|
| 1 | +package week5; |
| 2 | + |
| 3 | +import java.io.IOException; |
| 4 | +import java.io.Reader; |
| 5 | +import java.io.UncheckedIOException; |
| 6 | +import java.nio.charset.StandardCharsets; |
| 7 | +import java.nio.file.Files; |
| 8 | +import java.nio.file.Path; |
| 9 | +import java.nio.file.Paths; |
| 10 | +import java.util.ArrayList; |
| 11 | +import java.util.List; |
| 12 | + |
| 13 | +import static week5.AktkToken.Type.*; |
| 14 | + |
| 15 | +public class AktkHandwrittenLexer { |
| 16 | + |
| 17 | + // Siin on meie poolt mall ette antud, millega saad kasutada praktikumist tuttavad meetodid. |
| 18 | + // Selle malli kasutamine ei ole kohustuslik, aga sisend antakse ette Reader liidese kaudu |
| 19 | + // ja viimase testi läbimiseks ei tohi sisendis liiga palju ette vaadata. |
| 20 | + |
| 21 | + private static final char TERMINATOR = '\0'; |
| 22 | + private final Reader reader; |
| 23 | + private Character current; |
| 24 | + private int pos = 0; |
| 25 | + |
| 26 | + public AktkHandwrittenLexer(Reader input) { |
| 27 | + this.reader = input; |
| 28 | + } |
| 29 | + |
| 30 | + // See teeb meil testimise mugavamaks: |
| 31 | + public List<AktkToken> readAllTokens() { |
| 32 | + List<AktkToken> tokens = new ArrayList<>(); |
| 33 | + |
| 34 | + AktkToken token; |
| 35 | + do { |
| 36 | + token = this.readNextToken(); |
| 37 | + tokens.add(token); |
| 38 | + } while (token.type() != EOF); |
| 39 | + |
| 40 | + return tokens; |
| 41 | + } |
| 42 | + |
| 43 | + // Saame kasutada näiteks järgmiselt: |
| 44 | + static void main() throws IOException { |
| 45 | + Path path = Paths.get("inputs", "escaped.txt"); |
| 46 | + System.out.println(Files.readString(path)); |
| 47 | + System.out.println(new AktkHandwrittenLexer(Files.newBufferedReader(path, StandardCharsets.UTF_8)).readAllTokens()); |
| 48 | + } |
| 49 | + |
| 50 | + |
| 51 | + // Siin on abimeetodid, mida soovitame kasutada. |
| 52 | + // Consume on sisuliselt nagu pos++ kalalekseri näites ehk läheb järgmise tähe juurde. |
| 53 | + private void consume() { |
| 54 | + if (current == TERMINATOR) { |
| 55 | + throw new RuntimeException("Reading passed terminator!"); |
| 56 | + } |
| 57 | + read(); |
| 58 | + pos++; |
| 59 | + } |
| 60 | + |
| 61 | + // Peek on see, mis tagastab hetkel vaadeldavat tähte. |
| 62 | + // (Kõige algul võib see täht olla puudu.) |
| 63 | + private char peek() { |
| 64 | + if (current == null) read(); |
| 65 | + return current; |
| 66 | + } |
| 67 | + |
| 68 | + // See on pigem abimeetod, mis teostab tegeliku sisendist lugemist. |
| 69 | + // Meie enda lahenduses me seda otse ei kasuta. Kutsume ainult peek ja consume. |
| 70 | + private void read() { |
| 71 | + try { |
| 72 | + int i = reader.read(); |
| 73 | + current = (i == -1) ? TERMINATOR : (char) i; |
| 74 | + } catch (IOException e) { |
| 75 | + throw new UncheckedIOException(e); |
| 76 | + } |
| 77 | + } |
| 78 | + |
| 79 | + |
| 80 | + /** |
| 81 | + * See on nüüd see oluline meetod, mida peab ise implementeerima! |
| 82 | + */ |
| 83 | + public AktkToken readNextToken() { |
| 84 | + while (peek() != TERMINATOR) { |
| 85 | + if (Character.isDigit(peek())) { |
| 86 | + return readInteger(); |
| 87 | + } else if (Character.isWhitespace(peek())) { |
| 88 | + consume(); |
| 89 | + } else if (peek() == '"') { |
| 90 | + return readString(); |
| 91 | + } else if (Character.isLetter(peek()) || peek() == '_') { |
| 92 | + return readVariableOrKeyword(); |
| 93 | + } else if (peek() == '(') { |
| 94 | + consume(); |
| 95 | + return singleCharToken(LPAREN); |
| 96 | + } else if (peek() == ')') { |
| 97 | + consume(); |
| 98 | + return singleCharToken(RPAREN); |
| 99 | + } else if (peek() == '+') { |
| 100 | + consume(); |
| 101 | + return singleCharToken(PLUS); |
| 102 | + } else if (peek() == '-') { |
| 103 | + consume(); |
| 104 | + return singleCharToken(MINUS); |
| 105 | + } else if (peek() == '*') { |
| 106 | + consume(); |
| 107 | + return singleCharToken(TIMES); |
| 108 | + } else if (peek() == '/') { |
| 109 | + consume(); |
| 110 | + if (peek() == '/') { |
| 111 | + consume(); |
| 112 | + skipSingleLineComment(); |
| 113 | + } else if (peek() == '*') { |
| 114 | + consume(); |
| 115 | + skipMultiLineComment(); |
| 116 | + } else { |
| 117 | + return singleCharToken(DIV); |
| 118 | + } |
| 119 | + } else { |
| 120 | + throw new RuntimeException("Unexpected symbol"); |
| 121 | + } |
| 122 | + } |
| 123 | + return new AktkToken(EOF, pos, 0); |
| 124 | + } |
| 125 | + |
| 126 | + private AktkToken singleCharToken(AktkToken.Type type) { |
| 127 | + return new AktkToken(type, pos - 1, 1); |
| 128 | + } |
| 129 | + |
| 130 | + private void skipMultiLineComment() { |
| 131 | + do { |
| 132 | + while (peek() != '*') consume(); |
| 133 | + consume(); |
| 134 | + } while (peek() != '/'); |
| 135 | + consume(); |
| 136 | + } |
| 137 | + |
| 138 | + private void skipSingleLineComment() { |
| 139 | + while (peek() != '\n' && peek() != TERMINATOR) consume(); |
| 140 | + } |
| 141 | + |
| 142 | + private AktkToken readVariableOrKeyword() { |
| 143 | + assert Character.isLetter(peek()) || peek() == '_'; |
| 144 | + int initPos = this.pos; |
| 145 | + |
| 146 | + StringBuilder sb = new StringBuilder(); |
| 147 | + sb.append(peek()); |
| 148 | + consume(); |
| 149 | + |
| 150 | + while ((Character.isLetter(peek()) |
| 151 | + || Character.isDigit(peek()) |
| 152 | + || peek() == '_')) { |
| 153 | + sb.append(peek()); |
| 154 | + consume(); |
| 155 | + } |
| 156 | + |
| 157 | + String content = sb.toString(); |
| 158 | + return switch (content) { |
| 159 | + case "if" -> new AktkToken(IF, initPos, content.length()); |
| 160 | + case "while" -> new AktkToken(WHILE, initPos, content.length()); |
| 161 | + case "var" -> new AktkToken(VAR, initPos, content.length()); |
| 162 | + default -> new AktkToken(VARIABLE, content, initPos, content.length()); |
| 163 | + }; |
| 164 | + } |
| 165 | + |
| 166 | + private AktkToken readString() { |
| 167 | + assert peek() == '"'; |
| 168 | + int initPos = this.pos; |
| 169 | + |
| 170 | + StringBuilder sb = new StringBuilder(); |
| 171 | + |
| 172 | + consume(); |
| 173 | + while (peek() != '"') { |
| 174 | + if (peek() == '\\') { |
| 175 | + consume(); |
| 176 | + switch (peek()) { |
| 177 | + case 'n' -> sb.append('\n'); |
| 178 | + case 't' -> sb.append('\t'); |
| 179 | + case '"' -> sb.append('\"'); |
| 180 | + default -> throw new RuntimeException("Unknown character escape"); |
| 181 | + } |
| 182 | + } else { |
| 183 | + sb.append(peek()); |
| 184 | + } |
| 185 | + consume(); |
| 186 | + } |
| 187 | + consume(); |
| 188 | + |
| 189 | + return new AktkToken(STRING, sb.toString(), initPos, this.pos - initPos); |
| 190 | + } |
| 191 | + |
| 192 | + private AktkToken readInteger() { |
| 193 | + assert Character.isDigit(peek()); |
| 194 | + |
| 195 | + int initPos = this.pos; |
| 196 | + int result = 0; |
| 197 | + while (Character.isDigit(peek())) { |
| 198 | + result = 10 * result + Character.getNumericValue(peek()); |
| 199 | + consume(); |
| 200 | + } |
| 201 | + |
| 202 | + return new AktkToken(INTEGER, result, initPos, pos - initPos); |
| 203 | + } |
| 204 | +} |
0 commit comments