bool ON_wString::UrlDecode() { CopyArray(); bool rc = true; wchar_t c; wchar_t* s0 = Array(); if ( !s0 ) return true; wchar_t* s1 = s0; //const wchar_t* debg = s1; int i; for (i = Length(); i > 0; i-- ) { c = *s0++; if (0==c) break; if (i >= 3 && '%' == c && UrlDecodeHelper(s0) ) { s0++; *s1++ = *s0++; i -= 2; } else { *s1++ = c; if (rc) rc = IsValidUrlChar(c); } } *s1 = 0; SetLength(s1 - Array()); return rc; }
/** Parses URL from a token. Is used by SearchUrlL method and if a URL was found it's appended to item array. Note that parsing for generic URIs is done with SearchGenericUriL -method. @param aType a Type of URL to seach, i.e. www. wap. IP e.g.127.0.0.1 @param aTokenPtr Pointer to token that will be parsed @param aTextOffset Offset of the token (start position in the whole text) @leave KErrNone, if successful; otherwise one of the other system-wide error codes. @return ETrue if the parameter for phone number is valid, else returns EFalse */ TBool CTulAddressStringTokenizer::ParseUrlL(const TDesC& aType, const TPtrC& aTokenPtr, TInt aTextOffset) { TBool wasValidUrl = EFalse; TLex url; TInt position = aTokenPtr.FindF( aType ); if ( position != KErrNotFound ) { // address start found url = aTokenPtr.Right( aTokenPtr.Length() - position ); url.Inc( aType.Length() ); while( IsValidUrlChar( url.Peek() ) && !(url.Eos()) ) { if( url.Peek() == ':' ) { url.Inc(); if ( !url.Peek().IsDigit() ) { url.UnGet(); break; } } else url.Inc(); } // If a period or question mark was followed by a whitespace remove it if ( url.Eos() ) // Can't be followed by white space if it's { // the last character at token url.UnGet(); if ( url.Peek() != '.' && url.Peek() != '?' && url.Peek() != ',' ) // If it wasn't a period or question mark url.Inc(); } url.Mark(); wasValidUrl = ETrue; } if ( wasValidUrl && ( url.MarkedOffset() > aType.Length() ) ) { AddItemL( aTextOffset - aTokenPtr.Length() + position, url.MarkedOffset(), EFindItemSearchURLBin ); return ETrue; } return EFalse; }
/** Search fixed start URLs, i.e. URLs without schema (www., wap.). Also finds IPv4 addresses (*.*.*.*). As a special case, supports deprecated hardcoded schematic addresses finding (http://, https://, rtsp://) to make sure deprecated search cases work as they did previously. @param aText Text that will be parsed @param aFindFixedSchemas If true, will find old fixed schematic URLs also @return ETrue if any URL are found else returns EFalse @leave KErrNone, if successful; otherwise one of the other system-wide error codes. @panic ETulPanicDescriptorLength in debug build if item's position and/or length is out of the document's range. */ TBool CTulAddressStringTokenizer::SearchUrlL( const TDesC& aText, const TBool aFindFixedSchemas ) { TLex text = aText; while ( !text.Eos() ) { while( !(text.Eos()) && !IsValidUrlChar( text.Peek() ) ) text.Inc(); text.Mark(); while( !(text.Eos()) && IsValidUrlChar( text.Peek() ) ) text.Inc(); TPtrC tokenPtr = text.MarkedToken(); TBool wasValidUrl = EFalse; if ( aFindFixedSchemas ) // Search for http:// wasValidUrl = ParseUrlL( KHttpUrlAddress, tokenPtr, text.Offset() ); if (aFindFixedSchemas && !wasValidUrl) // Search for https:// wasValidUrl = ParseUrlL( KHttpsUrlAddress, tokenPtr, text.Offset() ); if (aFindFixedSchemas && !wasValidUrl) // Search for rtsp:// wasValidUrl = ParseUrlL( KRtspUrlAddress, tokenPtr, text.Offset() ); if ( !wasValidUrl ) // Search for www. wasValidUrl = ParseUrlL( KWwwUrlAddress, tokenPtr, text.Offset() ); if ( !wasValidUrl ) // Search for wap. wasValidUrl = ParseUrlL( KWapUrlAddress, tokenPtr, text.Offset() ); if ( !wasValidUrl ) // Search for IP-address (xxx.xxx.xxx.xxx) { if ( tokenPtr.Match( KIPAddress ) != KErrNotFound ) { TInt periods = 0; wasValidUrl = ETrue; TBool endWithPunctuation = EFalse; TBool betweenBrackets = EFalse; // First see if token ends with ",",".","!","?",";" or ":" TChar charac = tokenPtr[tokenPtr.Length() - 1]; TChar charac0 = tokenPtr[0]; if ( charac == ',' || charac == '.' || charac == '!' || charac == '?' || charac == ';' || charac == ':' ) { endWithPunctuation = ETrue; } // Or if it starts and ends with brackets or quotation marks else if ( ( charac0 == '(' && charac == ')' ) || ( charac0 == '"' && charac == '"' ) || ( charac0 == '[' && charac == ']' ) || ( charac0 == '<' && charac == '>' ) ) { betweenBrackets = ETrue; } TInt i = 0; TInt tokensEnd = tokenPtr.Length(); if ( endWithPunctuation ) tokensEnd--; else if ( betweenBrackets ) { i = 1; tokensEnd--; } // Take a closer look to see if a valid IP-address TBuf<3> ipPart; TInt numbers = 0; for ( ; i < tokensEnd; i++ ) { if ( !( ((TChar)tokenPtr[i]).IsDigit() || tokenPtr[i] == '.' ) ) { wasValidUrl = EFalse; break; } if ( tokenPtr[i] == '.' ) periods++; else numbers++; if ( numbers > KNumbersInIpAddress || periods > KDotsInIpAddress ) { wasValidUrl = EFalse; break; } if ( ((TChar)tokenPtr[i]).IsDigit() ) { ipPart.Append( tokenPtr[i] ); TBool checkInt = EFalse; if ( i + 1 < tokensEnd ) { if ( tokenPtr[i+1] == '.' ) checkInt = ETrue; } if ( i == tokensEnd - 1 || checkInt ) { TLex val = ipPart; TInt numberInt; TInt error = val.Val( numberInt ); if ( error != KErrNone || numberInt > 255 ) { wasValidUrl = EFalse; break; } numbers = 0; ipPart.Delete( 0, ipPart.Length() ); } } } if ( wasValidUrl && periods == KDotsInIpAddress ) { TInt startPos = text.Offset() - tokenPtr.Length(); TInt length = tokenPtr.Length(); // If there was a punctuation at the end or brackets, let's take it/them away if ( endWithPunctuation || betweenBrackets) { length--; if ( betweenBrackets ) { startPos++; length--; } } __ASSERT_DEBUG( startPos + length <= aText.Length(), Panic(ETulPanicDescriptorLength) ); AddItemL( startPos, length, EFindItemSearchURLBin ); } } } } return (iFoundItems->Count() > 0); }