lexhtml.cxx

一个可以提供语法高亮显示的编辑器

	for (int i = startPos; i < lengthDoc; i++) {
		const char chPrev2 = chPrev;
		chPrev = ch;
		if (ch != ' ' && ch != '\t')
			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:
				// 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_COMMENTLINE:
			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 :
				// 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 == '?')) {
			styler.ColourTo(i - 1, StateToPrint);
			beforePreProc = state;
			scriptLanguage = segIsScriptingIndicator(styler, styler.GetStartSegment() + 2, i + 10, eScriptPHP);
			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
			if (foldHTMLPreprocessor){
				levelCurrent++;
				if (scriptLanguage == eScriptXML)
					levelCurrent--; // no folding of the XML first tag (all XML-like tags in this case)
			}
			// should be better
			ch = styler.SafeGetCharAt(i);
			continue;
		}

		// handle the start of ASP pre-processor = Non-HTML
		else if (!isCommentASPState(state) && (ch == '<') && (chNext == '%')) {
			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))) {
			beforePreProc = state;
			styler.ColourTo(i - 2, StateToPrint);
			if ((chNext == '-') && (chNext2 == '-')) {
				state = SCE_H_COMMENT; // wait for a pending command
			}
			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 == eScriptPHP) && (ch == '?') && !isPHPStringState(state) && (state != SCE_HPHP_COMMENT)) ||
		                 ((scriptLanguage != eScriptNone) && !isStringState(state) &&
		                  (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);
				break;
			}
			if (scriptLanguage != eScriptSGML) {
				i++;
				visibleChars++;
			}
			if (ch == '%')
				styler.ColourTo(i, SCE_H_ASP);
			else if (scriptLanguage == eScriptXML)
				styler.ColourTo(i, SCE_H_XMLEND);
			else if (scriptLanguage == eScriptSGML)
				styler.ColourTo(i, SCE_H_SGML_DEFAULT);
			else
				styler.ColourTo(i, SCE_H_QUESTION);
			state = beforePreProc;
			if (inScriptType == eNonHtmlScriptPreProc)
				inScriptType = eNonHtmlScript;
			else
				inScriptType = eHtml;
			scriptLanguage = eScriptNone;
			// unfold all scripting languages
			if (foldHTMLPreprocessor)
				levelCurrent--;
			continue;
		}
		/////////////////////////////////////

		switch (state) {
		case SCE_H_DEFAULT:
			if (ch == '<') {
				// in HTML, fold on tag open and unfold on tag close
				tagOpened = true;
				tagClosing = (chNext == '/');
				styler.ColourTo(i - 1, StateToPrint);
				if (chNext != '!')
					state = SCE_H_TAGUNKNOWN;
			} else if (ch == '&') {
				styler.ColourTo(i - 1, SCE_H_DEFAULT);
				state = SCE_H_ENTITY;
			}
			break;
		case SCE_H_SGML_DEFAULT:
		case SCE_H_SGML_BLOCK_DEFAULT:
//			if (scriptLanguage == eScriptSGMLblock)
//				StateToPrint = SCE_H_SGML_BLOCK_DEFAULT;

			if (ch == '\"') {
				styler.ColourTo(i - 1, StateToPrint);
				state = SCE_H_SGML_DOUBLESTRING;
			} else if (ch == '\'') {
				styler.ColourTo(i - 1, StateToPrint);
				state = SCE_H_SGML_SIMPLESTRING;
			} else if ((ch == '-') && (chPrev == '-')) {
				styler.ColourTo(i - 2, StateToPrint);
				state = SCE_H_SGML_COMMENT;
			} else if (isalpha(ch) && (chPrev == '%')) {
				styler.ColourTo(i - 2, StateToPrint);
				state = SCE_H_SGML_ENTITY;
			} else if (ch == '#') {
				styler.ColourTo(i - 1, StateToPrint);
				state = SCE_H_SGML_SPECIAL;
			} else if (ch == '[') {
				styler.ColourTo(i - 1, StateToPrint);
				scriptLanguage = eScriptSGMLblock;
				state = SCE_H_SGML_BLOCK_DEFAULT;
			} else if (ch == ']') {
				if (scriptLanguage == eScriptSGMLblock) {
					styler.ColourTo(i, StateToPrint);
					scriptLanguage = eScriptSGML;
				} else {
					styler.ColourTo(i - 1, StateToPrint);
					styler.ColourTo(i, SCE_H_SGML_ERROR);
				}
				state = SCE_H_SGML_DEFAULT;
			} else if (scriptLanguage == eScriptSGMLblock) {
				if ((ch == '!') && (chPrev == '<')) {
					styler.ColourTo(i - 2, StateToPrint);
					styler.ColourTo(i, SCE_H_SGML_DEFAULT);
					state = SCE_H_SGML_COMMAND;
				} else if (ch == '>') {
					styler.ColourTo(i - 1, StateToPrint);
					styler.ColourTo(i, SCE_H_SGML_DEFAULT);
				}
			}
			break;
		case SCE_H_SGML_COMMAND:
			if ((ch == '-') && (chPrev == '-')) {
				styler.ColourTo(i - 2, StateToPrint);
				state = SCE_H_SGML_COMMENT;
			} else if (!issgmlwordchar(ch)) {
				if (isWordHSGML(styler.GetStartSegment(), i - 1, keywords6, styler)) {
					styler.ColourTo(i - 1, StateToPrint);
					state = SCE_H_SGML_1ST_PARAM;
				} else {
					state = SCE_H_SGML_ERROR;
				}
			}
			break;
		case SCE_H_SGML_1ST_PARAM:
			// wait for the beginning of the word
			if ((ch == '-') && (chPrev == '-')) {
				if (scriptLanguage == eScriptSGMLblock) {
					styler.ColourTo(i - 2, SCE_H_SGML_BLOCK_DEFAULT);
				} else {
					styler.ColourTo(i - 2, SCE_H_SGML_DEFAULT);
				}
				state = SCE_H_SGML_1ST_PARAM_COMMENT;
			} else if (issgmlwordchar(ch)) {
				if (scriptLanguage == eScriptSGMLblock) {
					styler.ColourTo(i - 1, SCE_H_SGML_BLOCK_DEFAULT);
				} else {
					styler.ColourTo(i - 1, SCE_H_SGML_DEFAULT);
				}
				// find the length of the word
				int size = 1;
				while (ishtmlwordchar(styler.SafeGetCharAt(i + size)))
					size++;
				styler.ColourTo(i + size - 1, StateToPrint);
				i += size - 1;
				visibleChars += size - 1;
				ch = styler.SafeGetCharAt(i);
				if (scriptLanguage == eScriptSGMLblock) {
					state = SCE_H_SGML_BLOCK_DEFAULT;
				} else {
					state = SCE_H_SGML_DEFAULT;
				}
				continue;
			}
			break;
		case SCE_H_SGML_ERROR:
			if ((ch == '-') && (chPrev == '-')) {
				styler.ColourTo(i - 2, StateToPrint);
				state = SCE_H_SGML_COMMENT;
			}
		case SCE_H_SGML_DOUBLESTRING:
			if (ch == '\"') {
				styler.ColourTo(i, StateToPrint);
				state = SCE_H_SGML_DEFAULT;
			}
			break;
		case SCE_H_SGML_SIMPLESTRING:
			if (ch == '\'') {
				styler.ColourTo(i, StateToPrint);
				state = SCE_H_SGML_DEFAULT;
			}
			break;
		case SCE_H_SGML_COMMENT:
			if ((ch == '-') && (chPrev == '-')) {
				styler.ColourTo(i, StateToPrint);
				state = SCE_H_SGML_DEFAULT;
			}
			break;
		case SCE_H_CDATA:
			if ((chPrev2 == ']') && (chPrev == ']') && (ch == '>')) {
				styler.ColourTo(i, StateToPrint);
				state = SCE_H_DEFAULT;
				levelCurrent--;
			}
			break;
		case SCE_H_COMMENT:
			if ((chPrev2 == '-') && (chPrev == '-') && (ch == '>')) {
				styler.ColourTo(i, StateToPrint);
				state = SCE_H_DEFAULT;
				levelCurrent--;
			}
			break;
		case SCE_H_SGML_1ST_PARAM_COMMENT:
			if ((ch == '-') && (chPrev == '-')) {
				styler.ColourTo(i, SCE_H_SGML_COMMENT);
				state = SCE_H_SGML_1ST_PARAM;
			}
			break;
		case SCE_H_SGML_SPECIAL:
			if (!isupper(ch)) {
				styler.ColourTo(i - 1, StateToPrint);
				if (isalnum(ch)) {
					state = SCE_H_SGML_ERROR;
				} else {
					state = SCE_H_SGML_DEFAULT;
				}
			}
			break;
		case SCE_H_SGML_ENTITY:
			if (ch == ';') {
				styler.ColourTo(i, StateToPrint);
				state = SCE_H_SGML_DEFAULT;
			} else if (!isalnum(ch) && ch != '-' && ch != '.') {
				styler.ColourTo(i, SCE_H_SGML_ERROR);
				state = SCE_H_SGML_DEFAULT;
			}
			break;
		case SCE_H_ENTITY:
			if (ch == ';') {
				styler.ColourTo(i, StateToPrint);
				state = SCE_H_DEFAULT;
			}
			if (ch != '#' && !isalnum(ch)) {	// Should check that '#' follows '&', but it is unlikely anyway...
				styler.ColourTo(i, SCE_H_TAGUNKNOWN);
				state = SCE_H_DEFAULT;
			}
			break;
		case SCE_H_TAGUNKNOWN:
			if (!ishtmlwordchar(ch) && !((ch == '/') && (chPrev == '<')) && ch != '[') {
				int eClass = classifyTagHTML(styler.GetStartSegment(),
					i - 1, keywords, styler, tagDontFold, caseSensitive);
				if (eClass == SCE_H_SCRIPT) {
					if (!tagClosing) {
						inScriptType = eNonHtmlScript;
						scriptLanguage = clientScript;
						eClass = SCE_H_TAG;
					} else {
						scriptLanguage = eScriptNone;
						eClass = SCE_H_TAG;
					}
				}
				if (ch == '>') {
					styler.ColourTo(i, eClass);
					if (inScriptType == eNonHtmlScript) {
						state = StateForScript(scriptLanguage);
					} else {
						state = SCE_H_DEFAULT;
					}
					tagOpened = false;
					if (!tagDontFold){