EXT-944 EXT-1026: cleaning up my quick fix for these issues.

I've now added an explicit LLUrlRegistry::findUrl() method
for LLWStrings. This deals with correcting the start/end
range for the url appropriately. Now the API can be used
without worrying about utf8/utf32 character offset issues.

Internal JIRAs: DEV-40127 DEV-39966.

indra/llui/lltextbox.cpp

@@ -302,11 +302,10 @@ LLWString LLTextBox::getWrappedText(const LLStringExplicit& in_text, F32 max_wid
 
 	// find the next Url in the text string
 	LLUrlMatch match;
-	while ( LLUrlRegistry::instance().findUrl(wstring_to_utf8str(wtext), match))
+	while ( LLUrlRegistry::instance().findUrl(wtext, match))
 	{
-		LLWString wurl = utf8str_to_wstring(match.getUrl());
-		S32 start = wtext.find(wurl);
-		S32 end = start + wurl.size();
+		S32 start = match.getStart();
+		S32 end = match.getEnd() + 1;
 
 		// perform word wrap on the text before the Url
 		final_wtext += wrapText(wtext.substr(0, start), hoffset, line_num, max_width);
@@ -579,14 +578,14 @@ void LLTextBox::updateDisplayTextAndSegments()
 	LLWString text = mText.getWString();
 		
 	// find the next Url in the text string
-	while ( LLUrlRegistry::instance().findUrl(wstring_to_utf8str(text), match,
+	while ( LLUrlRegistry::instance().findUrl(text, match,
 											  boost::bind(&LLTextBox::onUrlLabelUpdated, this, _1, _2)) )
 	{
 		// work out the char offset for the start/end of the url
-		LLWString wurl = utf8str_to_wstring(match.getUrl());
-		S32 url_start = text.find(wurl);
-		S32 url_end = url_start + wurl.size() - 1;
+		S32 url_start = match.getStart();
+		S32 url_end = match.getEnd();
 
+		// and the char offset for the label in the display text
 		S32 seg_start = mDisplayText.size();
 		S32 start = seg_start + url_start;
 		S32 end = start + match.getLabel().size();

indra/llui/llurlregistry.cpp

@@ -162,3 +162,31 @@ bool LLUrlRegistry::findUrl(const std::string &text, LLUrlMatch &match, const LL
 
 	return false;
 }
+
+bool LLUrlRegistry::findUrl(const LLWString &text, LLUrlMatch &match, const LLUrlLabelCallback &cb)
+{
+	// boost::regex_search() only works on char or wchar_t
+	// types, but wchar_t is only 2-bytes on Win32 (not 4).
+	// So we use UTF-8 to make this work the same everywhere.
+	std::string utf8_text = wstring_to_utf8str(text);
+	if (findUrl(utf8_text, match, cb))
+	{
+		// we cannot blindly return the start/end offsets from
+		// the UTF-8 string because it is a variable-length
+		// character encoding, so we need to update the start
+		// and end values to be correct for the wide string.
+		LLWString wurl = utf8str_to_wstring(match.getUrl());
+		S32 start = text.find(wurl);
+		if (start == std::string::npos)
+		{
+			return false;
+		}
+		S32 end = start + wurl.size() - 1;
+
+		match.setValues(start, end, match.getUrl(), match.getLabel(),
+						match.getTooltip(), match.getIcon(),
+						match.getMenuName(), match.getLocation());
+		return true;
+	}
+	return false;
+}

indra/llui/llurlregistry.h

@@ -37,10 +37,10 @@
 #include "llurlentry.h"
 #include "llurlmatch.h"
 #include "llsingleton.h"
+#include "llstring.h"
 
 #include <string>
 #include <vector>
-#include <map>
 
 /// This default callback for findUrl() simply ignores any label updates
 void LLUrlRegistryNullCallback(const std::string &url, const std::string &label);
@@ -77,6 +77,10 @@ public:
 	bool findUrl(const std::string &text, LLUrlMatch &match,
 				 const LLUrlLabelCallback &cb = &LLUrlRegistryNullCallback);
 
+	/// a slightly less efficient version of findUrl for wide strings
+	bool findUrl(const LLWString &text, LLUrlMatch &match,
+				 const LLUrlLabelCallback &cb = &LLUrlRegistryNullCallback);
+
 private:
 	LLUrlRegistry();
 	friend class LLSingleton<LLUrlRegistry>;