Fix #884, annoying wxString::Trim is destructive and there's no non-destructive variant. Adding a new function to utils.cpp to test whether a string is "blank", ie. empty or whitespace-only, as well as a function to check if a wchar_t is a whitspace character.

Trimming was used to test whether a string was blank or not, and this caused the source syllable texts to be altered when the kanji interpolation algorithm was run.

Originally committed to SVN as r3085.

aegisub/src/auto4_lua_assfile.cpp

@@ -42,6 +42,7 @@
 #include "ass_style.h"
 #include "ass_file.h"
 #include "ass_override.h"
+#include "utils.h"
 
 #ifdef __WINDOWS__
 #include "../../contrib/lua51/src/lualib.h"
@@ -78,7 +79,7 @@ namespace Automation4 {
 		lua_pushstring(L, raw.mb_str(wxConvUTF8));
 		lua_setfield(L, -2, "raw");
 
-		if (raw.Trim().IsEmpty()) {
+		if (StringEmptyOrWhitespace(raw)) {
 			lua_pushstring(L, "clear");
 
 		} else if (raw[0] == _T(';')) {

aegisub/src/auto4_ruby_assfile.cpp

@@ -46,6 +46,7 @@
 #include "ass_style.h"
 #include "ass_file.h"
 #include "ass_override.h"
+#include "utils.h"
 #include <assert.h>
 #include <algorithm>
 #include <ruby.h>
@@ -67,7 +68,7 @@ namespace Automation4 {
 			rb_hash_aset(ass_entry, STR2SYM("raw"), rb_str_new2(e->GetEntryData().mb_str(wxConvUTF8)));
 		VALUE entry_class;
 
-		if (raw.Trim().IsEmpty()) {
+		if (StringEmptyOrWhitespace(raw)) {
 			entry_class = STR2SYM("clear");
 
 		} else if (raw[0] == _T(';')) {

aegisub/src/dialog_kanji_timer.cpp

@@ -64,11 +64,11 @@ class KaraokeLineMatchDisplay : public wxControl {
 	struct MatchSyllable {
 		int dur;
 		wxString text;
-		MatchSyllable() : dur(0) { }
+		MatchSyllable(int _dur, const wxString &_text) : dur(_dur), text(_text) { }
 	};
 	struct MatchGroup {
-		std::vector<MatchSyllable> src;
+		std::vector<const MatchSyllable> src;
-		typedef std::vector<MatchSyllable>::iterator SrcIterator;
+		typedef std::vector<const MatchSyllable>::iterator SrcIterator;
 		wxString dst;
 		int duration;
 		int last_render_width;
@@ -79,8 +79,8 @@ class KaraokeLineMatchDisplay : public wxControl {
 	std::vector<MatchGroup> matched_groups;
 	typedef std::vector<MatchGroup>::iterator MatchedGroupIterator;
 	// Unmatched source syllables
-	std::deque<MatchSyllable> unmatched_source;
+	std::deque<const MatchSyllable> unmatched_source;
-	typedef std::deque<MatchSyllable>::iterator UnmatchedSourceIterator;
+	typedef std::deque<const MatchSyllable>::iterator UnmatchedSourceIterator;
 	// Unmatched destination text
 	wxString unmatched_destination;
 
@@ -374,10 +374,7 @@ void KaraokeLineMatchDisplay::SetInputData(const AssDialogue *src, const AssDial
 		// Start from 1 instead of 0: The first syllable is actually everything before the first
 		for (size_t i = 1; i < kara.size(); ++i)
 		{
-			MatchSyllable syl;
+			unmatched_source.push_back(MatchSyllable(kara[i].duration, kara[i].text));
-			syl.text = kara[i].text;
-			syl.dur = kara[i].duration;
-			unmatched_source.push_back(syl);
 		}
 		delete varsrc;
 	}
@@ -521,7 +518,7 @@ void KaraokeLineMatchDisplay::AutoMatchJapanese()
 
 	// Now the source syllable might consist of just whitespace.
 	// Eat all whitespace at the start of the destination.
-	if (src.Trim().size() == 0)
+	if (StringEmptyOrWhitespace(src))
 	{
 trycatchingmorespaces:
 		// ASCII space
@@ -594,7 +591,7 @@ trycatchingmorespaces:
 				// Yay! Time to interpolate.
 				// Special case: If the last source syllable before the matching one is
 				// empty or contains just whitespace, don't include that one.
-				if (src_lookahead_pos > 1 && unmatched_source[src_lookahead_pos-2].text.Trim().size() == 0)
+				if (src_lookahead_pos > 1 && StringEmptyOrWhitespace(unmatched_source[src_lookahead_pos-2].text))
 					src_lookahead_pos -= 1;
 				// Special case: Just one source syllable matching, pick all destination found
 				if (src_lookahead_pos == 2)

aegisub/src/utils.cpp

@@ -303,6 +303,39 @@ void GetWordBoundaries(const wxString text,IntPairVector &results,int start,int
 }
 
 
+/////////////////////////////////////////////////////////
+// Determine whether wchar 'c' is a whitespace character
+bool IsWhitespace(wchar_t c)
+{
+	const wchar_t whitespaces[] = {
+		// http://en.wikipedia.org/wiki/Space_(punctuation)#Table_of_spaces
+		0x0009, 0x000A, 0x000B, 0x000C, 0x000D, 0x0020, 0x0085, 0x00A0,
+		0x1680, 0x180E, 0x2000, 0x2001, 0x2002, 0x2003, 0x2004, 0x2005,
+		0x2006, 0x2007, 0x2008, 0x2009, 0x200A, 0x2028, 0x2029, 0x202F,
+		0x025F, 0x3000, 0xFEFF
+	};
+	const size_t num_chars = sizeof(whitespaces) / sizeof(whitespaces[0]);
+
+	for (size_t i = 0; i < num_chars; ++i)
+		if (whitespaces[i] == c)
+			return true;
+
+	return false;
+}
+
+
+///////////////////////////////////////////////////////////////
+// Returns true if str is empty of consists of only whitespace
+bool StringEmptyOrWhitespace(const wxString &str)
+{
+	for (size_t i = 0; i < str.size(); ++i)
+		if (!IsWhitespace(str[i]))
+			return false;
+
+	return true;
+}
+
+
 /////////////////////
 // String to integer
 // wxString::ToLong() is slow and not as flexible

aegisub/src/utils.h

@@ -68,6 +68,8 @@ wxString PrettySize(int bytes);
 wxMenuItem *AppendBitmapMenuItem (wxMenu* parentMenu,int id,wxString text,wxString help,wxBitmap bmp,int pos=-1);
 int SmallestPowerOf2(int x);
 void GetWordBoundaries(const wxString text,IntPairVector &results,int start=0,int end=-1);
+bool IsWhitespace(wchar_t c);
+bool StringEmptyOrWhitespace(const wxString &str);
 int AegiStringToInt(const wxString &str,int start=0,int end=-1);
 int AegiStringToFix(const wxString &str,size_t decimalPlaces,int start=0,int end=-1);
 wxIcon BitmapToIcon(wxBitmap bmp);