Bug #234936 --> expose method to find a url given a string. This allows the spell checker to

ignore urls in mail compose. Porting this functionality to seamonkey.

sr=bienvenu

extensions/spellcheck/src/mozEnglishWordUtils.cpp

@@ -50,6 +50,9 @@ NS_IMPL_ISUPPORTS1(mozEnglishWordUtils, mozISpellI18NUtil)
 mozEnglishWordUtils::mozEnglishWordUtils()
 {
   mLanguage.Assign(NS_LITERAL_STRING("en"));
+
+  nsresult rv;
+  mURLDetector = do_CreateInstance(MOZ_TXTTOHTMLCONV_CONTRACTID, &rv);
 }
 
 mozEnglishWordUtils::~mozEnglishWordUtils()
@@ -179,6 +182,36 @@ NS_IMETHODIMP mozEnglishWordUtils::FindNextWord(const PRUnichar *word, PRUint32
       { 
         p++;
       }
+    
+    // we could be trying to break down a url, we don't want to break a url into parts,
+    // instead we want to find out if it really is a url and if so, skip it, advancing startWord 
+    // to a point after the url.
+
+    // before we spend more time looking to see if the word is a url, look for a url identifer
+    // and make sure that identifer isn't the last character in the word fragment.
+    if ( (*p == ':' || *p == '@' || *p == '.') &&  p < endbuf - 1) {
+
+        // ok, we have a possible url...do more research to find out if we really have one
+        // and determine the length of the url so we can skip over it.
+       
+        if (mURLDetector)
+        {
+          PRInt32 startPos = -1;
+          PRInt32 endPos = -1;        
+
+          mURLDetector->FindURLInPlaintext(startWord, endbuf - startWord, p - startWord, &startPos, &endPos);
+
+          // ok, if we got a url, adjust the array bounds, skip the current url text and find the next word again
+          if (startPos != -1 && endPos != -1) { 
+            startWord = p + endPos + 1; // skip over the url
+            p = startWord; // reset p
+
+            // now recursively call FindNextWord to search for the next word now that we have skipped the url
+            return FindNextWord(word, length, startWord - word, begin, end);
+          }
+        }
+    }
+
     while((p > startWord)&&(*(p-1) == '\'')){  // trim trailing apostrophes
       p--;
     }

extensions/spellcheck/src/mozEnglishWordUtils.h

@@ -45,6 +45,8 @@
 #include "nsString.h"
 #include "nsICaseConversion.h"
 
+#include "mozITXTToHTMLConv.h" 
+
 class mozEnglishWordUtils : public mozISpellI18NUtil
 {
 public:
@@ -64,6 +66,7 @@ protected:
   nsString mLanguage;
   nsString mCharset;
   nsCOMPtr<nsICaseConversion> mCaseConv;
+  nsCOMPtr<mozITXTToHTMLConv> mURLDetector; // used to detect urls so the spell checker can skip them.
 };
 
 #endif

netwerk/streamconv/converters/mozTXTToHTMLConv.cpp

@@ -461,6 +461,19 @@ mozTXTToHTMLConv::CheckURLAndCreateHTML(
     return PR_FALSE;
 }
 
+NS_IMETHODIMP mozTXTToHTMLConv::FindURLInPlaintext(const PRUnichar * aInString, PRInt32 aInLength, PRInt32 aPos, PRInt32 * aStartPos, PRInt32 * aEndPos)
+{
+  // call FindURL on the passed in string
+  nsString outputHTML; // we'll ignore the generated output HTML
+  
+  *aStartPos = -1;
+  *aEndPos = -1;
+  
+  FindURL(aInString, aInLength, aPos, kURLs, outputHTML, *aStartPos, *aEndPos);
+
+  return NS_OK;
+}
+
 PRBool
 mozTXTToHTMLConv::FindURL(const PRUnichar * aInString, PRInt32 aInLength, const PRUint32 pos,
      const PRUint32 whathasbeendone,

netwerk/streamconv/public/mozITXTToHTMLConv.idl

@@ -105,4 +105,6 @@ interface mozITXTToHTMLConv : nsIStreamConverter {
  */
   unsigned long   citeLevelTXT([const] in wstring line,
                                out unsigned long logLineStart);
+
+  void findURLInPlaintext([const] in wstring text, in long aLength, in long aPos, out long aStartPos, out long aEndPos);
 };