savedFilters(stream, Array<std::wstring_view>{ L"|REGEX|", L"|BECOMES|", L"|MODIFIER|" });
concurrency::reader_writer_lock::scoped_lock_read readLock(m);
while (auto read = savedFilters.Next()) {
while (auto read = savedFilters.Next()) {
const auto& [regex, replacement, modifier] = read.value();
std::wregex regexp(regex, (modifier == L"i") ? std::regex::ECMAScript | std::regex::icase : std::regex::ECMAScript);
try { ::regex = regexp; }
if (modifier == L"g")
{
mod = std::regex::ECMAScript;
flag = std::regex_constants::format_default;
}
else if (modifier == L"gi" || modifier == L"ig")
{
mod = std::regex::icase;
flag = std::regex_constants::format_default;
}
else if (modifier == L"i")
{
mod = std::regex::icase;
flag = std::regex_constants::format_first_only;
}
else
{
mod = std::regex::ECMAScript;
flag = std::regex_constants::format_first_only;
}
try { ::regex = std::wregex(regex, mod); }
catch (std::regex_error) { continue; }
sentence = std::regex_replace(sentence, ::regex.value(), replacement);
sentence = std::regex_replace(sentence, ::regex.value(), replacement, flag);
}
return true;
}

17
text.cpp
View File

@ -226,9 +226,11 @@ Whitespace in original_text is ignored, but replacement_text can contain spaces,
This file must be encoded in Unicode (UTF-16 Little Endian).)";
const wchar_t* REGEX_REPLACER_INSTRUCTIONS = LR"(This file only does anything when the "Regex Replacer" extension is used.
Replacement commands must be formatted like this:
|REGEX|regular_expression|BECOMES|replacement_text|MODIFIER|i|END|
If the "MODIFIER" parameter is set to "i" the replacement ignores the case.
Any other value is not considered and the replacement is case sensitive.
|REGEX|regular_expression|BECOMES|replacement_text|MODIFIER|modifiers|END|
The "MODIFIER" parameter can contain the following modifiers:
"g" the replacement is global.
"i" the replacement ignores the case.
If the modifier is empty the replacement is only the first match and case sensitive.
All text in this file outside of a replacement command is ignored.
This file must be encoded in Unicode (UTF-16 Little Endian).
Learn, build, & test Regular Expressions: https://regexr.com/)";
@ -941,9 +943,12 @@ La spaziatura nel testo_originale è ignorato, ma testo_sostituito può contener
Questo file deve essere codificato in Unicode (UTF-16 Little Endian).)";
REGEX_REPLACER_INSTRUCTIONS = LR"(Questo file fa qualcosa solo quando l'estenzione "Regex Replacer" è utilizzata.
I comandi di sostituzione devono essere formattati cosi:
|REGEX|espressione_regolare|BECOMES|testo_sostituito|MODIFIER|i|END|
Se il parametro "MODIFIER" è impostato a "i" la sostituzione ignora maiuscole/minuscole
Qualsiasi altro valore non è considerato e la sostituzione è case sensitive.
|REGEX|espressione_regolare|BECOMES|testo_sostituito|MODIFIER|modificatori|END|
Il parametro "MODIFIER" può contenere i seguenti modificatori:
"g" la sostituzione è globale.
"i" la sostituzione ignora maiuscole/minuscole.
Se il modificatore è vuoto, la sostituzione viene applicata alla sola prima corrispondenza
e fa distinzione tra maiuscole e minuscole.
Tutto il testo in questo file all'infuori di un comando di sostituzione è ignorato.
Questo file deve essere codificato in Unicode (UTF-16 Little Endian).
Apprendere, creare e testare Espressioni Regolari: https://regexr.com/)";