lexhtml.cxx.svn-base

Notepad++ is a generic source code editor (it tries to be anyway) and Notepad replacement written in

}

static inline bool IsPhpWordStart(const unsigned char ch) {
	return (isascii(ch) && (isalpha(ch) || (ch == '_'))) || (ch >= 0x7f);
}

static inline bool IsPhpWordChar(char ch) {
	return IsADigit(ch) || IsPhpWordStart(ch);
}

static bool InTagState(int state) {
	return state == SCE_H_TAG || state == SCE_H_TAGUNKNOWN ||
	       state == SCE_H_SCRIPT ||
	       state == SCE_H_ATTRIBUTE || state == SCE_H_ATTRIBUTEUNKNOWN ||
	       state == SCE_H_NUMBER || state == SCE_H_OTHER ||
	       state == SCE_H_DOUBLESTRING || state == SCE_H_SINGLESTRING;
}

static bool IsCommentState(const int state) {
	return state == SCE_H_COMMENT || state == SCE_H_SGML_COMMENT;
}

static bool IsScriptCommentState(const int state) {
	return state == SCE_HJ_COMMENT || state == SCE_HJ_COMMENTLINE || state == SCE_HJA_COMMENT ||
		   state == SCE_HJA_COMMENTLINE || state == SCE_HB_COMMENTLINE || state == SCE_HBA_COMMENTLINE;
}

static bool isLineEnd(char ch) {
	return ch == '\r' || ch == '\n';
}

static bool isOKBeforeRE(char ch) {
	return (ch == '(') || (ch == '=') || (ch == ',');
}

static bool isPHPStringState(int state) {
	return
	    (state == SCE_HPHP_HSTRING) ||
	    (state == SCE_HPHP_SIMPLESTRING) ||
	    (state == SCE_HPHP_HSTRING_VARIABLE) ||
	    (state == SCE_HPHP_COMPLEX_VARIABLE);
}

static int FindPhpStringDelimiter(char *phpStringDelimiter, const int phpStringDelimiterSize, int i, const int lengthDoc, Accessor &styler) {
	int j;
	while (i < lengthDoc && (styler[i] == ' ' || styler[i] == '\t'))
		i++;
	phpStringDelimiter[0] = '\n';
	for (j = i; j < lengthDoc && styler[j] != '\n' && styler[j] != '\r'; j++) {
		if (j - i < phpStringDelimiterSize - 2)
			phpStringDelimiter[j-i+1] = styler[j];
		else
			i++;
	}
	phpStringDelimiter[j-i+1] = '\0';
	return j;
}

static void ColouriseHyperTextDoc(unsigned int startPos, int length, int initStyle, WordList *keywordlists[],
                                  Accessor &styler) {
	WordList &keywords = *keywordlists[0];
	WordList &keywords2 = *keywordlists[1];
	WordList &keywords3 = *keywordlists[2];
	WordList &keywords4 = *keywordlists[3];
	WordList &keywords5 = *keywordlists[4];
	WordList &keywords6 = *keywordlists[5]; // SGML (DTD) keywords

	// Lexer for HTML requires more lexical states (7 bits worth) than most lexers
	styler.StartAt(startPos, STYLE_MAX);
	char prevWord[200];
	prevWord[0] = '\0';
	char phpStringDelimiter[200]; // PHP is not limited in length, we are
	phpStringDelimiter[0] = '\0';
	int StateToPrint = initStyle;
	int state = stateForPrintState(StateToPrint);

	// If inside a tag, it may be a script tag, so reread from the start to ensure any language tags are seen
	if (InTagState(state)) {
		while ((startPos > 0) && (InTagState(styler.StyleAt(startPos - 1)))) {
			startPos--;
			length++;
		}
		state = SCE_H_DEFAULT;
	}
	// String can be heredoc, must find a delimiter first
	while (startPos > 0 && isPHPStringState(state) && state != SCE_HPHP_SIMPLESTRING) {
		startPos--;
		length++;
		state = styler.StyleAt(startPos);
	}
	styler.StartAt(startPos, STYLE_MAX);

	int lineCurrent = styler.GetLine(startPos);
	int lineState;
	if (lineCurrent > 0) {
		lineState = styler.GetLineState(lineCurrent);
	} else {
		// Default client and ASP scripting language is JavaScript
		lineState = eScriptJS << 8;
		// Don
		//lineState |= styler.GetPropertyInt("asp.default.language", eScriptJS) << 4;
		lineState |= styler.GetPropertyInt("asp.default.language", eScriptVBS) << 4;
		//nod
	}
	script_mode inScriptType = script_mode((lineState >> 0) & 0x03); // 2 bits of scripting mode
	bool tagOpened = (lineState >> 2) & 0x01; // 1 bit to know if we are in an opened tag
	bool tagClosing = (lineState >> 3) & 0x01; // 1 bit to know if we are in a closing tag
	bool tagDontFold = false; //some HTML tags should not be folded
	script_type aspScript = script_type((lineState >> 4) & 0x0F); // 4 bits of script name
	script_type clientScript = script_type((lineState >> 8) & 0x0F); // 4 bits of script name
	int beforePreProc = (lineState >> 12) & 0xFF; // 8 bits of state

	script_type scriptLanguage = ScriptOfState(state);

	const bool foldHTML = styler.GetPropertyInt("fold.html", 0) != 0;
	const bool fold = foldHTML && styler.GetPropertyInt("fold", 0);
	const bool foldHTMLPreprocessor = foldHTML && styler.GetPropertyInt("fold.html.preprocessor", 1);
	const bool foldCompact = styler.GetPropertyInt("fold.compact", 1) != 0;
	const bool caseSensitive = styler.GetPropertyInt("html.tags.case.sensitive", 0) != 0;

	int levelPrev = styler.LevelAt(lineCurrent) & SC_FOLDLEVELNUMBERMASK;
	int levelCurrent = levelPrev;
	int visibleChars = 0;

	char chPrev = ' ';
	char ch = ' ';
	char chPrevNonWhite = ' ';
	// look back to set chPrevNonWhite properly for better regex colouring
	if (scriptLanguage == eScriptJS && startPos > 0) {
		int back = startPos;
		int style = 0;
		while (--back) {
			style = styler.StyleAt(back);
			if (style < SCE_HJ_DEFAULT || style > SCE_HJ_COMMENTDOC)
				// includes SCE_HJ_COMMENT & SCE_HJ_COMMENTLINE
				break;
		}
		if (style == SCE_HJ_SYMBOLS) {
			chPrevNonWhite = styler.SafeGetCharAt(back);
		}
	}

	styler.StartSegment(startPos);
	const int lengthDoc = startPos + length;
	for (int i = startPos; i < lengthDoc; i++) {
		const char chPrev2 = chPrev;
		chPrev = ch;
		if (!isspacechar(ch) && state != SCE_HJ_COMMENT &&
			state != SCE_HJ_COMMENTLINE && state != SCE_HJ_COMMENTDOC)
			chPrevNonWhite = ch;
		ch = styler[i];
		char chNext = styler.SafeGetCharAt(i + 1);
		const char chNext2 = styler.SafeGetCharAt(i + 2);

		// Handle DBCS codepages
		if (styler.IsLeadByte(ch)) {
			chPrev = ' ';
			i += 1;
			continue;
		}

		if ((!isspacechar(ch) || !foldCompact) && fold)
			visibleChars++;

		// decide what is the current state to print (depending of the script tag)
		StateToPrint = statePrintForState(state, inScriptType);

		// handle script folding
		if (fold) {
			switch (scriptLanguage) {
			case eScriptJS:
			case eScriptPHP:
				//not currently supported				case eScriptVBS:

				if ((state != SCE_HPHP_COMMENT) && (state != SCE_HPHP_COMMENTLINE) && (state != SCE_HJ_COMMENT) && (state != SCE_HJ_COMMENTLINE) && (state != SCE_HJ_COMMENTDOC) && (!isStringState(state))) {
				//Platform::DebugPrintf("state=%d, StateToPrint=%d, initStyle=%d\n", state, StateToPrint, initStyle);
				//if ((state == SCE_HPHP_OPERATOR) || (state == SCE_HPHP_DEFAULT) || (state == SCE_HJ_SYMBOLS) || (state == SCE_HJ_START) || (state == SCE_HJ_DEFAULT)) {
					if ((ch == '{') || (ch == '}')) {
						levelCurrent += (ch == '{') ? 1 : -1;
					}
				}
				break;
			case eScriptPython:
				if (state != SCE_HP_COMMENTLINE) {
					if ((ch == ':') && ((chNext == '\n') || (chNext == '\r' && chNext2 == '\n'))) {
						levelCurrent++;
					} else if ((ch == '\n') && !((chNext == '\r') && (chNext2 == '\n')) && (chNext != '\n')) {
						// check if the number of tabs is lower than the level
						int Findlevel = (levelCurrent & ~SC_FOLDLEVELBASE) * 8;
						for (int j = 0; Findlevel > 0; j++) {
							char chTmp = styler.SafeGetCharAt(i + j + 1);
							if (chTmp == '\t') {
								Findlevel -= 8;
							} else if (chTmp == ' ') {
								Findlevel--;
							} else {
								break;
							}
						}

						if (Findlevel > 0) {
							levelCurrent -= Findlevel / 8;
							if (Findlevel % 8)
								levelCurrent--;
						}
					}
				}
				break;
			default:
				break;
			}
		}

		if ((ch == '\r' && chNext != '\n') || (ch == '\n')) {
			// Trigger on CR only (Mac style) or either on LF from CR+LF (Dos/Win) or on LF alone (Unix)
			// Avoid triggering two times on Dos/Win
			// New line -> record any line state onto /next/ line
			if (fold) {
				int lev = levelPrev;
				if (visibleChars == 0)
					lev |= SC_FOLDLEVELWHITEFLAG;
				if ((levelCurrent > levelPrev) && (visibleChars > 0))
					lev |= SC_FOLDLEVELHEADERFLAG;

				styler.SetLevel(lineCurrent, lev);
				visibleChars = 0;
				levelPrev = levelCurrent;
			}
			lineCurrent++;
			styler.SetLineState(lineCurrent,
			                    ((inScriptType & 0x03) << 0) |
			                    ((tagOpened & 0x01) << 2) |
			                    ((tagClosing & 0x01) << 3) |
			                    ((aspScript & 0x0F) << 4) |
			                    ((clientScript & 0x0F) << 8) |
			                    ((beforePreProc & 0xFF) << 12));
		}

		// generic end of script processing
		else if ((inScriptType == eNonHtmlScript) && (ch == '<') && (chNext == '/')) {
			// Check if it's the end of the script tag (or any other HTML tag)
			switch (state) {
				// in these cases, you can embed HTML tags (to confirm !!!!!!!!!!!!!!!!!!!!!!)
			case SCE_H_DOUBLESTRING:
			case SCE_H_SINGLESTRING:
			case SCE_HJ_COMMENT:
			case SCE_HJ_COMMENTDOC:
			//case SCE_HJ_COMMENTLINE: // removed as this is a common thing done to hide
			// the end of script marker from some JS interpreters.
			case SCE_HJ_DOUBLESTRING:
			case SCE_HJ_SINGLESTRING:
			case SCE_HJ_REGEX:
			case SCE_HB_STRING:
			case SCE_HP_STRING:
			case SCE_HP_TRIPLE:
			case SCE_HP_TRIPLEDOUBLE:
				break;
			default :
				// check if the closing tag is a script tag
				if (state == SCE_HJ_COMMENTLINE) {
					char tag[7]; // room for the <script> tag
					char chr;	// current char
					int j=0;
					chr = styler.SafeGetCharAt(i+2);
					while (j < 6 && !isspacechar(chr)) {
						tag[j++] = static_cast<char>(MakeLowerCase(chr));
						chr = styler.SafeGetCharAt(i+2+j);
					}
					tag[j] = '\0';
					if (strcmp(tag, "script") != 0) break;
				}
				// closing tag of the script (it's a closing HTML tag anyway)
				styler.ColourTo(i - 1, StateToPrint);
				state = SCE_H_TAGUNKNOWN;
				inScriptType = eHtml;
				scriptLanguage = eScriptNone;
				clientScript = eScriptJS;
				i += 2;
				visibleChars += 2;
				tagClosing = true;
				continue;
			}
		}

		/////////////////////////////////////
		// handle the start of PHP pre-processor = Non-HTML
		else if ((state != SCE_H_ASPAT) &&
		         !isPHPStringState(state) &&
		         (state != SCE_HPHP_COMMENT) &&
		         (ch == '<') &&
		         (chNext == '?') &&
				 !IsScriptCommentState(state) ) {
			scriptLanguage = segIsScriptingIndicator(styler, i + 2, i + 10, eScriptPHP);
			if (scriptLanguage != eScriptPHP && isStringState(state)) continue;
			styler.ColourTo(i - 1, StateToPrint);
			beforePreProc = state;
			i++;
			visibleChars++;
			i += PrintScriptingIndicatorOffset(styler, styler.GetStartSegment() + 2, i + 10);
			if (scriptLanguage == eScriptXML)
				styler.ColourTo(i, SCE_H_XMLSTART);
			else
				styler.ColourTo(i, SCE_H_QUESTION);
			state = StateForScript(scriptLanguage);
			if (inScriptType == eNonHtmlScript)
				inScriptType = eNonHtmlScriptPreProc;
			else
				inScriptType = eNonHtmlPreProc;
			// Fold whole script, but not if the XML first tag (all XML-like tags in this case)
			if (foldHTMLPreprocessor && (scriptLanguage != eScriptXML)) {
				levelCurrent++;
			}
			// should be better
			ch = styler.SafeGetCharAt(i);
			continue;
		}

		// handle the start of ASP pre-processor = Non-HTML
		else if (!isCommentASPState(state) && (ch == '<') && (chNext == '%') && !isPHPStringState(state)) {
			styler.ColourTo(i - 1, StateToPrint);
			beforePreProc = state;
			if (inScriptType == eNonHtmlScript)
				inScriptType = eNonHtmlScriptPreProc;
			else
				inScriptType = eNonHtmlPreProc;

			if (chNext2 == '@') {
				i += 2; // place as if it was the second next char treated
				visibleChars += 2;
				state = SCE_H_ASPAT;
			} else if ((chNext2 == '-') && (styler.SafeGetCharAt(i + 3) == '-')) {
				styler.ColourTo(i + 3, SCE_H_ASP);
				state = SCE_H_XCCOMMENT;
				scriptLanguage = eScriptVBS;
				continue;
			} else {
				if (chNext2 == '=') {
					i += 2; // place as if it was the second next char treated
					visibleChars += 2;
				} else {
					i++; // place as if it was the next char treated
					visibleChars++;
				}

				state = StateForScript(aspScript);
			}
			scriptLanguage = eScriptVBS;
			styler.ColourTo(i, SCE_H_ASP);
			// fold whole script
			if (foldHTMLPreprocessor)
				levelCurrent++;
			// should be better
			ch = styler.SafeGetCharAt(i);
			continue;
		}

		/////////////////////////////////////
		// handle the start of SGML language (DTD)
		else if (((scriptLanguage == eScriptNone) || (scriptLanguage == eScriptXML)) &&
				 (chPrev == '<') &&
				 (ch == '!') &&
				 (StateToPrint != SCE_H_CDATA) &&
				 (!IsCommentState(StateToPrint)) &&
				 (!IsScriptCommentState(StateToPrint)) ) {
			beforePreProc = state;
			styler.ColourTo(i - 2, StateToPrint);
			if ((chNext == '-') && (chNext2 == '-')) {
				state = SCE_H_COMMENT; // wait for a pending command
				styler.ColourTo(i + 2, SCE_H_COMMENT);
				i += 2; // follow styling after the --
			} else if (isWordCdata(i + 1, i + 7, styler)) {
				state = SCE_H_CDATA;
			} else {
				styler.ColourTo(i, SCE_H_SGML_DEFAULT); // <! is default
				scriptLanguage = eScriptSGML;
				state = SCE_H_SGML_COMMAND; // wait for a pending command
			}
			// fold whole tag (-- when closing the tag)
			if (foldHTMLPreprocessor)
				levelCurrent++;
			continue;
		}

		// handle the end of a pre-processor = Non-HTML
		else if ((
		             ((inScriptType == eNonHtmlPreProc)
		              || (inScriptType == eNonHtmlScriptPreProc)) && (
		                 ((scriptLanguage != eScriptNone) && stateAllowsTermination(state) && ((ch == '%') || (ch == '?')))
		             ) && (chNext == '>')) ||
		         ((scriptLanguage == eScriptSGML) && (ch == '>') && (state != SCE_H_SGML_COMMENT))) {
			if (state == SCE_H_ASPAT) {
				aspScript = segIsScriptingIndicator(styler,
				                                    styler.GetStartSegment(), i - 1, aspScript);
			}
			// Bounce out of any ASP mode
			switch (state) {
			case SCE_HJ_WORD:
				classifyWordHTJS(styler.GetStartSegment(), i - 1, keywords2, styler, inScriptType);
				break;
			case SCE_HB_WORD:
				classifyWordHTVB(styler.GetStartSegment(), i - 1, keywords3, styler, inScriptType);
				break;
			case SCE_HP_WORD:
				classifyWordHTPy(styler.GetStartSegment(), i - 1, keywords4, styler, prevWord, inScriptType);
				break;
			case SCE_HPHP_WORD:
				classifyWordHTPHP(styler.GetStartSegment(), i - 1, keywords5, styler);
				break;
			case SCE_H_XCCOMMENT:
				styler.ColourTo(i - 1, state);
				break;
			default :
				styler.ColourTo(i - 1, StateToPrint);