Tokenizer.java 8.5 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325
  1. package me.hammerle.snuviscript.tokenizer;
  2. import java.io.InputStream;
  3. import java.util.ArrayList;
  4. import me.hammerle.snuviscript.exceptions.PreScriptException;
  5. import static me.hammerle.snuviscript.tokenizer.TokenType.*;
  6. public class Tokenizer {
  7. private StreamCharReader stream = null;
  8. private final ArrayList<Token> tokens = new ArrayList<>();
  9. private int line = 1;
  10. private int next() {
  11. return stream.readChar();
  12. }
  13. private int peek() {
  14. return stream.peekChar();
  15. }
  16. private boolean next(char c) {
  17. if(peek() == c) {
  18. next();
  19. return true;
  20. }
  21. return false;
  22. }
  23. private void add(TokenType type) {
  24. tokens.add(new Token(type, line));
  25. }
  26. private void add(TokenType type, Object data) {
  27. tokens.add(new DataToken(type, line, data));
  28. }
  29. private void add(char c, TokenType t1, TokenType t2, TokenType t3, TokenType t4) {
  30. int peek = peek();
  31. if(peek == c) {
  32. next();
  33. if(peek() == '=') {
  34. next();
  35. add(t1);
  36. } else {
  37. add(t2);
  38. }
  39. } else if(peek == '=') {
  40. next();
  41. add(t3);
  42. } else {
  43. add(t4);
  44. }
  45. }
  46. public Token[] tokenize(InputStream... streams) {
  47. tokens.clear();
  48. int fileCounter = 0;
  49. for(InputStream in : streams) {
  50. fileCounter++;
  51. line = (fileCounter << 24) | 1;
  52. stream = new StreamCharReader(in);
  53. int c;
  54. while((c = Tokenizer.this.next()) != -1) {
  55. handleChar(c);
  56. }
  57. }
  58. add(EOF);
  59. return tokens.toArray(new Token[tokens.size()]);
  60. }
  61. private void handleChar(int c) {
  62. if(Character.isLetter(c) || c == '_' || c == '.') {
  63. handleLiteral(c, TokenType.LITERAL);
  64. } else if(Character.isDigit(c)) {
  65. handleNumber(c);
  66. } else {
  67. handleSpecial(c);
  68. }
  69. }
  70. private void handleLiteral(int c, TokenType type) {
  71. StringBuilder sb = new StringBuilder();
  72. sb.append((char) c);
  73. while(true) {
  74. int data = peek();
  75. if(!Character.isLetterOrDigit(data) && data != '_' && data != '.') {
  76. break;
  77. }
  78. sb.append((char) data);
  79. next();
  80. }
  81. String s = sb.toString();
  82. switch(s) {
  83. case "if":
  84. add(IF);
  85. break;
  86. case "else":
  87. add(ELSE);
  88. break;
  89. case "elseif":
  90. add(ELSEIF);
  91. break;
  92. case "while":
  93. add(WHILE);
  94. break;
  95. case "try":
  96. add(TRY);
  97. break;
  98. case "catch":
  99. add(CATCH);
  100. break;
  101. case "for":
  102. add(FOR);
  103. break;
  104. case "function":
  105. add(FUNCTION);
  106. break;
  107. case "break":
  108. add(BREAK);
  109. break;
  110. case "continue":
  111. add(CONTINUE);
  112. break;
  113. case "return":
  114. add(RETURN);
  115. break;
  116. case "true":
  117. add(TRUE);
  118. break;
  119. case "false":
  120. add(FALSE);
  121. break;
  122. case "null":
  123. add(NULL);
  124. break;
  125. default:
  126. add(type, s);
  127. }
  128. }
  129. private void handleNumber(int c) {
  130. StringBuilder sb = new StringBuilder();
  131. sb.append((char) c);
  132. while(true) {
  133. int data = peek();
  134. if(!Character.isLetterOrDigit(data) && data != '.') {
  135. break;
  136. }
  137. next();
  138. sb.append((char) data);
  139. }
  140. add(NUMBER, Double.parseDouble(sb.toString()));
  141. }
  142. private void handleSpecial(int c) {
  143. switch(c) {
  144. case ' ':
  145. case '\t':
  146. case '\r':
  147. break;
  148. case '\n':
  149. line++;
  150. break;
  151. case '"':
  152. handleString();
  153. break;
  154. case '(':
  155. add(OPEN_BRACKET);
  156. break;
  157. case ')':
  158. add(CLOSE_BRACKET);
  159. break;
  160. case '[':
  161. add(OPEN_SQUARE_BRACKET);
  162. break;
  163. case ']':
  164. add(CLOSE_SQUARE_BRACKET);
  165. break;
  166. case '{':
  167. add(OPEN_CURVED_BRACKET);
  168. break;
  169. case '}':
  170. add(CLOSE_CURVED_BRACKET);
  171. break;
  172. case '$':
  173. handleLiteral(c, LITERAL);
  174. break;
  175. case '@':
  176. handleLiteral(c, LABEL);
  177. break;
  178. case ';':
  179. add(SEMICOLON);
  180. break;
  181. case ',':
  182. add(COMMA);
  183. break;
  184. case '~':
  185. add(BIT_INVERT);
  186. break;
  187. case '+':
  188. add(next('=') ? ADD_SET : (next('+') ? INC : ADD));
  189. break;
  190. case '-':
  191. add(next('=') ? SUB_SET : (next('-') ? DEC : SUB));
  192. break;
  193. case '!':
  194. add(next('=') ? NOT_EQUAL : INVERT);
  195. break;
  196. case '=':
  197. add(next('=') ? EQUAL : SET);
  198. break;
  199. case '*':
  200. add(next('=') ? MUL_SET : MUL);
  201. break;
  202. case '/':
  203. handleSlash();
  204. break;
  205. case '%':
  206. add(next('=') ? MOD_SET : MOD);
  207. break;
  208. case '&':
  209. add(next('=') ? BIT_AND_SET : (next('&') ? AND : BIT_AND));
  210. break;
  211. case '|':
  212. add(next('=') ? BIT_OR_SET : (next('|') ? OR : BIT_OR));
  213. break;
  214. case '^':
  215. add(next('=') ? BIT_XOR_SET : BIT_XOR);
  216. break;
  217. case '<':
  218. add('<', LEFT_SHIFT_SET, LEFT_SHIFT, LESS_EQUAL, LESS);
  219. break;
  220. case '>':
  221. add('>', RIGHT_SHIFT_SET, RIGHT_SHIFT, GREATER_EQUAL, GREATER);
  222. break;
  223. default:
  224. throw new PreScriptException("unknown token " + c, line);
  225. }
  226. }
  227. private void handleString() {
  228. StringBuilder sb = new StringBuilder();
  229. int oldLine = line;
  230. while(true) {
  231. int data = next();
  232. if(data == -1) {
  233. throw new PreScriptException("non closed string literal", oldLine);
  234. }
  235. if(data == '"') {
  236. add(STRING, sb.toString());
  237. break;
  238. }
  239. if(data == '\n') {
  240. line++;
  241. }
  242. if(data == '\\') {
  243. int escape = next();
  244. switch(escape) {
  245. case 'n':
  246. data = '\n';
  247. break;
  248. case '\\':
  249. data = '\\';
  250. break;
  251. case '"':
  252. data = '"';
  253. break;
  254. default:
  255. throw new PreScriptException("invalid escaped character", line);
  256. }
  257. }
  258. sb.append((char) data);
  259. }
  260. }
  261. private void handleSlash() {
  262. switch(peek()) {
  263. case '/':
  264. next();
  265. handleOneLineComment();
  266. break;
  267. case '*':
  268. next();
  269. handleMultiLineComment();
  270. break;
  271. case '=':
  272. next();
  273. add(DIV_SET);
  274. break;
  275. default:
  276. add(DIV);
  277. }
  278. }
  279. private void handleOneLineComment() {
  280. while(true) {
  281. int data = next();
  282. if(data == -1 || data == '\n') {
  283. line++;
  284. break;
  285. }
  286. }
  287. }
  288. private void handleMultiLineComment() {
  289. int first;
  290. int sec = -1;
  291. while(true) {
  292. first = sec;
  293. sec = next();
  294. if(sec == -1 || (first == '*' && sec == '/')) {
  295. break;
  296. }
  297. if(sec == '\n') {
  298. line++;
  299. }
  300. }
  301. }
  302. }