Add an option to ignore whitespace for the character counter

aegisub/src/libresrc/default_config.json

@@ -350,6 +350,9 @@
 	},
 
 	"Subtitle" : {
+		"Character Counter" : {
+			"Ignore Whitespace" : false
+		},
 		"Character Limit" : 40,
 		"Default Resolution" : {
 			"Auto" : true,

aegisub/src/libresrc/osx/default_config.json

@@ -350,6 +350,9 @@
 	},
 
 	"Subtitle" : {
+		"Character Counter" : {
+			"Ignore Whitespace" : false
+		},
 		"Character Limit" : 40,
 		"Default Resolution" : {
 			"Auto" : true,

aegisub/src/preferences.cpp

@@ -18,24 +18,6 @@
 
 #include "config.h"
 
-#include <iterator>
-
-#include <wx/checkbox.h>
-#include <wx/combobox.h>
-#include <wx/event.h>
-#include <wx/filefn.h>
-#include <wx/listctrl.h>
-#include <wx/msgdlg.h>
-#include <wx/srchctrl.h>
-#include <wx/sizer.h>
-#include <wx/spinctrl.h>
-#include <wx/stattext.h>
-#include <wx/treebook.h>
-#include <wx/treebook.h>
-
-#include <libaegisub/exception.h>
-#include <libaegisub/hotkey.h>
-
 #include "preferences.h"
 
 #include "audio_renderer_waveform.h"
@@ -61,6 +43,23 @@
 #include <ffms.h>
 #endif
 
+#include <libaegisub/exception.h>
+#include <libaegisub/hotkey.h>
+
+#include <iterator>
+
+#include <wx/checkbox.h>
+#include <wx/combobox.h>
+#include <wx/event.h>
+#include <wx/filefn.h>
+#include <wx/listctrl.h>
+#include <wx/msgdlg.h>
+#include <wx/srchctrl.h>
+#include <wx/sizer.h>
+#include <wx/spinctrl.h>
+#include <wx/stattext.h>
+#include <wx/treebook.h>
+
 #define CLASS_PAGE(name)                             \
 class name: public OptionPage {                  \
 public:                                          \
@@ -192,7 +191,10 @@ Interface::Interface(wxTreebook *book, Preferences *parent): OptionPage(book, pa
 	OptionAdd(edit_box, _("Enable syntax highlighting"), "Subtitle/Highlight/Syntax");
 	OptionBrowse(edit_box, _("Dictionaries path"), "Path/Dictionary");
 	OptionFont(edit_box, "Subtitle/Edit Box/");
-	OptionAdd(edit_box, _("Maximum characters per line"), "Subtitle/Character Limit", 0, 1000);
+
+	wxFlexGridSizer *character_count = PageSizer(_("Character Counter"));
+	OptionAdd(character_count, _("Maximum characters per line"), "Subtitle/Character Limit", 0, 1000);
+	OptionAdd(character_count, _("Ignore whitespace"), "Subtitle/Character Counter/Ignore Whitespace");
 
 	wxFlexGridSizer *grid = PageSizer(_("Grid"));
 	OptionAdd(grid, _("Focus grid on click"), "Subtitle/Grid/Focus Allow");

aegisub/src/subs_edit_box.cpp

@@ -577,8 +577,9 @@ void SubsEditBox::CallCommand(const char *cmd_name) {
 }
 
 void SubsEditBox::UpdateCharacterCount(std::string const& text) {
+	auto ignore_whitespace = OPT_GET("Subtitle/Character Counter/Ignore Whitespace")->GetBool();
 	agi::dispatch::Background().Async([=]{
-		size_t length = MaxLineLength(text);
+		size_t length = MaxLineLength(text, ignore_whitespace);
 		agi::dispatch::Main().Async([=]{
 			char_count->SetValue(wxString::Format("%lu", (unsigned long)length));
 			size_t limit = (size_t)OPT_GET("Subtitle/Character Limit")->GetInt();

aegisub/src/utils.cpp

@@ -54,6 +54,8 @@
 #include <boost/locale/boundary.hpp>
 #include <boost/range/algorithm_ext.hpp>
 #include <map>
+#include <unicode/uchar.h>
+#include <unicode/utf8.h>
 
 #include <wx/clipbrd.h>
 #include <wx/filedlg.h>
@@ -222,7 +224,7 @@ void CleanCache(agi::fs::path const& directory, std::string const& file_type, ui
 	});
 }
 
-size_t MaxLineLength(std::string const& text) {
+size_t MaxLineLength(std::string const& text, bool ignore_whitespace) {
 	auto tokens = agi::ass::TokenizeDialogueBody(text);
 	agi::ass::MarkDrawings(text, tokens);
 
@@ -231,8 +233,10 @@ size_t MaxLineLength(std::string const& text) {
 	size_t current_line_length = 0;
 	for (auto token : tokens) {
 		if (token.type == agi::ass::DialogueTokenType::LINE_BREAK) {
-			if (text[pos + 1] == 'h')
+			if (text[pos + 1] == 'h') {
+				if (!ignore_whitespace)
 					current_line_length += 1;
+			}
 			else { // N or n
 				max_line_length = std::max(max_line_length, current_line_length);
 				current_line_length = 0;
@@ -241,7 +245,18 @@ size_t MaxLineLength(std::string const& text) {
 		else if (token.type == agi::ass::DialogueTokenType::TEXT) {
 			using namespace boost::locale::boundary;
 			const ssegment_index characters(character, begin(text) + pos, begin(text) + pos + token.length);
+			if (!ignore_whitespace)
 				current_line_length += boost::distance(characters);
+			else {
+				// characters.rule(word_any) doesn't seem to work for character indexes (everything is word_none)
+				for (auto const& chr : characters) {
+					UChar32 c;
+					int i = 0;
+					U8_NEXT_UNSAFE(chr.begin(), i, c);
+					if (!u_isUWhiteSpace(c))
+						++current_line_length;
+				}
+			}
 		}
 
 		pos += token.length;

aegisub/src/utils.h

@@ -62,7 +62,7 @@ void StatusTimeout(wxString const& msg, int ms = 10000);
 int SmallestPowerOf2(int x);
 
 /// Get the length in characters of the longest line in the given text
-size_t MaxLineLength(std::string const& text);
+size_t MaxLineLength(std::string const& text, bool ignore_whitespace);
 
 /// @brief Launch a new copy of Aegisub.
 ///