StringHelper.cs 12 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368
  1. using System.Diagnostics.CodeAnalysis;
  2. using System.Globalization;
  3. using System.Runtime.CompilerServices;
  4. using System.Text;
  5. using System.Text.RegularExpressions;
  6. namespace Lua.Internal;
  7. internal static class StringHelper
  8. {
  9. [MethodImpl(MethodImplOptions.AggressiveInlining)]
  10. public static ReadOnlySpan<char> Slice(string s, int i, int j)
  11. {
  12. if (i < 0) i = s.Length + i + 1;
  13. if (j < 0) j = s.Length + j + 1;
  14. if (i < 1) i = 1;
  15. if (j > s.Length) j = s.Length;
  16. return i > j ? "" : s.AsSpan()[(i - 1)..j];
  17. }
  18. public static bool TryFromStringLiteral(ReadOnlySpan<char> literal, [NotNullWhen(true)] out string? result)
  19. {
  20. var builder = new ValueStringBuilder(literal.Length);
  21. for (int i = 0; i < literal.Length; i++)
  22. {
  23. var c = literal[i];
  24. if (c is '\\' && i < literal.Length - 1)
  25. {
  26. i++;
  27. c = literal[i];
  28. switch (c)
  29. {
  30. case '\n':
  31. builder.Append('\n');
  32. break;
  33. case '\r':
  34. builder.Append('\n');
  35. // check CRLF
  36. if (i + 1 < literal.Length && literal[i + 1] is '\n')
  37. {
  38. i++;
  39. }
  40. break;
  41. case 'a':
  42. builder.Append('\a');
  43. break;
  44. case 'b':
  45. builder.Append('\b');
  46. break;
  47. case 'f':
  48. builder.Append('\f');
  49. break;
  50. case 'n':
  51. builder.Append('\n');
  52. break;
  53. case 'r':
  54. builder.Append('\r');
  55. break;
  56. case 't':
  57. builder.Append('\t');
  58. break;
  59. case 'v':
  60. builder.Append('\v');
  61. break;
  62. case '\\':
  63. builder.Append('\\');
  64. break;
  65. case '\"':
  66. builder.Append('\"');
  67. break;
  68. case '\'':
  69. builder.Append('\'');
  70. break;
  71. case '[':
  72. builder.Append('[');
  73. break;
  74. case ']':
  75. builder.Append(']');
  76. break;
  77. case 'x':
  78. i++;
  79. if (i >= literal.Length)
  80. {
  81. result = null;
  82. return false;
  83. }
  84. c = literal[i];
  85. if (IsDigit(c))
  86. {
  87. var start = i;
  88. for (int j = 0; j < 2; j++)
  89. {
  90. i++;
  91. if (i >= literal.Length) break;
  92. c = literal[i];
  93. if (!IsDigit(c)) break;
  94. }
  95. builder.Append((char)int.Parse(literal[start..i], NumberStyles.HexNumber));
  96. i--;
  97. }
  98. else
  99. {
  100. result = null;
  101. return false;
  102. }
  103. break;
  104. default:
  105. if (IsNumber(c))
  106. {
  107. var start = i;
  108. for (int j = 0; j < 3; j++)
  109. {
  110. i++;
  111. if (i >= literal.Length) break;
  112. c = literal[i];
  113. if (!IsNumber(c)) break;
  114. }
  115. builder.Append((char)int.Parse(literal[start..i]));
  116. i--;
  117. }
  118. else
  119. {
  120. result = null;
  121. return false;
  122. }
  123. break;
  124. }
  125. }
  126. else
  127. {
  128. builder.Append(c);
  129. }
  130. }
  131. result = builder.ToString();
  132. return true;
  133. }
  134. public static string Escape(ReadOnlySpan<char> str)
  135. {
  136. var builder = new ValueStringBuilder(str.Length);
  137. for (int i = 0; i < str.Length; i++)
  138. {
  139. var c = str[i];
  140. switch (c)
  141. {
  142. case '\a':
  143. builder.Append("\\\a");
  144. break;
  145. case '\b':
  146. builder.Append("\\\b");
  147. break;
  148. case '\f':
  149. builder.Append("\\\f");
  150. break;
  151. case '\n':
  152. builder.Append("\\\n");
  153. break;
  154. case '\r':
  155. builder.Append("\\\r");
  156. break;
  157. case '\t':
  158. builder.Append("\\\t");
  159. break;
  160. case '\v':
  161. builder.Append("\\\v");
  162. break;
  163. case '\\':
  164. builder.Append("\\\\");
  165. break;
  166. case '\"':
  167. builder.Append("\\\"");
  168. break;
  169. case '\'':
  170. builder.Append("\\\'");
  171. break;
  172. default:
  173. builder.Append(c);
  174. break;
  175. }
  176. }
  177. return builder.ToString();
  178. }
  179. public static Regex ToRegex(ReadOnlySpan<char> pattern)
  180. {
  181. var builder = new ValueStringBuilder();
  182. var isEscapeSequence = false;
  183. var isInSet = false;
  184. for (var i = 0; i < pattern.Length; i++)
  185. {
  186. var c = pattern[i];
  187. if (isEscapeSequence)
  188. {
  189. if (c == '%' || c == '_')
  190. {
  191. builder.Append(c);
  192. isEscapeSequence = false;
  193. }
  194. else
  195. {
  196. switch (c)
  197. {
  198. case 'a': // all letters
  199. builder.Append("\\p{L}");
  200. break;
  201. case 'A': // all Non letters
  202. builder.Append("\\P{L}");
  203. break;
  204. case 's': // all space characters
  205. builder.Append("\\s");
  206. break;
  207. case 'S': // all NON space characters
  208. builder.Append("\\S");
  209. break;
  210. case 'd': // all digits
  211. builder.Append("\\d");
  212. break;
  213. case 'D': // all NON digits
  214. builder.Append("\\D");
  215. break;
  216. case 'w': // all alphanumeric characters
  217. builder.Append("\\w");
  218. break;
  219. case 'W': // all NON alphanumeric characters
  220. builder.Append("\\W");
  221. break;
  222. case 'c': // all control characters
  223. builder.Append("\\p{C}");
  224. break;
  225. case 'C': // all NON control characters
  226. builder.Append("[\\P{C}]");
  227. break;
  228. case 'g': // all printable characters except space
  229. builder.Append("[^\\p{C}\\s]");
  230. break;
  231. case 'G': // all NON printable characters including space
  232. builder.Append("[\\p{C}\\s]");
  233. break;
  234. case 'p': // all punctuation characters
  235. builder.Append("\\p{P}");
  236. break;
  237. case 'P': // all NON punctuation characters
  238. builder.Append("\\P{P}");
  239. break;
  240. case 'l': // all lowercase letters
  241. builder.Append("\\p{Ll}");
  242. break;
  243. case 'L': // all NON lowercase letters
  244. builder.Append("\\P{Ll}");
  245. break;
  246. case 'u': // all uppercase letters
  247. builder.Append("\\p{Lu}");
  248. break;
  249. case 'U': // all NON uppercase letters
  250. builder.Append("\\P{Lu}");
  251. break;
  252. case 'x': // all hexadecimal digits
  253. builder.Append("[0-9A-Fa-f]");
  254. break;
  255. case 'X': // all NON hexadecimal digits
  256. builder.Append("[^0-9A-Fa-f]");
  257. break;
  258. case 'b':
  259. if (i < pattern.Length - 2)
  260. {
  261. var c1 = pattern[i + 1];
  262. var c2 = pattern[i + 2];
  263. var c1Escape = Regex.Escape(c1.ToString());
  264. var c2Escape = Regex.Escape(c2.ToString());
  265. builder.Append("(");
  266. builder.Append(c1Escape);
  267. builder.Append("(?>(?<n>");
  268. builder.Append(c1Escape);
  269. builder.Append(")|(?<-n>");
  270. builder.Append(c2Escape);
  271. builder.Append(")|(?:[^");
  272. builder.Append(c1Escape);
  273. builder.Append(c2Escape);
  274. builder.Append("]*))*");
  275. builder.Append(c2Escape);
  276. builder.Append("(?(n)(?!)))");
  277. i += 2;
  278. }
  279. else
  280. {
  281. throw new Exception(); // TODO: add message
  282. }
  283. break;
  284. default:
  285. builder.Append('\\');
  286. builder.Append(c);
  287. break;
  288. }
  289. isEscapeSequence = false;
  290. }
  291. }
  292. else if (c == '%')
  293. {
  294. isEscapeSequence = true;
  295. }
  296. else if (c == '\\')
  297. {
  298. builder.Append("\\\\");
  299. }
  300. else if (isInSet)
  301. {
  302. if (c == ']') isInSet = false;
  303. builder.Append(c);
  304. }
  305. else if (c == '-')
  306. {
  307. builder.Append("*?");
  308. }
  309. else if (c == '[')
  310. {
  311. builder.Append('[');
  312. isInSet = true;
  313. }
  314. else if (c == '^' && !isInSet)
  315. {
  316. builder.Append("\\G");
  317. }
  318. else if (c == '(')
  319. {
  320. builder.Append('(');
  321. }
  322. else
  323. {
  324. builder.Append(c);
  325. }
  326. }
  327. return new Regex(builder.ToString());
  328. }
  329. [MethodImpl(MethodImplOptions.AggressiveInlining)]
  330. public static bool IsNumber(char c)
  331. {
  332. return '0' <= c && c <= '9';
  333. }
  334. [MethodImpl(MethodImplOptions.AggressiveInlining)]
  335. public static bool IsDigit(char c)
  336. {
  337. return IsNumber(c) ||
  338. ('a' <= c && c <= 'f') ||
  339. ('A' <= c && c <= 'F');
  340. }
  341. }