Tokenizer.java 8.7 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333
  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 "for":
  96. add(FOR);
  97. break;
  98. case "function":
  99. add(FUNCTION);
  100. break;
  101. case "break":
  102. add(BREAK);
  103. break;
  104. case "continue":
  105. add(CONTINUE);
  106. break;
  107. case "return":
  108. add(RETURN);
  109. break;
  110. case "true":
  111. add(TRUE);
  112. break;
  113. case "false":
  114. add(FALSE);
  115. break;
  116. case "null":
  117. add(NULL);
  118. break;
  119. default:
  120. add(type, s);
  121. }
  122. }
  123. private void handleNumber(int c) {
  124. StringBuilder sb = new StringBuilder();
  125. sb.append((char) c);
  126. while(true) {
  127. int data = peek();
  128. if(!Character.isDigit(data)) {
  129. break;
  130. }
  131. next();
  132. sb.append((char) data);
  133. }
  134. if(peek() == '.') {
  135. sb.append((char) next());
  136. }
  137. while(true) {
  138. int data = peek();
  139. if(!Character.isDigit(data)) {
  140. break;
  141. }
  142. next();
  143. sb.append((char) data);
  144. }
  145. try {
  146. add(NUMBER, Double.parseDouble(sb.toString()));
  147. } catch(NumberFormatException ex) {
  148. throw new PreScriptException("invalid number", line);
  149. }
  150. }
  151. private void handleSpecial(int c) {
  152. switch(c) {
  153. case ' ':
  154. case '\t':
  155. case '\r':
  156. break;
  157. case '\n':
  158. line++;
  159. break;
  160. case '"':
  161. handleString();
  162. break;
  163. case '(':
  164. add(OPEN_BRACKET);
  165. break;
  166. case ')':
  167. add(CLOSE_BRACKET);
  168. break;
  169. case '[':
  170. add(OPEN_SQUARE_BRACKET);
  171. break;
  172. case ']':
  173. add(CLOSE_SQUARE_BRACKET);
  174. break;
  175. case '{':
  176. add(OPEN_CURVED_BRACKET);
  177. break;
  178. case '}':
  179. add(CLOSE_CURVED_BRACKET);
  180. break;
  181. case '$':
  182. handleLiteral(c, LITERAL);
  183. break;
  184. case '@':
  185. handleLiteral(c, LABEL);
  186. break;
  187. case ';':
  188. add(SEMICOLON);
  189. break;
  190. case ',':
  191. add(COMMA);
  192. break;
  193. case '~':
  194. add(BIT_INVERT);
  195. break;
  196. case '+':
  197. add(next('=') ? ADD_SET : (next('+') ? INC : ADD));
  198. break;
  199. case '-':
  200. add(next('=') ? SUB_SET : (next('-') ? DEC : SUB));
  201. break;
  202. case '!':
  203. add(next('=') ? NOT_EQUAL : INVERT);
  204. break;
  205. case '=':
  206. add(next('=') ? EQUAL : SET);
  207. break;
  208. case '*':
  209. add(next('=') ? MUL_SET : MUL);
  210. break;
  211. case '/':
  212. handleSlash();
  213. break;
  214. case '%':
  215. add(next('=') ? MOD_SET : MOD);
  216. break;
  217. case '&':
  218. add(next('=') ? BIT_AND_SET : (next('&') ? AND : BIT_AND));
  219. break;
  220. case '|':
  221. add(next('=') ? BIT_OR_SET : (next('|') ? OR : BIT_OR));
  222. break;
  223. case '^':
  224. add(next('=') ? BIT_XOR_SET : BIT_XOR);
  225. break;
  226. case '<':
  227. add('<', LEFT_SHIFT_SET, LEFT_SHIFT, LESS_EQUAL, LESS);
  228. break;
  229. case '>':
  230. add('>', RIGHT_SHIFT_SET, RIGHT_SHIFT, GREATER_EQUAL, GREATER);
  231. break;
  232. default:
  233. throw new PreScriptException("unknown token " + c, line);
  234. }
  235. }
  236. private void handleString() {
  237. StringBuilder sb = new StringBuilder();
  238. int oldLine = line;
  239. while(true) {
  240. int data = next();
  241. if(data == -1) {
  242. throw new PreScriptException("non closed string literal", oldLine);
  243. }
  244. if(data == '"') {
  245. add(STRING, sb.toString());
  246. break;
  247. }
  248. if(data == '\n') {
  249. line++;
  250. }
  251. if(data == '\\') {
  252. int escape = next();
  253. switch(escape) {
  254. case 'n':
  255. data = '\n';
  256. break;
  257. case '\\':
  258. data = '\\';
  259. break;
  260. case '"':
  261. data = '"';
  262. break;
  263. default:
  264. throw new PreScriptException("invalid escaped character", line);
  265. }
  266. }
  267. sb.append((char) data);
  268. }
  269. }
  270. private void handleSlash() {
  271. switch(peek()) {
  272. case '/':
  273. next();
  274. handleOneLineComment();
  275. break;
  276. case '*':
  277. next();
  278. handleMultiLineComment();
  279. break;
  280. case '=':
  281. next();
  282. add(DIV_SET);
  283. break;
  284. default:
  285. add(DIV);
  286. }
  287. }
  288. private void handleOneLineComment() {
  289. while(true) {
  290. int data = next();
  291. if(data == -1 || data == '\n') {
  292. line++;
  293. break;
  294. }
  295. }
  296. }
  297. private void handleMultiLineComment() {
  298. int first;
  299. int sec = -1;
  300. while(true) {
  301. first = sec;
  302. sec = next();
  303. if(sec == -1 || (first == '*' && sec == '/')) {
  304. break;
  305. }
  306. if(sec == '\n') {
  307. line++;
  308. }
  309. }
  310. }
  311. }