EXT-944 EXT-1026: converted the LLUrlRegistry::findUrl() method to

work on an LLWString instead of a std::string, so that we don't have
to worry about character offsets for variable-length-encoded UTF-8
strings.

This was causing crashes whenever we would try to show a textbox with
a URL and foreign characters (> 1 byte chars). Damn, I suck!

9 files changed, 61 insertions, 57 deletions

diff --git a/indra/llui/lltextbase.cpp b/indra/llui/lltextbase.cpp
index cb60b4fe36..325c7a8104 100644
--- a/indra/llui/lltextbase.cpp
+++ b/indra/llui/lltextbase.cpp
@@ -429,7 +429,7 @@ LLContextMenu *LLTextBase::createUrlContextMenu(const std::string &in_url)
 	// work out the XUI menu file to use for this url
 	LLUrlMatch match;
 	std::string url = in_url;
-	if (! LLUrlRegistry::instance().findUrl(url, match))
+	if (! LLUrlRegistry::instance().findUrl(utf8str_to_wstring(url), match))
 	{
 		return NULL;
 	}
diff --git a/indra/llui/lltextbox.cpp b/indra/llui/lltextbox.cpp
index 810626268f..3cde42d75c 100644
--- a/indra/llui/lltextbox.cpp
+++ b/indra/llui/lltextbox.cpp
@@ -297,7 +297,7 @@ 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))
 	{
 		S32 start = match.getStart();
 		S32 end = match.getEnd() + 1;
@@ -573,7 +573,7 @@ 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
diff --git a/indra/llui/lltexteditor.cpp b/indra/llui/lltexteditor.cpp
index 983777b747..f7680a0a2b 100644
--- a/indra/llui/lltexteditor.cpp
+++ b/indra/llui/lltexteditor.cpp
@@ -3525,7 +3525,7 @@ void LLTextEditor::appendStyledText(const std::string &new_text,
 
 		S32 start=0,end=0;
 		LLUrlMatch match;
-		std::string text = new_text;
+		LLWString text = utf8str_to_wstring(new_text);
 		while ( LLUrlRegistry::instance().findUrl(text, match,
 		        boost::bind(&LLTextEditor::onUrlLabelUpdated, this, _1, _2)) )
 		{
@@ -3549,8 +3549,8 @@ void LLTextEditor::appendStyledText(const std::string &new_text,
 				{
 					part = (S32)LLTextParser::MIDDLE;
 				}
-				std::string subtext=text.substr(0,start);
-				appendHighlightedText(subtext,allow_undo, prepend_newline, part, style_params); 
+				std::string subtext = wstring_to_utf8str(text.substr(0,start));
+				appendHighlightedText(subtext, allow_undo, prepend_newline, part, style_params); 
 				prepend_newline = false;
 			}
 
@@ -3595,7 +3595,8 @@ void LLTextEditor::appendStyledText(const std::string &new_text,
 			}
 		}
 		if (part != (S32)LLTextParser::WHOLE) part=(S32)LLTextParser::END;
-		if (end < (S32)text.length()) appendHighlightedText(text,allow_undo, prepend_newline, part, style_params);		
+		if (end < (S32)text.length()) appendHighlightedText(wstring_to_utf8str(text), allow_undo,
+															prepend_newline, part, style_params);		
 	}
 	else
 	{
@@ -4137,7 +4138,7 @@ void LLTextEditor::updateLinkSegments()
 			LLUrlMatch match;
 			LLStyleSP style = static_cast<LLStyleSP>(segment->getStyle());
 			std::string url_label = getText().substr(segment->getStart(), segment->getEnd()-segment->getStart());
-			if (LLUrlRegistry::instance().findUrl(url_label, match))
+			if (LLUrlRegistry::instance().findUrl(utf8str_to_wstring(url_label), match))
 			{
 				LLStringUtil::trim(url_label);
 				style->setLinkHREF(url_label);
diff --git a/indra/llui/llurlaction.cpp b/indra/llui/llurlaction.cpp
index 3b689b93c0..7721d0e502 100644
--- a/indra/llui/llurlaction.cpp
+++ b/indra/llui/llurlaction.cpp
@@ -112,7 +112,7 @@ void LLUrlAction::clickAction(std::string url)
 void LLUrlAction::teleportToLocation(std::string url)
 {
 	LLUrlMatch match;
-	if (LLUrlRegistry::instance().findUrl(url, match))
+	if (LLUrlRegistry::instance().findUrl(utf8str_to_wstring(url), match))
 	{
 		if (! match.getLocation().empty())
 		{
@@ -129,7 +129,7 @@ void LLUrlAction::copyURLToClipboard(std::string url)
 void LLUrlAction::copyLabelToClipboard(std::string url)
 {
 	LLUrlMatch match;
-	if (LLUrlRegistry::instance().findUrl(url, match))
+	if (LLUrlRegistry::instance().findUrl(utf8str_to_wstring(url), match))
 	{
 		LLView::getWindow()->copyTextToClipboard(utf8str_to_wstring(match.getLabel()));
 	}	
diff --git a/indra/llui/llurlentry.cpp b/indra/llui/llurlentry.cpp
index c20212c375..e4bcc27428 100644
--- a/indra/llui/llurlentry.cpp
+++ b/indra/llui/llurlentry.cpp
@@ -148,8 +148,8 @@ void LLUrlEntryBase::callObservers(const std::string &id, const std::string &lab
 //
 LLUrlEntryHTTP::LLUrlEntryHTTP()
 {
-	mPattern = boost::regex("https?://([-\\w\\.]+)+(:\\d+)?(:\\w+)?(@\\d+)?(@\\w+)?/?\\S*",
-							boost::regex::perl|boost::regex::icase);
+	mPattern = boost::wregex(L"https?://([-\\w\\.]+)+(:\\d+)?(:\\w+)?(@\\d+)?(@\\w+)?/?\\S*",
+							 boost::regex::perl|boost::regex::icase);
 	mMenuName = "menu_url_http.xml";
 	mTooltip = LLTrans::getString("TooltipHttpUrl");
 	//mIcon = "gear.tga";
@@ -166,8 +166,8 @@ std::string LLUrlEntryHTTP::getLabel(const std::string &url, const LLUrlLabelCal
 //
 LLUrlEntryHTTPLabel::LLUrlEntryHTTPLabel()
 {
-	mPattern = boost::regex("\\[https?://\\S+[ \t]+[^\\]]+\\]",
-							boost::regex::perl|boost::regex::icase);
+	mPattern = boost::wregex(L"\\[https?://\\S+[ \t]+[^\\]]+\\]",
+							 boost::regex::perl|boost::regex::icase);
 	mMenuName = "menu_url_http.xml";
 	mTooltip = LLTrans::getString("TooltipHttpUrl");
 }
@@ -188,8 +188,8 @@ std::string LLUrlEntryHTTPLabel::getUrl(const std::string &string)
 LLUrlEntrySLURL::LLUrlEntrySLURL()
 {
 	// see http://slurl.com/about.php for details on the SLURL format
-	mPattern = boost::regex("http://slurl.com/secondlife/\\S+/?(\\d+)?/?(\\d+)?/?(\\d+)?/?\\S*",
-							boost::regex::perl|boost::regex::icase);
+	mPattern = boost::wregex(L"http://slurl.com/secondlife/\\S+/?(\\d+)?/?(\\d+)?/?(\\d+)?/?\\S*",
+							 boost::regex::perl|boost::regex::icase);
 	mMenuName = "menu_url_slurl.xml";
 	mTooltip = LLTrans::getString("TooltipSLURL");
 }
@@ -260,8 +260,8 @@ std::string LLUrlEntrySLURL::getLocation(const std::string &url) const
 //
 LLUrlEntryAgent::LLUrlEntryAgent()
 {
-	mPattern = boost::regex("secondlife:///app/agent/[\\da-f-]+/about",
-							boost::regex::perl|boost::regex::icase);
+	mPattern = boost::wregex(L"secondlife:///app/agent/[\\da-f-]+/about",
+							 boost::regex::perl|boost::regex::icase);
 	mMenuName = "menu_url_agent.xml";
 	mTooltip = LLTrans::getString("TooltipAgentUrl");
 }
@@ -302,8 +302,8 @@ std::string LLUrlEntryAgent::getLabel(const std::string &url, const LLUrlLabelCa
 //
 LLUrlEntryGroup::LLUrlEntryGroup()
 {
-	mPattern = boost::regex("secondlife:///app/group/[\\da-f-]+/about",
-							boost::regex::perl|boost::regex::icase);
+	mPattern = boost::wregex(L"secondlife:///app/group/[\\da-f-]+/about",
+							 boost::regex::perl|boost::regex::icase);
 	mMenuName = "menu_url_group.xml";
 	mTooltip = LLTrans::getString("TooltipGroupUrl");
 }
@@ -344,8 +344,8 @@ std::string LLUrlEntryGroup::getLabel(const std::string &url, const LLUrlLabelCa
 ///
 LLUrlEntryParcel::LLUrlEntryParcel()
 {
-	mPattern = boost::regex("secondlife:///app/parcel/[\\da-f-]+/about",
-							boost::regex::perl|boost::regex::icase);
+	mPattern = boost::wregex(L"secondlife:///app/parcel/[\\da-f-]+/about",
+							 boost::regex::perl|boost::regex::icase);
 	mMenuName = "menu_url_parcel.xml";
 	mTooltip = LLTrans::getString("TooltipParcelUrl");
 }
@@ -360,8 +360,8 @@ std::string LLUrlEntryParcel::getLabel(const std::string &url, const LLUrlLabelC
 //
 LLUrlEntryPlace::LLUrlEntryPlace()
 {
-	mPattern = boost::regex("secondlife://\\S+/?(\\d+/\\d+/\\d+|\\d+/\\d+)/?",
-							boost::regex::perl|boost::regex::icase);
+	mPattern = boost::wregex(L"secondlife://\\S+/?(\\d+/\\d+/\\d+|\\d+/\\d+)/?",
+							 boost::regex::perl|boost::regex::icase);
 	mMenuName = "menu_url_slurl.xml";
 	mTooltip = LLTrans::getString("TooltipSLURL");
 }
@@ -416,8 +416,8 @@ std::string LLUrlEntryPlace::getLocation(const std::string &url) const
 //
 LLUrlEntryTeleport::LLUrlEntryTeleport()
 {
-	mPattern = boost::regex("secondlife:///app/teleport/\\S+(/\\d+)?(/\\d+)?(/\\d+)?/?\\S*",
-							boost::regex::perl|boost::regex::icase);
+	mPattern = boost::wregex(L"secondlife:///app/teleport/\\S+(/\\d+)?(/\\d+)?(/\\d+)?/?\\S*",
+							 boost::regex::perl|boost::regex::icase);
 	mMenuName = "menu_url_teleport.xml";
 	mTooltip = LLTrans::getString("TooltipTeleportUrl");
 }
@@ -488,8 +488,8 @@ std::string LLUrlEntryTeleport::getLocation(const std::string &url) const
 ///
 LLUrlEntryObjectIM::LLUrlEntryObjectIM()
 {
-	mPattern = boost::regex("secondlife:///app/objectim/[\\da-f-]+\\??\\S*",
-							boost::regex::perl|boost::regex::icase);
+	mPattern = boost::wregex(L"secondlife:///app/objectim/[\\da-f-]+\\??\\S*",
+							 boost::regex::perl|boost::regex::icase);
 	mMenuName = "menu_url_objectim.xml";
 	mTooltip = LLTrans::getString("TooltipObjectIMUrl");
 }
@@ -532,8 +532,8 @@ std::string LLUrlEntryObjectIM::getLocation(const std::string &url) const
 //
 LLUrlEntrySL::LLUrlEntrySL()
 {
-	mPattern = boost::regex("secondlife://(\\w+)?(:\\d+)?/\\S+",
-							boost::regex::perl|boost::regex::icase);
+	mPattern = boost::wregex(L"secondlife://(\\w+)?(:\\d+)?/\\S+",
+							 boost::regex::perl|boost::regex::icase);
 	mMenuName = "menu_url_slapp.xml";
 	mTooltip = LLTrans::getString("TooltipSLAPP");
 }
@@ -549,8 +549,8 @@ std::string LLUrlEntrySL::getLabel(const std::string &url, const LLUrlLabelCallb
 //
 LLUrlEntrySLLabel::LLUrlEntrySLLabel()
 {
-	mPattern = boost::regex("\\[secondlife://\\S+[ \t]+[^\\]]+\\]",
-							boost::regex::perl|boost::regex::icase);
+	mPattern = boost::wregex(L"\\[secondlife://\\S+[ \t]+[^\\]]+\\]",
+							 boost::regex::perl|boost::regex::icase);
 	mMenuName = "menu_url_slapp.xml";
 	mTooltip = LLTrans::getString("TooltipSLAPP");
 }
diff --git a/indra/llui/llurlentry.h b/indra/llui/llurlentry.h
index 54053872df..36758566ed 100644
--- a/indra/llui/llurlentry.h
+++ b/indra/llui/llurlentry.h
@@ -68,7 +68,7 @@ public:
 	virtual ~LLUrlEntryBase();
 	
 	/// Return the regex pattern that matches this Url 
-	boost::regex getPattern() const { return mPattern; }
+	boost::wregex getPattern() const { return mPattern; }
 
 	/// Return the url from a string that matched the regex
 	virtual std::string getUrl(const std::string &string);
@@ -102,7 +102,7 @@ protected:
 		LLUrlLabelSignal *signal;
 	} LLUrlEntryObserver;
 
-	boost::regex                                   mPattern;
+	boost::wregex                                  mPattern;
 	std::string                                    mIcon;
 	std::string                                    mMenuName;
 	std::string                                    mTooltip;
diff --git a/indra/llui/llurlregistry.cpp b/indra/llui/llurlregistry.cpp
index f2d340deb7..2ea3c59e03 100644
--- a/indra/llui/llurlregistry.cpp
+++ b/indra/llui/llurlregistry.cpp
@@ -75,9 +75,9 @@ void LLUrlRegistry::registerUrl(LLUrlEntryBase *url)
 	}
 }
 
-static bool matchRegex(const char *text, boost::regex regex, U32 &start, U32 &end)
+static bool matchRegex(const wchar_t *text, boost::wregex regex, U32 &start, U32 &end)
 {
-	boost::cmatch result;
+	boost::wcmatch result;
 	bool found;
 
 	// regex_search can potentially throw an exception, so check for it
@@ -107,7 +107,7 @@ static bool matchRegex(const char *text, boost::regex regex, U32 &start, U32 &en
 	}
 	// ignore a terminating ')' when Url contains no matching '('
 	// see DEV-19842 for details
-	else if (text[end] == ')' && std::string(text+start, end-start).find('(') == std::string::npos)
+	else if (text[end] == ')' && LLWString(text+start, end-start).find('(') == std::string::npos)
 	{
 		end--;
 	}
@@ -115,10 +115,10 @@ static bool matchRegex(const char *text, boost::regex regex, U32 &start, U32 &en
 	return true;
 }
 
-bool LLUrlRegistry::findUrl(const std::string &text, LLUrlMatch &match, const LLUrlLabelCallback &cb)
+bool LLUrlRegistry::findUrl(const LLWString &text, LLUrlMatch &match, const LLUrlLabelCallback &cb)
 {
 	// avoid costly regexes if there is clearly no URL in the text
-	if (text.find("://") == std::string::npos)
+	if (text.find(L"://") == std::string::npos)
 	{
 		return false;
 	}
@@ -149,7 +149,7 @@ bool LLUrlRegistry::findUrl(const std::string &text, LLUrlMatch &match, const LL
 	if (match_entry)
 	{
 		// fill in the LLUrlMatch object and return it
-		std::string url = text.substr(match_start, match_end - match_start + 1);
+		std::string url = wstring_to_utf8str(text.substr(match_start, match_end - match_start + 1));
 		match.setValues(match_start, match_end,
 						match_entry->getUrl(url),
 						match_entry->getLabel(url, cb),
diff --git a/indra/llui/llurlregistry.h b/indra/llui/llurlregistry.h
index 84b033036c..bf5257d630 100644
--- a/indra/llui/llurlregistry.h
+++ b/indra/llui/llurlregistry.h
@@ -37,6 +37,7 @@
 #include "llurlentry.h"
 #include "llurlmatch.h"
 #include "llsingleton.h"
+#include "llstring.h"
 
 #include <string>
 #include <vector>
@@ -74,7 +75,7 @@ public:
 
 	/// get the next Url in an input string, starting at a given character offset
 	/// your callback is invoked if the matched Url's label changes in the future
-	bool findUrl(const std::string &text, LLUrlMatch &match,
+	bool findUrl(const LLWString &text, LLUrlMatch &match,
 				 const LLUrlLabelCallback &cb = &LLUrlRegistryNullCallback);
 
 private:
diff --git a/indra/llui/tests/llurlentry_test.cpp b/indra/llui/tests/llurlentry_test.cpp
index 1e7a0f7f2c..89a80f1e73 100644
--- a/indra/llui/tests/llurlentry_test.cpp
+++ b/indra/llui/tests/llurlentry_test.cpp
@@ -44,17 +44,19 @@ namespace
 
 namespace tut
 {
-	void testRegex(const std::string &testname, boost::regex regex,
+	void testRegex(const std::string &testname, boost::wregex regex,
 				   const char *text, const std::string &expected)
 	{
 		std::string url = "";
-		boost::cmatch result;
-		bool found = boost::regex_search(text, result, regex);
+		boost::wcmatch result;
+		LLWString wtext = utf8str_to_wstring(text);
+		const wchar_t *wctext = wtext.c_str();
+		bool found = boost::regex_search(wctext, result, regex);
 		if (found)
 		{
-			S32 start = static_cast<U32>(result[0].first - text);
-			S32 end = static_cast<U32>(result[0].second - text);
-			url = std::string(text+start, end-start);
+			S32 start = static_cast<U32>(result[0].first - wctext);
+			S32 end = static_cast<U32>(result[0].second - wctext);
+			url = wstring_to_utf8str(wtext.substr(start, end-start));
 		}
 		ensure_equals(testname, url, expected);
 	}
@@ -66,7 +68,7 @@ namespace tut
 		// test LLUrlEntryHTTP - standard http Urls
 		//
 		LLUrlEntryHTTP url;
-		boost::regex r = url.getPattern();
+		boost::wregex r = url.getPattern();
 
 		testRegex("no valid url", r,
 				  "htp://slurl.com/",
@@ -145,7 +147,7 @@ namespace tut
 		// test LLUrlEntryHTTPLabel - wiki-style http Urls with labels
 		//
 		LLUrlEntryHTTPLabel url;
-		boost::regex r = url.getPattern();
+		boost::wregex r = url.getPattern();
 
 		testRegex("invalid wiki url [1]", r,
 				  "[http://www.example.org]",
@@ -187,7 +189,7 @@ namespace tut
 		// test LLUrlEntrySLURL - second life URLs
 		//
 		LLUrlEntrySLURL url;
-		boost::regex r = url.getPattern();
+		boost::wregex r = url.getPattern();
 
 		testRegex("no valid slurl [1]", r,
 				  "htp://slurl.com/secondlife/Ahern/50/50/50/",
@@ -259,7 +261,7 @@ namespace tut
 		// test LLUrlEntryAgent - secondlife://app/agent Urls
 		//
 		LLUrlEntryAgent url;
-		boost::regex r = url.getPattern();
+		boost::wregex r = url.getPattern();
 
 		testRegex("Invalid Agent Url", r,
 				  "secondlife:///app/agent/0e346d8b-4433-4d66-XXXX-fd37083abc4c/about",
@@ -285,7 +287,7 @@ namespace tut
 		// test LLUrlEntryGroup - secondlife://app/group Urls
 		//
 		LLUrlEntryGroup url;
-		boost::regex r = url.getPattern();
+		boost::wregex r = url.getPattern();
 
 		testRegex("Invalid Group Url", r,
 				  "secondlife:///app/group/00005ff3-4044-c79f-XXXX-fb28ae0df991/about",
@@ -311,7 +313,7 @@ namespace tut
 		// test LLUrlEntryPlace - secondlife://<location> URLs
 		//
 		LLUrlEntryPlace url;
-		boost::regex r = url.getPattern();
+		boost::wregex r = url.getPattern();
 
 		testRegex("no valid slurl [1]", r,
 				  "secondlife://Ahern/FOO/50/",
@@ -359,7 +361,7 @@ namespace tut
 		// test LLUrlEntryParcel - secondlife://app/parcel Urls
 		//
 		LLUrlEntryParcel url;
-		boost::regex r = url.getPattern();
+		boost::wregex r = url.getPattern();
 
 		testRegex("Invalid Classified Url", r,
 				  "secondlife:///app/parcel/0000060e-4b39-e00b-XXXX-d98b1934e3a8/about",
@@ -384,7 +386,7 @@ namespace tut
 		// test LLUrlEntryTeleport - secondlife://app/teleport URLs
 		//
 		LLUrlEntryTeleport url;
-		boost::regex r = url.getPattern();
+		boost::wregex r = url.getPattern();
 
 		testRegex("no valid teleport [1]", r,
 				  "http://slurl.com/secondlife/Ahern/50/50/50/",
@@ -460,7 +462,7 @@ namespace tut
 		// test LLUrlEntrySL - general secondlife:// URLs
 		//
 		LLUrlEntrySL url;
-		boost::regex r = url.getPattern();
+		boost::wregex r = url.getPattern();
 
 		testRegex("no valid slapp [1]", r,
 				  "http:///app/",
@@ -498,7 +500,7 @@ namespace tut
 		// test LLUrlEntrySLLabel - general secondlife:// URLs with labels
 		//
 		LLUrlEntrySLLabel url;
-		boost::regex r = url.getPattern();
+		boost::wregex r = url.getPattern();
 
 		testRegex("invalid wiki url [1]", r,
 				  "[secondlife:///app/]",