|
| 1 | +package lexer |
| 2 | + |
| 3 | +import ( |
| 4 | + "bufio" |
| 5 | + "bytes" |
| 6 | + "fmt" |
| 7 | + "io" |
| 8 | + "strconv" |
| 9 | + "strings" |
| 10 | +) |
| 11 | + |
| 12 | +type Options struct { |
| 13 | + NoPosition bool |
| 14 | +} |
| 15 | + |
| 16 | +func consumeCRLF(r *bufio.Reader, state *lexerState) error { |
| 17 | + b, err := r.ReadByte() |
| 18 | + if err != nil { |
| 19 | + return err |
| 20 | + } |
| 21 | + switch b { |
| 22 | + case '\r': |
| 23 | + b, err = r.ReadByte() |
| 24 | + if err != nil { |
| 25 | + return err |
| 26 | + } |
| 27 | + if b != '\n' { |
| 28 | + return fmt.Errorf("CR is not followed by LF") |
| 29 | + } |
| 30 | + fallthrough |
| 31 | + case '\n': |
| 32 | + state.Line++ |
| 33 | + state.Col = 0 |
| 34 | + return nil |
| 35 | + default: |
| 36 | + panic("consumeCRLF should not be called not on CR/LF") |
| 37 | + } |
| 38 | +} |
| 39 | + |
| 40 | +func Lex(r io.Reader, opts *Options) ([]Token, error) { |
| 41 | + if opts == nil { |
| 42 | + opts = &Options{} |
| 43 | + } |
| 44 | + toks, err := tokenStream(bufio.NewReader(r), opts) |
| 45 | + if err != nil { |
| 46 | + if err == io.EOF { |
| 47 | + return nil, io.ErrUnexpectedEOF |
| 48 | + } |
| 49 | + return nil, err |
| 50 | + } |
| 51 | + return toks, nil |
| 52 | +} |
| 53 | + |
| 54 | +type lexerState struct { |
| 55 | + Position |
| 56 | +} |
| 57 | + |
| 58 | +func tokenStream(r *bufio.Reader, opts *Options) ([]Token, error) { |
| 59 | + res := []Token{} |
| 60 | + state := &lexerState{} |
| 61 | + state.Line = 1 |
| 62 | + for { |
| 63 | + b, err := r.ReadByte() |
| 64 | + if err != nil { |
| 65 | + if err == io.EOF { |
| 66 | + break |
| 67 | + } |
| 68 | + return nil, err |
| 69 | + } |
| 70 | + if opts.NoPosition { |
| 71 | + state.Line = 0 |
| 72 | + state.Col = 0 |
| 73 | + } else { |
| 74 | + state.Col++ |
| 75 | + } |
| 76 | + switch b { |
| 77 | + case 0: |
| 78 | + return nil, fmt.Errorf("go-sieve/lexer: NUL is not allowed in input stream") |
| 79 | + case '[': |
| 80 | + res = append(res, ListStart{state.Position}) |
| 81 | + case ']': |
| 82 | + res = append(res, ListEnd{state.Position}) |
| 83 | + case '{': |
| 84 | + res = append(res, BlockStart{state.Position}) |
| 85 | + case '}': |
| 86 | + res = append(res, BlockEnd{state.Position}) |
| 87 | + case '(': |
| 88 | + res = append(res, TestListStart{state.Position}) |
| 89 | + case ')': |
| 90 | + res = append(res, TestListEnd{state.Position}) |
| 91 | + case ',': |
| 92 | + res = append(res, Comma{state.Position}) |
| 93 | + case ':': |
| 94 | + res = append(res, Colon{state.Position}) |
| 95 | + case ';': |
| 96 | + res = append(res, Semicolon{state.Position}) |
| 97 | + case ' ', '\t': |
| 98 | + continue |
| 99 | + case '\r', '\n': |
| 100 | + if err := r.UnreadByte(); err != nil { |
| 101 | + return nil, err |
| 102 | + } |
| 103 | + if err := consumeCRLF(r, state); err != nil { |
| 104 | + return nil, err |
| 105 | + } |
| 106 | + case '"': |
| 107 | + lineCol := state.Position |
| 108 | + str, err := quotedString(r, state) |
| 109 | + if err != nil { |
| 110 | + return nil, err |
| 111 | + } |
| 112 | + res = append(res, String{Position: lineCol, Text: str}) |
| 113 | + case '#': |
| 114 | + if err := hashComment(r, state); err != nil { |
| 115 | + return nil, err |
| 116 | + } |
| 117 | + case '/': |
| 118 | + b2, err := r.ReadByte() |
| 119 | + if err != nil { |
| 120 | + return nil, err |
| 121 | + } |
| 122 | + state.Col++ |
| 123 | + if b2 != '*' { |
| 124 | + return nil, fmt.Errorf("unexpected forward slash") |
| 125 | + } |
| 126 | + if err := multilineComment(r, state); err != nil { |
| 127 | + return nil, err |
| 128 | + } |
| 129 | + case 't': |
| 130 | + // "text:" |
| 131 | + lineCol := state.Position |
| 132 | + ext, err := r.Peek(4) |
| 133 | + if err != nil { |
| 134 | + return nil, err |
| 135 | + } |
| 136 | + if bytes.Equal(ext, []byte("ext:")) { |
| 137 | + if _, err := r.Discard(4); err != nil { |
| 138 | + return nil, err |
| 139 | + } |
| 140 | + state.Col += 4 |
| 141 | + // we consume whitespace and then build the multiline string |
| 142 | + wsLoop: |
| 143 | + for { |
| 144 | + b, err := r.ReadByte() |
| 145 | + if err != nil { |
| 146 | + return nil, err |
| 147 | + } |
| 148 | + state.Col++ |
| 149 | + switch b { |
| 150 | + case ' ', '\t': |
| 151 | + continue |
| 152 | + case '#': |
| 153 | + if err := hashComment(r, state); err != nil { |
| 154 | + return nil, err |
| 155 | + } |
| 156 | + break wsLoop |
| 157 | + case '\r', '\n': |
| 158 | + if err := r.UnreadByte(); err != nil { |
| 159 | + return nil, err |
| 160 | + } |
| 161 | + if err := consumeCRLF(r, state); err != nil { |
| 162 | + return nil, err |
| 163 | + } |
| 164 | + break wsLoop |
| 165 | + default: |
| 166 | + return nil, fmt.Errorf("unexpected character: %v", b) |
| 167 | + } |
| 168 | + } |
| 169 | + mlString, err := multilineString(r, state) |
| 170 | + if err != nil { |
| 171 | + return nil, err |
| 172 | + } |
| 173 | + res = append(res, String{Position: lineCol, Text: mlString}) |
| 174 | + } |
| 175 | + default: |
| 176 | + lineCol := state.Position |
| 177 | + if err := r.UnreadByte(); err != nil { |
| 178 | + return nil, err |
| 179 | + } |
| 180 | + state.Col-- |
| 181 | + |
| 182 | + if (b >= 'a' && b <= 'z') || (b >= 'A' && b <= 'Z') { |
| 183 | + str, err := identifier(r, state) |
| 184 | + if err != nil { |
| 185 | + return nil, err |
| 186 | + } |
| 187 | + res = append(res, Identifier{Position: lineCol, Text: str}) |
| 188 | + } else if b >= '0' && b <= '9' { |
| 189 | + num, err := number(r, state) |
| 190 | + if err != nil { |
| 191 | + return nil, err |
| 192 | + } |
| 193 | + num.Position = lineCol |
| 194 | + res = append(res, num) |
| 195 | + } else { |
| 196 | + return nil, fmt.Errorf("unexpected character: %v", b) |
| 197 | + } |
| 198 | + } |
| 199 | + } |
| 200 | + return res, nil |
| 201 | +} |
| 202 | + |
| 203 | +func identifier(r *bufio.Reader, state *lexerState) (string, error) { |
| 204 | + id := strings.Builder{} |
| 205 | + for { |
| 206 | + b, err := r.ReadByte() |
| 207 | + if err != nil { |
| 208 | + if err == io.EOF { |
| 209 | + break |
| 210 | + } |
| 211 | + return "", err |
| 212 | + } |
| 213 | + state.Col++ |
| 214 | + if (b >= 'a' && b <= 'z') || (b >= 'A' && b <= 'Z') || b == '_' { |
| 215 | + id.WriteByte(b) |
| 216 | + } else { |
| 217 | + if err := r.UnreadByte(); err != nil { |
| 218 | + return "", err |
| 219 | + } |
| 220 | + state.Col-- |
| 221 | + break |
| 222 | + } |
| 223 | + } |
| 224 | + return id.String(), nil |
| 225 | +} |
| 226 | + |
| 227 | +func number(r *bufio.Reader, state *lexerState) (Number, error) { |
| 228 | + num := strings.Builder{} |
| 229 | + q := None |
| 230 | +readLoop: |
| 231 | + for { |
| 232 | + b, err := r.ReadByte() |
| 233 | + if err != nil { |
| 234 | + if err == io.EOF { |
| 235 | + break |
| 236 | + } |
| 237 | + return Number{}, err |
| 238 | + } |
| 239 | + state.Col++ |
| 240 | + switch b { |
| 241 | + case 'K', 'G', 'M': |
| 242 | + q = Quantifier(b) |
| 243 | + break readLoop |
| 244 | + } |
| 245 | + if b >= '0' && b <= '9' { |
| 246 | + num.WriteByte(b) |
| 247 | + } else { |
| 248 | + if err := r.UnreadByte(); err != nil { |
| 249 | + return Number{}, err |
| 250 | + } |
| 251 | + state.Col-- |
| 252 | + break readLoop |
| 253 | + } |
| 254 | + } |
| 255 | + |
| 256 | + numParsed, err := strconv.Atoi(num.String()) |
| 257 | + if err != nil { |
| 258 | + return Number{}, err |
| 259 | + } |
| 260 | + return Number{Value: numParsed, Quantifier: q}, nil |
| 261 | +} |
| 262 | + |
| 263 | +func hashComment(r *bufio.Reader, state *lexerState) error { |
| 264 | + for { |
| 265 | + b, err := r.ReadByte() |
| 266 | + if err != nil { |
| 267 | + if err == io.EOF { |
| 268 | + break |
| 269 | + } |
| 270 | + return err |
| 271 | + } |
| 272 | + state.Col++ |
| 273 | + if b == '\r' || b == '\n' { |
| 274 | + if err := r.UnreadByte(); err != nil { |
| 275 | + return err |
| 276 | + } |
| 277 | + if err := consumeCRLF(r, state); err != nil { |
| 278 | + return err |
| 279 | + } |
| 280 | + break |
| 281 | + } |
| 282 | + } |
| 283 | + return nil |
| 284 | +} |
| 285 | + |
| 286 | +func multilineComment(r *bufio.Reader, state *lexerState) error { |
| 287 | + wasStar := false |
| 288 | + for { |
| 289 | + b, err := r.ReadByte() |
| 290 | + if err != nil { |
| 291 | + return err |
| 292 | + } |
| 293 | + state.Col++ |
| 294 | + if b == '\n' { |
| 295 | + state.Line++ |
| 296 | + state.Col = 0 |
| 297 | + } |
| 298 | + if wasStar && b == '/' { |
| 299 | + return nil |
| 300 | + } |
| 301 | + wasStar = b == '*' |
| 302 | + } |
| 303 | +} |
| 304 | + |
| 305 | +func quotedString(r *bufio.Reader, state *lexerState) (string, error) { |
| 306 | + str := strings.Builder{} |
| 307 | + atBackslash := false |
| 308 | + for { |
| 309 | + b, err := r.ReadByte() |
| 310 | + if err != nil { |
| 311 | + return "", err |
| 312 | + } |
| 313 | + state.Col++ |
| 314 | + switch b { |
| 315 | + case '\r', '\n': |
| 316 | + if err := r.UnreadByte(); err != nil { |
| 317 | + return "", err |
| 318 | + } |
| 319 | + if err := consumeCRLF(r, state); err != nil { |
| 320 | + return "", err |
| 321 | + } |
| 322 | + |
| 323 | + str.WriteByte('\r') |
| 324 | + str.WriteByte('\n') |
| 325 | + case '\\': |
| 326 | + if !atBackslash { |
| 327 | + atBackslash = true |
| 328 | + continue |
| 329 | + } |
| 330 | + str.WriteByte(b) |
| 331 | + case '"': |
| 332 | + if !atBackslash { |
| 333 | + return str.String(), nil |
| 334 | + } |
| 335 | + str.WriteByte(b) |
| 336 | + default: |
| 337 | + str.WriteByte(b) |
| 338 | + } |
| 339 | + atBackslash = false |
| 340 | + } |
| 341 | +} |
| 342 | + |
| 343 | +func multilineString(r *bufio.Reader, state *lexerState) (string, error) { |
| 344 | + atLF := false |
| 345 | + atLFHadDot := false |
| 346 | + var data strings.Builder |
| 347 | + for { |
| 348 | + b, err := r.ReadByte() |
| 349 | + if err != nil { |
| 350 | + return "", err |
| 351 | + } |
| 352 | + state.Col++ |
| 353 | + // We also normalize LF into CRLF while reading multiline strings. |
| 354 | + switch b { |
| 355 | + case '.': |
| 356 | + if atLF { |
| 357 | + atLFHadDot = true |
| 358 | + } else { |
| 359 | + data.WriteByte('.') |
| 360 | + } |
| 361 | + |
| 362 | + atLF = false |
| 363 | + case '\r', '\n': |
| 364 | + if err := r.UnreadByte(); err != nil { |
| 365 | + return "", err |
| 366 | + } |
| 367 | + if err := consumeCRLF(r, state); err != nil { |
| 368 | + return "", err |
| 369 | + } |
| 370 | + if atLFHadDot { |
| 371 | + return data.String(), nil |
| 372 | + } |
| 373 | + data.WriteByte('\r') |
| 374 | + data.WriteByte('\n') |
| 375 | + atLF = true |
| 376 | + default: |
| 377 | + atLF = false |
| 378 | + atLFHadDot = false |
| 379 | + } |
| 380 | + } |
| 381 | +} |
0 commit comments