PageRenderTime 44ms CodeModel.GetById 7ms RepoModel.GetById 0ms app.codeStats 1ms

/Branches/4.0/Scintilla/src/Lexers/LexHTML.cxx

#
C++ | 2131 lines | 1904 code | 118 blank | 109 comment | 1394 complexity | 7119e180093fefd3798b01ae02bf5b27 MD5 | raw file
Possible License(s): MIT

Large files files are truncated, but you can click here to view the full file

    

  1. // Scintilla source code edit control

    2. 

  2. /** @file LexHTML.cxx

    3. 

  3.  ** Lexer for HTML.

    4. 

  4.  **/

    5. 

  5. // Copyright 1998-2005 by Neil Hodgson <neilh@scintilla.org>

    6. 

  6. // The License.txt file describes the conditions under which this software may be distributed.

    7. 

  7. 

    8. 

  8. #include <stdlib.h>

    9. 

  9. #include <string.h>

    10. 

  10. #include <ctype.h>

    11. 

  11. #include <stdio.h>

    12. 

  12. #include <stdarg.h>

    13. 

  13. 

    14. 

  14. #include "Platform.h"

    15. 

  15. 

    16. 

  16. #include "PropSet.h"

    17. 

  17. #include "Accessor.h"

    18. 

  18. #include "StyleContext.h"

    19. 

  19. #include "KeyWords.h"

    20. 

  20. #include "Scintilla.h"

    21. 

  21. #include "SciLexer.h"

    22. 

  22. #include "CharacterSet.h"

    23. 

  23. 

    24. 

  24. #ifdef SCI_NAMESPACE

    25. 

  25. using namespace Scintilla;

    26. 

  26. #endif

    27. 

  27. 

    28. 

  28. #define SCE_HA_JS (SCE_HJA_START - SCE_HJ_START)

    29. 

  29. #define SCE_HA_VBS (SCE_HBA_START - SCE_HB_START)

    30. 

  30. #define SCE_HA_PYTHON (SCE_HPA_START - SCE_HP_START)

    31. 

  31. 

    32. 

  32. enum script_type { eScriptNone = 0, eScriptJS, eScriptVBS, eScriptPython, eScriptPHP, eScriptXML, eScriptSGML, eScriptSGMLblock, eScriptComment };

    33. 

  33. enum script_mode { eHtml = 0, eNonHtmlScript, eNonHtmlPreProc, eNonHtmlScriptPreProc };

    34. 

  34. 

    35. 

  35. static inline bool IsAWordChar(const int ch) {

    36. 

  36. 	return (ch < 0x80) && (isalnum(ch) || ch == '.' || ch == '_');

    37. 

  37. }

    38. 

  38. 

    39. 

  39. static inline bool IsAWordStart(const int ch) {

    40. 

  40. 	return (ch < 0x80) && (isalnum(ch) || ch == '_');

    41. 

  41. }

    42. 

  42. 

    43. 

  43. inline bool IsOperator(int ch) {

    44. 

  44. 	if (isascii(ch) && isalnum(ch))

    45. 

  45. 		return false;

    46. 

  46. 	// '.' left out as it is used to make up numbers

    47. 

  47. 	if (ch == '%' || ch == '^' || ch == '&' || ch == '*' ||

    48. 

  48. 	        ch == '(' || ch == ')' || ch == '-' || ch == '+' ||

    49. 

  49. 	        ch == '=' || ch == '|' || ch == '{' || ch == '}' ||

    50. 

  50. 	        ch == '[' || ch == ']' || ch == ':' || ch == ';' ||

    51. 

  51. 	        ch == '<' || ch == '>' || ch == ',' || ch == '/' ||

    52. 

  52. 	        ch == '?' || ch == '!' || ch == '.' || ch == '~')

    53. 

  53. 		return true;

    54. 

  54. 	return false;

    55. 

  55. }

    56. 

  56. 

    57. 

  57. static inline int MakeLowerCase(int ch) {

    58. 

  58. 	if (ch < 'A' || ch > 'Z')

    59. 

  59. 		return ch;

    60. 

  60. 	else

    61. 

  61. 		return ch - 'A' + 'a';

    62. 

  62. }

    63. 

  63. 

    64. 

  64. static void GetTextSegment(Accessor &styler, unsigned int start, unsigned int end, char *s, size_t len) {

    65. 

  65. 	size_t i = 0;

    66. 

  66. 	for (; (i < end - start + 1) && (i < len-1); i++) {

    67. 

  67. 		s[i] = static_cast<char>(MakeLowerCase(styler[start + i]));

    68. 

  68. 	}

    69. 

  69. 	s[i] = '\0';

    70. 

  70. }

    71. 

  71. 

    72. 

  72. static const char *GetNextWord(Accessor &styler, unsigned int start, char *s, size_t sLen) {

    73. 

  73. 

    74. 

  74. 	size_t i = 0;

    75. 

  75. 	for (; i < sLen-1; i++) {

    76. 

  76. 		char ch = static_cast<char>(styler.SafeGetCharAt(start + i));

    77. 

  77. 		if ((i == 0) && !IsAWordStart(ch))

    78. 

  78. 			break;

    79. 

  79. 		if ((i > 0) && !IsAWordChar(ch))

    80. 

  80. 			break;

    81. 

  81. 		s[i] = ch;

    82. 

  82. 	}

    83. 

  83. 	s[i] = '\0';

    84. 

  84. 

    85. 

  85. 	return s;

    86. 

  86. }

    87. 

  87. 

    88. 

  88. static script_type segIsScriptingIndicator(Accessor &styler, unsigned int start, unsigned int end, script_type prevValue) {

    89. 

  89. 	char s[100];

    90. 

  90. 	GetTextSegment(styler, start, end, s, sizeof(s));

    91. 

  91. 	//Platform::DebugPrintf("Scripting indicator [%s]\n", s);

    92. 

  92. 	if (strstr(s, "src"))	// External script

    93. 

  93. 		return eScriptNone;

    94. 

  94. 	if (strstr(s, "vbs"))

    95. 

  95. 		return eScriptVBS;

    96. 

  96. 	if (strstr(s, "pyth"))

    97. 

  97. 		return eScriptPython;

    98. 

  98. 	if (strstr(s, "javas"))

    99. 

  99. 		return eScriptJS;

    100. 

  100. 	if (strstr(s, "jscr"))

    101. 

  101. 		return eScriptJS;

    102. 

  102. 	if (strstr(s, "php"))

    103. 

  103. 		return eScriptPHP;

    104. 

  104. 	if (strstr(s, "xml")) {

    105. 

  105. 		const char *xml = strstr(s, "xml");

    106. 

  106. 		for (const char *t=s; t<xml; t++) {

    107. 

  107. 			if (!IsASpace(*t)) {

    108. 

  108. 				return prevValue;

    109. 

  109. 			}

    110. 

  110. 		}

    111. 

  111. 		return eScriptXML;

    112. 

  112. 	}

    113. 

  113. 

    114. 

  114. 	return prevValue;

    115. 

  115. }

    116. 

  116. 

    117. 

  117. static int PrintScriptingIndicatorOffset(Accessor &styler, unsigned int start, unsigned int end) {

    118. 

  118. 	int iResult = 0;

    119. 

  119. 	char s[100];

    120. 

  120. 	GetTextSegment(styler, start, end, s, sizeof(s));

    121. 

  121. 	if (0 == strncmp(s, "php", 3)) {

    122. 

  122. 		iResult = 3;

    123. 

  123. 	}

    124. 

  124. 

    125. 

  125. 	return iResult;

    126. 

  126. }

    127. 

  127. 

    128. 

  128. static script_type ScriptOfState(int state) {

    129. 

  129. 	if ((state >= SCE_HP_START) && (state <= SCE_HP_IDENTIFIER)) {

    130. 

  130. 		return eScriptPython;

    131. 

  131. 	} else if ((state >= SCE_HB_START) && (state <= SCE_HB_STRINGEOL)) {

    132. 

  132. 		return eScriptVBS;

    133. 

  133. 	} else if ((state >= SCE_HJ_START) && (state <= SCE_HJ_REGEX)) {

    134. 

  134. 		return eScriptJS;

    135. 

  135. 	} else if ((state >= SCE_HPHP_DEFAULT) && (state <= SCE_HPHP_COMMENTLINE)) {

    136. 

  136. 		return eScriptPHP;

    137. 

  137. 	} else if ((state >= SCE_H_SGML_DEFAULT) && (state < SCE_H_SGML_BLOCK_DEFAULT)) {

    138. 

  138. 		return eScriptSGML;

    139. 

  139. 	} else if (state == SCE_H_SGML_BLOCK_DEFAULT) {

    140. 

  140. 		return eScriptSGMLblock;

    141. 

  141. 	} else {

    142. 

  142. 		return eScriptNone;

    143. 

  143. 	}

    144. 

  144. }

    145. 

  145. 

    146. 

  146. static int statePrintForState(int state, script_mode inScriptType) {

    147. 

  147. 	int StateToPrint = state;

    148. 

  148. 

    149. 

  149. 	if (state >= SCE_HJ_START) {

    150. 

  150. 		if ((state >= SCE_HP_START) && (state <= SCE_HP_IDENTIFIER)) {

    151. 

  151. 			StateToPrint = state + ((inScriptType == eNonHtmlScript) ? 0 : SCE_HA_PYTHON);

    152. 

  152. 		} else if ((state >= SCE_HB_START) && (state <= SCE_HB_STRINGEOL)) {

    153. 

  153. 			StateToPrint = state + ((inScriptType == eNonHtmlScript) ? 0 : SCE_HA_VBS);

    154. 

  154. 		} else if ((state >= SCE_HJ_START) && (state <= SCE_HJ_REGEX)) {

    155. 

  155. 			StateToPrint = state + ((inScriptType == eNonHtmlScript) ? 0 : SCE_HA_JS);

    156. 

  156. 		}

    157. 

  157. 	}

    158. 

  158. 

    159. 

  159. 	return StateToPrint;

    160. 

  160. }

    161. 

  161. 

    162. 

  162. static int stateForPrintState(int StateToPrint) {

    163. 

  163. 	int state;

    164. 

  164. 

    165. 

  165. 	if ((StateToPrint >= SCE_HPA_START) && (StateToPrint <= SCE_HPA_IDENTIFIER)) {

    166. 

  166. 		state = StateToPrint - SCE_HA_PYTHON;

    167. 

  167. 	} else if ((StateToPrint >= SCE_HBA_START) && (StateToPrint <= SCE_HBA_STRINGEOL)) {

    168. 

  168. 		state = StateToPrint - SCE_HA_VBS;

    169. 

  169. 	} else if ((StateToPrint >= SCE_HJA_START) && (StateToPrint <= SCE_HJA_REGEX)) {

    170. 

  170. 		state = StateToPrint - SCE_HA_JS;

    171. 

  171. 	} else {

    172. 

  172. 		state = StateToPrint;

    173. 

  173. 	}

    174. 

  174. 

    175. 

  175. 	return state;

    176. 

  176. }

    177. 

  177. 

    178. 

  178. static inline bool IsNumber(unsigned int start, Accessor &styler) {

    179. 

  179. 	return IsADigit(styler[start]) || (styler[start] == '.') ||

    180. 

  180. 	       (styler[start] == '-') || (styler[start] == '#');

    181. 

  181. }

    182. 

  182. 

    183. 

  183. static inline bool isStringState(int state) {

    184. 

  184. 	bool bResult;

    185. 

  185. 

    186. 

  186. 	switch (state) {

    187. 

  187. 	case SCE_HJ_DOUBLESTRING:

    188. 

  188. 	case SCE_HJ_SINGLESTRING:

    189. 

  189. 	case SCE_HJA_DOUBLESTRING:

    190. 

  190. 	case SCE_HJA_SINGLESTRING:

    191. 

  191. 	case SCE_HB_STRING:

    192. 

  192. 	case SCE_HBA_STRING:

    193. 

  193. 	case SCE_HP_STRING:

    194. 

  194. 	case SCE_HP_CHARACTER:

    195. 

  195. 	case SCE_HP_TRIPLE:

    196. 

  196. 	case SCE_HP_TRIPLEDOUBLE:

    197. 

  197. 	case SCE_HPA_STRING:

    198. 

  198. 	case SCE_HPA_CHARACTER:

    199. 

  199. 	case SCE_HPA_TRIPLE:

    200. 

  200. 	case SCE_HPA_TRIPLEDOUBLE:

    201. 

  201. 	case SCE_HPHP_HSTRING:

    202. 

  202. 	case SCE_HPHP_SIMPLESTRING:

    203. 

  203. 	case SCE_HPHP_HSTRING_VARIABLE:

    204. 

  204. 	case SCE_HPHP_COMPLEX_VARIABLE:

    205. 

  205. 		bResult = true;

    206. 

  206. 		break;

    207. 

  207. 	default :

    208. 

  208. 		bResult = false;

    209. 

  209. 		break;

    210. 

  210. 	}

    211. 

  211. 	return bResult;

    212. 

  212. }

    213. 

  213. 

    214. 

  214. static inline bool stateAllowsTermination(int state) {

    215. 

  215. 	bool allowTermination = !isStringState(state);

    216. 

  216. 	if (allowTermination) {

    217. 

  217. 		switch (state) {

    218. 

  218. 		case SCE_HB_COMMENTLINE:

    219. 

  219. 		case SCE_HPHP_COMMENT:

    220. 

  220. 		case SCE_HP_COMMENTLINE:

    221. 

  221. 		case SCE_HPA_COMMENTLINE:

    222. 

  222. 			allowTermination = false;

    223. 

  223. 		}

    224. 

  224. 	}

    225. 

  225. 	return allowTermination;

    226. 

  226. }

    227. 

  227. 

    228. 

  228. // not really well done, since it's only comments that should lex the %> and <%

    229. 

  229. static inline bool isCommentASPState(int state) {

    230. 

  230. 	bool bResult;

    231. 

  231. 

    232. 

  232. 	switch (state) {

    233. 

  233. 	case SCE_HJ_COMMENT:

    234. 

  234. 	case SCE_HJ_COMMENTLINE:

    235. 

  235. 	case SCE_HJ_COMMENTDOC:

    236. 

  236. 	case SCE_HB_COMMENTLINE:

    237. 

  237. 	case SCE_HP_COMMENTLINE:

    238. 

  238. 	case SCE_HPHP_COMMENT:

    239. 

  239. 	case SCE_HPHP_COMMENTLINE:

    240. 

  240. 		bResult = true;

    241. 

  241. 		break;

    242. 

  242. 	default :

    243. 

  243. 		bResult = false;

    244. 

  244. 		break;

    245. 

  245. 	}

    246. 

  246. 	return bResult;

    247. 

  247. }

    248. 

  248. 

    249. 

  249. static void classifyAttribHTML(unsigned int start, unsigned int end, WordList &keywords, Accessor &styler) {

    250. 

  250. 	bool wordIsNumber = IsNumber(start, styler);

    251. 

  251. 	char chAttr = SCE_H_ATTRIBUTEUNKNOWN;

    252. 

  252. 	if (wordIsNumber) {

    253. 

  253. 		chAttr = SCE_H_NUMBER;

    254. 

  254. 	} else {

    255. 

  255. 		char s[100];

    256. 

  256. 		GetTextSegment(styler, start, end, s, sizeof(s));

    257. 

  257. 		if (keywords.InList(s))

    258. 

  258. 			chAttr = SCE_H_ATTRIBUTE;

    259. 

  259. 	}

    260. 

  260. 	if ((chAttr == SCE_H_ATTRIBUTEUNKNOWN) && !keywords)

    261. 

  261. 		// No keywords -> all are known

    262. 

  262. 		chAttr = SCE_H_ATTRIBUTE;

    263. 

  263. 	styler.ColourTo(end, chAttr);

    264. 

  264. }

    265. 

  265. 

    266. 

  266. static int classifyTagHTML(unsigned int start, unsigned int end,

    267. 

  267.                            WordList &keywords, Accessor &styler, bool &tagDontFold,

    268. 

  268. 			   bool caseSensitive, bool isXml, bool allowScripts) {

    269. 

  269. 	char s[30 + 2];

    270. 

  270. 	// Copy after the '<'

    271. 

  271. 	unsigned int i = 0;

    272. 

  272. 	for (unsigned int cPos = start; cPos <= end && i < 30; cPos++) {

    273. 

  273. 		char ch = styler[cPos];

    274. 

  274. 		if ((ch != '<') && (ch != '/')) {

    275. 

  275. 			s[i++] = caseSensitive ? ch : static_cast<char>(MakeLowerCase(ch));

    276. 

  276. 		}

    277. 

  277. 	}

    278. 

  278. 

    279. 

  279. 	//The following is only a quick hack, to see if this whole thing would work

    280. 

  280. 	//we first need the tagname with a trailing space...

    281. 

  281. 	s[i] = ' ';

    282. 

  282. 	s[i+1] = '\0';

    283. 

  283. 

    284. 

  284. 	// if the current language is XML, I can fold any tag

    285. 

  285. 	// if the current language is HTML, I don't want to fold certain tags (input, meta, etc.)

    286. 

  286. 	//...to find it in the list of no-container-tags

    287. 

  287. 	tagDontFold = (!isXml) && (NULL != strstr("meta link img area br hr input ", s));

    288. 

  288. 

    289. 

  289. 	//now we can remove the trailing space

    290. 

  290. 	s[i] = '\0';

    291. 

  291. 

    292. 

  292. 	// No keywords -> all are known

    293. 

  293. 	char chAttr = SCE_H_TAGUNKNOWN;

    294. 

  294. 	if (s[0] == '!') {

    295. 

  295. 		chAttr = SCE_H_SGML_DEFAULT;

    296. 

  296. 	} else if (!keywords || keywords.InList(s)) {

    297. 

  297. 		chAttr = SCE_H_TAG;

    298. 

  298. 	}

    299. 

  299. 	styler.ColourTo(end, chAttr);

    300. 

  300. 	if (chAttr == SCE_H_TAG) {

    301. 

  301. 		if (allowScripts && 0 == strcmp(s, "script")) {

    302. 

  302. 			// check to see if this is a self-closing tag by sniffing ahead

    303. 

  303. 			bool isSelfClose = false;

    304. 

  304. 			for (unsigned int cPos = end; cPos <= end + 100; cPos++) {

    305. 

  305. 				char ch = styler.SafeGetCharAt(cPos, '\0');

    306. 

  306. 				if (ch == '\0' || ch == '>')

    307. 

  307. 					break;

    308. 

  308. 				else if (ch == '/' && styler.SafeGetCharAt(cPos + 1, '\0') == '>') {

    309. 

  309. 					isSelfClose = true;

    310. 

  310. 					break;

    311. 

  311. 				}

    312. 

  312. 			}

    313. 

  313. 

    314. 

  314. 			// do not enter a script state if the tag self-closed

    315. 

  315. 			if (!isSelfClose)

    316. 

  316. 				chAttr = SCE_H_SCRIPT;

    317. 

  317. 		} else if (!isXml && 0 == strcmp(s, "comment")) {

    318. 

  318. 			chAttr = SCE_H_COMMENT;

    319. 

  319. 		}

    320. 

  320. 	}

    321. 

  321. 	return chAttr;

    322. 

  322. }

    323. 

  323. 

    324. 

  324. static void classifyWordHTJS(unsigned int start, unsigned int end,

    325. 

  325.                              WordList &keywords, Accessor &styler, script_mode inScriptType) {

    326. 

  326. 	char chAttr = SCE_HJ_WORD;

    327. 

  327. 	bool wordIsNumber = IsADigit(styler[start]) || (styler[start] == '.');

    328. 

  328. 	if (wordIsNumber)

    329. 

  329. 		chAttr = SCE_HJ_NUMBER;

    330. 

  330. 	else {

    331. 

  331. 		char s[30 + 1];

    332. 

  332. 		unsigned int i = 0;

    333. 

  333. 		for (; i < end - start + 1 && i < 30; i++) {

    334. 

  334. 			s[i] = styler[start + i];

    335. 

  335. 		}

    336. 

  336. 		s[i] = '\0';

    337. 

  337. 		if (keywords.InList(s))

    338. 

  338. 			chAttr = SCE_HJ_KEYWORD;

    339. 

  339. 	}

    340. 

  340. 	styler.ColourTo(end, statePrintForState(chAttr, inScriptType));

    341. 

  341. }

    342. 

  342. 

    343. 

  343. static int classifyWordHTVB(unsigned int start, unsigned int end, WordList &keywords, Accessor &styler, script_mode inScriptType) {

    344. 

  344. 	char chAttr = SCE_HB_IDENTIFIER;

    345. 

  345. 	bool wordIsNumber = IsADigit(styler[start]) || (styler[start] == '.');

    346. 

  346. 	if (wordIsNumber)

    347. 

  347. 		chAttr = SCE_HB_NUMBER;

    348. 

  348. 	else {

    349. 

  349. 		char s[100];

    350. 

  350. 		GetTextSegment(styler, start, end, s, sizeof(s));

    351. 

  351. 		if (keywords.InList(s)) {

    352. 

  352. 			chAttr = SCE_HB_WORD;

    353. 

  353. 			if (strcmp(s, "rem") == 0)

    354. 

  354. 				chAttr = SCE_HB_COMMENTLINE;

    355. 

  355. 		}

    356. 

  356. 	}

    357. 

  357. 	styler.ColourTo(end, statePrintForState(chAttr, inScriptType));

    358. 

  358. 	if (chAttr == SCE_HB_COMMENTLINE)

    359. 

  359. 		return SCE_HB_COMMENTLINE;

    360. 

  360. 	else

    361. 

  361. 		return SCE_HB_DEFAULT;

    362. 

  362. }

    363. 

  363. 

    364. 

  364. static void classifyWordHTPy(unsigned int start, unsigned int end, WordList &keywords, Accessor &styler, char *prevWord, script_mode inScriptType) {

    365. 

  365. 	bool wordIsNumber = IsADigit(styler[start]);

    366. 

  366. 	char s[30 + 1];

    367. 

  367. 	unsigned int i = 0;

    368. 

  368. 	for (; i < end - start + 1 && i < 30; i++) {

    369. 

  369. 		s[i] = styler[start + i];

    370. 

  370. 	}

    371. 

  371. 	s[i] = '\0';

    372. 

  372. 	char chAttr = SCE_HP_IDENTIFIER;

    373. 

  373. 	if (0 == strcmp(prevWord, "class"))

    374. 

  374. 		chAttr = SCE_HP_CLASSNAME;

    375. 

  375. 	else if (0 == strcmp(prevWord, "def"))

    376. 

  376. 		chAttr = SCE_HP_DEFNAME;

    377. 

  377. 	else if (wordIsNumber)

    378. 

  378. 		chAttr = SCE_HP_NUMBER;

    379. 

  379. 	else if (keywords.InList(s))

    380. 

  380. 		chAttr = SCE_HP_WORD;

    381. 

  381. 	styler.ColourTo(end, statePrintForState(chAttr, inScriptType));

    382. 

  382. 	strcpy(prevWord, s);

    383. 

  383. }

    384. 

  384. 

    385. 

  385. // Update the word colour to default or keyword

    386. 

  386. // Called when in a PHP word

    387. 

  387. static void classifyWordHTPHP(unsigned int start, unsigned int end, WordList &keywords, Accessor &styler) {

    388. 

  388. 	char chAttr = SCE_HPHP_DEFAULT;

    389. 

  389. 	bool wordIsNumber = IsADigit(styler[start]) || (styler[start] == '.' && start+1 <= end && IsADigit(styler[start+1]));

    390. 

  390. 	if (wordIsNumber)

    391. 

  391. 		chAttr = SCE_HPHP_NUMBER;

    392. 

  392. 	else {

    393. 

  393. 		char s[100];

    394. 

  394. 		GetTextSegment(styler, start, end, s, sizeof(s));

    395. 

  395. 		if (keywords.InList(s))

    396. 

  396. 			chAttr = SCE_HPHP_WORD;

    397. 

  397. 	}

    398. 

  398. 	styler.ColourTo(end, chAttr);

    399. 

  399. }

    400. 

  400. 

    401. 

  401. static bool isWordHSGML(unsigned int start, unsigned int end, WordList &keywords, Accessor &styler) {

    402. 

  402. 	char s[30 + 1];

    403. 

  403. 	unsigned int i = 0;

    404. 

  404. 	for (; i < end - start + 1 && i < 30; i++) {

    405. 

  405. 		s[i] = styler[start + i];

    406. 

  406. 	}

    407. 

  407. 	s[i] = '\0';

    408. 

  408. 	return keywords.InList(s);

    409. 

  409. }

    410. 

  410. 

    411. 

  411. static bool isWordCdata(unsigned int start, unsigned int end, Accessor &styler) {

    412. 

  412. 	char s[30 + 1];

    413. 

  413. 	unsigned int i = 0;

    414. 

  414. 	for (; i < end - start + 1 && i < 30; i++) {

    415. 

  415. 		s[i] = styler[start + i];

    416. 

  416. 	}

    417. 

  417. 	s[i] = '\0';

    418. 

  418. 	return (0 == strcmp(s, "[CDATA["));

    419. 

  419. }

    420. 

  420. 

    421. 

  421. // Return the first state to reach when entering a scripting language

    422. 

  422. static int StateForScript(script_type scriptLanguage) {

    423. 

  423. 	int Result;

    424. 

  424. 	switch (scriptLanguage) {

    425. 

  425. 	case eScriptVBS:

    426. 

  426. 		Result = SCE_HB_START;

    427. 

  427. 		break;

    428. 

  428. 	case eScriptPython:

    429. 

  429. 		Result = SCE_HP_START;

    430. 

  430. 		break;

    431. 

  431. 	case eScriptPHP:

    432. 

  432. 		Result = SCE_HPHP_DEFAULT;

    433. 

  433. 		break;

    434. 

  434. 	case eScriptXML:

    435. 

  435. 		Result = SCE_H_TAGUNKNOWN;

    436. 

  436. 		break;

    437. 

  437. 	case eScriptSGML:

    438. 

  438. 		Result = SCE_H_SGML_DEFAULT;

    439. 

  439. 		break;

    440. 

  440. 	case eScriptComment:

    441. 

  441. 		Result = SCE_H_COMMENT;

    442. 

  442. 		break;

    443. 

  443. 	default :

    444. 

  444. 		Result = SCE_HJ_START;

    445. 

  445. 		break;

    446. 

  446. 	}

    447. 

  447. 	return Result;

    448. 

  448. }

    449. 

  449. 

    450. 

  450. static inline bool ishtmlwordchar(int ch) {

    451. 

  451. 	return !isascii(ch) ||

    452. 

  452. 		(isalnum(ch) || ch == '.' || ch == '-' || ch == '_' || ch == ':' || ch == '!' || ch == '#');

    453. 

  453. }

    454. 

  454. 

    455. 

  455. static inline bool issgmlwordchar(int ch) {

    456. 

  456. 	return !isascii(ch) ||

    457. 

  457. 		(isalnum(ch) || ch == '.' || ch == '_' || ch == ':' || ch == '!' || ch == '#' || ch == '[');

    458. 

  458. }

    459. 

  459. 

    460. 

  460. static inline bool IsPhpWordStart(int ch) {

    461. 

  461. 	return (isascii(ch) && (isalpha(ch) || (ch == '_'))) || (ch >= 0x7f);

    462. 

  462. }

    463. 

  463. 

    464. 

  464. static inline bool IsPhpWordChar(int ch) {

    465. 

  465. 	return IsADigit(ch) || IsPhpWordStart(ch);

    466. 

  466. }

    467. 

  467. 

    468. 

  468. static bool InTagState(int state) {

    469. 

  469. 	return state == SCE_H_TAG || state == SCE_H_TAGUNKNOWN ||

    470. 

  470. 	       state == SCE_H_SCRIPT ||

    471. 

  471. 	       state == SCE_H_ATTRIBUTE || state == SCE_H_ATTRIBUTEUNKNOWN ||

    472. 

  472. 	       state == SCE_H_NUMBER || state == SCE_H_OTHER ||

    473. 

  473. 	       state == SCE_H_DOUBLESTRING || state == SCE_H_SINGLESTRING;

    474. 

  474. }

    475. 

  475. 

    476. 

  476. static bool IsCommentState(const int state) {

    477. 

  477. 	return state == SCE_H_COMMENT || state == SCE_H_SGML_COMMENT;

    478. 

  478. }

    479. 

  479. 

    480. 

  480. static bool IsScriptCommentState(const int state) {

    481. 

  481. 	return state == SCE_HJ_COMMENT || state == SCE_HJ_COMMENTLINE || state == SCE_HJA_COMMENT ||

    482. 

  482. 		   state == SCE_HJA_COMMENTLINE || state == SCE_HB_COMMENTLINE || state == SCE_HBA_COMMENTLINE;

    483. 

  483. }

    484. 

  484. 

    485. 

  485. static bool isLineEnd(int ch) {

    486. 

  486. 	return ch == '\r' || ch == '\n';

    487. 

  487. }

    488. 

  488. 

    489. 

  489. static bool isOKBeforeRE(int ch) {

    490. 

  490. 	return (ch == '(') || (ch == '=') || (ch == ',');

    491. 

  491. }

    492. 

  492. 

    493. 

  493. static bool isMakoBlockEnd(const int ch, const int chNext, const char *blockType) {

    494. 

  494. 	if (strlen(blockType) == 0) {

    495. 

  495. 		return ((ch == '%') && (chNext == '>'));

    496. 

  496. 	} else if ((0 == strcmp(blockType, "inherit")) ||

    497. 

  497. 			   (0 == strcmp(blockType, "namespace")) ||

    498. 

  498. 			   (0 == strcmp(blockType, "include")) ||

    499. 

  499. 			   (0 == strcmp(blockType, "page"))) {

    500. 

  500. 		return ((ch == '/') && (chNext == '>'));

    501. 

  501. 	} else if (0 == strcmp(blockType, "%")) {

    502. 

  502. 		return isLineEnd(ch);

    503. 

  503. 	} else if (0 == strcmp(blockType, "{")) {

    504. 

  504. 		return ch == '}';

    505. 

  505. 	} else {

    506. 

  506. 		return (ch == '>');

    507. 

  507. 	}

    508. 

  508. }

    509. 

  509. 

    510. 

  510. static bool isDjangoBlockEnd(const int ch, const int chNext, const char *blockType) {

    511. 

  511. 	if (strlen(blockType) == 0) {

    512. 

  512. 		return 0;

    513. 

  513. 	} else if (0 == strcmp(blockType, "%")) {

    514. 

  514. 		return ((ch == '%') && (chNext == '}'));

    515. 

  515. 	} else if (0 == strcmp(blockType, "{")) {

    516. 

  516. 		return ((ch == '}') && (chNext == '}'));

    517. 

  517. 	} else {

    518. 

  518. 		return 0;

    519. 

  519. 	}

    520. 

  520. }

    521. 

  521. 

    522. 

  522. static bool isPHPStringState(int state) {

    523. 

  523. 	return

    524. 

  524. 	    (state == SCE_HPHP_HSTRING) ||

    525. 

  525. 	    (state == SCE_HPHP_SIMPLESTRING) ||

    526. 

  526. 	    (state == SCE_HPHP_HSTRING_VARIABLE) ||

    527. 

  527. 	    (state == SCE_HPHP_COMPLEX_VARIABLE);

    528. 

  528. }

    529. 

  529. 

    530. 

  530. static int FindPhpStringDelimiter(char *phpStringDelimiter, const int phpStringDelimiterSize, int i, const int lengthDoc, Accessor &styler, bool &isSimpleString) {

    531. 

  531. 	int j;

    532. 

  532. 	const int beginning = i - 1;

    533. 

  533. 	bool isValidSimpleString = false;

    534. 

  534. 

    535. 

  535. 	while (i < lengthDoc && (styler[i] == ' ' || styler[i] == '\t'))

    536. 

  536. 		i++;

    537. 

  537. 

    538. 

  538. 	char ch = styler.SafeGetCharAt(i);

    539. 

  539. 	const char chNext = styler.SafeGetCharAt(i + 1);

    540. 

  540. 	if (!IsPhpWordStart(ch)) {

    541. 

  541. 		if (ch == '\'' && IsPhpWordStart(chNext)) {

    542. 

  542. 			i++;

    543. 

  543. 			ch = chNext;

    544. 

  544. 			isSimpleString = true;

    545. 

  545. 		} else {

    546. 

  546. 			phpStringDelimiter[0] = '\0';

    547. 

  547. 			return beginning;

    548. 

  548. 		}

    549. 

  549. 	}

    550. 

  550. 	phpStringDelimiter[0] = ch;

    551. 

  551. 	i++;

    552. 

  552. 

    553. 

  553. 	for (j = i; j < lengthDoc && !isLineEnd(styler[j]); j++) {

    554. 

  554. 		if (!IsPhpWordChar(styler[j])) {

    555. 

  555. 			if (isSimpleString && (styler[j] == '\'') && isLineEnd(styler.SafeGetCharAt(j + 1))) {

    556. 

  556. 				isValidSimpleString = true;

    557. 

  557. 				j++;

    558. 

  558. 				break;

    559. 

  559. 			} else {

    560. 

  560. 				phpStringDelimiter[0] = '\0';

    561. 

  561. 				return beginning;

    562. 

  562. 			}

    563. 

  563. 		}

    564. 

  564. 		if (j - i < phpStringDelimiterSize - 2)

    565. 

  565. 			phpStringDelimiter[j-i+1] = styler[j];

    566. 

  566. 		else

    567. 

  567. 			i++;

    568. 

  568. 	}

    569. 

  569. 	if (isSimpleString && !isValidSimpleString) {

    570. 

  570. 		phpStringDelimiter[0] = '\0';

    571. 

  571. 		return beginning;

    572. 

  572. 	}

    573. 

  573. 	phpStringDelimiter[j-i+1 - (isSimpleString ? 1 : 0)] = '\0';

    574. 

  574. 	return j - 1;

    575. 

  575. }

    576. 

  576. 

    577. 

  577. static void ColouriseHyperTextDoc(unsigned int startPos, int length, int initStyle, WordList *keywordlists[],

    578. 

  578.                                   Accessor &styler, bool isXml) {

    579. 

  579. 	WordList &keywords = *keywordlists[0];

    580. 

  580. 	WordList &keywords2 = *keywordlists[1];

    581. 

  581. 	WordList &keywords3 = *keywordlists[2];

    582. 

  582. 	WordList &keywords4 = *keywordlists[3];

    583. 

  583. 	WordList &keywords5 = *keywordlists[4];

    584. 

  584. 	WordList &keywords6 = *keywordlists[5]; // SGML (DTD) keywords

    585. 

  585. 

    586. 

  586. 	// Lexer for HTML requires more lexical states (8 bits worth) than most lexers

    587. 

  587. 	styler.StartAt(startPos, static_cast<char>(STYLE_MAX));

    588. 

  588. 	char prevWord[200];

    589. 

  589. 	prevWord[0] = '\0';

    590. 

  590. 	char phpStringDelimiter[200]; // PHP is not limited in length, we are

    591. 

  591. 	phpStringDelimiter[0] = '\0';

    592. 

  592. 	int StateToPrint = initStyle;

    593. 

  593. 	int state = stateForPrintState(StateToPrint);

    594. 

  594. 	char makoBlockType[200];

    595. 

  595. 	makoBlockType[0] = '\0';

    596. 

  596. 	char djangoBlockType[2];

    597. 

  597. 	djangoBlockType[0] = '\0';

    598. 

  598. 

    599. 

  599. 	// If inside a tag, it may be a script tag, so reread from the start to ensure any language tags are seen

    600. 

  600. 	if (InTagState(state)) {

    601. 

  601. 		while ((startPos > 0) && (InTagState(styler.StyleAt(startPos - 1)))) {

    602. 

  602. 			startPos--;

    603. 

  603. 			length++;

    604. 

  604. 		}

    605. 

  605. 		state = SCE_H_DEFAULT;

    606. 

  606. 	}

    607. 

  607. 	// String can be heredoc, must find a delimiter first. Reread from beginning of line containing the string, to get the correct lineState

    608. 

  608. 	if (isPHPStringState(state)) {

    609. 

  609. 		while (startPos > 0 && (isPHPStringState(state) || !isLineEnd(styler[startPos - 1]))) {

    610. 

  610. 			startPos--;

    611. 

  611. 			length++;

    612. 

  612. 			state = styler.StyleAt(startPos);

    613. 

  613. 		}

    614. 

  614. 		if (startPos == 0)

    615. 

  615. 			state = SCE_H_DEFAULT;

    616. 

  616. 	}

    617. 

  617. 	styler.StartAt(startPos, static_cast<char>(STYLE_MAX));

    618. 

  618. 

    619. 

  619. 	int lineCurrent = styler.GetLine(startPos);

    620. 

  620. 	int lineState;

    621. 

  621. 	if (lineCurrent > 0) {

    622. 

  622. 		lineState = styler.GetLineState(lineCurrent);

    623. 

  623. 	} else {

    624. 

  624. 		// Default client and ASP scripting language is JavaScript

    625. 

  625. 		lineState = eScriptJS << 8;

    626. 

  626. 

    627. 

  627. 		// property asp.default.language

    628. 

  628. 		//	Script in ASP code is initially assumed to be in JavaScript.

    629. 

  629. 		//	To change this to VBScript set asp.default.language to 2. Python is 3.

    630. 

  630. 		lineState |= styler.GetPropertyInt("asp.default.language", eScriptJS) << 4;

    631. 

  631. 	}

    632. 

  632. 	script_mode inScriptType = script_mode((lineState >> 0) & 0x03); // 2 bits of scripting mode

    633. 

  633. 	bool tagOpened = (lineState >> 2) & 0x01; // 1 bit to know if we are in an opened tag

    634. 

  634. 	bool tagClosing = (lineState >> 3) & 0x01; // 1 bit to know if we are in a closing tag

    635. 

  635. 	bool tagDontFold = false; //some HTML tags should not be folded

    636. 

  636. 	script_type aspScript = script_type((lineState >> 4) & 0x0F); // 4 bits of script name

    637. 

  637. 	script_type clientScript = script_type((lineState >> 8) & 0x0F); // 4 bits of script name

    638. 

  638. 	int beforePreProc = (lineState >> 12) & 0xFF; // 8 bits of state

    639. 

  639. 

    640. 

  640. 	script_type scriptLanguage = ScriptOfState(state);

    641. 

  641. 	// If eNonHtmlScript coincides with SCE_H_COMMENT, assume eScriptComment

    642. 

  642. 	if (inScriptType == eNonHtmlScript && state == SCE_H_COMMENT) {

    643. 

  643. 		scriptLanguage = eScriptComment;

    644. 

  644. 	}

    645. 

  645. 	script_type beforeLanguage = ScriptOfState(beforePreProc);

    646. 

  646. 

    647. 

  647. 	// property fold.html

    648. 

  648. 	//	Folding is turned on or off for HTML and XML files with this option.

    649. 

  649. 	//	The fold option must also be on for folding to occur.

    650. 

  650. 	const bool foldHTML = styler.GetPropertyInt("fold.html", 0) != 0;

    651. 

  651. 

    652. 

  652. 	const bool fold = foldHTML && styler.GetPropertyInt("fold", 0);

    653. 

  653. 

    654. 

  654. 	// property fold.html.preprocessor

    655. 

  655. 	//	Folding is turned on or off for scripts embedded in HTML files with this option.

    656. 

  656. 	//	The default is on.

    657. 

  657. 	const bool foldHTMLPreprocessor = foldHTML && styler.GetPropertyInt("fold.html.preprocessor", 1);

    658. 

  658. 

    659. 

  659. 	const bool foldCompact = styler.GetPropertyInt("fold.compact", 1) != 0;

    660. 

  660. 

    661. 

  661. 	// property fold.hypertext.comment

    662. 

  662. 	//	Allow folding for comments in scripts embedded in HTML.

    663. 

  663. 	//	The default is off.

    664. 

  664. 	const bool foldComment = fold && styler.GetPropertyInt("fold.hypertext.comment", 0) != 0;

    665. 

  665. 

    666. 

  666. 	// property fold.hypertext.heredoc

    667. 

  667. 	//	Allow folding for heredocs in scripts embedded in HTML.

    668. 

  668. 	//	The default is off.

    669. 

  669. 	const bool foldHeredoc = fold && styler.GetPropertyInt("fold.hypertext.heredoc", 0) != 0;

    670. 

  670. 

    671. 

  671. 	// property html.tags.case.sensitive

    672. 

  672. 	//	For XML and HTML, setting this property to 1 will make tags match in a case

    673. 

  673. 	//	sensitive way which is the expected behaviour for XML and XHTML.

    674. 

  674. 	const bool caseSensitive = styler.GetPropertyInt("html.tags.case.sensitive", 0) != 0;

    675. 

  675. 

    676. 

  676. 	// property lexer.xml.allow.scripts

    677. 

  677. 	//	Set to 0 to disable scripts in XML.

    678. 

  678. 	const bool allowScripts = styler.GetPropertyInt("lexer.xml.allow.scripts", 1) != 0;

    679. 

  679. 

    680. 

  680. 	// property lexer.html.mako

    681. 

  681. 	//	Set to 1 to enable the mako template language.

    682. 

  682. 	const bool isMako = styler.GetPropertyInt("lexer.html.mako", 0) != 0;

    683. 

  683. 

    684. 

  684. 	// property lexer.html.django

    685. 

  685. 	//	Set to 1 to enable the django template language.

    686. 

  686. 	const bool isDjango = styler.GetPropertyInt("lexer.html.django", 0) != 0;

    687. 

  687. 

    688. 

  688. 	const CharacterSet setHTMLWord(CharacterSet::setAlphaNum, ".-_:!#", 0x80, true);

    689. 

  689. 	const CharacterSet setTagContinue(CharacterSet::setAlphaNum, ".-_:!#[", 0x80, true);

    690. 

  690. 	const CharacterSet setAttributeContinue(CharacterSet::setAlphaNum, ".-_:!#/", 0x80, true);

    691. 

  691. 

    692. 

  692. 	int levelPrev = styler.LevelAt(lineCurrent) & SC_FOLDLEVELNUMBERMASK;

    693. 

  693. 	int levelCurrent = levelPrev;

    694. 

  694. 	int visibleChars = 0;

    695. 

  695. 	int lineStartVisibleChars = 0;

    696. 

  696. 

    697. 

  697. 	int chPrev = ' ';

    698. 

  698. 	int ch = ' ';

    699. 

  699. 	int chPrevNonWhite = ' ';

    700. 

  700. 	// look back to set chPrevNonWhite properly for better regex colouring

    701. 

  701. 	if (scriptLanguage == eScriptJS && startPos > 0) {

    702. 

  702. 		int back = startPos;

    703. 

  703. 		int style = 0;

    704. 

  704. 		while (--back) {

    705. 

  705. 			style = styler.StyleAt(back);

    706. 

  706. 			if (style < SCE_HJ_DEFAULT || style > SCE_HJ_COMMENTDOC)

    707. 

  707. 				// includes SCE_HJ_COMMENT & SCE_HJ_COMMENTLINE

    708. 

  708. 				break;

    709. 

  709. 		}

    710. 

  710. 		if (style == SCE_HJ_SYMBOLS) {

    711. 

  711. 			chPrevNonWhite = static_cast<unsigned char>(styler.SafeGetCharAt(back));

    712. 

  712. 		}

    713. 

  713. 	}

    714. 

  714. 

    715. 

  715. 	styler.StartSegment(startPos);

    716. 

  716. 	const int lengthDoc = startPos + length;

    717. 

  717. 	for (int i = startPos; i < lengthDoc; i++) {

    718. 

  718. 		const int chPrev2 = chPrev;

    719. 

  719. 		chPrev = ch;

    720. 

  720. 		if (!IsASpace(ch) && state != SCE_HJ_COMMENT &&

    721. 

  721. 			state != SCE_HJ_COMMENTLINE && state != SCE_HJ_COMMENTDOC)

    722. 

  722. 			chPrevNonWhite = ch;

    723. 

  723. 		ch = static_cast<unsigned char>(styler[i]);

    724. 

  724. 		int chNext = static_cast<unsigned char>(styler.SafeGetCharAt(i + 1));

    725. 

  725. 		const int chNext2 = static_cast<unsigned char>(styler.SafeGetCharAt(i + 2));

    726. 

  726. 

    727. 

  727. 		// Handle DBCS codepages

    728. 

  728. 		if (styler.IsLeadByte(static_cast<char>(ch))) {

    729. 

  729. 			chPrev = ' ';

    730. 

  730. 			i += 1;

    731. 

  731. 			continue;

    732. 

  732. 		}

    733. 

  733. 

    734. 

  734. 		if ((!IsASpace(ch) || !foldCompact) && fold)

    735. 

  735. 			visibleChars++;

    736. 

  736. 		if (!IsASpace(ch))

    737. 

  737. 			lineStartVisibleChars++;

    738. 

  738. 

    739. 

  739. 		// decide what is the current state to print (depending of the script tag)

    740. 

  740. 		StateToPrint = statePrintForState(state, inScriptType);

    741. 

  741. 

    742. 

  742. 		// handle script folding

    743. 

  743. 		if (fold) {

    744. 

  744. 			switch (scriptLanguage) {

    745. 

  745. 			case eScriptJS:

    746. 

  746. 			case eScriptPHP:

    747. 

  747. 				//not currently supported				case eScriptVBS:

    748. 

  748. 

    749. 

  749. 				if ((state != SCE_HPHP_COMMENT) && (state != SCE_HPHP_COMMENTLINE) && (state != SCE_HJ_COMMENT) && (state != SCE_HJ_COMMENTLINE) && (state != SCE_HJ_COMMENTDOC) && (!isStringState(state))) {

    750. 

  750. 				//Platform::DebugPrintf("state=%d, StateToPrint=%d, initStyle=%d\n", state, StateToPrint, initStyle);

    751. 

  751. 				//if ((state == SCE_HPHP_OPERATOR) || (state == SCE_HPHP_DEFAULT) || (state == SCE_HJ_SYMBOLS) || (state == SCE_HJ_START) || (state == SCE_HJ_DEFAULT)) {

    752. 

  752. 					if ((ch == '{') || (ch == '}') || (foldComment && (ch == '/') && (chNext == '*'))) {

    753. 

  753. 						levelCurrent += ((ch == '{') || (ch == '/')) ? 1 : -1;

    754. 

  754. 					}

    755. 

  755. 				} else if (((state == SCE_HPHP_COMMENT) || (state == SCE_HJ_COMMENT)) && foldComment && (ch == '*') && (chNext == '/')) {

    756. 

  756. 					levelCurrent--;

    757. 

  757. 				}

    758. 

  758. 				break;

    759. 

  759. 			case eScriptPython:

    760. 

  760. 				if (state != SCE_HP_COMMENTLINE) {

    761. 

  761. 					if ((ch == ':') && ((chNext == '\n') || (chNext == '\r' && chNext2 == '\n'))) {

    762. 

  762. 						levelCurrent++;

    763. 

  763. 					} else if ((ch == '\n') && !((chNext == '\r') && (chNext2 == '\n')) && (chNext != '\n')) {

    764. 

  764. 						// check if the number of tabs is lower than the level

    765. 

  765. 						int Findlevel = (levelCurrent & ~SC_FOLDLEVELBASE) * 8;

    766. 

  766. 						for (int j = 0; Findlevel > 0; j++) {

    767. 

  767. 							char chTmp = styler.SafeGetCharAt(i + j + 1);

    768. 

  768. 							if (chTmp == '\t') {

    769. 

  769. 								Findlevel -= 8;

    770. 

  770. 							} else if (chTmp == ' ') {

    771. 

  771. 								Findlevel--;

    772. 

  772. 							} else {

    773. 

  773. 								break;

    774. 

  774. 							}

    775. 

  775. 						}

    776. 

  776. 

    777. 

  777. 						if (Findlevel > 0) {

    778. 

  778. 							levelCurrent -= Findlevel / 8;

    779. 

  779. 							if (Findlevel % 8)

    780. 

  780. 								levelCurrent--;

    781. 

  781. 						}

    782. 

  782. 					}

    783. 

  783. 				}

    784. 

  784. 				break;

    785. 

  785. 			default:

    786. 

  786. 				break;

    787. 

  787. 			}

    788. 

  788. 		}

    789. 

  789. 

    790. 

  790. 		if ((ch == '\r' && chNext != '\n') || (ch == '\n')) {

    791. 

  791. 			// Trigger on CR only (Mac style) or either on LF from CR+LF (Dos/Win) or on LF alone (Unix)

    792. 

  792. 			// Avoid triggering two times on Dos/Win

    793. 

  793. 			// New line -> record any line state onto /next/ line

    794. 

  794. 			if (fold) {

    795. 

  795. 				int lev = levelPrev;

    796. 

  796. 				if (visibleChars == 0)

    797. 

  797. 					lev |= SC_FOLDLEVELWHITEFLAG;

    798. 

  798. 				if ((levelCurrent > levelPrev) && (visibleChars > 0))

    799. 

  799. 					lev |= SC_FOLDLEVELHEADERFLAG;

    800. 

  800. 

    801. 

  801. 				styler.SetLevel(lineCurrent, lev);

    802. 

  802. 				visibleChars = 0;

    803. 

  803. 				levelPrev = levelCurrent;

    804. 

  804. 			}

    805. 

  805. 			lineCurrent++;

    806. 

  806. 			lineStartVisibleChars = 0;

    807. 

  807. 			styler.SetLineState(lineCurrent,

    808. 

  808. 			                    ((inScriptType & 0x03) << 0) |

    809. 

  809. 			                    ((tagOpened & 0x01) << 2) |

    810. 

  810. 			                    ((tagClosing & 0x01) << 3) |

    811. 

  811. 			                    ((aspScript & 0x0F) << 4) |

    812. 

  812. 			                    ((clientScript & 0x0F) << 8) |

    813. 

  813. 			                    ((beforePreProc & 0xFF) << 12));

    814. 

  814. 		}

    815. 

  815. 

    816. 

  816. 		// Allow falling through to mako handling code if newline is going to end a block

    817. 

  817. 		if (((ch == '\r' && chNext != '\n') || (ch == '\n')) &&

    818. 

  818. 			(!isMako || (0 != strcmp(makoBlockType, "%")))) {

    819. 

  819. 		}

    820. 

  820. 

    821. 

  821. 		// generic end of script processing

    822. 

  822. 		else if ((inScriptType == eNonHtmlScript) && (ch == '<') && (chNext == '/')) {

    823. 

  823. 			// Check if it's the end of the script tag (or any other HTML tag)

    824. 

  824. 			switch (state) {

    825. 

  825. 				// in these cases, you can embed HTML tags (to confirm !!!!!!!!!!!!!!!!!!!!!!)

    826. 

  826. 			case SCE_H_DOUBLESTRING:

    827. 

  827. 			case SCE_H_SINGLESTRING:

    828. 

  828. 			case SCE_HJ_COMMENT:

    829. 

  829. 			case SCE_HJ_COMMENTDOC:

    830. 

  830. 			//case SCE_HJ_COMMENTLINE: // removed as this is a common thing done to hide

    831. 

  831. 			// the end of script marker from some JS interpreters.

    832. 

  832. 			case SCE_HB_COMMENTLINE:

    833. 

  833. 			case SCE_HBA_COMMENTLINE:

    834. 

  834. 			case SCE_HJ_DOUBLESTRING:

    835. 

  835. 			case SCE_HJ_SINGLESTRING:

    836. 

  836. 			case SCE_HJ_REGEX:

    837. 

  837. 			case SCE_HB_STRING:

    838. 

  838. 			case SCE_HBA_STRING:

    839. 

  839. 			case SCE_HP_STRING:

    840. 

  840. 			case SCE_HP_TRIPLE:

    841. 

  841. 			case SCE_HP_TRIPLEDOUBLE:

    842. 

  842. 			case SCE_HPHP_HSTRING:

    843. 

  843. 			case SCE_HPHP_SIMPLESTRING:

    844. 

  844. 			case SCE_HPHP_COMMENT:

    845. 

  845. 			case SCE_HPHP_COMMENTLINE:

    846. 

  846. 				break;

    847. 

  847. 			default :

    848. 

  848. 				// check if the closing tag is a script tag

    849. 

  849. 				if (const char *tag =

    850. 

  850. 						state == SCE_HJ_COMMENTLINE || isXml ? "script" :

    851. 

  851. 						state == SCE_H_COMMENT ? "comment" : 0) {

    852. 

  852. 					int j = i + 2;

    853. 

  853. 					int chr;

    854. 

  854. 					do {

    855. 

  855. 						chr = static_cast<int>(*tag++);

    856. 

  856. 					} while (chr != 0 && chr == MakeLowerCase(styler.SafeGetCharAt(j++)));

    857. 

  857. 					if (chr != 0) break;

    858. 

  858. 				}

    859. 

  859. 				// closing tag of the script (it's a closing HTML tag anyway)

    860. 

  860. 				styler.ColourTo(i - 1, StateToPrint);

    861. 

  861. 				state = SCE_H_TAGUNKNOWN;

    862. 

  862. 				inScriptType = eHtml;

    863. 

  863. 				scriptLanguage = eScriptNone;

    864. 

  864. 				clientScript = eScriptJS;

    865. 

  865. 				i += 2;

    866. 

  866. 				visibleChars += 2;

    867. 

  867. 				tagClosing = true;

    868. 

  868. 				continue;

    869. 

  869. 			}

    870. 

  870. 		}

    871. 

  871. 

    872. 

  872. 		/////////////////////////////////////

    873. 

  873. 		// handle the start of PHP pre-processor = Non-HTML

    874. 

  874. 		else if ((state != SCE_H_ASPAT) &&

    875. 

  875. 		         !isPHPStringState(state) &&

    876. 

  876. 		         (state != SCE_HPHP_COMMENT) &&

    877. 

  877. 		         (ch == '<') &&

    878. 

  878. 		         (chNext == '?') &&

    879. 

  879. 				 !IsScriptCommentState(state) ) {

    880. 

  880. 			scriptLanguage = segIsScriptingIndicator(styler, i + 2, i + 6, eScriptPHP);

    881. 

  881. 			if (scriptLanguage != eScriptPHP && isStringState(state)) continue;

    882. 

  882. 			styler.ColourTo(i - 1, StateToPrint);

    883. 

  883. 			beforePreProc = state;

    884. 

  884. 			i++;

    885. 

  885. 			visibleChars++;

    886. 

  886. 			i += PrintScriptingIndicatorOffset(styler, styler.GetStartSegment() + 2, i + 6);

    887. 

  887. 			if (scriptLanguage == eScriptXML)

    888. 

  888. 				styler.ColourTo(i, SCE_H_XMLSTART);

    889. 

  889. 			else

    890. 

  890. 				styler.ColourTo(i, SCE_H_QUESTION);

    891. 

  891. 			state = StateForScript(scriptLanguage);

    892. 

  892. 			if (inScriptType == eNonHtmlScript)

    893. 

  893. 				inScriptType = eNonHtmlScriptPreProc;

    894. 

  894. 			else

    895. 

  895. 				inScriptType = eNonHtmlPreProc;

    896. 

  896. 			// Fold whole script, but not if the XML first tag (all XML-like tags in this case)

    897. 

  897. 			if (foldHTMLPreprocessor && (scriptLanguage != eScriptXML)) {

    898. 

  898. 				levelCurrent++;

    899. 

  899. 			}

    900. 

  900. 			// should be better

    901. 

  901. 			ch = static_cast<unsigned char>(styler.SafeGetCharAt(i));

    902. 

  902. 			continue;

    903. 

  903. 		}

    904. 

  904. 

    905. 

  905. 		// handle the start Mako template Python code

    906. 

  906. 		else if (isMako && scriptLanguage == eScriptNone && ((ch == '<' && chNext == '%') ||

    907. 

  907. 															 (lineStartVisibleChars == 1 && ch == '%') ||

    908. 

  908. 															 (ch == '$' && chNext == '{') ||

    909. 

  909. 															 (ch == '<' && chNext == '/' && chNext2 == '%'))) {

    910. 

  910. 			if (ch == '%')

    911. 

  911. 				strcpy(makoBlockType, "%");

    912. 

  912. 			else if (ch == '$')

    913. 

  913. 				strcpy(makoBlockType, "{");

    914. 

  914. 			else if (chNext == '/')

    915. 

  915. 				GetNextWord(styler, i+3, makoBlockType, sizeof(makoBlockType));

    916. 

  916. 			else

    917. 

  917. 				GetNextWord(styler, i+2, makoBlockType, sizeof(makoBlockType));

    918. 

  918. 			styler.ColourTo(i - 1, StateToPrint);

    919. 

  919. 			beforePreProc = state;

    920. 

  920. 			if (inScriptType == eNonHtmlScript)

    921. 

  921. 				inScriptType = eNonHtmlScriptPreProc;

    922. 

  922. 			else

    923. 

  923. 				inScriptType = eNonHtmlPreProc;

    924. 

  924. 

    925. 

  925. 			if (chNext == '/') {

    926. 

  926. 				i += 2;

    927. 

  927. 				visibleChars += 2;

    928. 

  928. 			} else if (ch != '%') {

    929. 

  929. 				i++;

    930. 

  930. 				visibleChars++;

    931. 

  931. 			}

    932. 

  932. 			state = SCE_HP_START;

    933. 

  933. 			scriptLanguage = eScriptPython;

    934. 

  934. 			styler.ColourTo(i, SCE_H_ASP);

    935. 

  935. 			if (foldHTMLPreprocessor && ch == '<')

    936. 

  936. 				levelCurrent++;

    937. 

  937. 

    938. 

  938. 			if (ch != '%' && ch != '$') {

    939. 

  939. 				i += strlen(makoBlockType);

    940. 

  940. 				visibleChars += strlen(makoBlockType);

    941. 

  941. 				if (keywords4.InList(makoBlockType))

    942. 

  942. 					styler.ColourTo(i, SCE_HP_WORD);

    943. 

  943. 				else

    944. 

  944. 					styler.ColourTo(i, SCE_H_TAGUNKNOWN);

    945. 

  945. 			}

    946. 

  946. 

    947. 

  947. 			ch = static_cast<unsigned char>(styler.SafeGetCharAt(i));

    948. 

  948. 			continue;

    949. 

  949. 		}

    950. 

  950. 

    951. 

  951. 		// handle the start Django template code

    952. 

  952. 		else if (isDjango && scriptLanguage != eScriptPython && (ch == '{' && (chNext == '%' ||  chNext == '{'))) {

    953. 

  953. 			if (chNext == '%')

    954. 

  954. 				strcpy(djangoBlockType, "%");

    955. 

  955. 			else

    956. 

  956. 				strcpy(djangoBlockType, "{");

    957. 

  957. 			styler.ColourTo(i - 1, StateToPrint);

    958. 

  958. 			beforePreProc = state;

    959. 

  959. 			if (inScriptType == eNonHtmlScript)

    960. 

  960. 				inScriptType = eNonHtmlScriptPreProc;

    961. 

  961. 			else

    962. 

  962. 				inScriptType = eNonHtmlPreProc;

    963. 

  963. 

    964. 

  964. 			i += 1;

    965. 

  965. 			visibleChars += 1;

    966. 

  966. 			state = SCE_HP_START;

    967. 

  967. 			beforeLanguage = scriptLanguage;

    968. 

  968. 			scriptLanguage = eScriptPython;

    969. 

  969. 			styler.ColourTo(i, SCE_H_ASP);

    970. 

  970. 			if (foldHTMLPreprocessor && chNext == '%')

    971. 

  971. 				levelCurrent++;

    972. 

  972. 

    973. 

  973. 			ch = static_cast<unsigned char>(styler.SafeGetCharAt(i));

    974. 

  974. 			continue;

    975. 

  975. 		}

    976. 

  976. 

    977. 

  977. 		// handle the start of ASP pre-processor = Non-HTML

    978. 

  978. 		else if (!isMako && !isDjango && !isCommentASPState(state) && (ch == '<') && (chNext == '%') && !isPHPStringState(state)) {

    979. 

  979. 			styler.ColourTo(i - 1, StateToPrint);

    980. 

  980. 			beforePreProc = state;

    981. 

  981. 			if (inScriptType == eNonHtmlScript)

    982. 

  982. 				inScriptType = eNonHtmlScriptPreProc;

    983. 

  983. 			else

    984. 

  984. 				inScriptType = eNonHtmlPreProc;

    985. 

  985. 

    986. 

  986. 			if (chNext2 == '@') {

    987. 

  987. 				i += 2; // place as if it was the second next char treated

    988. 

  988. 				visibleChars += 2;

    989. 

  989. 				state = SCE_H_ASPAT;

    990. 

  990. 			} else if ((chNext2 == '-') && (styler.SafeGetCharAt(i + 3) == '-')) {

    991. 

  991. 				styler.ColourTo(i + 3, SCE_H_ASP);

    992. 

  992. 				state = SCE_H_XCCOMMENT;

    993. 

  993. 				scriptLanguage = eScriptVBS;

    994. 

  994. 				continue;

    995. 

  995. 			} else {

    996. 

  996. 				if (chNext2 == '=') {

    997. 

  997. 					i += 2; // place as if it was the second next char treated

    998. 

  998. 					visibleChars += 2;

    999. 

  999. 				} else {

    1000. 

  1000. 					i++; // place as if it was the next char treated

    1001. 

  1001. 					visibleChars++;

    1002. 

  1002. 				}

    1003. 

  1003. 

    1004. 

  1004. 				state = StateForScript(aspScript);

    1005. 

  1005. 			}

    1006. 

  1006. 			scriptLanguage = eScriptVBS;

    1007. 

  1007. 			styler.ColourTo(i, SCE_H_ASP);

    1008. 

  1008. 			// fold whole script

    1009. 

  1009. 			if (foldHTMLPreprocessor)

    1010. 

  1010. 				levelCurrent++;

    1011. 

  1011. 			// should be better

    1012. 

  1012. 			ch = static_cast<unsigned char>(styler.SafeGetCharAt(i));

    1013. 

  1013. 			continue;

    1014. 

  1014. 		}

    1015. 

  1015. 

    1016. 

  1016. 		/////////////////////////////////////

    1017. 

  1017. 		// handle the start of SGML language (DTD)

    1018. 

  1018. 		else if (((scriptLanguage == eScriptNone) || (scriptLanguage == eScriptXML)) &&

    1019. 

  1019. 				 (chPrev == '<') &&

    1020. 

  1020. 				 (ch == '!') &&

    1021. 

  1021. 				 (StateToPrint != SCE_H_CDATA) &&

    1022. 

  1022. 				 (!IsCommentState(StateToPrint)) &&

    1023. 

  1023. 				 (!IsScriptCommentState(StateToPrint)) ) {

    1024. 

  1024. 			beforePreProc = state;

    1025. 

  1025. 			styler.ColourTo(i - 2, StateToPrint);

    1026. 

  1026. 			if ((chNext == '-') && (chNext2 == '-')) {

    1027. 

  1027. 				state = SCE_H_COMMENT; // wait for a pending command

    1028. 

  1028. 				styler.ColourTo(i + 2, SCE_H_COMMENT);

    1029. 

  1029. 				i += 2; // follow styling after the --

    1030. 

  1030. 			} else if (isWordCdata(i + 1, i + 7, styler)) {

    1031. 

  1031. 				state = SCE_H_CDATA;

    1032. 

  1032. 			} else {

    1033. 

  1033. 				styler.ColourTo(i, SCE_H_SGML_DEFAULT); // <! is default

    1034. 

  1034. 				scriptLanguage = eScriptSGML;

    1035. 

  1035. 				state = SCE_H_SGML_COMMAND; // wait for a pending command

    1036. 

  1036. 			}

    1037. 

  1037. 			// fold whole tag (-- when closing the tag)

    1038. 

  1038. 			if (foldHTMLPreprocessor || (state == SCE_H_COMMENT))

    1039. 

  1039. 				levelCurrent++;

    1040. 

  1040. 			continue;

    1041. 

  1041. 		}

    1042. 

  1042. 

    1043. 

  1043. 		// handle the end of Mako Python code

    1044. 

  1044. 		else if (isMako &&

    1045. 

  1045. 			     ((inScriptType == eNonHtmlPreProc) || (inScriptType == eNonHtmlScriptPreProc)) &&

    1046. 

  1046. 				 (scriptLanguage != eScriptNone) && stateAllowsTermination(state) &&

    1047. 

  1047. 				 isMakoBlockEnd(ch, chNext, makoBlockType)) {

    1048. 

  1048. 			if (state == SCE_H_ASPAT) {

    1049. 

  1049. 				aspScript = segIsScriptingIndicator(styler,

    1050. 

  1050. 				                                    styler.GetStartSegment(), i - 1, aspScript);

    1051. 

  1051. 			}

    1052. 

  1052. 			if (state == SCE_HP_WORD) {

    1053. 

  1053. 				classifyWordHTPy(styler.GetStartSegment(), i - 1, keywords4, styler, prevWord, inScriptType);

    1054. 

  1054. 			} else {

    1055. 

  1055. 				styler.ColourTo(i - 1, StateToPrint);

    1056. 

  1056. 			}

    1057. 

  1057. 			if (0 != strcmp(makoBlockType, "%") && (0 != strcmp(makoBlockType, "{")) && ch != '>') {

    1058. 

  1058. 				i++;

    1059. 

  1059. 				visibleChars++;

    1060. 

  1060. 		    }

    1061. 

  1061. 			if (0 != strcmp(makoBlockType, "%")) {

    1062. 

  1062. 				styler.ColourTo(i, SCE_H_ASP);

    1063. 

  1063. 			}

    1064. 

  1064. 			state = beforePreProc;

    1065. 

  1065. 			if (inScriptType == eNonHtmlScriptPreProc)

    1066. 

  1066. 				inScriptType = eNonHtmlScript;

    1067. 

  1067. 			else

    1068. 

  1068. 				inScriptType = eHtml;

    1069. 

  1069. 			if (foldHTMLPreprocessor && ch != '\n' && ch != '\r') {

    1070. 

  1070. 				levelCurrent--;

    1071. 

  1071. 			}

    1072. 

  1072. 			scriptLanguage = eScriptNone;

    1073. 

  1073. 			continue;

    1074. 

  1074. 		}

    1075. 

  1075. 

    1076. 

  1076. 		// handle the end of Django template code

    1077. 

  1077. 		else if (isDjango &&

    1078. 

  1078. 			     ((inScriptType == eNonHtmlPreProc) || (inScriptType == eNonHtmlScriptPreProc)) &&

    1079. 

  1079. 				 (scriptLanguage != eScriptNone) && stateAllowsTermination(state) &&

    1080. 

  1080. 				 isDjangoBlockEnd(ch, chNext, djangoBlockType)) {

    1081. 

  1081. 			if (state == SCE_H_ASPAT) {

    1082. 

  1082. 				aspScript = segIsScriptingIndicator(styler,

    1083. 

  1083. 				                                    styler.GetStartSegment(), i - 1, aspScript);

    1084. 

  1084. 			}

    1085. 

  1085. 			if (state == SCE_HP_WORD) {

    1086. 

  1086. 				classifyWordHTPy(styler.GetStartSegment(), i - 1, keywords4, styler, prevWord, inScriptType);

    1087. 

  1087. 			} else {

    1088. 

  1088. 				styler.ColourTo(i - 1, StateToPrint);

    1089. 

  1089. 			}

    1090. 

  1090. 			i += 1;

    1091. 

  1091. 			visibleChars += 1;

    1092. 

  1092. 			styler.ColourTo(i, SCE_H_ASP);

    1093. 

  1093. 			state = beforePreProc;

    1094. 

  1094. 			if (inScriptType == eNonHtmlScriptPreProc)

    1095. 

  1095. 				inScriptType = eNonHtmlScript;

    1096. 

  1096. 			else

    1097. 

  1097. 				inScriptType = eHtml;

    1098. 

  1098. 			if (foldHTMLPreprocessor) {

    1099. 

  1099. 				levelCurrent--;

    1100. 

  1100. 			}

    1101. 

  1101. 			scriptLanguage = beforeLanguage;

    1102. 

  1102. 			continue;

    1103. 

  1103. 		}

    1104. 

  1104. 

    1105. 

  1105. 		// handle the end of a pre-processor = Non-HTML

    1106. 

  1106. 		else if ((!isMako && !isDjango && ((inScriptType == eNonHtmlPreProc) || (inScriptType == eNonHtmlScriptPreProc)) &&

    1107. 

  1107. 				  (((scriptLanguage != eScriptNone) && stateAllowsTermination(state))) &&

    1108. 

  1108. 				  (((ch == '%') || (ch == '?')) && (chNext == '>'))) ||

    1109. 

  1109. 		         ((scriptLanguage == eScriptSGML) && (ch == '>') && (state != SCE_H_SGML_COMMENT))) {

    1110. 

  1110. 			if (state == SCE_H_ASPAT) {

    1111. 

  1111. 				aspScript = segIsScriptingIndicator(styler,

    1112. 

  1112. 				                                    styler.GetStartSegment(), i - 1, aspScript);

    1113. 

  1113. 			}

    1114. 

  1114. 			// Bounce out of any ASP mode

    1115. 

  1115. 			switch (state) {

    1116. 

  1116. 			case SCE_HJ_WORD:

    1117. 

  1117. 				classifyWordHTJS(styler.GetStartSegment(), i - 1, keywords2, styler, inScriptType);

    1118. 

  1118. 				break;

    1119. 

  1119. 			case SCE_HB_WORD:

    1120. 

  1120. 				classifyWordHTVB(styler.GetStartSegment(), i - 1, keywords3, styler, inScriptType);

    1121. 

  1121. 				break;

    1122. 

  1122. 			case SCE_HP_WORD:

    1123. 

  1123. 				classifyWordHTPy(styler.GetStartSegment(), i - 1, keywords4, styler, prevWord, inScriptType);

    1124. 

  1124. 				break;

    1125. 

  1125. 			case SCE_HPHP_WORD:

    1126. 

  1126. 				classifyWordHTPHP(styler.GetStartSegment(), i - 1, keywords5, styler);

    1127. 

  1127. 				break;

    1128. 

  1128. 			case SCE_H_XCCOMMENT:

    1129. 

  1129. 				styler.ColourTo(i - 1, state);

    1130. 

  1130. 				break;

    1131. 

  1131. 			default :

    1132. 

  1132. 				styler.ColourTo(i - 1, StateToPrint);

    1133. 

  1133. 				break;

    1134. 

  1134. 			}

    1135. 

  1135. 			if (scriptLanguage != eScriptSGML) {

    1136. 

  1136. 				i++;

    1137. 

  1137. 				visibleChars++;

    1138. 

  1138. 			}

    1139. 

  1139. 			if (ch == '%')

    1140. 

  1140. 				styler.ColourTo(i, SCE_H_ASP);

    1141. 

  1141. 			else if (scriptLanguage == eScriptXML)

    1142. 

  1142. 				styler.ColourTo(i, SCE_H_XMLEND);

    1143. 

  1143. 			else if (scriptLanguage == eScriptSGML)

    1144. 

  1144. 				styler.ColourTo(i, SCE_H_SGML_DEFAULT);

    1145. 

  1145. 			else

    1146. 

  1146. 				styler.ColourTo(i, SCE_H_QUESTION);

    1147. 

  1147. 			state = beforePreProc;

    1148. 

  1148. 			if (inScriptType == eNonHtmlScriptPreProc)

    1149. 

  1149. 				inScriptType = eNonHtmlScript;

    1150. 

  1150. 			else

    1151. 

  1151. 				inScriptType = eHtml;

    1152. 

  1152. 			// Unfold all scripting languages, except for XML tag

    1153. 

  1153. 			if (foldHTMLPreprocessor && (scriptLanguage != eScriptXML)) {

    1154. 

  1154. 				levelCurrent--;

    1155. 

  1155. 			}

    1156. 

  1156. 			scriptLanguage = eScriptNone;

    1157. 

  1157. 			continue;

    1158. 

  1158. 		}

    1159. 

  1159. 		/////////////////////////////////////

    1160. 

  1160. 

    1161. 

  1161. 		switch (state) {

    1162. 

  1162. 		case SCE_H_DEFAULT:

    1163. 

  1163. 			if (ch == '<') {

    1164. 

  1164. 				// in HTML, fold on tag open and unfold on tag close

    1165. 

  1165. 				tagOpened = true;

    1166. 

  1166. 				tagClosing = (chNext == '/');

    1167. 

  1167. 				styler.ColourTo(i - 1, StateToPrint);

    1168. 

  1168. 				if (chNext != '!')

    1169. 

  1169. 					state = SCE_H_TAGUNKNOWN;

    1170. 

  1170. 			} else if (ch == '&') {

    1171. 

  1171. 				styler.ColourTo(i - 1, SCE_H_DEFAULT);

    1172. 

  1172. 				state = SCE_H_ENTITY;

    1173. 

  1173. 			}

    1174. 

  1174. 			break;

    1175. 

  1175. 		case SCE_H_SGML_DEFAULT:

    1176. 

  1176. 		case SCE_H_SGML_BLOCK_DEFAULT:

    1177. 

  1177. //			if (scriptLanguage == eScriptSGMLblock)

    1178. 

  1178. //				StateToPrint = SCE_H_SGML_BLOCK_DEFAULT;

    1179. 

  1179. 

    1180. 

  1180. 			if (ch == '\"') {

    1181. 

  1181. 				styler.ColourTo(i - 1, StateToPrint);

    1182. 

  1182. 				state = SCE_H_SGML_DOUBLESTRING;

    1183. 

  1183. 			} else if (ch == '\'') {

    1184. 

  1184. 				styler.ColourTo(i - 1, StateToPrint);

    1185. 

  1185. 				state = SCE_H_SGML_SIMPLESTRING;

    1186. 

  1186. 			} else if ((ch == '-') && (chPrev == '-')) {

    1187. 

  1187. 				if (static_cast<int>(styler.GetStartSegment()) <= (i - 2)) {

    1188. 

  1188. 					styler.ColourTo(i - 2, StateToPrint);

    1189. 

  1189. 				}

    1190. 

  1190. 				state = SCE_H_SGML_COMMENT;

    1191. 

  1191. 			} else if (isascii(ch) && isalpha(ch) && (chPrev == '%')) {

    1192. 

  1192. 				styler.ColourTo(i - 2, StateToPrint);

    1193. 

  1193. 				state = SCE_H_SGML_ENTITY;

    1194. 

  1194. 			} else if (ch == '#') {

    1195. 

  1195. 				styler.ColourTo(i - 1, StateToPrint);

    1196. 

  1196. 				state = SCE_H_SGML_SPECIAL;

    1197. 

  1197. 			} else if (ch == '[') {

    1198. 

  1198. 				styler.ColourTo(i - 1, StateToPrint);

    1199. 

  1199. 				scriptLanguage = eScriptSGMLblock;

    1200. 

  1200. 				state = SCE_H_SGML_BLOCK_DEFAULT;

    1201. 

  1201. 			} else if (ch == ']') {

    1202. 

  1202. 				if (scriptLanguage == eScriptSGMLblock) {

    1203. 

  1203. 					styler.ColourTo(i, StateToPrint);

    1204. 

  1204. 					scriptLanguage = eScriptSGML;

    1205. 

  1205. 				} else {

    1206. 

  1206. 					styler.ColourTo(i - 1, StateToPrint);

    1207. 

  1207. 					styler.ColourTo(i, SCE_H_SGML_ERROR);

    1208. 

  1208. 				}

    1209. 

  1209. 				state = SCE_H_SGML_DEFAULT;

    1210. 

  1210. 			} else if (scriptLanguage == eScriptSGMLblock) {

    1211. 

  1211. 				if ((ch == '!') && (chPrev == '<')) {

    1212. 

  1212. 					styler.ColourTo(i - 2, StateToPrint);

    1213. 

  1213. 					styler.ColourTo(i, SCE_H_SGML_DEFAULT);

    1214. 

  1214. 					state = SCE_H_SGML_COMMAND;

    1215. 

  1215. 				} else if (ch == '>') {

    1216. 

  1216. 					styler.ColourTo(i - 1, StateToPrint);

    1217. 

  1217. 					styler.ColourTo(i, SCE_H_SGML_DEFAULT);

    1218. 

  1218. 				}

    1219. 

  1219. 			}

    1220. 

  1220. 			break;

    1221. 

  1221. 		case SCE_H_SGML_COMMAND:

    1222. 

  1222. 			if ((ch == '-') && (chPrev == '-')) {

    1223. 

  1223. 				styler.ColourTo(i - 2, StateToPrint);

    1224. 

  1224. 				state = SCE_H_SGML_COMMENT;

    1225. 

  1225. 			} else if (!issgmlwordchar(ch)) {

    1226. 

  1226. 				if (isWordHSGML(styler.GetStartSegment(), i - 1, keywords6, styler)) {

    1227. 

  1227. 					styler.ColourTo(i - 1, StateToPrint);

    1228. 

  1228. 					state = SCE_H_SGML_1ST_PARAM;

    1229. 

  1229. 				} else {

    1230. 

  1230. 					state = SCE_H_SGML_ERROR;

    1231. 

  1231. 				}

    1232. 

  1232. 			}

    1233. 

  1233. 			break;

    1234. 

  1234. 		case SCE_H_SGML_1ST_PARAM:

    1235. 

  1235. 			// wait for the beginning of the word

    1236. 

  1236. 			if ((ch == '-') && (chPrev == '-')) {

    1237. 

  1237. 				if (scriptLanguage == eScriptSGMLblock) {

    1238. 

  1238. 					styler.ColourTo(i - 2, SCE_H_SGML_BLOCK_DEFAULT);

    1239. 

  1239. 				} else {

    1240. 

  1240. 					styler.ColourTo(i - 2, SCE_H_SGML_DEFAULT);

    1241. 

  1241. 				}

    1242. 

  1242. 				state = SCE_H_SGML_1ST_PARAM_COMMENT;

    1243. 

  1243. 			} else if (issgmlwordchar(ch)) {

    1244. 

  1244. 				if (scriptLanguage == eScriptSGMLblock) {

    1245. 

  1245. 					styler.ColourTo(i - 1, SCE_H_SGML_BLOCK_DEFAULT);

    1246. 

  1246. 				} else {

    1247. 

  1247. 					styler.ColourTo(i - 1, SCE_H_SGML_DEFAULT);

    1248. 

  1248. 				}

    1249. 

  1249. 				// find the length of the word

    1250. 

  1250. 				int size = 1;

    1251. 

  1251. 				while (setHTMLWord.Contains(static_cast<unsigned char>(styler.SafeGetCharAt(i + size))))

    1252. 

  1252. 					size++;

    1253. 

  1253. 				styler.ColourTo(i + size - 1, StateToPrint);

    1254. 

  1254. 				i += size - 1;

    1255. 

  1255. 				visibleChars += size - 1;

    1256. 

  1256. 				ch = static_cast<unsigned char>(styler.SafeGetCharAt(i));

    1257. 

  1257. 				if (scriptLanguage == eScriptSGMLblock) {

    1258. 

  1258. 					state = SCE_H_SGML_BLOCK_DEFAULT;

    1259. 

  1259. 				} else {

    1260. 

  1260. 					state = SCE_H_SGML_DEFAULT;

    1261. 

  1261. 				}

    1262. 

  1262. 				continue;

    1263. 

  1263. 			}

    1264. 

  1264. 			break;

    1265. 

  1265. 		case SCE_H_SGML_ERROR:

    1266. 

  1266. 			if ((ch == '-') && (chPrev == '-')) {

    1267. 

  1267. 				styler.ColourTo(i - 2, StateToPrint);

    1268. 

  1268. 				state = SCE_H_SGML_COMMENT;

    1269. 

  1269. 			}

    1270. 

  1270. 		case SCE_H_SGML_DOUBLESTRING:

    1271. 

  1271. 			if (ch == '\"') {

    1272. 

  1272. 				styler.ColourTo(i, StateToPrint);

    1273. 

  1273. 				state = SCE_H_SGML_DEFAULT;

    1274. 

  1274. 			}

    1275. 

  1275. 			break;

    1276. 

  1276. 		case SCE_H_SGML_SIMPLESTRING:

    1277. 

  1277. 			if (ch == '\'') {

    1278. 

  1278. 				styler.ColourTo(i, StateToPrint);

    1279. 

  1279. 				state = SCE_H_SGML_DEFAULT;

    1280. 

  1280. 			}

    1281. 

  1281. 			break;

    1282. 

  1282. 		case SCE_H_SGML_COMMENT:

    1283. 

  1283. 			if ((ch == '-') && (chPrev == '-')) {

    1284. 

  1284. 				styler.ColourTo(i, StateToPrint);

    1285. 

  1285. 				state = SCE_H_SGML_DEFAULT;

    1286. 

  1286. 			}

    1287. 

  1287. 			break;

    1288. 

  1288. 		case SCE_H_CDATA:

    1289. 

  1289. 			if ((chPrev2 == ']') && (chPrev == ']') && (ch == '>')) {

    1290. 

  1290. 				styler.ColourTo(i, StateToPrint);

    1291. 

  1291. 				state = SCE_H_DEFAULT;

    1292. 

  1292. 				levelCurrent--;

    1293. 

  1293. 			}

    1294. 

  1294. 			break;

    1295. 

  1295. 		case SCE_H_COMMENT:

    1296. 

  1296. 			if ((scriptLanguage != eScriptComment) && (chPrev2 == '-') && (chPrev == '-') && (ch == '>')) {

    1297. 

  1297. 				styler.ColourTo(i, StateToPrint);

    1298. 

  1298. 				state = SCE_H_DEFAULT;

    1299. 

  1299. 				levelCurrent--;

    1300. 

  1300. 			}

    1301. 

  1301. 			break;

    1302. 

  1302. 		case SCE_H_SGML_1ST_PARAM_COMMENT:

    1303. 

  1303. 			if ((ch == '-') && (chPrev == '-')) {

    1304. 

  1304. 				styler.ColourTo(i, SCE_H_SGML_COMMENT);

    1305. 

  1305. 				state = SCE_H_SGML_1ST_PARAM;

    1306. 

  1306. 			}

    1307. 

  1307. 			break;

    1308. 

  1308. 		case SCE_H_SGML_SPECIAL:

    1309. 

  1309. 			if (!(isascii(ch) && isupper(ch))) {

    1310. 

  1310. 				styler.ColourTo(i - 1, StateToPrint);

    1311. 

  1311. 				if (isalnum(ch)) {

    1312. 

  1312. 					state = SCE_H_SGML_ERROR;

    1313. 

  1313. 				} else {

    1314. 

  1314. 					state = SCE_H_SGML_DEFAULT;

    1315. 

  1315. 				}

    1316. 

  1316. 			}

    1317. 

  1317. 			break;

    1318. 

  1318. 		case SCE_H_SGML_ENTITY:

    1319. 

  1319. 			if (ch == ';') {

    1320. 

  1320. 				styler.ColourTo(i, StateToPrint);

    1321. 

  1321. 				state = SCE_H_SGML_DEFAULT;

    1322. 

  1322. 			} else if (!(isascii(ch) && isalnum(ch)) && ch != '-' && ch != '.') {

    1323. 

  1323. 				styler.ColourTo(i, SCE_H_SGML_ERROR);

    1324. 

  1324. 				state = SCE_H_SGML_DEFAULT;

    1325. 

  1325. 			}

    1326. 

  1326. 			break;

    1327. 

  1327. 		case SCE_H_ENTITY:

    1328. 

  1328. 			if (ch == ';') {

    1329. 

  1329. 				styler.ColourTo(i, StateToPrint);

    1330. 

  1330. 				state = SCE_H_DEFAULT;

    1331. 

  1331. 			}

    1332. 

  1332. 			if (ch != '#' && !(isascii(ch) && isalnum(ch))	// Should check that '#' follows '&', but it is unlikely anyway...

    1333. 

  1333. 				&& ch != '.' && ch != '-' && ch != '_' && ch != ':') { // valid in XML

    1334. 

  1334. 				if (!isascii(ch))	// Possibly start of a multibyte character so don't allow this byte to be in entity style

    1335. 

  1335. 					styler.ColourTo(i-1, SCE_H_TAGUNKNOWN);

    1336. 

  1336. 				else

    1337. 

  1337. 					styler.ColourTo(i, SCE_H_TAGUNKNOWN);

    1338. 

  1338. 				state = SCE_H_DEFAULT;

    1339. 

  1339. 			}

    1340. 

  1340. 			break;

    1341. 

  1341. 		case SCE_H_TAGUNKNOWN:

    1342. 

  1342. 			if (!setTagContinue.Contains(ch) && !((ch == '/') && (chPrev == '<'))) {

    1343. 

  1343. 				int eClass = classifyTagHTML(styler.GetStartSegment(),

    1344. 

  1344. 					i - 1, keywords, styler, tagDontFold, caseSensitive, isXml, allowScripts);

    1345. 

  1345. 				if (eClass == SCE_H_SCRIPT || eClass == SCE_H_COMMENT) {

    1346. 

  1346. 					if (!tagClosing) {

    1347. 

  1347. 						inScriptType = eNonHtmlScript;

    1348. 

  1348. 						scriptLanguage = eClass == SCE_H_SCRIPT ? clientScript : eScriptComment;

    1349. 

  1349. 					} else {

    1350. 

  1350. 						scriptLanguage = eScriptNone;

    1351. 

  1351. 					}

    1352. 

  1352. 					eClass = SCE_H_TAG;

    1353. 

  1353. 				}

    1354. 

  1354. 				if (ch == '>') {

    1355. 

  1355. 					styler.ColourTo(i, eClass);

    1356. 

  1356. 					if (inScriptType == eNonHtmlScript) {

    1357. 

  1357. 						state = StateForScript(scriptLanguage);

    1358. 

  1358. 					} else {

    1359. 

  1359. 						state = SCE_H_DEFAULT;

    1360. 

  1360. 					}

    1361. 

  1361. 					tagOpened = false;

    1362. 

  1362. 					if (!tagDontFold) {

    1363. 

  1363. 						if (tagClosing) {

    1364. 

  1364. 							levelCurrent--;

    1365. 

  1365. 						} else {

    1366. 

  1366. 							levelCurrent++;

    1367. 

  1367. 						}

    1368. 

  1368. 					}

    1369. 

  1369. 					tagClosing = false;

    1370. 

  1370. 				} else if (ch == '/' && chNext == '>') {

    1371. 

  1371. 					if (eClass == SCE_H_TAGUNKNOWN) {

    1372. 

  1372. 						styler.ColourTo(i + 1, SCE_H_TAGUNKNOWN);

    1373. 

  1373. 					} else {

    1374. 

  1374. 						styler.ColourTo(i - 1, StateToPrint);

    1375. 

  1375. 						styler.ColourTo(i + 1, SCE_H_TAGEND);

    1376. 

  1376. 					}

    1377. 

  1377. 					i++;

    1378. 

  1378. 					ch = chNext;

    1379. 

  1379. 					state = SCE_H_DEFAULT;

    1380. 

  1380. 					tagOpened = false;

    1381. 

  1381. 				} else {

    1382. 

  1382. 					if (eClass != SCE_H_TAGUNKNOWN) {

    1383. 

  1383. 						if (eClass == SCE_H_SGML_DEFAULT) {

    1384. 

  1384. 							state = SCE_H_SGML_DEFAULT;

    1385. 

  1385. 						} else {

    1386. 

  1386. 							state = SCE_H_OTHER;

    1387. 

  1387. 						}

    1388. 

  1388. 					}

    1389. 

  1389. 				}

    1390. 

  1390. 			}

    1391. 

  1391. 			break;

    1392. 

  1392. 		case SCE_H_ATTRIBUTE:

    1393. 

  1393. 			if (!setAttributeContinue.Contains(ch)) {

    1394. 

  1394. 				if (inScriptType == eNonHtmlScript) {

    1395. 

  1395. 					int scriptLanguagePrev = scriptLanguage;

    1396. 

  1396. 					clientScript = segIsScriptingIndicator(styler, styler.GetStartSegment(), i - 1, scriptLanguage);

    1397. 

  1397. 					scriptLanguage = clientScript;

    1398. 

  1398. 					if ((scriptLanguagePrev != scriptLanguage) && (scriptLanguage == eScriptNone))

    1399. 

  1399. 						inScriptType = eHtml;

    1400. 

  1400. 				}

    1401. 

  1401. 				classifyAttribHTML(styler.GetStartSegment(), i - 1, keywords, styler);

    1402. 

  1402. 				if (ch == '>') {

    1403. 

  1403. 					styler.ColourTo(i, SCE_H_TAG);

    1404. 

  1404. 					if (inScriptType == eNonHtmlScript) {

    1405. 

  1405. 						state = StateForScript(scriptLanguage);

    1406. 

  1406. 					} else {

    1407. 

  1407. 						state = SCE_H_DEFAULT;

    1408. 

  1408. 					}

    1409. 

  1409. 					tagOpened = false;

    1410. 

  1410. 					if (!tagDontFold) {

    1411. 

  1411. 						if (tagClosing) {

    1412. 

  1412. 							levelCurrent--;

    1413. 

  1413. 						} else {

    1414. 

  1414. 							levelCurrent++;

    1415. 

  1415. 						}

    1416. 

  1416. 					}

    1417. 

  1417. 					tagClosing = false;

    1418. 

  1418. 				} else if (ch == '=') {

    1419. 

  1419. 					styler.ColourTo(i, SCE_H_OTHER);

    1420. 

  1420. 					state = SCE_H_VALUE;

    1421. 

  1421. 				} else {

    1422. 

  1422. 					state = SCE_H_OTHER;

    1423. 

  1423. 				}

    1424. 

  1424. 			}

    1425. 

  1425. 			break;

    1426. 

  1426. 		case SCE_H_OTHER:

    1427. 

  1427. 			if (ch == '>') {

    1428. 

  1428. 				styler.ColourTo(i - 1, StateToPrint);

    1429. 

  1429. 				styler.ColourTo(i, SCE_H_TAG);

    1430. 

  1430. 				if (inScriptType == eNonHtmlScript) {

    1431. 

  1431. 					state = StateForScript(scriptLanguage);

    1432. 

  1432. 				} else {

    1433. 

  1433. 					state = SCE_H_DEFAULT;

    1434. 

  1434. 				}

    1435. 

  1435. 				tagOpened = false;

    1436. 

  1436. 				if (!tagDontFold) {

    1437. 

  1437. 					if (tagClosing) {

    1438. 

  1438. 						levelCurrent--;

    1439. 

  1439. 					} else {

    1440. 

  1440. 						levelCurrent++;

    1441. 

  1441. 					}

    1442. 

  1442. 				}

    1443. 

  1443. 				tagClosing = false;

    1444. 

  1444. 			} else if (ch == '\"') {

    1445. 

  1445. 				styler.ColourTo(i - 1, StateToPrint);

    1446. 

  1446. 				state = SCE_H_DOUBLESTRING;

    1447. 

  1447. 			} else if (ch == '\'') {

    1448. 

  1448. 				styler.ColourTo(i - 1, StateToPrint);

    1449. 

  1449. 				state = SCE_H_SINGLESTRING;

    1450. 

  1450. 			} else if (ch == '=') {

    1451. 

  1451. 				styler.ColourTo(i, StateToPrint);

    1452. 

  1452. 				state = SCE_H_VALUE;

    1453. 

  1453. 			} else if (ch == '/' && chNext == '>') {

    1454. 

  1454. 				styler.ColourTo(i - 1, StateToPrint);

    1455. 

  1455. 				styler.ColourTo(i + 1, SCE_H_TAGEND);

    1456. 

  1456. 				i++;

    1457. 

  1457. 				ch = chNext;

    1458. 

  1458. 				state = SCE_H_DEFAULT;

    1459. 

  1459. 				tagOpened = false;

    1460. 

  1460. 			} else if (ch == '?' && chNext == '>') {

    1461. 

  1461. 				styler.ColourTo(i - 1, StateToPrint);

    1462. 

  1462. 				styler.ColourTo(i + 1, SCE_H_XMLEND);

    1463. 

  1463. 				i++;

    1464. 

  1464. 				ch = chNext;

    1465. 

  1465. 				state = SCE_H_DEFAULT;

    1466. 

  1466. 			} else if (setHTMLWord.Contains(ch)) {

    1467. 

  1467. 				styler.ColourTo(i - 1, StateToPrint);

    1468. 

  1468. 				state = SCE_H_ATTRIBUTE;

    1469. 

  1469. 			}

    1470. 

  1470. 			break;

    1471. 

  1471. 		case SCE_H_DOUBLESTRING:

    1472. 

  1472. 			if (ch == '\"') {

    1473. 

  1473. 				if (inScriptType == eNonHtmlScript) {

    1474. 

  1474. 					scriptLanguage = segIsScriptingIndicator(styler, styler.GetStartSegment(), i, scriptLanguage);

    1475. 

  1475. 				}

    1476. 

  1476. 				styler.ColourTo(i, SCE_H_DOUBLESTRING);

    1477. 

  1477. 				state = SCE_H_OTHER;

    1478. 

  1478. 			}

    1479. 

  1479. 			break;

    1480. 

  1480. 		case SCE_H_SINGLESTRING:

    1481. 

  1481. 			if (ch == '\'') {

    1482. 

  1482. 				if (inScriptType == eNonHtmlScript) {

    1483. 

  1483. 					scriptLanguage = segIsScriptingIndicator(styler, styler.GetStartSegment(), i, scriptLanguage);

    1484. 

  1484. 				}

    1485. 

  1485. 				styler.ColourTo(i, SCE_H_SINGLESTRING);

    1486. 

  1486. 				state = SCE_H_OTHER;

    1487. 

  1487. 			}

    1488. 

  1488. 			break;

    1489. 

  1489. 		case SCE_H_VALUE:

    1490. 

  1490. 			if (!setHTMLWord.Contains(ch)) {

    1491. 

  1491. 				if (ch == '\"' && chPrev == '=') {

    1492. 

  1492. 					// Should really test for being first character

    1493. 

  1493. 					state = SCE_H_DOUBLESTRING;

    1494. 

  1494. 				} else if (ch == '\'' && chPrev == '=') {

    1495. 

  1495. 					state = SCE_H_SINGLESTRING;

    1496. 

  1496. 				} else {

    1497. 

  1497. 					if (IsNumber(styler.GetStartSegment(), styler)) {

    1498. 

  1498. 						styler.ColourTo(i - 1, SCE_H_NUMBER);

    1499. 

  1499. 					} else {

    1500. 

  1500. 						styler.ColourTo(i - 1, StateToPrint);

    1501. 

  1501. 					}

    1502. 

  1502. 					if (ch == '>') {

    1503. 

  1503. 						styler.ColourTo(i, SCE_H_TAG);

    1504. 

  1504. 						if (inScriptType == eNonHtmlScript) {

    1505. 

  1505. 							state = StateForScript(scriptLanguage);

    1506. 

  1506. 						} else {

    1507. 

  1507. 							state = SCE_H_DEFAULT;

    1508. 

  1508. 						}

    1509. 

  1509. 						tagOpened = false;

    1510. 

  1510. 						if (!tagDontFold) {

    1511. 

  1511. 							if (tagClosing) {

    1512. 

  1512. 								levelCurrent--;

    1513. 

  1513. 							} else {

    1514. 

  1514. 								levelCurrent++;

    1515. 

  1515. 							}

    1516. 

  1516. 						}

    1517. 

  1517. 						tagClosing = false;

    1518. 

  1518. 					} else {

    1519. 

  1519. 						state = SCE_H_OTHER;

    1520. 

  1520. 					}

    1521. 

  1521. 				}

    1522. 

  1522. 			}

    1523. 

  1523. 			break;

    1524. 

  1524. 		case SCE_HJ_DEFAULT:

    1525. 

  1525. 		case SCE_HJ_START:

    1526. 

  1526. 		case SCE_HJ_SYMBOLS:

    1527. 

  1527. 			if (IsAWordStart(ch)) {

    1528. 

  1528. 				styler.ColourTo(i - 1, StateToPrint);

    1529. 

  1529. 				state = SCE_HJ_WORD;

    1530. 

  1530. 			} else if (ch == '/' && chNext == '*') {

    1531. 

  1531. 				styler.ColourTo(i - 1, StateToPrint);

    1532. 

  1532. 				if (chNext2 == '*')

    1533. 

  1533. 					state = SCE_HJ_COMMENTDOC;

    1534. 

  1534. 				else

    1535. 

  1535. 					state = SCE_HJ_COMMENT;

    1536. 

  1536. 			} else if (ch == '/' && chNext == '/') {

    1537. 

  1537. 				styler.ColourTo(i - 1, StateToPrint);

    1538. 

  1538. 				state = SCE_HJ_COMMENTLINE;

    1539. 

  1539. 			} else if (ch == '/' && isOKBeforeRE(chPrevNonWhite)) {

    1540. 

  1540. 				styler.ColourTo(i - 1, StateToPrint);

    1541. 

  1541. 				state = SCE_HJ_REGEX;

    1542. 

  1542. 			} else if (ch == '\"') {

    1543. 

  1543. 				styler.ColourTo(i - 1, StateToPrint);

    1544. 

  1544. 				state = SCE_HJ_DOUBLESTRING;

    1545. 

  1545. 			} else if (ch == '\'') {

    1546. 

  1546. 				styler.ColourTo(i - 1, StateToPrint);

    1547. 

  1547. 				state = SCE_HJ_SIN…
    1548.

Large files files are truncated, but you can click here to view the full file