| 1 | ! ===================================== |
| 2 | ! Lexer Module - Phase 1 of Grammar-Aware Parser |
| 3 | ! ===================================== |
| 4 | ! Tokenizes shell input into meaningful units |
| 5 | ! Part of the parser rewrite project |
| 6 | ! |
| 7 | ! Status: PHASE 1 - Full implementation |
| 8 | ! Author: Parser Rewrite Team |
| 9 | ! Created: 2025-11-05 |
| 10 | |
| 11 | module lexer |
| 12 | use iso_fortran_env |
| 13 | use shell_types |
| 14 | use shell_types, only: QUOTE_NONE, QUOTE_SINGLE, QUOTE_DOUBLE |
| 15 | implicit none |
| 16 | private |
| 17 | |
| 18 | ! Public interface |
| 19 | public :: tokenize |
| 20 | public :: next_token |
| 21 | public :: peek_token |
| 22 | public :: is_keyword |
| 23 | public :: is_operator |
| 24 | |
| 25 | ! Lexer state enumeration |
| 26 | integer, parameter :: LEX_NORMAL = 1 |
| 27 | integer, parameter :: LEX_IN_SINGLE_QUOTE = 2 |
| 28 | integer, parameter :: LEX_IN_DOUBLE_QUOTE = 3 |
| 29 | integer, parameter :: LEX_IN_WORD = 4 |
| 30 | integer, parameter :: LEX_IN_OPERATOR = 5 |
| 31 | integer, parameter :: LEX_IN_DOLLAR_SINGLE_QUOTE = 6 |
| 32 | |
| 33 | ! Context tracking for [[ ]] test expressions |
| 34 | ! Inside [[ ]], && || < > are test operators, not shell operators |
| 35 | logical :: in_double_bracket_context = .false. |
| 36 | |
| 37 | contains |
| 38 | |
| 39 | ! ===================================== |
| 40 | ! Character Classification Helpers |
| 41 | ! ===================================== |
| 42 | |
| 43 | pure function is_whitespace(ch) result(is_ws) |
| 44 | character(len=1), intent(in) :: ch |
| 45 | logical :: is_ws |
| 46 | is_ws = (ch == ' ' .or. ch == char(9) .or. ch == char(13)) ! space, tab, CR |
| 47 | end function is_whitespace |
| 48 | |
| 49 | pure function is_operator_start(ch) result(is_op) |
| 50 | character(len=1), intent(in) :: ch |
| 51 | logical :: is_op |
| 52 | is_op = (ch == '|' .or. ch == '&' .or. ch == ';' .or. & |
| 53 | ch == '<' .or. ch == '>' .or. ch == '(' .or. ch == ')') |
| 54 | end function is_operator_start |
| 55 | |
| 56 | pure function is_word_char(ch) result(is_wc) |
| 57 | character(len=1), intent(in) :: ch |
| 58 | logical :: is_wc |
| 59 | ! Word characters: anything that's not whitespace, operator, or special |
| 60 | is_wc = .not. (is_whitespace(ch) .or. is_operator_start(ch) .or. & |
| 61 | ch == char(10) .or. ch == '#' .or. ch == '"' .or. & |
| 62 | ch == "'" .or. ch == '\') |
| 63 | end function is_word_char |
| 64 | |
| 65 | ! ===================================== |
| 66 | ! Operator Recognition |
| 67 | ! ===================================== |
| 68 | |
| 69 | function is_operator(str) result(is_op) |
| 70 | character(len=*), intent(in) :: str |
| 71 | logical :: is_op |
| 72 | |
| 73 | select case(trim(str)) |
| 74 | ! Logical operators |
| 75 | case('&&', '||') |
| 76 | is_op = .true. |
| 77 | ! Pipe and background |
| 78 | case('|', '&') |
| 79 | is_op = .true. |
| 80 | ! Separators |
| 81 | case(';', ';;') |
| 82 | is_op = .true. |
| 83 | ! Redirections |
| 84 | case('<', '>', '>>', '<>', '>&', '<&', '>|', '<<', '<<-', '<<<') |
| 85 | is_op = .true. |
| 86 | ! Grouping |
| 87 | case('(', ')', '{', '}') |
| 88 | is_op = .true. |
| 89 | case default |
| 90 | is_op = .false. |
| 91 | end select |
| 92 | end function is_operator |
| 93 | |
| 94 | ! ===================================== |
| 95 | ! is_keyword - Check if word is a shell keyword |
| 96 | ! ===================================== |
| 97 | function is_keyword(word) result(is_kw) |
| 98 | character(len=*), intent(in) :: word |
| 99 | logical :: is_kw |
| 100 | |
| 101 | select case(trim(word)) |
| 102 | ! Control flow keywords |
| 103 | case('if', 'then', 'else', 'elif', 'fi') |
| 104 | is_kw = .true. |
| 105 | case('for', 'in', 'do', 'done') |
| 106 | is_kw = .true. |
| 107 | case('while', 'until') |
| 108 | is_kw = .true. |
| 109 | case('case', 'esac') |
| 110 | is_kw = .true. |
| 111 | ! Other keywords |
| 112 | case('function', 'select', 'time', 'coproc') |
| 113 | is_kw = .true. |
| 114 | case('{', '}') |
| 115 | is_kw = .true. |
| 116 | case('!') ! Negation operator (context-dependent) |
| 117 | is_kw = .true. |
| 118 | case default |
| 119 | is_kw = .false. |
| 120 | end select |
| 121 | end function is_keyword |
| 122 | |
| 123 | ! ===================================== |
| 124 | ! tokenize - Main entry point for lexical analysis |
| 125 | ! ===================================== |
| 126 | subroutine tokenize(input, tokens, num_tokens) |
| 127 | character(len=*), intent(in) :: input |
| 128 | type(token_t), intent(out) :: tokens(:) |
| 129 | integer, intent(out) :: num_tokens |
| 130 | |
| 131 | integer :: pos, input_len, state, token_start |
| 132 | character(len=1) :: ch, next_ch |
| 133 | character(len=MAX_TOKEN_LEN) :: current_token |
| 134 | integer :: token_len, paren_depth |
| 135 | logical :: in_escape, continuing_word, token_has_quoted_part |
| 136 | |
| 137 | num_tokens = 0 |
| 138 | pos = 1 |
| 139 | input_len = len_trim(input) |
| 140 | state = LEX_NORMAL |
| 141 | token_start = 1 |
| 142 | current_token = '' |
| 143 | token_len = 0 |
| 144 | in_escape = .false. |
| 145 | paren_depth = 0 |
| 146 | continuing_word = .false. |
| 147 | token_has_quoted_part = .false. |
| 148 | in_double_bracket_context = .false. |
| 149 | |
| 150 | do while (pos <= input_len .and. num_tokens < size(tokens)) |
| 151 | ch = input(pos:pos) |
| 152 | |
| 153 | ! Get next character for lookahead (if available) |
| 154 | if (pos < input_len) then |
| 155 | next_ch = input(pos+1:pos+1) |
| 156 | else |
| 157 | next_ch = ' ' |
| 158 | end if |
| 159 | |
| 160 | select case(state) |
| 161 | |
| 162 | ! ============ NORMAL STATE ============ |
| 163 | case(LEX_NORMAL) |
| 164 | |
| 165 | ! Skip whitespace |
| 166 | if (is_whitespace(ch)) then |
| 167 | pos = pos + 1 |
| 168 | cycle |
| 169 | end if |
| 170 | |
| 171 | ! Newline - significant token |
| 172 | if (ch == char(10)) then |
| 173 | call add_token(tokens, num_tokens, TOKEN_NEWLINE, char(10), pos, pos, .false.) |
| 174 | pos = pos + 1 |
| 175 | cycle |
| 176 | end if |
| 177 | |
| 178 | ! Comments: # to end of line |
| 179 | if (ch == '#') then |
| 180 | ! Skip until newline or end of input |
| 181 | do while (pos <= input_len .and. input(pos:pos) /= char(10)) |
| 182 | pos = pos + 1 |
| 183 | end do |
| 184 | cycle |
| 185 | end if |
| 186 | |
| 187 | ! Single quote: literal string |
| 188 | if (ch == "'") then |
| 189 | state = LEX_IN_SINGLE_QUOTE |
| 190 | ! Only reset token if we're NOT continuing a word |
| 191 | if (.not. continuing_word) then |
| 192 | token_start = pos |
| 193 | token_len = 0 |
| 194 | current_token = '' |
| 195 | end if |
| 196 | pos = pos + 1 |
| 197 | cycle |
| 198 | end if |
| 199 | |
| 200 | ! Double quote: expandable string |
| 201 | if (ch == '"') then |
| 202 | state = LEX_IN_DOUBLE_QUOTE |
| 203 | ! Only reset token if we're NOT continuing a word |
| 204 | if (.not. continuing_word) then |
| 205 | token_start = pos |
| 206 | token_len = 0 |
| 207 | current_token = '' |
| 208 | end if |
| 209 | pos = pos + 1 |
| 210 | cycle |
| 211 | end if |
| 212 | |
| 213 | ! Backslash escape |
| 214 | if (ch == '\') then |
| 215 | if (pos < input_len) then |
| 216 | ! Start a word token with the escaped character |
| 217 | state = LEX_IN_WORD |
| 218 | token_start = pos |
| 219 | in_escape = .true. ! Mark this token as escaped |
| 220 | ! For characters that would trigger expansion ($, `, etc), preserve backslash |
| 221 | ! so the expansion phase knows not to expand them |
| 222 | if (next_ch == '$' .or. next_ch == '`') then |
| 223 | token_len = 2 |
| 224 | current_token(1:2) = '\' // next_ch |
| 225 | else |
| 226 | token_len = 1 |
| 227 | current_token = next_ch |
| 228 | end if |
| 229 | pos = pos + 2 ! Skip backslash and next char |
| 230 | cycle |
| 231 | end if |
| 232 | end if |
| 233 | |
| 234 | ! Multi-character operators |
| 235 | ! Inside [[ ]], treat & | < > ( ) as word characters (test operators) |
| 236 | if (in_double_bracket_context .and. & |
| 237 | (ch == '&' .or. ch == '|' .or. ch == '<' .or. ch == '>' .or. & |
| 238 | ch == '(' .or. ch == ')')) then |
| 239 | state = LEX_IN_WORD |
| 240 | token_start = pos |
| 241 | token_len = 1 |
| 242 | current_token = ch |
| 243 | pos = pos + 1 |
| 244 | cycle |
| 245 | end if |
| 246 | if (is_operator_start(ch)) then |
| 247 | state = LEX_IN_OPERATOR |
| 248 | token_start = pos |
| 249 | token_len = 1 |
| 250 | current_token = ch |
| 251 | pos = pos + 1 |
| 252 | cycle |
| 253 | end if |
| 254 | |
| 255 | ! Check for $'...' ANSI-C quoting |
| 256 | if (ch == '$' .and. pos < input_len .and. next_ch == "'") then |
| 257 | state = LEX_IN_DOLLAR_SINGLE_QUOTE |
| 258 | token_start = pos |
| 259 | token_len = 0 |
| 260 | current_token = '' |
| 261 | continuing_word = .false. |
| 262 | token_has_quoted_part = .true. |
| 263 | pos = pos + 2 ! Skip $' |
| 264 | cycle |
| 265 | end if |
| 266 | |
| 267 | ! Check for $( or $(( - these should be kept in word tokens for expansion |
| 268 | if (ch == '$' .and. pos < input_len .and. next_ch == '(') then |
| 269 | ! This is command substitution or arithmetic - include in word |
| 270 | state = LEX_IN_WORD |
| 271 | token_start = pos |
| 272 | token_len = 2 |
| 273 | current_token = '$(' |
| 274 | paren_depth = 1 ! Track that we're inside $( |
| 275 | pos = pos + 2 |
| 276 | cycle |
| 277 | end if |
| 278 | |
| 279 | ! Check for ${ - parameter expansion should be kept in word tokens |
| 280 | if (ch == '$' .and. pos < input_len .and. next_ch == '{') then |
| 281 | ! This is parameter expansion - include in word |
| 282 | state = LEX_IN_WORD |
| 283 | token_start = pos |
| 284 | token_len = 2 |
| 285 | current_token = '${' |
| 286 | paren_depth = 1 ! Track that we're inside ${ |
| 287 | pos = pos + 2 |
| 288 | cycle |
| 289 | end if |
| 290 | |
| 291 | ! Check for $' - ANSI-C quoting |
| 292 | if (ch == '$' .and. pos < input_len .and. next_ch == "'") then |
| 293 | state = LEX_IN_WORD |
| 294 | token_start = pos |
| 295 | token_len = 0 |
| 296 | current_token = '' |
| 297 | token_has_quoted_part = .true. |
| 298 | pos = pos + 2 ! Skip $' |
| 299 | call process_ansi_c_quote(input, pos, input_len, current_token, token_len) |
| 300 | cycle |
| 301 | end if |
| 302 | |
| 303 | ! Assignment detection: VAR=value |
| 304 | ! (This is complex - we'll detect it as WORD and let parser handle it) |
| 305 | |
| 306 | ! Start of word |
| 307 | state = LEX_IN_WORD |
| 308 | token_start = pos |
| 309 | token_len = 1 |
| 310 | current_token = ch |
| 311 | pos = pos + 1 |
| 312 | |
| 313 | ! ============ SINGLE QUOTE STATE ============ |
| 314 | case(LEX_IN_SINGLE_QUOTE) |
| 315 | if (ch == "'") then |
| 316 | ! End of single-quoted string |
| 317 | ! Add sentinel char(3) to mark end of single-quoted literal |
| 318 | if (token_len < MAX_TOKEN_LEN) then |
| 319 | token_len = token_len + 1 |
| 320 | current_token(token_len:token_len) = char(3) |
| 321 | end if |
| 322 | pos = pos + 1 ! Move past closing quote |
| 323 | ! Check if next character continues the word (adjacent quote, word char, or escape) |
| 324 | if (pos <= input_len) then |
| 325 | next_ch = input(pos:pos) |
| 326 | if (next_ch == "'" .or. next_ch == '"') then |
| 327 | ! Adjacent quote follows - continue building this token |
| 328 | state = LEX_IN_WORD |
| 329 | continuing_word = .false. |
| 330 | cycle |
| 331 | else if (next_ch == '\') then |
| 332 | ! Backslash escape follows - continue building this token |
| 333 | state = LEX_IN_WORD |
| 334 | continuing_word = .false. |
| 335 | cycle |
| 336 | else if (is_word_char(next_ch)) then |
| 337 | ! Word character follows - continue building this token |
| 338 | state = LEX_IN_WORD |
| 339 | continuing_word = .false. |
| 340 | cycle |
| 341 | end if |
| 342 | end if |
| 343 | ! No adjacent quote or word char - finalize token |
| 344 | if (continuing_word) then |
| 345 | ! We're building a multi-part word - go back to LEX_IN_WORD |
| 346 | state = LEX_IN_WORD |
| 347 | continuing_word = .false. |
| 348 | else |
| 349 | ! Standalone quoted string - emit token |
| 350 | call add_token(tokens, num_tokens, TOKEN_WORD, current_token(1:token_len), & |
| 351 | token_start, pos-1, .true., quote_type=QUOTE_SINGLE) |
| 352 | state = LEX_NORMAL |
| 353 | end if |
| 354 | else |
| 355 | ! Add character to token (everything is literal) |
| 356 | if (token_len < MAX_TOKEN_LEN) then |
| 357 | token_len = token_len + 1 |
| 358 | current_token(token_len:token_len) = ch |
| 359 | end if |
| 360 | pos = pos + 1 |
| 361 | end if |
| 362 | |
| 363 | ! ============ DOLLAR SINGLE QUOTE STATE ($'...') ============ |
| 364 | case(LEX_IN_DOLLAR_SINGLE_QUOTE) |
| 365 | if (ch == "'") then |
| 366 | ! End of $'...' string — add sentinels to mark as quoted |
| 367 | if (token_len < MAX_TOKEN_LEN) then |
| 368 | token_len = token_len + 1 |
| 369 | current_token(token_len:token_len) = char(3) ! end sentinel |
| 370 | end if |
| 371 | pos = pos + 1 |
| 372 | ! Check if next character continues the word |
| 373 | if (pos <= input_len) then |
| 374 | next_ch = input(pos:pos) |
| 375 | if (next_ch == "'" .or. next_ch == '"' .or. next_ch == '\' .or. & |
| 376 | is_word_char(next_ch)) then |
| 377 | state = LEX_IN_WORD |
| 378 | continuing_word = .false. |
| 379 | cycle |
| 380 | end if |
| 381 | end if |
| 382 | if (continuing_word) then |
| 383 | state = LEX_IN_WORD |
| 384 | continuing_word = .false. |
| 385 | else |
| 386 | call add_token(tokens, num_tokens, TOKEN_WORD, current_token(1:token_len), & |
| 387 | token_start, pos-1, .true., quote_type=QUOTE_SINGLE) |
| 388 | state = LEX_NORMAL |
| 389 | end if |
| 390 | else if (ch == '\' .and. pos < input_len) then |
| 391 | ! Escape sequences in $'...' |
| 392 | next_ch = input(pos+1:pos+1) |
| 393 | if (token_len < MAX_TOKEN_LEN) then |
| 394 | select case(next_ch) |
| 395 | case('a') |
| 396 | token_len = token_len + 1 |
| 397 | current_token(token_len:token_len) = char(7) ! bell |
| 398 | case('b') |
| 399 | token_len = token_len + 1 |
| 400 | current_token(token_len:token_len) = char(8) ! backspace |
| 401 | case('e', 'E') |
| 402 | token_len = token_len + 1 |
| 403 | current_token(token_len:token_len) = char(27) ! escape |
| 404 | case('f') |
| 405 | token_len = token_len + 1 |
| 406 | current_token(token_len:token_len) = char(12) ! form feed |
| 407 | case('n') |
| 408 | token_len = token_len + 1 |
| 409 | current_token(token_len:token_len) = char(10) ! newline |
| 410 | case('r') |
| 411 | token_len = token_len + 1 |
| 412 | current_token(token_len:token_len) = char(13) ! carriage return |
| 413 | case('t') |
| 414 | token_len = token_len + 1 |
| 415 | current_token(token_len:token_len) = char(9) ! tab |
| 416 | case('v') |
| 417 | token_len = token_len + 1 |
| 418 | current_token(token_len:token_len) = char(11) ! vertical tab |
| 419 | case('\') |
| 420 | token_len = token_len + 1 |
| 421 | current_token(token_len:token_len) = '\' |
| 422 | case("'") |
| 423 | token_len = token_len + 1 |
| 424 | current_token(token_len:token_len) = "'" |
| 425 | case('"') |
| 426 | token_len = token_len + 1 |
| 427 | current_token(token_len:token_len) = '"' |
| 428 | case('x') |
| 429 | ! Hex escape: \xHH (up to 2 hex digits) |
| 430 | block |
| 431 | integer :: hval, hdigits |
| 432 | character :: hch |
| 433 | hval = 0; hdigits = 0 |
| 434 | pos = pos + 2 ! skip \x |
| 435 | do while (pos <= input_len .and. hdigits < 2) |
| 436 | hch = input(pos:pos) |
| 437 | if (hch >= '0' .and. hch <= '9') then |
| 438 | hval = hval * 16 + (ichar(hch) - ichar('0')) |
| 439 | else if (hch >= 'a' .and. hch <= 'f') then |
| 440 | hval = hval * 16 + (ichar(hch) - ichar('a') + 10) |
| 441 | else if (hch >= 'A' .and. hch <= 'F') then |
| 442 | hval = hval * 16 + (ichar(hch) - ichar('A') + 10) |
| 443 | else |
| 444 | exit |
| 445 | end if |
| 446 | pos = pos + 1 |
| 447 | hdigits = hdigits + 1 |
| 448 | end do |
| 449 | if (hdigits > 0 .and. hval <= 255) then |
| 450 | token_len = token_len + 1 |
| 451 | current_token(token_len:token_len) = char(hval) |
| 452 | end if |
| 453 | cycle ! pos already advanced past hex digits |
| 454 | end block |
| 455 | case('0', '1', '2', '3', '4', '5', '6', '7') |
| 456 | ! Octal escape: \nnn (up to 3 octal digits) |
| 457 | block |
| 458 | integer :: oval, odigits |
| 459 | character :: och |
| 460 | oval = 0; odigits = 0 |
| 461 | pos = pos + 1 ! skip backslash only |
| 462 | do while (pos <= input_len .and. odigits < 3) |
| 463 | och = input(pos:pos) |
| 464 | if (och >= '0' .and. och <= '7') then |
| 465 | oval = oval * 8 + (ichar(och) - ichar('0')) |
| 466 | else |
| 467 | exit |
| 468 | end if |
| 469 | pos = pos + 1 |
| 470 | odigits = odigits + 1 |
| 471 | end do |
| 472 | if (odigits > 0 .and. oval <= 255) then |
| 473 | token_len = token_len + 1 |
| 474 | current_token(token_len:token_len) = char(oval) |
| 475 | end if |
| 476 | cycle ! pos already advanced past octal digits |
| 477 | end block |
| 478 | case default |
| 479 | ! Unknown escape — keep both chars |
| 480 | token_len = token_len + 1 |
| 481 | current_token(token_len:token_len) = ch |
| 482 | token_len = token_len + 1 |
| 483 | current_token(token_len:token_len) = next_ch |
| 484 | end select |
| 485 | end if |
| 486 | pos = pos + 2 |
| 487 | else |
| 488 | ! Regular character — add literally |
| 489 | if (token_len < MAX_TOKEN_LEN) then |
| 490 | token_len = token_len + 1 |
| 491 | current_token(token_len:token_len) = ch |
| 492 | end if |
| 493 | pos = pos + 1 |
| 494 | end if |
| 495 | |
| 496 | ! ============ DOUBLE QUOTE STATE ============ |
| 497 | case(LEX_IN_DOUBLE_QUOTE) |
| 498 | if (ch == '\' .and. pos < input_len) then |
| 499 | ! Backslash escape in double quotes (only for $, `, ", \, newline) |
| 500 | if (next_ch == '$' .or. next_ch == '`') then |
| 501 | ! For \$ and \` - keep BOTH chars so expansion can see the escape |
| 502 | if (token_len < MAX_TOKEN_LEN - 1) then |
| 503 | token_len = token_len + 1 |
| 504 | current_token(token_len:token_len) = ch |
| 505 | token_len = token_len + 1 |
| 506 | current_token(token_len:token_len) = next_ch |
| 507 | end if |
| 508 | pos = pos + 2 |
| 509 | else if (next_ch == '"' .or. next_ch == '\' .or. next_ch == char(10)) then |
| 510 | ! For \" and \\ and \newline - add only escaped character |
| 511 | if (token_len < MAX_TOKEN_LEN) then |
| 512 | token_len = token_len + 1 |
| 513 | current_token(token_len:token_len) = next_ch |
| 514 | end if |
| 515 | pos = pos + 2 |
| 516 | else |
| 517 | ! Backslash is literal |
| 518 | if (token_len < MAX_TOKEN_LEN) then |
| 519 | token_len = token_len + 1 |
| 520 | current_token(token_len:token_len) = ch |
| 521 | end if |
| 522 | pos = pos + 1 |
| 523 | end if |
| 524 | else if (ch == '$' .and. pos < input_len .and. next_ch == '(') then |
| 525 | ! Command substitution inside double quotes - need to find matching ) |
| 526 | ! while ignoring quotes inside $() |
| 527 | if (token_len < MAX_TOKEN_LEN - 1) then |
| 528 | token_len = token_len + 1 |
| 529 | current_token(token_len:token_len) = '$' |
| 530 | token_len = token_len + 1 |
| 531 | current_token(token_len:token_len) = '(' |
| 532 | end if |
| 533 | pos = pos + 2 |
| 534 | paren_depth = 1 |
| 535 | ! Scan to find matching ), respecting nested parens and quotes |
| 536 | do while (pos <= input_len .and. paren_depth > 0) |
| 537 | ch = input(pos:pos) |
| 538 | if (ch == '"') then |
| 539 | ! Skip double-quoted string inside command substitution |
| 540 | if (token_len < MAX_TOKEN_LEN) then |
| 541 | token_len = token_len + 1 |
| 542 | current_token(token_len:token_len) = ch |
| 543 | end if |
| 544 | pos = pos + 1 |
| 545 | do while (pos <= input_len) |
| 546 | ch = input(pos:pos) |
| 547 | if (ch == '\' .and. pos < input_len) then |
| 548 | ! Skip escaped char |
| 549 | if (token_len < MAX_TOKEN_LEN - 1) then |
| 550 | token_len = token_len + 1 |
| 551 | current_token(token_len:token_len) = ch |
| 552 | token_len = token_len + 1 |
| 553 | current_token(token_len:token_len) = input(pos+1:pos+1) |
| 554 | end if |
| 555 | pos = pos + 2 |
| 556 | else if (ch == '"') then |
| 557 | if (token_len < MAX_TOKEN_LEN) then |
| 558 | token_len = token_len + 1 |
| 559 | current_token(token_len:token_len) = ch |
| 560 | end if |
| 561 | pos = pos + 1 |
| 562 | exit |
| 563 | else |
| 564 | if (token_len < MAX_TOKEN_LEN) then |
| 565 | token_len = token_len + 1 |
| 566 | current_token(token_len:token_len) = ch |
| 567 | end if |
| 568 | pos = pos + 1 |
| 569 | end if |
| 570 | end do |
| 571 | else if (ch == "'") then |
| 572 | ! Skip single-quoted string |
| 573 | if (token_len < MAX_TOKEN_LEN) then |
| 574 | token_len = token_len + 1 |
| 575 | current_token(token_len:token_len) = ch |
| 576 | end if |
| 577 | pos = pos + 1 |
| 578 | do while (pos <= input_len .and. input(pos:pos) /= "'") |
| 579 | if (token_len < MAX_TOKEN_LEN) then |
| 580 | token_len = token_len + 1 |
| 581 | current_token(token_len:token_len) = input(pos:pos) |
| 582 | end if |
| 583 | pos = pos + 1 |
| 584 | end do |
| 585 | if (pos <= input_len) then |
| 586 | if (token_len < MAX_TOKEN_LEN) then |
| 587 | token_len = token_len + 1 |
| 588 | current_token(token_len:token_len) = "'" |
| 589 | end if |
| 590 | pos = pos + 1 |
| 591 | end if |
| 592 | else if (ch == '(') then |
| 593 | paren_depth = paren_depth + 1 |
| 594 | if (token_len < MAX_TOKEN_LEN) then |
| 595 | token_len = token_len + 1 |
| 596 | current_token(token_len:token_len) = ch |
| 597 | end if |
| 598 | pos = pos + 1 |
| 599 | else if (ch == ')') then |
| 600 | paren_depth = paren_depth - 1 |
| 601 | if (token_len < MAX_TOKEN_LEN) then |
| 602 | token_len = token_len + 1 |
| 603 | current_token(token_len:token_len) = ch |
| 604 | end if |
| 605 | pos = pos + 1 |
| 606 | else |
| 607 | if (token_len < MAX_TOKEN_LEN) then |
| 608 | token_len = token_len + 1 |
| 609 | current_token(token_len:token_len) = ch |
| 610 | end if |
| 611 | pos = pos + 1 |
| 612 | end if |
| 613 | end do |
| 614 | else if (ch == '"') then |
| 615 | ! End of double-quoted string |
| 616 | pos = pos + 1 ! Move past closing quote |
| 617 | ! Check if next character continues the word (adjacent quote, word char, or escape) |
| 618 | if (pos <= input_len) then |
| 619 | next_ch = input(pos:pos) |
| 620 | if (next_ch == "'" .or. next_ch == '"') then |
| 621 | ! Adjacent quote follows - continue building this token |
| 622 | ! Add sentinel to mark quote boundary (so expansion knows where quoted part ends) |
| 623 | if (token_len < MAX_TOKEN_LEN) then |
| 624 | token_len = token_len + 1 |
| 625 | current_token(token_len:token_len) = char(1) ! ASCII SOH as sentinel |
| 626 | end if |
| 627 | state = LEX_IN_WORD |
| 628 | continuing_word = .false. |
| 629 | cycle |
| 630 | else if (next_ch == '\') then |
| 631 | ! Backslash escape follows - continue building this token |
| 632 | ! Add sentinel to mark quote boundary |
| 633 | if (token_len < MAX_TOKEN_LEN) then |
| 634 | token_len = token_len + 1 |
| 635 | current_token(token_len:token_len) = char(1) ! ASCII SOH as sentinel |
| 636 | end if |
| 637 | state = LEX_IN_WORD |
| 638 | continuing_word = .false. |
| 639 | cycle |
| 640 | else if (is_word_char(next_ch)) then |
| 641 | ! Word character follows - continue building this token |
| 642 | ! Add sentinel to mark quote boundary (so expansion knows where quoted part ends) |
| 643 | if (token_len < MAX_TOKEN_LEN) then |
| 644 | token_len = token_len + 1 |
| 645 | current_token(token_len:token_len) = char(1) ! ASCII SOH as sentinel |
| 646 | end if |
| 647 | state = LEX_IN_WORD |
| 648 | continuing_word = .false. |
| 649 | cycle |
| 650 | end if |
| 651 | end if |
| 652 | ! No adjacent quote or word char - finalize token |
| 653 | if (continuing_word) then |
| 654 | ! We're building a multi-part word - go back to LEX_IN_WORD |
| 655 | state = LEX_IN_WORD |
| 656 | continuing_word = .false. |
| 657 | else |
| 658 | ! Standalone quoted string - emit token |
| 659 | call add_token(tokens, num_tokens, TOKEN_WORD, current_token(1:token_len), & |
| 660 | token_start, pos-1, .true., quote_type=QUOTE_DOUBLE) |
| 661 | state = LEX_NORMAL |
| 662 | end if |
| 663 | else |
| 664 | ! Add character to token |
| 665 | if (token_len < MAX_TOKEN_LEN) then |
| 666 | token_len = token_len + 1 |
| 667 | current_token(token_len:token_len) = ch |
| 668 | end if |
| 669 | pos = pos + 1 |
| 670 | end if |
| 671 | |
| 672 | ! ============ WORD STATE ============ |
| 673 | case(LEX_IN_WORD) |
| 674 | ! Check if we're inside $() - if so, keep EVERYTHING including spaces |
| 675 | ! IMPORTANT: Also check paren_depth > 0 to ensure we're actually inside the $() |
| 676 | if (index(current_token(1:token_len), '$(') > 0 .and. paren_depth > 0) then |
| 677 | ! Inside command substitution - track paren depth |
| 678 | if (ch == '(') then |
| 679 | paren_depth = paren_depth + 1 |
| 680 | if (token_len < MAX_TOKEN_LEN) then |
| 681 | token_len = token_len + 1 |
| 682 | current_token(token_len:token_len) = ch |
| 683 | end if |
| 684 | pos = pos + 1 |
| 685 | else if (ch == ')') then |
| 686 | paren_depth = paren_depth - 1 |
| 687 | if (token_len < MAX_TOKEN_LEN) then |
| 688 | token_len = token_len + 1 |
| 689 | current_token(token_len:token_len) = ch |
| 690 | end if |
| 691 | pos = pos + 1 |
| 692 | ! If paren_depth hits 0, we closed the $(...) |
| 693 | if (paren_depth == 0) then |
| 694 | ! End of command substitution - finish token |
| 695 | call add_word_or_keyword(tokens, num_tokens, current_token(1:token_len), & |
| 696 | token_start, pos-1, token_has_quoted_part, in_escape) |
| 697 | state = LEX_NORMAL |
| 698 | in_escape = .false. |
| 699 | token_has_quoted_part = .false. |
| 700 | end if |
| 701 | else |
| 702 | ! Inside $() - keep EVERYTHING including spaces |
| 703 | if (token_len < MAX_TOKEN_LEN) then |
| 704 | token_len = token_len + 1 |
| 705 | current_token(token_len:token_len) = ch |
| 706 | end if |
| 707 | pos = pos + 1 |
| 708 | end if |
| 709 | ! Check if we're inside ${ - if so, keep EVERYTHING until closing } |
| 710 | ! IMPORTANT: Also check paren_depth > 0 to ensure we're actually inside the ${} |
| 711 | else if (index(current_token(1:token_len), '${') > 0 .and. paren_depth > 0) then |
| 712 | ! Inside parameter expansion - track brace depth |
| 713 | if (ch == '{') then |
| 714 | paren_depth = paren_depth + 1 |
| 715 | if (token_len < MAX_TOKEN_LEN) then |
| 716 | token_len = token_len + 1 |
| 717 | current_token(token_len:token_len) = ch |
| 718 | end if |
| 719 | pos = pos + 1 |
| 720 | else if (ch == '}') then |
| 721 | paren_depth = paren_depth - 1 |
| 722 | if (token_len < MAX_TOKEN_LEN) then |
| 723 | token_len = token_len + 1 |
| 724 | current_token(token_len:token_len) = ch |
| 725 | end if |
| 726 | pos = pos + 1 |
| 727 | ! If paren_depth hits 0, we closed the ${...} |
| 728 | if (paren_depth == 0) then |
| 729 | ! Check if next character continues the word (e.g., ${A}${B}) |
| 730 | ! Don't end token if next char is $ or other word character |
| 731 | if (pos <= input_len) then |
| 732 | next_ch = input(pos:pos) |
| 733 | ! If next character starts a new expansion or is alphanumeric, continue token |
| 734 | if (next_ch == '$' .or. next_ch == '{' .or. & |
| 735 | (next_ch >= 'a' .and. next_ch <= 'z') .or. & |
| 736 | (next_ch >= 'A' .and. next_ch <= 'Z') .or. & |
| 737 | (next_ch >= '0' .and. next_ch <= '9') .or. & |
| 738 | next_ch == '_' .or. next_ch == '-' .or. next_ch == '.') then |
| 739 | ! Continue building the same token - don't end it yet |
| 740 | ! state stays LEX_WORD |
| 741 | else |
| 742 | ! End of parameter expansion - finish token |
| 743 | call add_word_or_keyword(tokens, num_tokens, current_token(1:token_len), & |
| 744 | token_start, pos-1, token_has_quoted_part, in_escape) |
| 745 | state = LEX_NORMAL |
| 746 | in_escape = .false. |
| 747 | token_has_quoted_part = .false. |
| 748 | end if |
| 749 | else |
| 750 | ! End of input - finish token |
| 751 | call add_word_or_keyword(tokens, num_tokens, current_token(1:token_len), & |
| 752 | token_start, pos-1, token_has_quoted_part, in_escape) |
| 753 | state = LEX_NORMAL |
| 754 | in_escape = .false. |
| 755 | token_has_quoted_part = .false. |
| 756 | end if |
| 757 | end if |
| 758 | else |
| 759 | ! Inside ${ - keep EVERYTHING including spaces |
| 760 | if (token_len < MAX_TOKEN_LEN) then |
| 761 | token_len = token_len + 1 |
| 762 | current_token(token_len:token_len) = ch |
| 763 | end if |
| 764 | pos = pos + 1 |
| 765 | end if |
| 766 | else if (ch == '\' .and. pos < input_len) then |
| 767 | ! Backslash escape in word |
| 768 | ! For expansion-triggering chars, preserve backslash |
| 769 | if (next_ch == '$' .or. next_ch == '`') then |
| 770 | if (token_len < MAX_TOKEN_LEN - 1) then |
| 771 | token_len = token_len + 1 |
| 772 | current_token(token_len:token_len) = '\' |
| 773 | token_len = token_len + 1 |
| 774 | current_token(token_len:token_len) = next_ch |
| 775 | end if |
| 776 | else |
| 777 | if (token_len < MAX_TOKEN_LEN) then |
| 778 | token_len = token_len + 1 |
| 779 | current_token(token_len:token_len) = next_ch |
| 780 | end if |
| 781 | end if |
| 782 | pos = pos + 2 |
| 783 | else if (ch == "'" .or. ch == '"') then |
| 784 | ! Check for $' (ANSI-C quoting) when last char in token is $ |
| 785 | if (ch == "'" .and. token_len >= 1 .and. & |
| 786 | current_token(token_len:token_len) == '$') then |
| 787 | ! Remove trailing $ and process ANSI-C quoted string |
| 788 | token_len = token_len - 1 |
| 789 | token_has_quoted_part = .true. |
| 790 | pos = pos + 1 ! Skip opening ' |
| 791 | call process_ansi_c_quote(input, pos, input_len, current_token, token_len) |
| 792 | cycle |
| 793 | end if |
| 794 | ! Quote in middle of word - continue building the same token |
| 795 | ! Mark that we're continuing a word so quote handler doesn't reset the token |
| 796 | continuing_word = .true. |
| 797 | token_has_quoted_part = .true. ! Track that this word contains quoted content |
| 798 | ! Transition to appropriate quote state |
| 799 | if (ch == "'") then |
| 800 | ! Check for $' (ANSI-C quoting) — the $ is already in the token |
| 801 | if (token_len >= 1 .and. current_token(token_len:token_len) == '$') then |
| 802 | token_len = token_len - 1 ! Remove the $ from token |
| 803 | state = LEX_IN_DOLLAR_SINGLE_QUOTE |
| 804 | pos = pos + 1 ! Skip the opening quote |
| 805 | cycle |
| 806 | end if |
| 807 | ! Add sentinel char(2) to mark start of single-quoted literal (no expansion) |
| 808 | if (token_len < MAX_TOKEN_LEN) then |
| 809 | token_len = token_len + 1 |
| 810 | current_token(token_len:token_len) = char(2) |
| 811 | end if |
| 812 | state = LEX_IN_SINGLE_QUOTE |
| 813 | else |
| 814 | state = LEX_IN_DOUBLE_QUOTE |
| 815 | end if |
| 816 | pos = pos + 1 ! Skip the opening quote |
| 817 | else if (ch == '#') then |
| 818 | ! # is normally comment, but in $# it's part of variable |
| 819 | ! Keep it if current token is just $ |
| 820 | if (token_len == 1 .and. current_token(1:1) == '$') then |
| 821 | token_len = token_len + 1 |
| 822 | current_token(token_len:token_len) = ch |
| 823 | pos = pos + 1 |
| 824 | else |
| 825 | ! End word, let # start a comment |
| 826 | call add_word_or_keyword(tokens, num_tokens, current_token(1:token_len), & |
| 827 | token_start, pos-1, token_has_quoted_part, in_escape) |
| 828 | state = LEX_NORMAL |
| 829 | in_escape = .false. |
| 830 | token_has_quoted_part = .false. |
| 831 | end if |
| 832 | else if (ch == '$' .and. pos < input_len .and. next_ch == '(') then |
| 833 | ! $( for command/arithmetic substitution - keep in word |
| 834 | if (token_len < MAX_TOKEN_LEN - 1) then |
| 835 | token_len = token_len + 1 |
| 836 | current_token(token_len:token_len) = ch |
| 837 | token_len = token_len + 1 |
| 838 | current_token(token_len:token_len) = next_ch |
| 839 | paren_depth = 1 ! Track that we're inside $( |
| 840 | end if |
| 841 | pos = pos + 2 |
| 842 | else if (ch == '$' .and. pos < input_len .and. next_ch == '{') then |
| 843 | ! ${ for parameter expansion - keep in word |
| 844 | if (token_len < MAX_TOKEN_LEN - 1) then |
| 845 | token_len = token_len + 1 |
| 846 | current_token(token_len:token_len) = ch |
| 847 | token_len = token_len + 1 |
| 848 | current_token(token_len:token_len) = next_ch |
| 849 | paren_depth = 1 ! Track that we're inside ${ |
| 850 | end if |
| 851 | pos = pos + 2 |
| 852 | else if ((ch >= '0' .and. ch <= '9') .or. ch == '+' .or. ch == '-' .or. & |
| 853 | ch == '*' .or. ch == '/' .or. ch == '%') then |
| 854 | ! Keep these chars in word (for variables and arithmetic) |
| 855 | if (token_len < MAX_TOKEN_LEN) then |
| 856 | token_len = token_len + 1 |
| 857 | current_token(token_len:token_len) = ch |
| 858 | end if |
| 859 | pos = pos + 1 |
| 860 | else if (ch == '(' .or. ch == ')') then |
| 861 | ! Inside [[ ]], keep parens as part of word (regex patterns, grouping) |
| 862 | if (in_double_bracket_context) then |
| 863 | if (token_len < MAX_TOKEN_LEN) then |
| 864 | token_len = token_len + 1 |
| 865 | current_token(token_len:token_len) = ch |
| 866 | end if |
| 867 | pos = pos + 1 |
| 868 | ! Parentheses: Keep ONLY if inside $(( or $( |
| 869 | ! Check if current token ends with $ (for x=$(cmd) or just $(cmd)) |
| 870 | ! NOTE: Only for '(' - ')' after $ (like $$) should end the word |
| 871 | else if (ch == '(' .and. token_len >= 1 .and. current_token(token_len:token_len) == '$') then |
| 872 | ! Just added $, now seeing ( - this is $( substitution - keep both |
| 873 | if (token_len < MAX_TOKEN_LEN) then |
| 874 | token_len = token_len + 1 |
| 875 | current_token(token_len:token_len) = ch |
| 876 | end if |
| 877 | pos = pos + 1 |
| 878 | else if (token_len >= 2 .and. index(current_token(1:token_len), '$(') > 0) then |
| 879 | ! Already inside $(...) - keep parens |
| 880 | if (token_len < MAX_TOKEN_LEN) then |
| 881 | token_len = token_len + 1 |
| 882 | current_token(token_len:token_len) = ch |
| 883 | end if |
| 884 | pos = pos + 1 |
| 885 | else if (ch == '(' .and. token_len >= 1 .and. & |
| 886 | current_token(token_len:token_len) == '=') then |
| 887 | ! Array assignment: VAR=(...) - include the parenthesized content |
| 888 | ! Scan for matching ) respecting quotes and nested parens |
| 889 | if (token_len < MAX_TOKEN_LEN) then |
| 890 | token_len = token_len + 1 |
| 891 | current_token(token_len:token_len) = '(' |
| 892 | end if |
| 893 | pos = pos + 1 |
| 894 | paren_depth = 1 |
| 895 | do while (pos <= input_len .and. paren_depth > 0) |
| 896 | ch = input(pos:pos) |
| 897 | if (ch == '"') then |
| 898 | ! Skip double-quoted string |
| 899 | if (token_len < MAX_TOKEN_LEN) then |
| 900 | token_len = token_len + 1 |
| 901 | current_token(token_len:token_len) = ch |
| 902 | end if |
| 903 | pos = pos + 1 |
| 904 | do while (pos <= input_len .and. input(pos:pos) /= '"') |
| 905 | if (input(pos:pos) == '\' .and. pos < input_len) then |
| 906 | if (token_len < MAX_TOKEN_LEN - 1) then |
| 907 | token_len = token_len + 1 |
| 908 | current_token(token_len:token_len) = input(pos:pos) |
| 909 | token_len = token_len + 1 |
| 910 | current_token(token_len:token_len) = input(pos+1:pos+1) |
| 911 | end if |
| 912 | pos = pos + 2 |
| 913 | else |
| 914 | if (token_len < MAX_TOKEN_LEN) then |
| 915 | token_len = token_len + 1 |
| 916 | current_token(token_len:token_len) = input(pos:pos) |
| 917 | end if |
| 918 | pos = pos + 1 |
| 919 | end if |
| 920 | end do |
| 921 | if (pos <= input_len) then |
| 922 | if (token_len < MAX_TOKEN_LEN) then |
| 923 | token_len = token_len + 1 |
| 924 | current_token(token_len:token_len) = '"' |
| 925 | end if |
| 926 | pos = pos + 1 |
| 927 | end if |
| 928 | else if (ch == "'") then |
| 929 | ! Skip single-quoted string |
| 930 | if (token_len < MAX_TOKEN_LEN) then |
| 931 | token_len = token_len + 1 |
| 932 | current_token(token_len:token_len) = ch |
| 933 | end if |
| 934 | pos = pos + 1 |
| 935 | do while (pos <= input_len .and. input(pos:pos) /= "'") |
| 936 | if (token_len < MAX_TOKEN_LEN) then |
| 937 | token_len = token_len + 1 |
| 938 | current_token(token_len:token_len) = input(pos:pos) |
| 939 | end if |
| 940 | pos = pos + 1 |
| 941 | end do |
| 942 | if (pos <= input_len) then |
| 943 | if (token_len < MAX_TOKEN_LEN) then |
| 944 | token_len = token_len + 1 |
| 945 | current_token(token_len:token_len) = "'" |
| 946 | end if |
| 947 | pos = pos + 1 |
| 948 | end if |
| 949 | else if (ch == '(') then |
| 950 | paren_depth = paren_depth + 1 |
| 951 | if (token_len < MAX_TOKEN_LEN) then |
| 952 | token_len = token_len + 1 |
| 953 | current_token(token_len:token_len) = ch |
| 954 | end if |
| 955 | pos = pos + 1 |
| 956 | else if (ch == ')') then |
| 957 | paren_depth = paren_depth - 1 |
| 958 | if (token_len < MAX_TOKEN_LEN) then |
| 959 | token_len = token_len + 1 |
| 960 | current_token(token_len:token_len) = ch |
| 961 | end if |
| 962 | pos = pos + 1 |
| 963 | else |
| 964 | if (token_len < MAX_TOKEN_LEN) then |
| 965 | token_len = token_len + 1 |
| 966 | current_token(token_len:token_len) = ch |
| 967 | end if |
| 968 | pos = pos + 1 |
| 969 | end if |
| 970 | end do |
| 971 | ! Token complete with closing ) |
| 972 | call add_word_or_keyword(tokens, num_tokens, current_token(1:token_len), & |
| 973 | token_start, pos-1, token_has_quoted_part, in_escape) |
| 974 | state = LEX_NORMAL |
| 975 | in_escape = .false. |
| 976 | token_has_quoted_part = .false. |
| 977 | else |
| 978 | ! Not in substitution - end word, let paren be operator |
| 979 | call add_word_or_keyword(tokens, num_tokens, current_token(1:token_len), & |
| 980 | token_start, pos-1, token_has_quoted_part, in_escape) |
| 981 | state = LEX_NORMAL |
| 982 | in_escape = .false. |
| 983 | token_has_quoted_part = .false. |
| 984 | end if |
| 985 | else if (ch == '{' .or. ch == '}') then |
| 986 | ! Braces: Keep in word for brace expansion (e.g., {1,2,3} or file{a,b}.txt) |
| 987 | ! They're only command group operators when surrounded by whitespace |
| 988 | if (token_len < MAX_TOKEN_LEN) then |
| 989 | token_len = token_len + 1 |
| 990 | current_token(token_len:token_len) = ch |
| 991 | end if |
| 992 | pos = pos + 1 |
| 993 | else if (in_double_bracket_context .and. & |
| 994 | (ch == '&' .or. ch == '|' .or. ch == '<' .or. ch == '>' .or. & |
| 995 | ch == '(' .or. ch == ')')) then |
| 996 | ! Inside [[ ]], these are test operators, not shell operators |
| 997 | if (token_len < MAX_TOKEN_LEN) then |
| 998 | token_len = token_len + 1 |
| 999 | current_token(token_len:token_len) = ch |
| 1000 | end if |
| 1001 | pos = pos + 1 |
| 1002 | else if (is_word_char(ch)) then |
| 1003 | ! Continue word |
| 1004 | if (token_len < MAX_TOKEN_LEN) then |
| 1005 | token_len = token_len + 1 |
| 1006 | current_token(token_len:token_len) = ch |
| 1007 | end if |
| 1008 | pos = pos + 1 |
| 1009 | else |
| 1010 | ! End of word |
| 1011 | call add_word_or_keyword(tokens, num_tokens, current_token(1:token_len), & |
| 1012 | token_start, pos-1, token_has_quoted_part, in_escape) |
| 1013 | state = LEX_NORMAL |
| 1014 | in_escape = .false. |
| 1015 | token_has_quoted_part = .false. |
| 1016 | ! Don't increment pos, let NORMAL state handle this character |
| 1017 | end if |
| 1018 | |
| 1019 | ! ============ OPERATOR STATE ============ |
| 1020 | case(LEX_IN_OPERATOR) |
| 1021 | ! Try to match multi-character operators |
| 1022 | if (token_len == 1) then |
| 1023 | select case(current_token(1:1)) |
| 1024 | case('&') |
| 1025 | if (ch == '&') then |
| 1026 | current_token(2:2) = ch |
| 1027 | token_len = 2 |
| 1028 | call add_token(tokens, num_tokens, TOKEN_OPERATOR, '&&', token_start, pos, .false.) |
| 1029 | state = LEX_NORMAL |
| 1030 | pos = pos + 1 |
| 1031 | else |
| 1032 | call add_token(tokens, num_tokens, TOKEN_OPERATOR, '&', token_start, pos-1, .false.) |
| 1033 | state = LEX_NORMAL |
| 1034 | end if |
| 1035 | case('|') |
| 1036 | if (ch == '|') then |
| 1037 | current_token(2:2) = ch |
| 1038 | token_len = 2 |
| 1039 | call add_token(tokens, num_tokens, TOKEN_OPERATOR, '||', token_start, pos, .false.) |
| 1040 | state = LEX_NORMAL |
| 1041 | pos = pos + 1 |
| 1042 | else |
| 1043 | call add_token(tokens, num_tokens, TOKEN_OPERATOR, '|', token_start, pos-1, .false.) |
| 1044 | state = LEX_NORMAL |
| 1045 | end if |
| 1046 | case('>') |
| 1047 | if (ch == '>') then |
| 1048 | current_token(2:2) = ch |
| 1049 | token_len = 2 |
| 1050 | call add_token(tokens, num_tokens, TOKEN_REDIRECT, '>>', token_start, pos, .false.) |
| 1051 | state = LEX_NORMAL |
| 1052 | pos = pos + 1 |
| 1053 | else if (ch == '&') then |
| 1054 | current_token(2:2) = ch |
| 1055 | token_len = 2 |
| 1056 | call add_token(tokens, num_tokens, TOKEN_REDIRECT, '>&', token_start, pos, .false.) |
| 1057 | state = LEX_NORMAL |
| 1058 | pos = pos + 1 |
| 1059 | else if (ch == '|') then |
| 1060 | current_token(2:2) = ch |
| 1061 | token_len = 2 |
| 1062 | call add_token(tokens, num_tokens, TOKEN_REDIRECT, '>|', token_start, pos, .false.) |
| 1063 | state = LEX_NORMAL |
| 1064 | pos = pos + 1 |
| 1065 | else |
| 1066 | call add_token(tokens, num_tokens, TOKEN_REDIRECT, '>', token_start, pos-1, .false.) |
| 1067 | state = LEX_NORMAL |
| 1068 | end if |
| 1069 | case('<') |
| 1070 | if (ch == '<') then |
| 1071 | ! Could be << or <<< or <<- |
| 1072 | if (pos < input_len .and. next_ch == '<') then |
| 1073 | current_token(2:3) = '<<' |
| 1074 | token_len = 3 |
| 1075 | call add_token(tokens, num_tokens, TOKEN_REDIRECT, '<<<', token_start, pos+1, .false.) |
| 1076 | state = LEX_NORMAL |
| 1077 | pos = pos + 2 |
| 1078 | else if (pos < input_len .and. next_ch == '-') then |
| 1079 | current_token(2:3) = '<-' |
| 1080 | token_len = 3 |
| 1081 | call add_token(tokens, num_tokens, TOKEN_REDIRECT, '<<-', token_start, pos+1, .false.) |
| 1082 | state = LEX_NORMAL |
| 1083 | pos = pos + 2 |
| 1084 | else |
| 1085 | current_token(2:2) = ch |
| 1086 | token_len = 2 |
| 1087 | call add_token(tokens, num_tokens, TOKEN_REDIRECT, '<<', token_start, pos, .false.) |
| 1088 | state = LEX_NORMAL |
| 1089 | pos = pos + 1 |
| 1090 | end if |
| 1091 | else if (ch == '>') then |
| 1092 | current_token(2:2) = ch |
| 1093 | token_len = 2 |
| 1094 | call add_token(tokens, num_tokens, TOKEN_REDIRECT, '<>', token_start, pos, .false.) |
| 1095 | state = LEX_NORMAL |
| 1096 | pos = pos + 1 |
| 1097 | else if (ch == '&') then |
| 1098 | current_token(2:2) = ch |
| 1099 | token_len = 2 |
| 1100 | call add_token(tokens, num_tokens, TOKEN_REDIRECT, '<&', token_start, pos, .false.) |
| 1101 | state = LEX_NORMAL |
| 1102 | pos = pos + 1 |
| 1103 | else |
| 1104 | call add_token(tokens, num_tokens, TOKEN_REDIRECT, '<', token_start, pos-1, .false.) |
| 1105 | state = LEX_NORMAL |
| 1106 | end if |
| 1107 | case(';') |
| 1108 | if (ch == ';') then |
| 1109 | current_token(2:2) = ch |
| 1110 | token_len = 2 |
| 1111 | call add_token(tokens, num_tokens, TOKEN_OPERATOR, ';;', token_start, pos, .false.) |
| 1112 | state = LEX_NORMAL |
| 1113 | pos = pos + 1 |
| 1114 | else |
| 1115 | call add_token(tokens, num_tokens, TOKEN_OPERATOR, ';', token_start, pos-1, .false.) |
| 1116 | state = LEX_NORMAL |
| 1117 | end if |
| 1118 | case('(', ')') |
| 1119 | call add_token(tokens, num_tokens, TOKEN_OPERATOR, current_token(1:1), & |
| 1120 | token_start, pos-1, .false.) |
| 1121 | state = LEX_NORMAL |
| 1122 | case default |
| 1123 | ! Single-character operator |
| 1124 | call add_token(tokens, num_tokens, TOKEN_OPERATOR, current_token(1:1), & |
| 1125 | token_start, pos-1, .false.) |
| 1126 | state = LEX_NORMAL |
| 1127 | end select |
| 1128 | else |
| 1129 | ! Multi-character operator complete |
| 1130 | state = LEX_NORMAL |
| 1131 | end if |
| 1132 | |
| 1133 | end select |
| 1134 | end do |
| 1135 | |
| 1136 | ! Flush any remaining token |
| 1137 | if (state == LEX_IN_WORD .and. token_len > 0) then |
| 1138 | call add_word_or_keyword(tokens, num_tokens, current_token(1:token_len), & |
| 1139 | token_start, input_len, token_has_quoted_part, in_escape) |
| 1140 | else if (state == LEX_IN_SINGLE_QUOTE .or. state == LEX_IN_DOUBLE_QUOTE) then |
| 1141 | ! Unterminated quote - add as word with error marker |
| 1142 | if (state == LEX_IN_SINGLE_QUOTE) then |
| 1143 | call add_token(tokens, num_tokens, TOKEN_WORD, current_token(1:token_len), & |
| 1144 | token_start, input_len, .true., quote_type=QUOTE_SINGLE) |
| 1145 | else |
| 1146 | call add_token(tokens, num_tokens, TOKEN_WORD, current_token(1:token_len), & |
| 1147 | token_start, input_len, .true., quote_type=QUOTE_DOUBLE) |
| 1148 | end if |
| 1149 | else if (state == LEX_IN_OPERATOR .and. token_len > 0) then |
| 1150 | ! Flush operator |
| 1151 | call add_token(tokens, num_tokens, TOKEN_OPERATOR, current_token(1:token_len), & |
| 1152 | token_start, input_len, .false.) |
| 1153 | end if |
| 1154 | |
| 1155 | ! Add EOF token |
| 1156 | call add_token(tokens, num_tokens, TOKEN_EOF, '', input_len+1, input_len+1, .false.) |
| 1157 | |
| 1158 | end subroutine tokenize |
| 1159 | |
| 1160 | ! ===================================== |
| 1161 | ! Helper: Add token to array |
| 1162 | ! ===================================== |
| 1163 | subroutine add_token(tokens, num_tokens, tok_type, value, start_pos, end_pos, quoted, escaped, quote_type) |
| 1164 | use shell_types, only: QUOTE_NONE |
| 1165 | type(token_t), intent(inout) :: tokens(:) |
| 1166 | integer, intent(inout) :: num_tokens |
| 1167 | integer, intent(in) :: tok_type, start_pos, end_pos |
| 1168 | character(len=*), intent(in) :: value |
| 1169 | logical, intent(in) :: quoted |
| 1170 | logical, intent(in), optional :: escaped |
| 1171 | integer, intent(in), optional :: quote_type |
| 1172 | |
| 1173 | if (num_tokens < size(tokens)) then |
| 1174 | num_tokens = num_tokens + 1 |
| 1175 | tokens(num_tokens)%token_type = tok_type |
| 1176 | tokens(num_tokens)%value = value |
| 1177 | tokens(num_tokens)%value_length = len(value) ! Store actual content length |
| 1178 | tokens(num_tokens)%start_pos = start_pos |
| 1179 | tokens(num_tokens)%end_pos = end_pos |
| 1180 | tokens(num_tokens)%quoted = quoted |
| 1181 | if (present(escaped)) then |
| 1182 | tokens(num_tokens)%escaped = escaped |
| 1183 | else |
| 1184 | tokens(num_tokens)%escaped = .false. |
| 1185 | end if |
| 1186 | if (present(quote_type)) then |
| 1187 | tokens(num_tokens)%quote_type = quote_type |
| 1188 | else |
| 1189 | tokens(num_tokens)%quote_type = QUOTE_NONE |
| 1190 | end if |
| 1191 | end if |
| 1192 | end subroutine add_token |
| 1193 | |
| 1194 | ! ===================================== |
| 1195 | ! Helper: Add word or keyword token |
| 1196 | ! ===================================== |
| 1197 | subroutine add_word_or_keyword(tokens, num_tokens, value, start_pos, end_pos, quoted, escaped) |
| 1198 | type(token_t), intent(inout) :: tokens(:) |
| 1199 | integer, intent(inout) :: num_tokens |
| 1200 | character(len=*), intent(in) :: value |
| 1201 | integer, intent(in) :: start_pos, end_pos |
| 1202 | logical, intent(in) :: quoted |
| 1203 | logical, intent(in), optional :: escaped |
| 1204 | |
| 1205 | integer :: tok_type |
| 1206 | |
| 1207 | ! Quoted strings are always words, never keywords |
| 1208 | if (quoted) then |
| 1209 | tok_type = TOKEN_WORD |
| 1210 | else if (is_keyword(value)) then |
| 1211 | tok_type = TOKEN_KEYWORD |
| 1212 | else |
| 1213 | tok_type = TOKEN_WORD |
| 1214 | end if |
| 1215 | |
| 1216 | ! Track [[ ]] context: inside test expressions, operators become words |
| 1217 | if (.not. quoted) then |
| 1218 | if (trim(value) == '[[') in_double_bracket_context = .true. |
| 1219 | if (trim(value) == ']]') in_double_bracket_context = .false. |
| 1220 | end if |
| 1221 | |
| 1222 | call add_token(tokens, num_tokens, tok_type, value, start_pos, end_pos, quoted, escaped) |
| 1223 | end subroutine add_word_or_keyword |
| 1224 | |
| 1225 | ! ===================================== |
| 1226 | ! next_token - Get next token from stream |
| 1227 | ! ===================================== |
| 1228 | function next_token(tokens, pos) result(tok) |
| 1229 | type(token_t), intent(in) :: tokens(:) |
| 1230 | integer, intent(inout) :: pos |
| 1231 | type(token_t) :: tok |
| 1232 | |
| 1233 | if (pos <= size(tokens) .and. pos > 0) then |
| 1234 | tok = tokens(pos) |
| 1235 | pos = pos + 1 |
| 1236 | else |
| 1237 | ! Return EOF token |
| 1238 | tok%token_type = TOKEN_EOF |
| 1239 | tok%value = '' |
| 1240 | tok%start_pos = 0 |
| 1241 | tok%end_pos = 0 |
| 1242 | tok%quoted = .false. |
| 1243 | end if |
| 1244 | end function next_token |
| 1245 | |
| 1246 | ! ===================================== |
| 1247 | ! peek_token - Look ahead without consuming |
| 1248 | ! ===================================== |
| 1249 | function peek_token(tokens, pos) result(tok) |
| 1250 | type(token_t), intent(in) :: tokens(:) |
| 1251 | integer, intent(in) :: pos |
| 1252 | type(token_t) :: tok |
| 1253 | |
| 1254 | if (pos <= size(tokens) .and. pos > 0) then |
| 1255 | tok = tokens(pos) |
| 1256 | else |
| 1257 | ! Return EOF token |
| 1258 | tok%token_type = TOKEN_EOF |
| 1259 | tok%value = '' |
| 1260 | tok%start_pos = 0 |
| 1261 | tok%end_pos = 0 |
| 1262 | tok%quoted = .false. |
| 1263 | end if |
| 1264 | end function peek_token |
| 1265 | |
| 1266 | ! ===================================== |
| 1267 | ! process_ansi_c_quote - Handle $'...' ANSI-C quoting |
| 1268 | ! Reads characters from input starting at pos (after $'), |
| 1269 | ! processes escape sequences, appends to current_token. |
| 1270 | ! Wraps output in char(2)/char(3) sentinels to prevent expansion. |
| 1271 | ! ===================================== |
| 1272 | subroutine process_ansi_c_quote(input, pos, input_len, current_token, token_len) |
| 1273 | character(len=*), intent(in) :: input |
| 1274 | integer, intent(inout) :: pos |
| 1275 | integer, intent(in) :: input_len |
| 1276 | character(len=*), intent(inout) :: current_token |
| 1277 | integer, intent(inout) :: token_len |
| 1278 | |
| 1279 | character :: ch, esc_ch |
| 1280 | integer :: oct_val, hex_val, n_digits, i |
| 1281 | |
| 1282 | ! Add sentinel to mark start of quoted content (no expansion) |
| 1283 | if (token_len < MAX_TOKEN_LEN) then |
| 1284 | token_len = token_len + 1 |
| 1285 | current_token(token_len:token_len) = char(2) |
| 1286 | end if |
| 1287 | |
| 1288 | do while (pos <= input_len) |
| 1289 | ch = input(pos:pos) |
| 1290 | |
| 1291 | if (ch == "'") then |
| 1292 | ! Closing quote |
| 1293 | pos = pos + 1 |
| 1294 | ! Add sentinel to mark end of quoted content |
| 1295 | if (token_len < MAX_TOKEN_LEN) then |
| 1296 | token_len = token_len + 1 |
| 1297 | current_token(token_len:token_len) = char(3) |
| 1298 | end if |
| 1299 | return |
| 1300 | end if |
| 1301 | |
| 1302 | if (ch == '\' .and. pos < input_len) then |
| 1303 | ! Escape sequence |
| 1304 | esc_ch = input(pos+1:pos+1) |
| 1305 | select case(esc_ch) |
| 1306 | case('a') ! Alert (bell) |
| 1307 | if (token_len < MAX_TOKEN_LEN) then |
| 1308 | token_len = token_len + 1 |
| 1309 | current_token(token_len:token_len) = char(7) |
| 1310 | end if |
| 1311 | pos = pos + 2 |
| 1312 | case('b') ! Backspace |
| 1313 | if (token_len < MAX_TOKEN_LEN) then |
| 1314 | token_len = token_len + 1 |
| 1315 | current_token(token_len:token_len) = char(8) |
| 1316 | end if |
| 1317 | pos = pos + 2 |
| 1318 | case('e', 'E') ! Escape |
| 1319 | if (token_len < MAX_TOKEN_LEN) then |
| 1320 | token_len = token_len + 1 |
| 1321 | current_token(token_len:token_len) = char(27) |
| 1322 | end if |
| 1323 | pos = pos + 2 |
| 1324 | case('f') ! Form feed |
| 1325 | if (token_len < MAX_TOKEN_LEN) then |
| 1326 | token_len = token_len + 1 |
| 1327 | current_token(token_len:token_len) = char(12) |
| 1328 | end if |
| 1329 | pos = pos + 2 |
| 1330 | case('n') ! Newline |
| 1331 | if (token_len < MAX_TOKEN_LEN) then |
| 1332 | token_len = token_len + 1 |
| 1333 | current_token(token_len:token_len) = char(10) |
| 1334 | end if |
| 1335 | pos = pos + 2 |
| 1336 | case('r') ! Carriage return |
| 1337 | if (token_len < MAX_TOKEN_LEN) then |
| 1338 | token_len = token_len + 1 |
| 1339 | current_token(token_len:token_len) = char(13) |
| 1340 | end if |
| 1341 | pos = pos + 2 |
| 1342 | case('t') ! Horizontal tab |
| 1343 | if (token_len < MAX_TOKEN_LEN) then |
| 1344 | token_len = token_len + 1 |
| 1345 | current_token(token_len:token_len) = char(9) |
| 1346 | end if |
| 1347 | pos = pos + 2 |
| 1348 | case('v') ! Vertical tab |
| 1349 | if (token_len < MAX_TOKEN_LEN) then |
| 1350 | token_len = token_len + 1 |
| 1351 | current_token(token_len:token_len) = char(11) |
| 1352 | end if |
| 1353 | pos = pos + 2 |
| 1354 | case('\') ! Literal backslash |
| 1355 | if (token_len < MAX_TOKEN_LEN) then |
| 1356 | token_len = token_len + 1 |
| 1357 | current_token(token_len:token_len) = '\' |
| 1358 | end if |
| 1359 | pos = pos + 2 |
| 1360 | case("'") ! Literal single quote |
| 1361 | if (token_len < MAX_TOKEN_LEN) then |
| 1362 | token_len = token_len + 1 |
| 1363 | current_token(token_len:token_len) = "'" |
| 1364 | end if |
| 1365 | pos = pos + 2 |
| 1366 | case('"') ! Literal double quote |
| 1367 | if (token_len < MAX_TOKEN_LEN) then |
| 1368 | token_len = token_len + 1 |
| 1369 | current_token(token_len:token_len) = '"' |
| 1370 | end if |
| 1371 | pos = pos + 2 |
| 1372 | case('0', '1', '2', '3', '4', '5', '6', '7') |
| 1373 | ! Octal: \nnn (up to 3 digits) |
| 1374 | oct_val = 0 |
| 1375 | n_digits = 0 |
| 1376 | pos = pos + 1 ! skip backslash |
| 1377 | do while (pos <= input_len .and. n_digits < 3) |
| 1378 | ch = input(pos:pos) |
| 1379 | if (ch >= '0' .and. ch <= '7') then |
| 1380 | oct_val = oct_val * 8 + (ichar(ch) - ichar('0')) |
| 1381 | pos = pos + 1 |
| 1382 | n_digits = n_digits + 1 |
| 1383 | else |
| 1384 | exit |
| 1385 | end if |
| 1386 | end do |
| 1387 | if (oct_val > 0 .and. oct_val <= 127 .and. token_len < MAX_TOKEN_LEN) then |
| 1388 | token_len = token_len + 1 |
| 1389 | current_token(token_len:token_len) = char(oct_val) |
| 1390 | end if |
| 1391 | case('x') |
| 1392 | ! Hex: \xHH (up to 2 digits) |
| 1393 | hex_val = 0 |
| 1394 | n_digits = 0 |
| 1395 | pos = pos + 2 ! skip \x |
| 1396 | do while (pos <= input_len .and. n_digits < 2) |
| 1397 | ch = input(pos:pos) |
| 1398 | if (ch >= '0' .and. ch <= '9') then |
| 1399 | hex_val = hex_val * 16 + (ichar(ch) - ichar('0')) |
| 1400 | pos = pos + 1 |
| 1401 | n_digits = n_digits + 1 |
| 1402 | else if (ch >= 'a' .and. ch <= 'f') then |
| 1403 | hex_val = hex_val * 16 + (ichar(ch) - ichar('a') + 10) |
| 1404 | pos = pos + 1 |
| 1405 | n_digits = n_digits + 1 |
| 1406 | else if (ch >= 'A' .and. ch <= 'F') then |
| 1407 | hex_val = hex_val * 16 + (ichar(ch) - ichar('A') + 10) |
| 1408 | pos = pos + 1 |
| 1409 | n_digits = n_digits + 1 |
| 1410 | else |
| 1411 | exit |
| 1412 | end if |
| 1413 | end do |
| 1414 | if (hex_val > 0 .and. hex_val <= 127 .and. token_len < MAX_TOKEN_LEN) then |
| 1415 | token_len = token_len + 1 |
| 1416 | current_token(token_len:token_len) = char(hex_val) |
| 1417 | end if |
| 1418 | case default |
| 1419 | ! Unknown escape: include backslash and character literally |
| 1420 | if (token_len < MAX_TOKEN_LEN - 1) then |
| 1421 | token_len = token_len + 1 |
| 1422 | current_token(token_len:token_len) = '\' |
| 1423 | token_len = token_len + 1 |
| 1424 | current_token(token_len:token_len) = esc_ch |
| 1425 | end if |
| 1426 | pos = pos + 2 |
| 1427 | end select |
| 1428 | else |
| 1429 | ! Regular character |
| 1430 | if (token_len < MAX_TOKEN_LEN) then |
| 1431 | token_len = token_len + 1 |
| 1432 | current_token(token_len:token_len) = ch |
| 1433 | end if |
| 1434 | pos = pos + 1 |
| 1435 | end if |
| 1436 | end do |
| 1437 | |
| 1438 | ! Unterminated $'...' - add sentinel anyway |
| 1439 | if (token_len < MAX_TOKEN_LEN) then |
| 1440 | token_len = token_len + 1 |
| 1441 | current_token(token_len:token_len) = char(3) |
| 1442 | end if |
| 1443 | end subroutine process_ansi_c_quote |
| 1444 | |
| 1445 | end module lexer |
| 1446 |