XmlChar.cs 12 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358
  1. // -*- Mode: C; tab-width: 8; indent-tabs-mode: t; c-basic-offset: 8 -*-
  2. //
  3. // System.Xml.XmlChar.cs
  4. //
  5. // Author:
  6. // Jason Diamond ([email protected])
  7. //
  8. // (C) 2001 Jason Diamond http://injektilo.org/
  9. //
  10. namespace System.Xml
  11. {
  12. // Now, "XmlChar" and "XmlConstructs" are made as equivalent, so
  13. // I dicided to rename XmlConstruts class as "XmlChar" and use it
  14. // for default build.
  15. // However, this class will be used for the future compact framework
  16. // (XmlConstruts class uses not a little memory).
  17. internal class XmlCharCompact
  18. {
  19. public static char [] WhitespaceChars = new char [] {' ', '\n', '\t', '\r'};
  20. public static bool IsWhitespace (int ch)
  21. {
  22. return ch == 0x20 || ch == 0x9 || ch == 0xD || ch == 0xA;
  23. }
  24. public static bool IsWhitespace (string str)
  25. {
  26. for (int i = 0; i < str.Length; i++)
  27. if (!IsWhitespace (str [i])) return false;
  28. return true;
  29. }
  30. public static bool IsFirstNameChar (int ch)
  31. {
  32. bool result = false;
  33. if (ch >= 0 && ch <= 0xFFFF)
  34. {
  35. result = (nameBitmap[(firstNamePages[ch >> 8] << 3) + ((ch & 0xFF) >> 5)] & (1 << (ch & 0x1F))) != 0;
  36. }
  37. return result;
  38. }
  39. public static bool IsValid (int ch)
  40. {
  41. return !IsInvalid (ch);
  42. }
  43. public static bool IsInvalid (int ch)
  44. {
  45. switch (ch) {
  46. case 9:
  47. case 10:
  48. case 13:
  49. return false;
  50. }
  51. if (ch < 32)
  52. return true;
  53. if (ch < 0xD800)
  54. return false;
  55. if (ch < 0xE000)
  56. return true;
  57. if (ch < 0xFFFE)
  58. return false;
  59. if (ch < 0x10000)
  60. return true;
  61. if (ch < 0x110000)
  62. return false;
  63. else
  64. return true;
  65. }
  66. public static bool IsNameChar (int ch)
  67. {
  68. bool result = false;
  69. if (ch >= 0 && ch <= 0xFFFF)
  70. {
  71. result = (nameBitmap[(namePages[ch >> 8] << 3) + ((ch & 0xFF) >> 5)] & (1 << (ch & 0x1F))) != 0;
  72. }
  73. return result;
  74. }
  75. public static bool IsNCNameChar (int ch)
  76. {
  77. bool result = false;
  78. if (ch >= 0 && ch <= 0xFFFF && ch != ':')
  79. {
  80. result = (nameBitmap[(namePages[ch >> 8] << 3) + ((ch & 0xFF) >> 5)] & (1 << (ch & 0x1F))) != 0;
  81. }
  82. return result;
  83. }
  84. public static bool IsName (string str)
  85. {
  86. if (str.Length == 0)
  87. return false;
  88. if (!IsFirstNameChar (str [0]))
  89. return false;
  90. for (int i = 1; i < str.Length; i++)
  91. if (!IsNameChar (str [i]))
  92. return false;
  93. return true;
  94. }
  95. public static bool IsNCName (string str)
  96. {
  97. if (str.Length == 0)
  98. return false;
  99. if (!IsFirstNameChar (str [0]))
  100. return false;
  101. for (int i = 0; i < str.Length; i++)
  102. if (!IsNCNameChar (str [i]))
  103. return false;
  104. return true;
  105. }
  106. public static bool IsNmToken (string str)
  107. {
  108. if (str.Length == 0)
  109. return false;
  110. for (int i = 0; i < str.Length; i++)
  111. if (!IsNameChar (str [i]))
  112. return false;
  113. return true;
  114. }
  115. public static bool IsPubidChar (int ch)
  116. {
  117. return (IsWhitespace(ch) && ch != '\t') | ('a' <= ch && ch <= 'z') | ('A' <= ch && ch <= 'Z') | ('0' <= ch && ch <= '9') | "-'()+,./:=?;!*#@$_%".IndexOf((char)ch) >= 0;
  118. }
  119. public static bool IsPubid (string str)
  120. {
  121. for (int i = 0; i < str.Length; i++)
  122. if (!IsPubidChar (str [i]))
  123. return false;
  124. return true;
  125. }
  126. // encodings (copied from XmlConstructs.cs)
  127. /// <summary>
  128. /// Returns true if the encoding name is a valid IANA encoding.
  129. /// This method does not verify that there is a decoder available
  130. /// for this encoding, only that the characters are valid for an
  131. /// IANA encoding name.
  132. /// </summary>
  133. /// <param name="ianaEncoding">The encoding to check.</param>
  134. /// <returns></returns>
  135. public static bool IsValidIANAEncoding (String ianaEncoding)
  136. {
  137. if (ianaEncoding != null)
  138. {
  139. int length = ianaEncoding.Length;
  140. if (length > 0)
  141. {
  142. char c = ianaEncoding[0];
  143. if ((c >= 'A' && c <= 'Z') || (c >= 'a' && c <= 'z'))
  144. {
  145. for (int i = 1; i < length; i++)
  146. {
  147. c = ianaEncoding[i];
  148. if ((c < 'A' || c > 'Z') && (c < 'a' || c > 'z') &&
  149. (c < '0' || c > '9') && c != '.' && c != '_' &&
  150. c != '-')
  151. {
  152. return false;
  153. }
  154. }
  155. return true;
  156. }
  157. }
  158. }
  159. return false;
  160. }
  161. public static int GetPredefinedEntity (string name)
  162. {
  163. switch (name) {
  164. case "amp":
  165. return '&';
  166. case "lt":
  167. return '<';
  168. case "gt":
  169. return '>';
  170. case "quot":
  171. return '"';
  172. case "apos":
  173. return '\'';
  174. default:
  175. return -1;
  176. }
  177. }
  178. private static byte[] firstNamePages =
  179. {
  180. 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 0x08, 0x00,
  181. 0x00, 0x09, 0x0A, 0x0B, 0x0C, 0x0D, 0x0E, 0x0F,
  182. 0x10, 0x11, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
  183. 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x12, 0x13,
  184. 0x00, 0x14, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
  185. 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
  186. 0x15, 0x16, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
  187. 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
  188. 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
  189. 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x01, 0x01,
  190. 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01,
  191. 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01,
  192. 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01,
  193. 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01,
  194. 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01,
  195. 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01,
  196. 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01,
  197. 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01,
  198. 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01,
  199. 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x17,
  200. 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
  201. 0x00, 0x00, 0x00, 0x00, 0x01, 0x01, 0x01, 0x01,
  202. 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01,
  203. 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01,
  204. 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01,
  205. 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01,
  206. 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x18,
  207. 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
  208. 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
  209. 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
  210. 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
  211. 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00
  212. };
  213. private static byte[] namePages =
  214. {
  215. 0x19, 0x03, 0x1A, 0x1B, 0x1C, 0x1D, 0x1E, 0x00,
  216. 0x00, 0x1F, 0x20, 0x21, 0x22, 0x23, 0x24, 0x25,
  217. 0x10, 0x11, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
  218. 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x12, 0x13,
  219. 0x26, 0x14, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
  220. 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
  221. 0x27, 0x16, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
  222. 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
  223. 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
  224. 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x01, 0x01,
  225. 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01,
  226. 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01,
  227. 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01,
  228. 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01,
  229. 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01,
  230. 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01,
  231. 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01,
  232. 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01,
  233. 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01,
  234. 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x17,
  235. 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
  236. 0x00, 0x00, 0x00, 0x00, 0x01, 0x01, 0x01, 0x01,
  237. 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01,
  238. 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01,
  239. 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01,
  240. 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01,
  241. 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x18,
  242. 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
  243. 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
  244. 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
  245. 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
  246. 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00
  247. };
  248. private static uint[] nameBitmap =
  249. {
  250. 0x00000000, 0x00000000, 0x00000000, 0x00000000,
  251. 0x00000000, 0x00000000, 0x00000000, 0x00000000,
  252. 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
  253. 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
  254. 0x00000000, 0x04000000, 0x87FFFFFE, 0x07FFFFFE,
  255. 0x00000000, 0x00000000, 0xFF7FFFFF, 0xFF7FFFFF,
  256. 0xFFFFFFFF, 0x7FF3FFFF, 0xFFFFFDFE, 0x7FFFFFFF,
  257. 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFE00F, 0xFC31FFFF,
  258. 0x00FFFFFF, 0x00000000, 0xFFFF0000, 0xFFFFFFFF,
  259. 0xFFFFFFFF, 0xF80001FF, 0x00000003, 0x00000000,
  260. 0x00000000, 0x00000000, 0x00000000, 0x00000000,
  261. 0xFFFFD740, 0xFFFFFFFB, 0x547F7FFF, 0x000FFFFD,
  262. 0xFFFFDFFE, 0xFFFFFFFF, 0xDFFEFFFF, 0xFFFFFFFF,
  263. 0xFFFF0003, 0xFFFFFFFF, 0xFFFF199F, 0x033FCFFF,
  264. 0x00000000, 0xFFFE0000, 0x027FFFFF, 0xFFFFFFFE,
  265. 0x0000007F, 0x00000000, 0xFFFF0000, 0x000707FF,
  266. 0x00000000, 0x07FFFFFE, 0x000007FE, 0xFFFE0000,
  267. 0xFFFFFFFF, 0x7CFFFFFF, 0x002F7FFF, 0x00000060,
  268. 0xFFFFFFE0, 0x23FFFFFF, 0xFF000000, 0x00000003,
  269. 0xFFF99FE0, 0x03C5FDFF, 0xB0000000, 0x00030003,
  270. 0xFFF987E0, 0x036DFDFF, 0x5E000000, 0x001C0000,
  271. 0xFFFBAFE0, 0x23EDFDFF, 0x00000000, 0x00000001,
  272. 0xFFF99FE0, 0x23CDFDFF, 0xB0000000, 0x00000003,
  273. 0xD63DC7E0, 0x03BFC718, 0x00000000, 0x00000000,
  274. 0xFFFDDFE0, 0x03EFFDFF, 0x00000000, 0x00000003,
  275. 0xFFFDDFE0, 0x03EFFDFF, 0x40000000, 0x00000003,
  276. 0xFFFDDFE0, 0x03FFFDFF, 0x00000000, 0x00000003,
  277. 0x00000000, 0x00000000, 0x00000000, 0x00000000,
  278. 0xFFFFFFFE, 0x000D7FFF, 0x0000003F, 0x00000000,
  279. 0xFEF02596, 0x200D6CAE, 0x0000001F, 0x00000000,
  280. 0x00000000, 0x00000000, 0xFFFFFEFF, 0x000003FF,
  281. 0x00000000, 0x00000000, 0x00000000, 0x00000000,
  282. 0x00000000, 0x00000000, 0x00000000, 0x00000000,
  283. 0x00000000, 0xFFFFFFFF, 0xFFFF003F, 0x007FFFFF,
  284. 0x0007DAED, 0x50000000, 0x82315001, 0x002C62AB,
  285. 0x40000000, 0xF580C900, 0x00000007, 0x02010800,
  286. 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
  287. 0x0FFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0x03FFFFFF,
  288. 0x3F3FFFFF, 0xFFFFFFFF, 0xAAFF3F3F, 0x3FFFFFFF,
  289. 0xFFFFFFFF, 0x5FDFFFFF, 0x0FCF1FDC, 0x1FDC1FFF,
  290. 0x00000000, 0x00004C40, 0x00000000, 0x00000000,
  291. 0x00000007, 0x00000000, 0x00000000, 0x00000000,
  292. 0x00000080, 0x000003FE, 0xFFFFFFFE, 0xFFFFFFFF,
  293. 0x001FFFFF, 0xFFFFFFFE, 0xFFFFFFFF, 0x07FFFFFF,
  294. 0xFFFFFFE0, 0x00001FFF, 0x00000000, 0x00000000,
  295. 0x00000000, 0x00000000, 0x00000000, 0x00000000,
  296. 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
  297. 0xFFFFFFFF, 0x0000003F, 0x00000000, 0x00000000,
  298. 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF,
  299. 0xFFFFFFFF, 0x0000000F, 0x00000000, 0x00000000,
  300. 0x00000000, 0x07FF6000, 0x87FFFFFE, 0x07FFFFFE,
  301. 0x00000000, 0x00800000, 0xFF7FFFFF, 0xFF7FFFFF,
  302. 0x00FFFFFF, 0x00000000, 0xFFFF0000, 0xFFFFFFFF,
  303. 0xFFFFFFFF, 0xF80001FF, 0x00030003, 0x00000000,
  304. 0xFFFFFFFF, 0xFFFFFFFF, 0x0000003F, 0x00000003,
  305. 0xFFFFD7C0, 0xFFFFFFFB, 0x547F7FFF, 0x000FFFFD,
  306. 0xFFFFDFFE, 0xFFFFFFFF, 0xDFFEFFFF, 0xFFFFFFFF,
  307. 0xFFFF007B, 0xFFFFFFFF, 0xFFFF199F, 0x033FCFFF,
  308. 0x00000000, 0xFFFE0000, 0x027FFFFF, 0xFFFFFFFE,
  309. 0xFFFE007F, 0xBBFFFFFB, 0xFFFF0016, 0x000707FF,
  310. 0x00000000, 0x07FFFFFE, 0x0007FFFF, 0xFFFF03FF,
  311. 0xFFFFFFFF, 0x7CFFFFFF, 0xFFEF7FFF, 0x03FF3DFF,
  312. 0xFFFFFFEE, 0xF3FFFFFF, 0xFF1E3FFF, 0x0000FFCF,
  313. 0xFFF99FEE, 0xD3C5FDFF, 0xB080399F, 0x0003FFCF,
  314. 0xFFF987E4, 0xD36DFDFF, 0x5E003987, 0x001FFFC0,
  315. 0xFFFBAFEE, 0xF3EDFDFF, 0x00003BBF, 0x0000FFC1,
  316. 0xFFF99FEE, 0xF3CDFDFF, 0xB0C0398F, 0x0000FFC3,
  317. 0xD63DC7EC, 0xC3BFC718, 0x00803DC7, 0x0000FF80,
  318. 0xFFFDDFEE, 0xC3EFFDFF, 0x00603DDF, 0x0000FFC3,
  319. 0xFFFDDFEC, 0xC3EFFDFF, 0x40603DDF, 0x0000FFC3,
  320. 0xFFFDDFEC, 0xC3FFFDFF, 0x00803DCF, 0x0000FFC3,
  321. 0x00000000, 0x00000000, 0x00000000, 0x00000000,
  322. 0xFFFFFFFE, 0x07FF7FFF, 0x03FF7FFF, 0x00000000,
  323. 0xFEF02596, 0x3BFF6CAE, 0x03FF3F5F, 0x00000000,
  324. 0x03000000, 0xC2A003FF, 0xFFFFFEFF, 0xFFFE03FF,
  325. 0xFEBF0FDF, 0x02FE3FFF, 0x00000000, 0x00000000,
  326. 0x00000000, 0x00000000, 0x00000000, 0x00000000,
  327. 0x00000000, 0x00000000, 0x1FFF0000, 0x00000002,
  328. 0x000000A0, 0x003EFFFE, 0xFFFFFFFE, 0xFFFFFFFF,
  329. 0x661FFFFF, 0xFFFFFFFE, 0xFFFFFFFF, 0x77FFFFFF
  330. };
  331. }
  332. }