]> git.ipfire.org Git - thirdparty/bash.git/blob - parse.y
Bash-4.2 patch 44
[thirdparty/bash.git] / parse.y
1 /* parse.y - Yacc grammar for bash. */
2
3 /* Copyright (C) 1989-2010 Free Software Foundation, Inc.
4
5 This file is part of GNU Bash, the Bourne Again SHell.
6
7 Bash is free software: you can redistribute it and/or modify
8 it under the terms of the GNU General Public License as published by
9 the Free Software Foundation, either version 3 of the License, or
10 (at your option) any later version.
11
12 Bash is distributed in the hope that it will be useful,
13 but WITHOUT ANY WARRANTY; without even the implied warranty of
14 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 GNU General Public License for more details.
16
17 You should have received a copy of the GNU General Public License
18 along with Bash. If not, see <http://www.gnu.org/licenses/>.
19 */
20
21 %{
22 #include "config.h"
23
24 #include "bashtypes.h"
25 #include "bashansi.h"
26
27 #include "filecntl.h"
28
29 #if defined (HAVE_UNISTD_H)
30 # include <unistd.h>
31 #endif
32
33 #if defined (HAVE_LOCALE_H)
34 # include <locale.h>
35 #endif
36
37 #include <stdio.h>
38 #include "chartypes.h"
39 #include <signal.h>
40
41 #include "memalloc.h"
42
43 #include "bashintl.h"
44
45 #define NEED_STRFTIME_DECL /* used in externs.h */
46
47 #include "shell.h"
48 #include "trap.h"
49 #include "flags.h"
50 #include "parser.h"
51 #include "mailcheck.h"
52 #include "test.h"
53 #include "builtins.h"
54 #include "builtins/common.h"
55 #include "builtins/builtext.h"
56
57 #include "shmbutil.h"
58
59 #if defined (READLINE)
60 # include "bashline.h"
61 # include <readline/readline.h>
62 #endif /* READLINE */
63
64 #if defined (HISTORY)
65 # include "bashhist.h"
66 # include <readline/history.h>
67 #endif /* HISTORY */
68
69 #if defined (JOB_CONTROL)
70 # include "jobs.h"
71 #endif /* JOB_CONTROL */
72
73 #if defined (ALIAS)
74 # include "alias.h"
75 #else
76 typedef void *alias_t;
77 #endif /* ALIAS */
78
79 #if defined (PROMPT_STRING_DECODE)
80 # ifndef _MINIX
81 # include <sys/param.h>
82 # endif
83 # include <time.h>
84 # if defined (TM_IN_SYS_TIME)
85 # include <sys/types.h>
86 # include <sys/time.h>
87 # endif /* TM_IN_SYS_TIME */
88 # include "maxpath.h"
89 #endif /* PROMPT_STRING_DECODE */
90
91 #define RE_READ_TOKEN -99
92 #define NO_EXPANSION -100
93
94 #ifdef DEBUG
95 # define YYDEBUG 1
96 #else
97 # define YYDEBUG 0
98 #endif
99
100 #if defined (HANDLE_MULTIBYTE)
101 # define last_shell_getc_is_singlebyte \
102 ((shell_input_line_index > 1) \
103 ? shell_input_line_property[shell_input_line_index - 1] \
104 : 1)
105 # define MBTEST(x) ((x) && last_shell_getc_is_singlebyte)
106 #else
107 # define last_shell_getc_is_singlebyte 1
108 # define MBTEST(x) ((x))
109 #endif
110
111 #if defined (EXTENDED_GLOB)
112 extern int extended_glob;
113 #endif
114
115 extern int eof_encountered;
116 extern int no_line_editing, running_under_emacs;
117 extern int current_command_number;
118 extern int sourcelevel, parse_and_execute_level;
119 extern int posixly_correct;
120 extern int last_command_exit_value;
121 extern pid_t last_command_subst_pid;
122 extern char *shell_name, *current_host_name;
123 extern char *dist_version;
124 extern int patch_level;
125 extern int dump_translatable_strings, dump_po_strings;
126 extern sh_builtin_func_t *last_shell_builtin, *this_shell_builtin;
127 #if defined (BUFFERED_INPUT)
128 extern int bash_input_fd_changed;
129 #endif
130
131 extern int errno;
132 /* **************************************************************** */
133 /* */
134 /* "Forward" declarations */
135 /* */
136 /* **************************************************************** */
137
138 #ifdef DEBUG
139 static void debug_parser __P((int));
140 #endif
141
142 static int yy_getc __P((void));
143 static int yy_ungetc __P((int));
144
145 #if defined (READLINE)
146 static int yy_readline_get __P((void));
147 static int yy_readline_unget __P((int));
148 #endif
149
150 static int yy_string_get __P((void));
151 static int yy_string_unget __P((int));
152 static void rewind_input_string __P((void));
153 static int yy_stream_get __P((void));
154 static int yy_stream_unget __P((int));
155
156 static int shell_getc __P((int));
157 static void shell_ungetc __P((int));
158 static void discard_until __P((int));
159
160 #if defined (ALIAS) || defined (DPAREN_ARITHMETIC)
161 static void push_string __P((char *, int, alias_t *));
162 static void pop_string __P((void));
163 static void free_string_list __P((void));
164 #endif
165
166 static char *read_a_line __P((int));
167
168 static int reserved_word_acceptable __P((int));
169 static int yylex __P((void));
170 static int alias_expand_token __P((char *));
171 static int time_command_acceptable __P((void));
172 static int special_case_tokens __P((char *));
173 static int read_token __P((int));
174 static char *parse_matched_pair __P((int, int, int, int *, int));
175 static char *parse_comsub __P((int, int, int, int *, int));
176 #if defined (ARRAY_VARS)
177 static char *parse_compound_assignment __P((int *));
178 #endif
179 #if defined (DPAREN_ARITHMETIC) || defined (ARITH_FOR_COMMAND)
180 static int parse_dparen __P((int));
181 static int parse_arith_cmd __P((char **, int));
182 #endif
183 #if defined (COND_COMMAND)
184 static void cond_error __P((void));
185 static COND_COM *cond_expr __P((void));
186 static COND_COM *cond_or __P((void));
187 static COND_COM *cond_and __P((void));
188 static COND_COM *cond_term __P((void));
189 static int cond_skip_newlines __P((void));
190 static COMMAND *parse_cond_command __P((void));
191 #endif
192 #if defined (ARRAY_VARS)
193 static int token_is_assignment __P((char *, int));
194 static int token_is_ident __P((char *, int));
195 #endif
196 static int read_token_word __P((int));
197 static void discard_parser_constructs __P((int));
198
199 static char *error_token_from_token __P((int));
200 static char *error_token_from_text __P((void));
201 static void print_offending_line __P((void));
202 static void report_syntax_error __P((char *));
203
204 static void handle_eof_input_unit __P((void));
205 static void prompt_again __P((void));
206 #if 0
207 static void reset_readline_prompt __P((void));
208 #endif
209 static void print_prompt __P((void));
210
211 #if defined (HANDLE_MULTIBYTE)
212 static void set_line_mbstate __P((void));
213 static char *shell_input_line_property = NULL;
214 #else
215 # define set_line_mbstate()
216 #endif
217
218 extern int yyerror __P((const char *));
219
220 #ifdef DEBUG
221 extern int yydebug;
222 #endif
223
224 /* Default prompt strings */
225 char *primary_prompt = PPROMPT;
226 char *secondary_prompt = SPROMPT;
227
228 /* PROMPT_STRING_POINTER points to one of these, never to an actual string. */
229 char *ps1_prompt, *ps2_prompt;
230
231 /* Handle on the current prompt string. Indirectly points through
232 ps1_ or ps2_prompt. */
233 char **prompt_string_pointer = (char **)NULL;
234 char *current_prompt_string;
235
236 /* Non-zero means we expand aliases in commands. */
237 int expand_aliases = 0;
238
239 /* If non-zero, the decoded prompt string undergoes parameter and
240 variable substitution, command substitution, arithmetic substitution,
241 string expansion, process substitution, and quote removal in
242 decode_prompt_string. */
243 int promptvars = 1;
244
245 /* If non-zero, $'...' and $"..." are expanded when they appear within
246 a ${...} expansion, even when the expansion appears within double
247 quotes. */
248 int extended_quote = 1;
249
250 /* The number of lines read from input while creating the current command. */
251 int current_command_line_count;
252
253 /* The number of lines in a command saved while we run parse_and_execute */
254 int saved_command_line_count;
255
256 /* The token that currently denotes the end of parse. */
257 int shell_eof_token;
258
259 /* The token currently being read. */
260 int current_token;
261
262 /* The current parser state. */
263 int parser_state;
264
265 /* Variables to manage the task of reading here documents, because we need to
266 defer the reading until after a complete command has been collected. */
267 static REDIRECT *redir_stack[10];
268 int need_here_doc;
269
270 /* Where shell input comes from. History expansion is performed on each
271 line when the shell is interactive. */
272 static char *shell_input_line = (char *)NULL;
273 static int shell_input_line_index;
274 static int shell_input_line_size; /* Amount allocated for shell_input_line. */
275 static int shell_input_line_len; /* strlen (shell_input_line) */
276
277 /* Either zero or EOF. */
278 static int shell_input_line_terminator;
279
280 /* The line number in a script on which a function definition starts. */
281 static int function_dstart;
282
283 /* The line number in a script on which a function body starts. */
284 static int function_bstart;
285
286 /* The line number in a script at which an arithmetic for command starts. */
287 static int arith_for_lineno;
288
289 /* The decoded prompt string. Used if READLINE is not defined or if
290 editing is turned off. Analogous to current_readline_prompt. */
291 static char *current_decoded_prompt;
292
293 /* The last read token, or NULL. read_token () uses this for context
294 checking. */
295 static int last_read_token;
296
297 /* The token read prior to last_read_token. */
298 static int token_before_that;
299
300 /* The token read prior to token_before_that. */
301 static int two_tokens_ago;
302
303 static int global_extglob;
304
305 /* The line number in a script where the word in a `case WORD', `select WORD'
306 or `for WORD' begins. This is a nested command maximum, since the array
307 index is decremented after a case, select, or for command is parsed. */
308 #define MAX_CASE_NEST 128
309 static int word_lineno[MAX_CASE_NEST];
310 static int word_top = -1;
311
312 /* If non-zero, it is the token that we want read_token to return
313 regardless of what text is (or isn't) present to be read. This
314 is reset by read_token. If token_to_read == WORD or
315 ASSIGNMENT_WORD, yylval.word should be set to word_desc_to_read. */
316 static int token_to_read;
317 static WORD_DESC *word_desc_to_read;
318
319 static REDIRECTEE source;
320 static REDIRECTEE redir;
321 %}
322
323 %union {
324 WORD_DESC *word; /* the word that we read. */
325 int number; /* the number that we read. */
326 WORD_LIST *word_list;
327 COMMAND *command;
328 REDIRECT *redirect;
329 ELEMENT element;
330 PATTERN_LIST *pattern;
331 }
332
333 /* Reserved words. Members of the first group are only recognized
334 in the case that they are preceded by a list_terminator. Members
335 of the second group are for [[...]] commands. Members of the
336 third group are recognized only under special circumstances. */
337 %token IF THEN ELSE ELIF FI CASE ESAC FOR SELECT WHILE UNTIL DO DONE FUNCTION COPROC
338 %token COND_START COND_END COND_ERROR
339 %token IN BANG TIME TIMEOPT TIMEIGN
340
341 /* More general tokens. yylex () knows how to make these. */
342 %token <word> WORD ASSIGNMENT_WORD REDIR_WORD
343 %token <number> NUMBER
344 %token <word_list> ARITH_CMD ARITH_FOR_EXPRS
345 %token <command> COND_CMD
346 %token AND_AND OR_OR GREATER_GREATER LESS_LESS LESS_AND LESS_LESS_LESS
347 %token GREATER_AND SEMI_SEMI SEMI_AND SEMI_SEMI_AND
348 %token LESS_LESS_MINUS AND_GREATER AND_GREATER_GREATER LESS_GREATER
349 %token GREATER_BAR BAR_AND
350
351 /* The types that the various syntactical units return. */
352
353 %type <command> inputunit command pipeline pipeline_command
354 %type <command> list list0 list1 compound_list simple_list simple_list1
355 %type <command> simple_command shell_command
356 %type <command> for_command select_command case_command group_command
357 %type <command> arith_command
358 %type <command> cond_command
359 %type <command> arith_for_command
360 %type <command> coproc
361 %type <command> function_def function_body if_command elif_clause subshell
362 %type <redirect> redirection redirection_list
363 %type <element> simple_command_element
364 %type <word_list> word_list pattern
365 %type <pattern> pattern_list case_clause_sequence case_clause
366 %type <number> timespec
367 %type <number> list_terminator
368
369 %start inputunit
370
371 %left '&' ';' '\n' yacc_EOF
372 %left AND_AND OR_OR
373 %right '|' BAR_AND
374 %%
375
376 inputunit: simple_list simple_list_terminator
377 {
378 /* Case of regular command. Discard the error
379 safety net,and return the command just parsed. */
380 global_command = $1;
381 eof_encountered = 0;
382 /* discard_parser_constructs (0); */
383 if (parser_state & PST_CMDSUBST)
384 parser_state |= PST_EOFTOKEN;
385 YYACCEPT;
386 }
387 | '\n'
388 {
389 /* Case of regular command, but not a very
390 interesting one. Return a NULL command. */
391 global_command = (COMMAND *)NULL;
392 if (parser_state & PST_CMDSUBST)
393 parser_state |= PST_EOFTOKEN;
394 YYACCEPT;
395 }
396 | error '\n'
397 {
398 /* Error during parsing. Return NULL command. */
399 global_command = (COMMAND *)NULL;
400 eof_encountered = 0;
401 /* discard_parser_constructs (1); */
402 if (interactive && parse_and_execute_level == 0)
403 {
404 YYACCEPT;
405 }
406 else
407 {
408 YYABORT;
409 }
410 }
411 | yacc_EOF
412 {
413 /* Case of EOF seen by itself. Do ignoreeof or
414 not. */
415 global_command = (COMMAND *)NULL;
416 handle_eof_input_unit ();
417 YYACCEPT;
418 }
419 ;
420
421 word_list: WORD
422 { $$ = make_word_list ($1, (WORD_LIST *)NULL); }
423 | word_list WORD
424 { $$ = make_word_list ($2, $1); }
425 ;
426
427 redirection: '>' WORD
428 {
429 source.dest = 1;
430 redir.filename = $2;
431 $$ = make_redirection (source, r_output_direction, redir, 0);
432 }
433 | '<' WORD
434 {
435 source.dest = 0;
436 redir.filename = $2;
437 $$ = make_redirection (source, r_input_direction, redir, 0);
438 }
439 | NUMBER '>' WORD
440 {
441 source.dest = $1;
442 redir.filename = $3;
443 $$ = make_redirection (source, r_output_direction, redir, 0);
444 }
445 | NUMBER '<' WORD
446 {
447 source.dest = $1;
448 redir.filename = $3;
449 $$ = make_redirection (source, r_input_direction, redir, 0);
450 }
451 | REDIR_WORD '>' WORD
452 {
453 source.filename = $1;
454 redir.filename = $3;
455 $$ = make_redirection (source, r_output_direction, redir, REDIR_VARASSIGN);
456 }
457 | REDIR_WORD '<' WORD
458 {
459 source.filename = $1;
460 redir.filename = $3;
461 $$ = make_redirection (source, r_input_direction, redir, REDIR_VARASSIGN);
462 }
463 | GREATER_GREATER WORD
464 {
465 source.dest = 1;
466 redir.filename = $2;
467 $$ = make_redirection (source, r_appending_to, redir, 0);
468 }
469 | NUMBER GREATER_GREATER WORD
470 {
471 source.dest = $1;
472 redir.filename = $3;
473 $$ = make_redirection (source, r_appending_to, redir, 0);
474 }
475 | REDIR_WORD GREATER_GREATER WORD
476 {
477 source.filename = $1;
478 redir.filename = $3;
479 $$ = make_redirection (source, r_appending_to, redir, REDIR_VARASSIGN);
480 }
481 | GREATER_BAR WORD
482 {
483 source.dest = 1;
484 redir.filename = $2;
485 $$ = make_redirection (source, r_output_force, redir, 0);
486 }
487 | NUMBER GREATER_BAR WORD
488 {
489 source.dest = $1;
490 redir.filename = $3;
491 $$ = make_redirection (source, r_output_force, redir, 0);
492 }
493 | REDIR_WORD GREATER_BAR WORD
494 {
495 source.filename = $1;
496 redir.filename = $3;
497 $$ = make_redirection (source, r_output_force, redir, REDIR_VARASSIGN);
498 }
499 | LESS_GREATER WORD
500 {
501 source.dest = 0;
502 redir.filename = $2;
503 $$ = make_redirection (source, r_input_output, redir, 0);
504 }
505 | NUMBER LESS_GREATER WORD
506 {
507 source.dest = $1;
508 redir.filename = $3;
509 $$ = make_redirection (source, r_input_output, redir, 0);
510 }
511 | REDIR_WORD LESS_GREATER WORD
512 {
513 source.filename = $1;
514 redir.filename = $3;
515 $$ = make_redirection (source, r_input_output, redir, REDIR_VARASSIGN);
516 }
517 | LESS_LESS WORD
518 {
519 source.dest = 0;
520 redir.filename = $2;
521 $$ = make_redirection (source, r_reading_until, redir, 0);
522 redir_stack[need_here_doc++] = $$;
523 }
524 | NUMBER LESS_LESS WORD
525 {
526 source.dest = $1;
527 redir.filename = $3;
528 $$ = make_redirection (source, r_reading_until, redir, 0);
529 redir_stack[need_here_doc++] = $$;
530 }
531 | REDIR_WORD LESS_LESS WORD
532 {
533 source.filename = $1;
534 redir.filename = $3;
535 $$ = make_redirection (source, r_reading_until, redir, REDIR_VARASSIGN);
536 redir_stack[need_here_doc++] = $$;
537 }
538 | LESS_LESS_MINUS WORD
539 {
540 source.dest = 0;
541 redir.filename = $2;
542 $$ = make_redirection (source, r_deblank_reading_until, redir, 0);
543 redir_stack[need_here_doc++] = $$;
544 }
545 | NUMBER LESS_LESS_MINUS WORD
546 {
547 source.dest = $1;
548 redir.filename = $3;
549 $$ = make_redirection (source, r_deblank_reading_until, redir, 0);
550 redir_stack[need_here_doc++] = $$;
551 }
552 | REDIR_WORD LESS_LESS_MINUS WORD
553 {
554 source.filename = $1;
555 redir.filename = $3;
556 $$ = make_redirection (source, r_deblank_reading_until, redir, REDIR_VARASSIGN);
557 redir_stack[need_here_doc++] = $$;
558 }
559 | LESS_LESS_LESS WORD
560 {
561 source.dest = 0;
562 redir.filename = $2;
563 $$ = make_redirection (source, r_reading_string, redir, 0);
564 }
565 | NUMBER LESS_LESS_LESS WORD
566 {
567 source.dest = $1;
568 redir.filename = $3;
569 $$ = make_redirection (source, r_reading_string, redir, 0);
570 }
571 | REDIR_WORD LESS_LESS_LESS WORD
572 {
573 source.filename = $1;
574 redir.filename = $3;
575 $$ = make_redirection (source, r_reading_string, redir, REDIR_VARASSIGN);
576 }
577 | LESS_AND NUMBER
578 {
579 source.dest = 0;
580 redir.dest = $2;
581 $$ = make_redirection (source, r_duplicating_input, redir, 0);
582 }
583 | NUMBER LESS_AND NUMBER
584 {
585 source.dest = $1;
586 redir.dest = $3;
587 $$ = make_redirection (source, r_duplicating_input, redir, 0);
588 }
589 | REDIR_WORD LESS_AND NUMBER
590 {
591 source.filename = $1;
592 redir.dest = $3;
593 $$ = make_redirection (source, r_duplicating_input, redir, REDIR_VARASSIGN);
594 }
595 | GREATER_AND NUMBER
596 {
597 source.dest = 1;
598 redir.dest = $2;
599 $$ = make_redirection (source, r_duplicating_output, redir, 0);
600 }
601 | NUMBER GREATER_AND NUMBER
602 {
603 source.dest = $1;
604 redir.dest = $3;
605 $$ = make_redirection (source, r_duplicating_output, redir, 0);
606 }
607 | REDIR_WORD GREATER_AND NUMBER
608 {
609 source.filename = $1;
610 redir.dest = $3;
611 $$ = make_redirection (source, r_duplicating_output, redir, REDIR_VARASSIGN);
612 }
613 | LESS_AND WORD
614 {
615 source.dest = 0;
616 redir.filename = $2;
617 $$ = make_redirection (source, r_duplicating_input_word, redir, 0);
618 }
619 | NUMBER LESS_AND WORD
620 {
621 source.dest = $1;
622 redir.filename = $3;
623 $$ = make_redirection (source, r_duplicating_input_word, redir, 0);
624 }
625 | REDIR_WORD LESS_AND WORD
626 {
627 source.filename = $1;
628 redir.filename = $3;
629 $$ = make_redirection (source, r_duplicating_input_word, redir, REDIR_VARASSIGN);
630 }
631 | GREATER_AND WORD
632 {
633 source.dest = 1;
634 redir.filename = $2;
635 $$ = make_redirection (source, r_duplicating_output_word, redir, 0);
636 }
637 | NUMBER GREATER_AND WORD
638 {
639 source.dest = $1;
640 redir.filename = $3;
641 $$ = make_redirection (source, r_duplicating_output_word, redir, 0);
642 }
643 | REDIR_WORD GREATER_AND WORD
644 {
645 source.filename = $1;
646 redir.filename = $3;
647 $$ = make_redirection (source, r_duplicating_output_word, redir, REDIR_VARASSIGN);
648 }
649 | GREATER_AND '-'
650 {
651 source.dest = 1;
652 redir.dest = 0;
653 $$ = make_redirection (source, r_close_this, redir, 0);
654 }
655 | NUMBER GREATER_AND '-'
656 {
657 source.dest = $1;
658 redir.dest = 0;
659 $$ = make_redirection (source, r_close_this, redir, 0);
660 }
661 | REDIR_WORD GREATER_AND '-'
662 {
663 source.filename = $1;
664 redir.dest = 0;
665 $$ = make_redirection (source, r_close_this, redir, REDIR_VARASSIGN);
666 }
667 | LESS_AND '-'
668 {
669 source.dest = 0;
670 redir.dest = 0;
671 $$ = make_redirection (source, r_close_this, redir, 0);
672 }
673 | NUMBER LESS_AND '-'
674 {
675 source.dest = $1;
676 redir.dest = 0;
677 $$ = make_redirection (source, r_close_this, redir, 0);
678 }
679 | REDIR_WORD LESS_AND '-'
680 {
681 source.filename = $1;
682 redir.dest = 0;
683 $$ = make_redirection (source, r_close_this, redir, REDIR_VARASSIGN);
684 }
685 | AND_GREATER WORD
686 {
687 source.dest = 1;
688 redir.filename = $2;
689 $$ = make_redirection (source, r_err_and_out, redir, 0);
690 }
691 | AND_GREATER_GREATER WORD
692 {
693 source.dest = 1;
694 redir.filename = $2;
695 $$ = make_redirection (source, r_append_err_and_out, redir, 0);
696 }
697 ;
698
699 simple_command_element: WORD
700 { $$.word = $1; $$.redirect = 0; }
701 | ASSIGNMENT_WORD
702 { $$.word = $1; $$.redirect = 0; }
703 | redirection
704 { $$.redirect = $1; $$.word = 0; }
705 ;
706
707 redirection_list: redirection
708 {
709 $$ = $1;
710 }
711 | redirection_list redirection
712 {
713 register REDIRECT *t;
714
715 for (t = $1; t->next; t = t->next)
716 ;
717 t->next = $2;
718 $$ = $1;
719 }
720 ;
721
722 simple_command: simple_command_element
723 { $$ = make_simple_command ($1, (COMMAND *)NULL); }
724 | simple_command simple_command_element
725 { $$ = make_simple_command ($2, $1); }
726 ;
727
728 command: simple_command
729 { $$ = clean_simple_command ($1); }
730 | shell_command
731 { $$ = $1; }
732 | shell_command redirection_list
733 {
734 COMMAND *tc;
735
736 tc = $1;
737 if (tc->redirects)
738 {
739 register REDIRECT *t;
740 for (t = tc->redirects; t->next; t = t->next)
741 ;
742 t->next = $2;
743 }
744 else
745 tc->redirects = $2;
746 $$ = $1;
747 }
748 | function_def
749 { $$ = $1; }
750 | coproc
751 { $$ = $1; }
752 ;
753
754 shell_command: for_command
755 { $$ = $1; }
756 | case_command
757 { $$ = $1; }
758 | WHILE compound_list DO compound_list DONE
759 { $$ = make_while_command ($2, $4); }
760 | UNTIL compound_list DO compound_list DONE
761 { $$ = make_until_command ($2, $4); }
762 | select_command
763 { $$ = $1; }
764 | if_command
765 { $$ = $1; }
766 | subshell
767 { $$ = $1; }
768 | group_command
769 { $$ = $1; }
770 | arith_command
771 { $$ = $1; }
772 | cond_command
773 { $$ = $1; }
774 | arith_for_command
775 { $$ = $1; }
776 ;
777
778 for_command: FOR WORD newline_list DO compound_list DONE
779 {
780 $$ = make_for_command ($2, add_string_to_list ("\"$@\"", (WORD_LIST *)NULL), $5, word_lineno[word_top]);
781 if (word_top > 0) word_top--;
782 }
783 | FOR WORD newline_list '{' compound_list '}'
784 {
785 $$ = make_for_command ($2, add_string_to_list ("\"$@\"", (WORD_LIST *)NULL), $5, word_lineno[word_top]);
786 if (word_top > 0) word_top--;
787 }
788 | FOR WORD ';' newline_list DO compound_list DONE
789 {
790 $$ = make_for_command ($2, add_string_to_list ("\"$@\"", (WORD_LIST *)NULL), $6, word_lineno[word_top]);
791 if (word_top > 0) word_top--;
792 }
793 | FOR WORD ';' newline_list '{' compound_list '}'
794 {
795 $$ = make_for_command ($2, add_string_to_list ("\"$@\"", (WORD_LIST *)NULL), $6, word_lineno[word_top]);
796 if (word_top > 0) word_top--;
797 }
798 | FOR WORD newline_list IN word_list list_terminator newline_list DO compound_list DONE
799 {
800 $$ = make_for_command ($2, REVERSE_LIST ($5, WORD_LIST *), $9, word_lineno[word_top]);
801 if (word_top > 0) word_top--;
802 }
803 | FOR WORD newline_list IN word_list list_terminator newline_list '{' compound_list '}'
804 {
805 $$ = make_for_command ($2, REVERSE_LIST ($5, WORD_LIST *), $9, word_lineno[word_top]);
806 if (word_top > 0) word_top--;
807 }
808 | FOR WORD newline_list IN list_terminator newline_list DO compound_list DONE
809 {
810 $$ = make_for_command ($2, (WORD_LIST *)NULL, $8, word_lineno[word_top]);
811 if (word_top > 0) word_top--;
812 }
813 | FOR WORD newline_list IN list_terminator newline_list '{' compound_list '}'
814 {
815 $$ = make_for_command ($2, (WORD_LIST *)NULL, $8, word_lineno[word_top]);
816 if (word_top > 0) word_top--;
817 }
818 ;
819
820 arith_for_command: FOR ARITH_FOR_EXPRS list_terminator newline_list DO compound_list DONE
821 {
822 $$ = make_arith_for_command ($2, $6, arith_for_lineno);
823 if (word_top > 0) word_top--;
824 }
825 | FOR ARITH_FOR_EXPRS list_terminator newline_list '{' compound_list '}'
826 {
827 $$ = make_arith_for_command ($2, $6, arith_for_lineno);
828 if (word_top > 0) word_top--;
829 }
830 | FOR ARITH_FOR_EXPRS DO compound_list DONE
831 {
832 $$ = make_arith_for_command ($2, $4, arith_for_lineno);
833 if (word_top > 0) word_top--;
834 }
835 | FOR ARITH_FOR_EXPRS '{' compound_list '}'
836 {
837 $$ = make_arith_for_command ($2, $4, arith_for_lineno);
838 if (word_top > 0) word_top--;
839 }
840 ;
841
842 select_command: SELECT WORD newline_list DO list DONE
843 {
844 $$ = make_select_command ($2, add_string_to_list ("\"$@\"", (WORD_LIST *)NULL), $5, word_lineno[word_top]);
845 if (word_top > 0) word_top--;
846 }
847 | SELECT WORD newline_list '{' list '}'
848 {
849 $$ = make_select_command ($2, add_string_to_list ("\"$@\"", (WORD_LIST *)NULL), $5, word_lineno[word_top]);
850 if (word_top > 0) word_top--;
851 }
852 | SELECT WORD ';' newline_list DO list DONE
853 {
854 $$ = make_select_command ($2, add_string_to_list ("\"$@\"", (WORD_LIST *)NULL), $6, word_lineno[word_top]);
855 if (word_top > 0) word_top--;
856 }
857 | SELECT WORD ';' newline_list '{' list '}'
858 {
859 $$ = make_select_command ($2, add_string_to_list ("\"$@\"", (WORD_LIST *)NULL), $6, word_lineno[word_top]);
860 if (word_top > 0) word_top--;
861 }
862 | SELECT WORD newline_list IN word_list list_terminator newline_list DO list DONE
863 {
864 $$ = make_select_command ($2, REVERSE_LIST ($5, WORD_LIST *), $9, word_lineno[word_top]);
865 if (word_top > 0) word_top--;
866 }
867 | SELECT WORD newline_list IN word_list list_terminator newline_list '{' list '}'
868 {
869 $$ = make_select_command ($2, REVERSE_LIST ($5, WORD_LIST *), $9, word_lineno[word_top]);
870 if (word_top > 0) word_top--;
871 }
872 ;
873
874 case_command: CASE WORD newline_list IN newline_list ESAC
875 {
876 $$ = make_case_command ($2, (PATTERN_LIST *)NULL, word_lineno[word_top]);
877 if (word_top > 0) word_top--;
878 }
879 | CASE WORD newline_list IN case_clause_sequence newline_list ESAC
880 {
881 $$ = make_case_command ($2, $5, word_lineno[word_top]);
882 if (word_top > 0) word_top--;
883 }
884 | CASE WORD newline_list IN case_clause ESAC
885 {
886 $$ = make_case_command ($2, $5, word_lineno[word_top]);
887 if (word_top > 0) word_top--;
888 }
889 ;
890
891 function_def: WORD '(' ')' newline_list function_body
892 { $$ = make_function_def ($1, $5, function_dstart, function_bstart); }
893
894 | FUNCTION WORD '(' ')' newline_list function_body
895 { $$ = make_function_def ($2, $6, function_dstart, function_bstart); }
896
897 | FUNCTION WORD newline_list function_body
898 { $$ = make_function_def ($2, $4, function_dstart, function_bstart); }
899 ;
900
901 function_body: shell_command
902 { $$ = $1; }
903 | shell_command redirection_list
904 {
905 COMMAND *tc;
906
907 tc = $1;
908 /* According to Posix.2 3.9.5, redirections
909 specified after the body of a function should
910 be attached to the function and performed when
911 the function is executed, not as part of the
912 function definition command. */
913 /* XXX - I don't think it matters, but we might
914 want to change this in the future to avoid
915 problems differentiating between a function
916 definition with a redirection and a function
917 definition containing a single command with a
918 redirection. The two are semantically equivalent,
919 though -- the only difference is in how the
920 command printing code displays the redirections. */
921 if (tc->redirects)
922 {
923 register REDIRECT *t;
924 for (t = tc->redirects; t->next; t = t->next)
925 ;
926 t->next = $2;
927 }
928 else
929 tc->redirects = $2;
930 $$ = $1;
931 }
932 ;
933
934 subshell: '(' compound_list ')'
935 {
936 $$ = make_subshell_command ($2);
937 $$->flags |= CMD_WANT_SUBSHELL;
938 }
939 ;
940
941 coproc: COPROC shell_command
942 {
943 $$ = make_coproc_command ("COPROC", $2);
944 $$->flags |= CMD_WANT_SUBSHELL|CMD_COPROC_SUBSHELL;
945 }
946 | COPROC shell_command redirection_list
947 {
948 COMMAND *tc;
949
950 tc = $2;
951 if (tc->redirects)
952 {
953 register REDIRECT *t;
954 for (t = tc->redirects; t->next; t = t->next)
955 ;
956 t->next = $3;
957 }
958 else
959 tc->redirects = $3;
960 $$ = make_coproc_command ("COPROC", $2);
961 $$->flags |= CMD_WANT_SUBSHELL|CMD_COPROC_SUBSHELL;
962 }
963 | COPROC WORD shell_command
964 {
965 $$ = make_coproc_command ($2->word, $3);
966 $$->flags |= CMD_WANT_SUBSHELL|CMD_COPROC_SUBSHELL;
967 }
968 | COPROC WORD shell_command redirection_list
969 {
970 COMMAND *tc;
971
972 tc = $3;
973 if (tc->redirects)
974 {
975 register REDIRECT *t;
976 for (t = tc->redirects; t->next; t = t->next)
977 ;
978 t->next = $4;
979 }
980 else
981 tc->redirects = $4;
982 $$ = make_coproc_command ($2->word, $3);
983 $$->flags |= CMD_WANT_SUBSHELL|CMD_COPROC_SUBSHELL;
984 }
985 | COPROC simple_command
986 {
987 $$ = make_coproc_command ("COPROC", clean_simple_command ($2));
988 $$->flags |= CMD_WANT_SUBSHELL|CMD_COPROC_SUBSHELL;
989 }
990 ;
991
992 if_command: IF compound_list THEN compound_list FI
993 { $$ = make_if_command ($2, $4, (COMMAND *)NULL); }
994 | IF compound_list THEN compound_list ELSE compound_list FI
995 { $$ = make_if_command ($2, $4, $6); }
996 | IF compound_list THEN compound_list elif_clause FI
997 { $$ = make_if_command ($2, $4, $5); }
998 ;
999
1000
1001 group_command: '{' compound_list '}'
1002 { $$ = make_group_command ($2); }
1003 ;
1004
1005 arith_command: ARITH_CMD
1006 { $$ = make_arith_command ($1); }
1007 ;
1008
1009 cond_command: COND_START COND_CMD COND_END
1010 { $$ = $2; }
1011 ;
1012
1013 elif_clause: ELIF compound_list THEN compound_list
1014 { $$ = make_if_command ($2, $4, (COMMAND *)NULL); }
1015 | ELIF compound_list THEN compound_list ELSE compound_list
1016 { $$ = make_if_command ($2, $4, $6); }
1017 | ELIF compound_list THEN compound_list elif_clause
1018 { $$ = make_if_command ($2, $4, $5); }
1019 ;
1020
1021 case_clause: pattern_list
1022 | case_clause_sequence pattern_list
1023 { $2->next = $1; $$ = $2; }
1024 ;
1025
1026 pattern_list: newline_list pattern ')' compound_list
1027 { $$ = make_pattern_list ($2, $4); }
1028 | newline_list pattern ')' newline_list
1029 { $$ = make_pattern_list ($2, (COMMAND *)NULL); }
1030 | newline_list '(' pattern ')' compound_list
1031 { $$ = make_pattern_list ($3, $5); }
1032 | newline_list '(' pattern ')' newline_list
1033 { $$ = make_pattern_list ($3, (COMMAND *)NULL); }
1034 ;
1035
1036 case_clause_sequence: pattern_list SEMI_SEMI
1037 { $$ = $1; }
1038 | case_clause_sequence pattern_list SEMI_SEMI
1039 { $2->next = $1; $$ = $2; }
1040 | pattern_list SEMI_AND
1041 { $1->flags |= CASEPAT_FALLTHROUGH; $$ = $1; }
1042 | case_clause_sequence pattern_list SEMI_AND
1043 { $2->flags |= CASEPAT_FALLTHROUGH; $2->next = $1; $$ = $2; }
1044 | pattern_list SEMI_SEMI_AND
1045 { $1->flags |= CASEPAT_TESTNEXT; $$ = $1; }
1046 | case_clause_sequence pattern_list SEMI_SEMI_AND
1047 { $2->flags |= CASEPAT_TESTNEXT; $2->next = $1; $$ = $2; }
1048 ;
1049
1050 pattern: WORD
1051 { $$ = make_word_list ($1, (WORD_LIST *)NULL); }
1052 | pattern '|' WORD
1053 { $$ = make_word_list ($3, $1); }
1054 ;
1055
1056 /* A list allows leading or trailing newlines and
1057 newlines as operators (equivalent to semicolons).
1058 It must end with a newline or semicolon.
1059 Lists are used within commands such as if, for, while. */
1060
1061 list: newline_list list0
1062 {
1063 $$ = $2;
1064 if (need_here_doc)
1065 gather_here_documents ();
1066 }
1067 ;
1068
1069 compound_list: list
1070 | newline_list list1
1071 {
1072 $$ = $2;
1073 }
1074 ;
1075
1076 list0: list1 '\n' newline_list
1077 | list1 '&' newline_list
1078 {
1079 if ($1->type == cm_connection)
1080 $$ = connect_async_list ($1, (COMMAND *)NULL, '&');
1081 else
1082 $$ = command_connect ($1, (COMMAND *)NULL, '&');
1083 }
1084 | list1 ';' newline_list
1085
1086 ;
1087
1088 list1: list1 AND_AND newline_list list1
1089 { $$ = command_connect ($1, $4, AND_AND); }
1090 | list1 OR_OR newline_list list1
1091 { $$ = command_connect ($1, $4, OR_OR); }
1092 | list1 '&' newline_list list1
1093 {
1094 if ($1->type == cm_connection)
1095 $$ = connect_async_list ($1, $4, '&');
1096 else
1097 $$ = command_connect ($1, $4, '&');
1098 }
1099 | list1 ';' newline_list list1
1100 { $$ = command_connect ($1, $4, ';'); }
1101 | list1 '\n' newline_list list1
1102 { $$ = command_connect ($1, $4, ';'); }
1103 | pipeline_command
1104 { $$ = $1; }
1105 ;
1106
1107 simple_list_terminator: '\n'
1108 | yacc_EOF
1109 ;
1110
1111 list_terminator:'\n'
1112 { $$ = '\n'; }
1113 | ';'
1114 { $$ = ';'; }
1115 | yacc_EOF
1116 { $$ = yacc_EOF; }
1117 ;
1118
1119 newline_list:
1120 | newline_list '\n'
1121 ;
1122
1123 /* A simple_list is a list that contains no significant newlines
1124 and no leading or trailing newlines. Newlines are allowed
1125 only following operators, where they are not significant.
1126
1127 This is what an inputunit consists of. */
1128
1129 simple_list: simple_list1
1130 {
1131 $$ = $1;
1132 if (need_here_doc)
1133 gather_here_documents ();
1134 if ((parser_state & PST_CMDSUBST) && current_token == shell_eof_token)
1135 {
1136 global_command = $1;
1137 eof_encountered = 0;
1138 rewind_input_string ();
1139 YYACCEPT;
1140 }
1141 }
1142 | simple_list1 '&'
1143 {
1144 if ($1->type == cm_connection)
1145 $$ = connect_async_list ($1, (COMMAND *)NULL, '&');
1146 else
1147 $$ = command_connect ($1, (COMMAND *)NULL, '&');
1148 if (need_here_doc)
1149 gather_here_documents ();
1150 if ((parser_state & PST_CMDSUBST) && current_token == shell_eof_token)
1151 {
1152 global_command = $1;
1153 eof_encountered = 0;
1154 rewind_input_string ();
1155 YYACCEPT;
1156 }
1157 }
1158 | simple_list1 ';'
1159 {
1160 $$ = $1;
1161 if (need_here_doc)
1162 gather_here_documents ();
1163 if ((parser_state & PST_CMDSUBST) && current_token == shell_eof_token)
1164 {
1165 global_command = $1;
1166 eof_encountered = 0;
1167 rewind_input_string ();
1168 YYACCEPT;
1169 }
1170 }
1171 ;
1172
1173 simple_list1: simple_list1 AND_AND newline_list simple_list1
1174 { $$ = command_connect ($1, $4, AND_AND); }
1175 | simple_list1 OR_OR newline_list simple_list1
1176 { $$ = command_connect ($1, $4, OR_OR); }
1177 | simple_list1 '&' simple_list1
1178 {
1179 if ($1->type == cm_connection)
1180 $$ = connect_async_list ($1, $3, '&');
1181 else
1182 $$ = command_connect ($1, $3, '&');
1183 }
1184 | simple_list1 ';' simple_list1
1185 { $$ = command_connect ($1, $3, ';'); }
1186
1187 | pipeline_command
1188 { $$ = $1; }
1189 ;
1190
1191 pipeline_command: pipeline
1192 { $$ = $1; }
1193 | BANG pipeline_command
1194 {
1195 if ($2)
1196 $2->flags ^= CMD_INVERT_RETURN; /* toggle */
1197 $$ = $2;
1198 }
1199 | timespec pipeline_command
1200 {
1201 if ($2)
1202 $2->flags |= $1;
1203 $$ = $2;
1204 }
1205 | timespec list_terminator
1206 {
1207 ELEMENT x;
1208
1209 /* Boy, this is unclean. `time' by itself can
1210 time a null command. We cheat and push a
1211 newline back if the list_terminator was a newline
1212 to avoid the double-newline problem (one to
1213 terminate this, one to terminate the command) */
1214 x.word = 0;
1215 x.redirect = 0;
1216 $$ = make_simple_command (x, (COMMAND *)NULL);
1217 $$->flags |= $1;
1218 /* XXX - let's cheat and push a newline back */
1219 if ($2 == '\n')
1220 token_to_read = '\n';
1221 }
1222 | BANG list_terminator
1223 {
1224 ELEMENT x;
1225
1226 /* This is just as unclean. Posix says that `!'
1227 by itself should be equivalent to `false'.
1228 We cheat and push a
1229 newline back if the list_terminator was a newline
1230 to avoid the double-newline problem (one to
1231 terminate this, one to terminate the command) */
1232 x.word = 0;
1233 x.redirect = 0;
1234 $$ = make_simple_command (x, (COMMAND *)NULL);
1235 $$->flags |= CMD_INVERT_RETURN;
1236 /* XXX - let's cheat and push a newline back */
1237 if ($2 == '\n')
1238 token_to_read = '\n';
1239 }
1240 ;
1241
1242 pipeline: pipeline '|' newline_list pipeline
1243 { $$ = command_connect ($1, $4, '|'); }
1244 | pipeline BAR_AND newline_list pipeline
1245 {
1246 /* Make cmd1 |& cmd2 equivalent to cmd1 2>&1 | cmd2 */
1247 COMMAND *tc;
1248 REDIRECTEE rd, sd;
1249 REDIRECT *r;
1250
1251 tc = $1->type == cm_simple ? (COMMAND *)$1->value.Simple : $1;
1252 sd.dest = 2;
1253 rd.dest = 1;
1254 r = make_redirection (sd, r_duplicating_output, rd, 0);
1255 if (tc->redirects)
1256 {
1257 register REDIRECT *t;
1258 for (t = tc->redirects; t->next; t = t->next)
1259 ;
1260 t->next = r;
1261 }
1262 else
1263 tc->redirects = r;
1264
1265 $$ = command_connect ($1, $4, '|');
1266 }
1267 | command
1268 { $$ = $1; }
1269 ;
1270
1271 timespec: TIME
1272 { $$ = CMD_TIME_PIPELINE; }
1273 | TIME TIMEOPT
1274 { $$ = CMD_TIME_PIPELINE|CMD_TIME_POSIX; }
1275 | TIME TIMEOPT TIMEIGN
1276 { $$ = CMD_TIME_PIPELINE|CMD_TIME_POSIX; }
1277 ;
1278 %%
1279
1280 /* Initial size to allocate for tokens, and the
1281 amount to grow them by. */
1282 #define TOKEN_DEFAULT_INITIAL_SIZE 496
1283 #define TOKEN_DEFAULT_GROW_SIZE 512
1284
1285 /* Should we call prompt_again? */
1286 #define SHOULD_PROMPT() \
1287 (interactive && (bash_input.type == st_stdin || bash_input.type == st_stream))
1288
1289 #if defined (ALIAS)
1290 # define expanding_alias() (pushed_string_list && pushed_string_list->expander)
1291 #else
1292 # define expanding_alias() 0
1293 #endif
1294
1295 /* Global var is non-zero when end of file has been reached. */
1296 int EOF_Reached = 0;
1297
1298 #ifdef DEBUG
1299 static void
1300 debug_parser (i)
1301 int i;
1302 {
1303 #if YYDEBUG != 0
1304 yydebug = i;
1305 #endif
1306 }
1307 #endif
1308
1309 /* yy_getc () returns the next available character from input or EOF.
1310 yy_ungetc (c) makes `c' the next character to read.
1311 init_yy_io (get, unget, type, location) makes the function GET the
1312 installed function for getting the next character, makes UNGET the
1313 installed function for un-getting a character, sets the type of stream
1314 (either string or file) from TYPE, and makes LOCATION point to where
1315 the input is coming from. */
1316
1317 /* Unconditionally returns end-of-file. */
1318 int
1319 return_EOF ()
1320 {
1321 return (EOF);
1322 }
1323
1324 /* Variable containing the current get and unget functions.
1325 See ./input.h for a clearer description. */
1326 BASH_INPUT bash_input;
1327
1328 /* Set all of the fields in BASH_INPUT to NULL. Free bash_input.name if it
1329 is non-null, avoiding a memory leak. */
1330 void
1331 initialize_bash_input ()
1332 {
1333 bash_input.type = st_none;
1334 FREE (bash_input.name);
1335 bash_input.name = (char *)NULL;
1336 bash_input.location.file = (FILE *)NULL;
1337 bash_input.location.string = (char *)NULL;
1338 bash_input.getter = (sh_cget_func_t *)NULL;
1339 bash_input.ungetter = (sh_cunget_func_t *)NULL;
1340 }
1341
1342 /* Set the contents of the current bash input stream from
1343 GET, UNGET, TYPE, NAME, and LOCATION. */
1344 void
1345 init_yy_io (get, unget, type, name, location)
1346 sh_cget_func_t *get;
1347 sh_cunget_func_t *unget;
1348 enum stream_type type;
1349 const char *name;
1350 INPUT_STREAM location;
1351 {
1352 bash_input.type = type;
1353 FREE (bash_input.name);
1354 bash_input.name = name ? savestring (name) : (char *)NULL;
1355
1356 /* XXX */
1357 #if defined (CRAY)
1358 memcpy((char *)&bash_input.location.string, (char *)&location.string, sizeof(location));
1359 #else
1360 bash_input.location = location;
1361 #endif
1362 bash_input.getter = get;
1363 bash_input.ungetter = unget;
1364 }
1365
1366 char *
1367 yy_input_name ()
1368 {
1369 return (bash_input.name ? bash_input.name : "stdin");
1370 }
1371
1372 /* Call this to get the next character of input. */
1373 static int
1374 yy_getc ()
1375 {
1376 return (*(bash_input.getter)) ();
1377 }
1378
1379 /* Call this to unget C. That is, to make C the next character
1380 to be read. */
1381 static int
1382 yy_ungetc (c)
1383 int c;
1384 {
1385 return (*(bash_input.ungetter)) (c);
1386 }
1387
1388 #if defined (BUFFERED_INPUT)
1389 #ifdef INCLUDE_UNUSED
1390 int
1391 input_file_descriptor ()
1392 {
1393 switch (bash_input.type)
1394 {
1395 case st_stream:
1396 return (fileno (bash_input.location.file));
1397 case st_bstream:
1398 return (bash_input.location.buffered_fd);
1399 case st_stdin:
1400 default:
1401 return (fileno (stdin));
1402 }
1403 }
1404 #endif
1405 #endif /* BUFFERED_INPUT */
1406
1407 /* **************************************************************** */
1408 /* */
1409 /* Let input be read from readline (). */
1410 /* */
1411 /* **************************************************************** */
1412
1413 #if defined (READLINE)
1414 char *current_readline_prompt = (char *)NULL;
1415 char *current_readline_line = (char *)NULL;
1416 int current_readline_line_index = 0;
1417
1418 static int
1419 yy_readline_get ()
1420 {
1421 SigHandler *old_sigint;
1422 int line_len;
1423 unsigned char c;
1424
1425 if (!current_readline_line)
1426 {
1427 if (!bash_readline_initialized)
1428 initialize_readline ();
1429
1430 #if defined (JOB_CONTROL)
1431 if (job_control)
1432 give_terminal_to (shell_pgrp, 0);
1433 #endif /* JOB_CONTROL */
1434
1435 old_sigint = (SigHandler *)IMPOSSIBLE_TRAP_HANDLER;
1436 if (signal_is_ignored (SIGINT) == 0)
1437 {
1438 interrupt_immediately++;
1439 old_sigint = (SigHandler *)set_signal_handler (SIGINT, sigint_sighandler);
1440 }
1441 terminate_immediately = 1;
1442
1443 current_readline_line = readline (current_readline_prompt ?
1444 current_readline_prompt : "");
1445
1446 terminate_immediately = 0;
1447 if (signal_is_ignored (SIGINT) == 0)
1448 {
1449 interrupt_immediately--;
1450 if (old_sigint != IMPOSSIBLE_TRAP_HANDLER)
1451 set_signal_handler (SIGINT, old_sigint);
1452 }
1453
1454 #if 0
1455 /* Reset the prompt to the decoded value of prompt_string_pointer. */
1456 reset_readline_prompt ();
1457 #endif
1458
1459 if (current_readline_line == 0)
1460 return (EOF);
1461
1462 current_readline_line_index = 0;
1463 line_len = strlen (current_readline_line);
1464
1465 current_readline_line = (char *)xrealloc (current_readline_line, 2 + line_len);
1466 current_readline_line[line_len++] = '\n';
1467 current_readline_line[line_len] = '\0';
1468 }
1469
1470 if (current_readline_line[current_readline_line_index] == 0)
1471 {
1472 free (current_readline_line);
1473 current_readline_line = (char *)NULL;
1474 return (yy_readline_get ());
1475 }
1476 else
1477 {
1478 c = current_readline_line[current_readline_line_index++];
1479 return (c);
1480 }
1481 }
1482
1483 static int
1484 yy_readline_unget (c)
1485 int c;
1486 {
1487 if (current_readline_line_index && current_readline_line)
1488 current_readline_line[--current_readline_line_index] = c;
1489 return (c);
1490 }
1491
1492 void
1493 with_input_from_stdin ()
1494 {
1495 INPUT_STREAM location;
1496
1497 if (bash_input.type != st_stdin && stream_on_stack (st_stdin) == 0)
1498 {
1499 location.string = current_readline_line;
1500 init_yy_io (yy_readline_get, yy_readline_unget,
1501 st_stdin, "readline stdin", location);
1502 }
1503 }
1504
1505 #else /* !READLINE */
1506
1507 void
1508 with_input_from_stdin ()
1509 {
1510 with_input_from_stream (stdin, "stdin");
1511 }
1512 #endif /* !READLINE */
1513
1514 /* **************************************************************** */
1515 /* */
1516 /* Let input come from STRING. STRING is zero terminated. */
1517 /* */
1518 /* **************************************************************** */
1519
1520 static int
1521 yy_string_get ()
1522 {
1523 register char *string;
1524 register unsigned char c;
1525
1526 string = bash_input.location.string;
1527
1528 /* If the string doesn't exist, or is empty, EOF found. */
1529 if (string && *string)
1530 {
1531 c = *string++;
1532 bash_input.location.string = string;
1533 return (c);
1534 }
1535 else
1536 return (EOF);
1537 }
1538
1539 static int
1540 yy_string_unget (c)
1541 int c;
1542 {
1543 *(--bash_input.location.string) = c;
1544 return (c);
1545 }
1546
1547 void
1548 with_input_from_string (string, name)
1549 char *string;
1550 const char *name;
1551 {
1552 INPUT_STREAM location;
1553
1554 location.string = string;
1555 init_yy_io (yy_string_get, yy_string_unget, st_string, name, location);
1556 }
1557
1558 /* Count the number of characters we've consumed from bash_input.location.string
1559 and read into shell_input_line, but have not returned from shell_getc.
1560 That is the true input location. Rewind bash_input.location.string by
1561 that number of characters, so it points to the last character actually
1562 consumed by the parser. */
1563 static void
1564 rewind_input_string ()
1565 {
1566 int xchars;
1567
1568 /* number of unconsumed characters in the input -- XXX need to take newlines
1569 into account, e.g., $(...\n) */
1570 xchars = shell_input_line_len - shell_input_line_index;
1571 if (bash_input.location.string[-1] == '\n')
1572 xchars++;
1573
1574 /* XXX - how to reflect bash_input.location.string back to string passed to
1575 parse_and_execute or xparse_dolparen? xparse_dolparen needs to know how
1576 far into the string we parsed. parse_and_execute knows where bash_input.
1577 location.string is, and how far from orig_string that is -- that's the
1578 number of characters the command consumed. */
1579
1580 /* bash_input.location.string - xchars should be where we parsed to */
1581 /* need to do more validation on xchars value for sanity -- test cases. */
1582 bash_input.location.string -= xchars;
1583 }
1584
1585 /* **************************************************************** */
1586 /* */
1587 /* Let input come from STREAM. */
1588 /* */
1589 /* **************************************************************** */
1590
1591 /* These two functions used to test the value of the HAVE_RESTARTABLE_SYSCALLS
1592 define, and just use getc/ungetc if it was defined, but since bash
1593 installs its signal handlers without the SA_RESTART flag, some signals
1594 (like SIGCHLD, SIGWINCH, etc.) received during a read(2) will not cause
1595 the read to be restarted. We need to restart it ourselves. */
1596
1597 static int
1598 yy_stream_get ()
1599 {
1600 int result;
1601
1602 result = EOF;
1603 if (bash_input.location.file)
1604 {
1605 if (interactive)
1606 {
1607 interrupt_immediately++;
1608 terminate_immediately++;
1609 }
1610 result = getc_with_restart (bash_input.location.file);
1611 if (interactive)
1612 {
1613 interrupt_immediately--;
1614 terminate_immediately--;
1615 }
1616 }
1617 return (result);
1618 }
1619
1620 static int
1621 yy_stream_unget (c)
1622 int c;
1623 {
1624 return (ungetc_with_restart (c, bash_input.location.file));
1625 }
1626
1627 void
1628 with_input_from_stream (stream, name)
1629 FILE *stream;
1630 const char *name;
1631 {
1632 INPUT_STREAM location;
1633
1634 location.file = stream;
1635 init_yy_io (yy_stream_get, yy_stream_unget, st_stream, name, location);
1636 }
1637
1638 typedef struct stream_saver {
1639 struct stream_saver *next;
1640 BASH_INPUT bash_input;
1641 int line;
1642 #if defined (BUFFERED_INPUT)
1643 BUFFERED_STREAM *bstream;
1644 #endif /* BUFFERED_INPUT */
1645 } STREAM_SAVER;
1646
1647 /* The globally known line number. */
1648 int line_number = 0;
1649
1650 /* The line number offset set by assigning to LINENO. Not currently used. */
1651 int line_number_base = 0;
1652
1653 #if defined (COND_COMMAND)
1654 static int cond_lineno;
1655 static int cond_token;
1656 #endif
1657
1658 STREAM_SAVER *stream_list = (STREAM_SAVER *)NULL;
1659
1660 void
1661 push_stream (reset_lineno)
1662 int reset_lineno;
1663 {
1664 STREAM_SAVER *saver = (STREAM_SAVER *)xmalloc (sizeof (STREAM_SAVER));
1665
1666 xbcopy ((char *)&bash_input, (char *)&(saver->bash_input), sizeof (BASH_INPUT));
1667
1668 #if defined (BUFFERED_INPUT)
1669 saver->bstream = (BUFFERED_STREAM *)NULL;
1670 /* If we have a buffered stream, clear out buffers[fd]. */
1671 if (bash_input.type == st_bstream && bash_input.location.buffered_fd >= 0)
1672 saver->bstream = set_buffered_stream (bash_input.location.buffered_fd,
1673 (BUFFERED_STREAM *)NULL);
1674 #endif /* BUFFERED_INPUT */
1675
1676 saver->line = line_number;
1677 bash_input.name = (char *)NULL;
1678 saver->next = stream_list;
1679 stream_list = saver;
1680 EOF_Reached = 0;
1681 if (reset_lineno)
1682 line_number = 0;
1683 }
1684
1685 void
1686 pop_stream ()
1687 {
1688 if (!stream_list)
1689 EOF_Reached = 1;
1690 else
1691 {
1692 STREAM_SAVER *saver = stream_list;
1693
1694 EOF_Reached = 0;
1695 stream_list = stream_list->next;
1696
1697 init_yy_io (saver->bash_input.getter,
1698 saver->bash_input.ungetter,
1699 saver->bash_input.type,
1700 saver->bash_input.name,
1701 saver->bash_input.location);
1702
1703 #if defined (BUFFERED_INPUT)
1704 /* If we have a buffered stream, restore buffers[fd]. */
1705 /* If the input file descriptor was changed while this was on the
1706 save stack, update the buffered fd to the new file descriptor and
1707 re-establish the buffer <-> bash_input fd correspondence. */
1708 if (bash_input.type == st_bstream && bash_input.location.buffered_fd >= 0)
1709 {
1710 if (bash_input_fd_changed)
1711 {
1712 bash_input_fd_changed = 0;
1713 if (default_buffered_input >= 0)
1714 {
1715 bash_input.location.buffered_fd = default_buffered_input;
1716 saver->bstream->b_fd = default_buffered_input;
1717 SET_CLOSE_ON_EXEC (default_buffered_input);
1718 }
1719 }
1720 /* XXX could free buffered stream returned as result here. */
1721 set_buffered_stream (bash_input.location.buffered_fd, saver->bstream);
1722 }
1723 #endif /* BUFFERED_INPUT */
1724
1725 line_number = saver->line;
1726
1727 FREE (saver->bash_input.name);
1728 free (saver);
1729 }
1730 }
1731
1732 /* Return 1 if a stream of type TYPE is saved on the stack. */
1733 int
1734 stream_on_stack (type)
1735 enum stream_type type;
1736 {
1737 register STREAM_SAVER *s;
1738
1739 for (s = stream_list; s; s = s->next)
1740 if (s->bash_input.type == type)
1741 return 1;
1742 return 0;
1743 }
1744
1745 /* Save the current token state and return it in a malloced array. */
1746 int *
1747 save_token_state ()
1748 {
1749 int *ret;
1750
1751 ret = (int *)xmalloc (4 * sizeof (int));
1752 ret[0] = last_read_token;
1753 ret[1] = token_before_that;
1754 ret[2] = two_tokens_ago;
1755 ret[3] = current_token;
1756 return ret;
1757 }
1758
1759 void
1760 restore_token_state (ts)
1761 int *ts;
1762 {
1763 if (ts == 0)
1764 return;
1765 last_read_token = ts[0];
1766 token_before_that = ts[1];
1767 two_tokens_ago = ts[2];
1768 current_token = ts[3];
1769 }
1770
1771 /*
1772 * This is used to inhibit alias expansion and reserved word recognition
1773 * inside case statement pattern lists. A `case statement pattern list' is:
1774 *
1775 * everything between the `in' in a `case word in' and the next ')'
1776 * or `esac'
1777 * everything between a `;;' and the next `)' or `esac'
1778 */
1779
1780 #if defined (ALIAS) || defined (DPAREN_ARITHMETIC)
1781
1782 #define END_OF_ALIAS 0
1783
1784 /*
1785 * Pseudo-global variables used in implementing token-wise alias expansion.
1786 */
1787
1788 /*
1789 * Pushing and popping strings. This works together with shell_getc to
1790 * implement alias expansion on a per-token basis.
1791 */
1792
1793 typedef struct string_saver {
1794 struct string_saver *next;
1795 int expand_alias; /* Value to set expand_alias to when string is popped. */
1796 char *saved_line;
1797 #if defined (ALIAS)
1798 alias_t *expander; /* alias that caused this line to be pushed. */
1799 #endif
1800 int saved_line_size, saved_line_index, saved_line_terminator;
1801 } STRING_SAVER;
1802
1803 STRING_SAVER *pushed_string_list = (STRING_SAVER *)NULL;
1804
1805 /*
1806 * Push the current shell_input_line onto a stack of such lines and make S
1807 * the current input. Used when expanding aliases. EXPAND is used to set
1808 * the value of expand_next_token when the string is popped, so that the
1809 * word after the alias in the original line is handled correctly when the
1810 * alias expands to multiple words. TOKEN is the token that was expanded
1811 * into S; it is saved and used to prevent infinite recursive expansion.
1812 */
1813 static void
1814 push_string (s, expand, ap)
1815 char *s;
1816 int expand;
1817 alias_t *ap;
1818 {
1819 STRING_SAVER *temp = (STRING_SAVER *)xmalloc (sizeof (STRING_SAVER));
1820
1821 temp->expand_alias = expand;
1822 temp->saved_line = shell_input_line;
1823 temp->saved_line_size = shell_input_line_size;
1824 temp->saved_line_index = shell_input_line_index;
1825 temp->saved_line_terminator = shell_input_line_terminator;
1826 #if defined (ALIAS)
1827 temp->expander = ap;
1828 #endif
1829 temp->next = pushed_string_list;
1830 pushed_string_list = temp;
1831
1832 #if defined (ALIAS)
1833 if (ap)
1834 ap->flags |= AL_BEINGEXPANDED;
1835 #endif
1836
1837 shell_input_line = s;
1838 shell_input_line_size = strlen (s);
1839 shell_input_line_index = 0;
1840 shell_input_line_terminator = '\0';
1841 #if 0
1842 parser_state &= ~PST_ALEXPNEXT; /* XXX */
1843 #endif
1844
1845 set_line_mbstate ();
1846 }
1847
1848 /*
1849 * Make the top of the pushed_string stack be the current shell input.
1850 * Only called when there is something on the stack. Called from shell_getc
1851 * when it thinks it has consumed the string generated by an alias expansion
1852 * and needs to return to the original input line.
1853 */
1854 static void
1855 pop_string ()
1856 {
1857 STRING_SAVER *t;
1858
1859 FREE (shell_input_line);
1860 shell_input_line = pushed_string_list->saved_line;
1861 shell_input_line_index = pushed_string_list->saved_line_index;
1862 shell_input_line_size = pushed_string_list->saved_line_size;
1863 shell_input_line_terminator = pushed_string_list->saved_line_terminator;
1864
1865 if (pushed_string_list->expand_alias)
1866 parser_state |= PST_ALEXPNEXT;
1867 else
1868 parser_state &= ~PST_ALEXPNEXT;
1869
1870 t = pushed_string_list;
1871 pushed_string_list = pushed_string_list->next;
1872
1873 #if defined (ALIAS)
1874 if (t->expander)
1875 t->expander->flags &= ~AL_BEINGEXPANDED;
1876 #endif
1877
1878 free ((char *)t);
1879
1880 set_line_mbstate ();
1881 }
1882
1883 static void
1884 free_string_list ()
1885 {
1886 register STRING_SAVER *t, *t1;
1887
1888 for (t = pushed_string_list; t; )
1889 {
1890 t1 = t->next;
1891 FREE (t->saved_line);
1892 #if defined (ALIAS)
1893 if (t->expander)
1894 t->expander->flags &= ~AL_BEINGEXPANDED;
1895 #endif
1896 free ((char *)t);
1897 t = t1;
1898 }
1899 pushed_string_list = (STRING_SAVER *)NULL;
1900 }
1901
1902 #endif /* ALIAS || DPAREN_ARITHMETIC */
1903
1904 void
1905 free_pushed_string_input ()
1906 {
1907 #if defined (ALIAS) || defined (DPAREN_ARITHMETIC)
1908 free_string_list ();
1909 #endif
1910 }
1911
1912 /* Return a line of text, taken from wherever yylex () reads input.
1913 If there is no more input, then we return NULL. If REMOVE_QUOTED_NEWLINE
1914 is non-zero, we remove unquoted \<newline> pairs. This is used by
1915 read_secondary_line to read here documents. */
1916 static char *
1917 read_a_line (remove_quoted_newline)
1918 int remove_quoted_newline;
1919 {
1920 static char *line_buffer = (char *)NULL;
1921 static int buffer_size = 0;
1922 int indx, c, peekc, pass_next;
1923
1924 #if defined (READLINE)
1925 if (no_line_editing && SHOULD_PROMPT ())
1926 #else
1927 if (SHOULD_PROMPT ())
1928 #endif
1929 print_prompt ();
1930
1931 pass_next = indx = 0;
1932 while (1)
1933 {
1934 /* Allow immediate exit if interrupted during input. */
1935 QUIT;
1936
1937 c = yy_getc ();
1938
1939 /* Ignore null bytes in input. */
1940 if (c == 0)
1941 {
1942 #if 0
1943 internal_warning ("read_a_line: ignored null byte in input");
1944 #endif
1945 continue;
1946 }
1947
1948 /* If there is no more input, then we return NULL. */
1949 if (c == EOF)
1950 {
1951 if (interactive && bash_input.type == st_stream)
1952 clearerr (stdin);
1953 if (indx == 0)
1954 return ((char *)NULL);
1955 c = '\n';
1956 }
1957
1958 /* `+2' in case the final character in the buffer is a newline. */
1959 RESIZE_MALLOCED_BUFFER (line_buffer, indx, 2, buffer_size, 128);
1960
1961 /* IF REMOVE_QUOTED_NEWLINES is non-zero, we are reading a
1962 here document with an unquoted delimiter. In this case,
1963 the line will be expanded as if it were in double quotes.
1964 We allow a backslash to escape the next character, but we
1965 need to treat the backslash specially only if a backslash
1966 quoting a backslash-newline pair appears in the line. */
1967 if (pass_next)
1968 {
1969 line_buffer[indx++] = c;
1970 pass_next = 0;
1971 }
1972 else if (c == '\\' && remove_quoted_newline)
1973 {
1974 QUIT;
1975 peekc = yy_getc ();
1976 if (peekc == '\n')
1977 {
1978 line_number++;
1979 continue; /* Make the unquoted \<newline> pair disappear. */
1980 }
1981 else
1982 {
1983 yy_ungetc (peekc);
1984 pass_next = 1;
1985 line_buffer[indx++] = c; /* Preserve the backslash. */
1986 }
1987 }
1988 else
1989 line_buffer[indx++] = c;
1990
1991 if (c == '\n')
1992 {
1993 line_buffer[indx] = '\0';
1994 return (line_buffer);
1995 }
1996 }
1997 }
1998
1999 /* Return a line as in read_a_line (), but insure that the prompt is
2000 the secondary prompt. This is used to read the lines of a here
2001 document. REMOVE_QUOTED_NEWLINE is non-zero if we should remove
2002 newlines quoted with backslashes while reading the line. It is
2003 non-zero unless the delimiter of the here document was quoted. */
2004 char *
2005 read_secondary_line (remove_quoted_newline)
2006 int remove_quoted_newline;
2007 {
2008 char *ret;
2009 int n, c;
2010
2011 prompt_string_pointer = &ps2_prompt;
2012 if (SHOULD_PROMPT())
2013 prompt_again ();
2014 ret = read_a_line (remove_quoted_newline);
2015 #if defined (HISTORY)
2016 if (ret && remember_on_history && (parser_state & PST_HEREDOC))
2017 {
2018 /* To make adding the the here-document body right, we need to rely
2019 on history_delimiting_chars() returning \n for the first line of
2020 the here-document body and the null string for the second and
2021 subsequent lines, so we avoid double newlines.
2022 current_command_line_count == 2 for the first line of the body. */
2023
2024 current_command_line_count++;
2025 maybe_add_history (ret);
2026 }
2027 #endif /* HISTORY */
2028 return ret;
2029 }
2030
2031 /* **************************************************************** */
2032 /* */
2033 /* YYLEX () */
2034 /* */
2035 /* **************************************************************** */
2036
2037 /* Reserved words. These are only recognized as the first word of a
2038 command. */
2039 STRING_INT_ALIST word_token_alist[] = {
2040 { "if", IF },
2041 { "then", THEN },
2042 { "else", ELSE },
2043 { "elif", ELIF },
2044 { "fi", FI },
2045 { "case", CASE },
2046 { "esac", ESAC },
2047 { "for", FOR },
2048 #if defined (SELECT_COMMAND)
2049 { "select", SELECT },
2050 #endif
2051 { "while", WHILE },
2052 { "until", UNTIL },
2053 { "do", DO },
2054 { "done", DONE },
2055 { "in", IN },
2056 { "function", FUNCTION },
2057 #if defined (COMMAND_TIMING)
2058 { "time", TIME },
2059 #endif
2060 { "{", '{' },
2061 { "}", '}' },
2062 { "!", BANG },
2063 #if defined (COND_COMMAND)
2064 { "[[", COND_START },
2065 { "]]", COND_END },
2066 #endif
2067 #if defined (COPROCESS_SUPPORT)
2068 { "coproc", COPROC },
2069 #endif
2070 { (char *)NULL, 0}
2071 };
2072
2073 /* other tokens that can be returned by read_token() */
2074 STRING_INT_ALIST other_token_alist[] = {
2075 /* Multiple-character tokens with special values */
2076 { "--", TIMEIGN },
2077 { "-p", TIMEOPT },
2078 { "&&", AND_AND },
2079 { "||", OR_OR },
2080 { ">>", GREATER_GREATER },
2081 { "<<", LESS_LESS },
2082 { "<&", LESS_AND },
2083 { ">&", GREATER_AND },
2084 { ";;", SEMI_SEMI },
2085 { ";&", SEMI_AND },
2086 { ";;&", SEMI_SEMI_AND },
2087 { "<<-", LESS_LESS_MINUS },
2088 { "<<<", LESS_LESS_LESS },
2089 { "&>", AND_GREATER },
2090 { "&>>", AND_GREATER_GREATER },
2091 { "<>", LESS_GREATER },
2092 { ">|", GREATER_BAR },
2093 { "|&", BAR_AND },
2094 { "EOF", yacc_EOF },
2095 /* Tokens whose value is the character itself */
2096 { ">", '>' },
2097 { "<", '<' },
2098 { "-", '-' },
2099 { "{", '{' },
2100 { "}", '}' },
2101 { ";", ';' },
2102 { "(", '(' },
2103 { ")", ')' },
2104 { "|", '|' },
2105 { "&", '&' },
2106 { "newline", '\n' },
2107 { (char *)NULL, 0}
2108 };
2109
2110 /* others not listed here:
2111 WORD look at yylval.word
2112 ASSIGNMENT_WORD look at yylval.word
2113 NUMBER look at yylval.number
2114 ARITH_CMD look at yylval.word_list
2115 ARITH_FOR_EXPRS look at yylval.word_list
2116 COND_CMD look at yylval.command
2117 */
2118
2119 /* These are used by read_token_word, but appear up here so that shell_getc
2120 can use them to decide when to add otherwise blank lines to the history. */
2121
2122 /* The primary delimiter stack. */
2123 struct dstack dstack = { (char *)NULL, 0, 0 };
2124
2125 /* A temporary delimiter stack to be used when decoding prompt strings.
2126 This is needed because command substitutions in prompt strings (e.g., PS2)
2127 can screw up the parser's quoting state. */
2128 static struct dstack temp_dstack = { (char *)NULL, 0, 0 };
2129
2130 /* Macro for accessing the top delimiter on the stack. Returns the
2131 delimiter or zero if none. */
2132 #define current_delimiter(ds) \
2133 (ds.delimiter_depth ? ds.delimiters[ds.delimiter_depth - 1] : 0)
2134
2135 #define push_delimiter(ds, character) \
2136 do \
2137 { \
2138 if (ds.delimiter_depth + 2 > ds.delimiter_space) \
2139 ds.delimiters = (char *)xrealloc \
2140 (ds.delimiters, (ds.delimiter_space += 10) * sizeof (char)); \
2141 ds.delimiters[ds.delimiter_depth] = character; \
2142 ds.delimiter_depth++; \
2143 } \
2144 while (0)
2145
2146 #define pop_delimiter(ds) ds.delimiter_depth--
2147
2148 /* Return the next shell input character. This always reads characters
2149 from shell_input_line; when that line is exhausted, it is time to
2150 read the next line. This is called by read_token when the shell is
2151 processing normal command input. */
2152
2153 /* This implements one-character lookahead/lookbehind across physical input
2154 lines, to avoid something being lost because it's pushed back with
2155 shell_ungetc when we're at the start of a line. */
2156 static int eol_ungetc_lookahead = 0;
2157
2158 static int
2159 shell_getc (remove_quoted_newline)
2160 int remove_quoted_newline;
2161 {
2162 register int i;
2163 int c;
2164 unsigned char uc;
2165
2166 QUIT;
2167
2168 if (sigwinch_received)
2169 {
2170 sigwinch_received = 0;
2171 get_new_window_size (0, (int *)0, (int *)0);
2172 }
2173
2174 if (eol_ungetc_lookahead)
2175 {
2176 c = eol_ungetc_lookahead;
2177 eol_ungetc_lookahead = 0;
2178 return (c);
2179 }
2180
2181 #if defined (ALIAS) || defined (DPAREN_ARITHMETIC)
2182 /* If shell_input_line[shell_input_line_index] == 0, but there is
2183 something on the pushed list of strings, then we don't want to go
2184 off and get another line. We let the code down below handle it. */
2185
2186 if (!shell_input_line || ((!shell_input_line[shell_input_line_index]) &&
2187 (pushed_string_list == (STRING_SAVER *)NULL)))
2188 #else /* !ALIAS && !DPAREN_ARITHMETIC */
2189 if (!shell_input_line || !shell_input_line[shell_input_line_index])
2190 #endif /* !ALIAS && !DPAREN_ARITHMETIC */
2191 {
2192 line_number++;
2193
2194 restart_read:
2195
2196 /* Allow immediate exit if interrupted during input. */
2197 QUIT;
2198
2199 i = 0;
2200 shell_input_line_terminator = 0;
2201
2202 /* If the shell is interatctive, but not currently printing a prompt
2203 (interactive_shell && interactive == 0), we don't want to print
2204 notifies or cleanup the jobs -- we want to defer it until we do
2205 print the next prompt. */
2206 if (interactive_shell == 0 || SHOULD_PROMPT())
2207 {
2208 #if defined (JOB_CONTROL)
2209 /* This can cause a problem when reading a command as the result
2210 of a trap, when the trap is called from flush_child. This call
2211 had better not cause jobs to disappear from the job table in
2212 that case, or we will have big trouble. */
2213 notify_and_cleanup ();
2214 #else /* !JOB_CONTROL */
2215 cleanup_dead_jobs ();
2216 #endif /* !JOB_CONTROL */
2217 }
2218
2219 #if defined (READLINE)
2220 if (no_line_editing && SHOULD_PROMPT())
2221 #else
2222 if (SHOULD_PROMPT())
2223 #endif
2224 print_prompt ();
2225
2226 if (bash_input.type == st_stream)
2227 clearerr (stdin);
2228
2229 while (1)
2230 {
2231 c = yy_getc ();
2232
2233 /* Allow immediate exit if interrupted during input. */
2234 QUIT;
2235
2236 if (c == '\0')
2237 {
2238 #if 0
2239 internal_warning ("shell_getc: ignored null byte in input");
2240 #endif
2241 continue;
2242 }
2243
2244 RESIZE_MALLOCED_BUFFER (shell_input_line, i, 2, shell_input_line_size, 256);
2245
2246 if (c == EOF)
2247 {
2248 if (bash_input.type == st_stream)
2249 clearerr (stdin);
2250
2251 if (i == 0)
2252 shell_input_line_terminator = EOF;
2253
2254 shell_input_line[i] = '\0';
2255 break;
2256 }
2257
2258 shell_input_line[i++] = c;
2259
2260 if (c == '\n')
2261 {
2262 shell_input_line[--i] = '\0';
2263 current_command_line_count++;
2264 break;
2265 }
2266 }
2267
2268 shell_input_line_index = 0;
2269 shell_input_line_len = i; /* == strlen (shell_input_line) */
2270
2271 set_line_mbstate ();
2272
2273 #if defined (HISTORY)
2274 if (remember_on_history && shell_input_line && shell_input_line[0])
2275 {
2276 char *expansions;
2277 # if defined (BANG_HISTORY)
2278 int old_hist;
2279
2280 /* If the current delimiter is a single quote, we should not be
2281 performing history expansion, even if we're on a different
2282 line from the original single quote. */
2283 old_hist = history_expansion_inhibited;
2284 if (current_delimiter (dstack) == '\'')
2285 history_expansion_inhibited = 1;
2286 # endif
2287 expansions = pre_process_line (shell_input_line, 1, 1);
2288 # if defined (BANG_HISTORY)
2289 history_expansion_inhibited = old_hist;
2290 # endif
2291 if (expansions != shell_input_line)
2292 {
2293 free (shell_input_line);
2294 shell_input_line = expansions;
2295 shell_input_line_len = shell_input_line ?
2296 strlen (shell_input_line) : 0;
2297 if (shell_input_line_len == 0)
2298 current_command_line_count--;
2299
2300 /* We have to force the xrealloc below because we don't know
2301 the true allocated size of shell_input_line anymore. */
2302 shell_input_line_size = shell_input_line_len;
2303
2304 set_line_mbstate ();
2305 }
2306 }
2307 /* Try to do something intelligent with blank lines encountered while
2308 entering multi-line commands. XXX - this is grotesque */
2309 else if (remember_on_history && shell_input_line &&
2310 shell_input_line[0] == '\0' &&
2311 current_command_line_count > 1)
2312 {
2313 if (current_delimiter (dstack))
2314 /* We know shell_input_line[0] == 0 and we're reading some sort of
2315 quoted string. This means we've got a line consisting of only
2316 a newline in a quoted string. We want to make sure this line
2317 gets added to the history. */
2318 maybe_add_history (shell_input_line);
2319 else
2320 {
2321 char *hdcs;
2322 hdcs = history_delimiting_chars (shell_input_line);
2323 if (hdcs && hdcs[0] == ';')
2324 maybe_add_history (shell_input_line);
2325 }
2326 }
2327
2328 #endif /* HISTORY */
2329
2330 if (shell_input_line)
2331 {
2332 /* Lines that signify the end of the shell's input should not be
2333 echoed. */
2334 if (echo_input_at_read && (shell_input_line[0] ||
2335 shell_input_line_terminator != EOF))
2336 fprintf (stderr, "%s\n", shell_input_line);
2337 }
2338 else
2339 {
2340 shell_input_line_size = 0;
2341 prompt_string_pointer = &current_prompt_string;
2342 if (SHOULD_PROMPT ())
2343 prompt_again ();
2344 goto restart_read;
2345 }
2346
2347 /* Add the newline to the end of this string, iff the string does
2348 not already end in an EOF character. */
2349 if (shell_input_line_terminator != EOF)
2350 {
2351 if (shell_input_line_len + 3 > shell_input_line_size)
2352 shell_input_line = (char *)xrealloc (shell_input_line,
2353 1 + (shell_input_line_size += 2));
2354
2355 shell_input_line[shell_input_line_len] = '\n';
2356 shell_input_line[shell_input_line_len + 1] = '\0';
2357
2358 set_line_mbstate ();
2359 }
2360 }
2361
2362 next_alias_char:
2363 uc = shell_input_line[shell_input_line_index];
2364
2365 if (uc)
2366 shell_input_line_index++;
2367
2368 #if defined (ALIAS) || defined (DPAREN_ARITHMETIC)
2369 /* If UC is NULL, we have reached the end of the current input string. If
2370 pushed_string_list is non-empty, it's time to pop to the previous string
2371 because we have fully consumed the result of the last alias expansion.
2372 Do it transparently; just return the next character of the string popped
2373 to. */
2374 pop_alias:
2375 if (uc == 0 && (pushed_string_list != (STRING_SAVER *)NULL))
2376 {
2377 pop_string ();
2378 uc = shell_input_line[shell_input_line_index];
2379 if (uc)
2380 shell_input_line_index++;
2381 }
2382 #endif /* ALIAS || DPAREN_ARITHMETIC */
2383
2384 if MBTEST(uc == '\\' && remove_quoted_newline && shell_input_line[shell_input_line_index] == '\n')
2385 {
2386 if (SHOULD_PROMPT ())
2387 prompt_again ();
2388 line_number++;
2389 /* What do we do here if we're expanding an alias whose definition
2390 includes an escaped newline? If that's the last character in the
2391 alias expansion, we just pop the pushed string list (recall that
2392 we inhibit the appending of a space in mk_alexpansion() if newline
2393 is the last character). If it's not the last character, we need
2394 to consume the quoted newline and move to the next character in
2395 the expansion. */
2396 #if defined (ALIAS)
2397 if (expanding_alias () && shell_input_line[shell_input_line_index+1] == '\0')
2398 {
2399 uc = 0;
2400 goto pop_alias;
2401 }
2402 else if (expanding_alias () && shell_input_line[shell_input_line_index+1] != '\0')
2403 {
2404 shell_input_line_index++; /* skip newline */
2405 goto next_alias_char; /* and get next character */
2406 }
2407 else
2408 #endif
2409 goto restart_read;
2410 }
2411
2412 if (uc == 0 && shell_input_line_terminator == EOF)
2413 return ((shell_input_line_index != 0) ? '\n' : EOF);
2414
2415 return (uc);
2416 }
2417
2418 /* Put C back into the input for the shell. This might need changes for
2419 HANDLE_MULTIBYTE around EOLs. Since we (currently) never push back a
2420 character different than we read, shell_input_line_property doesn't need
2421 to change when manipulating shell_input_line. The define for
2422 last_shell_getc_is_singlebyte should take care of it, though. */
2423 static void
2424 shell_ungetc (c)
2425 int c;
2426 {
2427 if (shell_input_line && shell_input_line_index)
2428 shell_input_line[--shell_input_line_index] = c;
2429 else
2430 eol_ungetc_lookahead = c;
2431 }
2432
2433 #ifdef INCLUDE_UNUSED
2434 /* Back the input pointer up by one, effectively `ungetting' a character. */
2435 static void
2436 shell_ungetchar ()
2437 {
2438 if (shell_input_line && shell_input_line_index)
2439 shell_input_line_index--;
2440 }
2441 #endif
2442
2443 /* Discard input until CHARACTER is seen, then push that character back
2444 onto the input stream. */
2445 static void
2446 discard_until (character)
2447 int character;
2448 {
2449 int c;
2450
2451 while ((c = shell_getc (0)) != EOF && c != character)
2452 ;
2453
2454 if (c != EOF)
2455 shell_ungetc (c);
2456 }
2457
2458 void
2459 execute_variable_command (command, vname)
2460 char *command, *vname;
2461 {
2462 char *last_lastarg;
2463 sh_parser_state_t ps;
2464
2465 save_parser_state (&ps);
2466 last_lastarg = get_string_value ("_");
2467 if (last_lastarg)
2468 last_lastarg = savestring (last_lastarg);
2469
2470 parse_and_execute (savestring (command), vname, SEVAL_NONINT|SEVAL_NOHIST);
2471
2472 restore_parser_state (&ps);
2473 bind_variable ("_", last_lastarg, 0);
2474 FREE (last_lastarg);
2475
2476 if (token_to_read == '\n') /* reset_parser was called */
2477 token_to_read = 0;
2478 }
2479
2480 /* Place to remember the token. We try to keep the buffer
2481 at a reasonable size, but it can grow. */
2482 static char *token = (char *)NULL;
2483
2484 /* Current size of the token buffer. */
2485 static int token_buffer_size;
2486
2487 /* Command to read_token () explaining what we want it to do. */
2488 #define READ 0
2489 #define RESET 1
2490 #define prompt_is_ps1 \
2491 (!prompt_string_pointer || prompt_string_pointer == &ps1_prompt)
2492
2493 /* Function for yyparse to call. yylex keeps track of
2494 the last two tokens read, and calls read_token. */
2495 static int
2496 yylex ()
2497 {
2498 if (interactive && (current_token == 0 || current_token == '\n'))
2499 {
2500 /* Before we print a prompt, we might have to check mailboxes.
2501 We do this only if it is time to do so. Notice that only here
2502 is the mail alarm reset; nothing takes place in check_mail ()
2503 except the checking of mail. Please don't change this. */
2504 if (prompt_is_ps1 && parse_and_execute_level == 0 && time_to_check_mail ())
2505 {
2506 check_mail ();
2507 reset_mail_timer ();
2508 }
2509
2510 /* Avoid printing a prompt if we're not going to read anything, e.g.
2511 after resetting the parser with read_token (RESET). */
2512 if (token_to_read == 0 && SHOULD_PROMPT ())
2513 prompt_again ();
2514 }
2515
2516 two_tokens_ago = token_before_that;
2517 token_before_that = last_read_token;
2518 last_read_token = current_token;
2519 current_token = read_token (READ);
2520
2521 if ((parser_state & PST_EOFTOKEN) && current_token == shell_eof_token)
2522 {
2523 current_token = yacc_EOF;
2524 if (bash_input.type == st_string)
2525 rewind_input_string ();
2526 }
2527 parser_state &= ~PST_EOFTOKEN;
2528
2529 return (current_token);
2530 }
2531
2532 /* When non-zero, we have read the required tokens
2533 which allow ESAC to be the next one read. */
2534 static int esacs_needed_count;
2535
2536 void
2537 gather_here_documents ()
2538 {
2539 int r;
2540
2541 r = 0;
2542 while (need_here_doc)
2543 {
2544 parser_state |= PST_HEREDOC;
2545 make_here_document (redir_stack[r++], line_number);
2546 parser_state &= ~PST_HEREDOC;
2547 need_here_doc--;
2548 }
2549 }
2550
2551 /* When non-zero, an open-brace used to create a group is awaiting a close
2552 brace partner. */
2553 static int open_brace_count;
2554
2555 #define command_token_position(token) \
2556 (((token) == ASSIGNMENT_WORD) || (parser_state&PST_REDIRLIST) || \
2557 ((token) != SEMI_SEMI && (token) != SEMI_AND && (token) != SEMI_SEMI_AND && reserved_word_acceptable(token)))
2558
2559 #define assignment_acceptable(token) \
2560 (command_token_position(token) && ((parser_state & PST_CASEPAT) == 0))
2561
2562 /* Check to see if TOKEN is a reserved word and return the token
2563 value if it is. */
2564 #define CHECK_FOR_RESERVED_WORD(tok) \
2565 do { \
2566 if (!dollar_present && !quoted && \
2567 reserved_word_acceptable (last_read_token)) \
2568 { \
2569 int i; \
2570 for (i = 0; word_token_alist[i].word != (char *)NULL; i++) \
2571 if (STREQ (tok, word_token_alist[i].word)) \
2572 { \
2573 if ((parser_state & PST_CASEPAT) && (word_token_alist[i].token != ESAC)) \
2574 break; \
2575 if (word_token_alist[i].token == TIME && time_command_acceptable () == 0) \
2576 break; \
2577 if (word_token_alist[i].token == ESAC) \
2578 parser_state &= ~(PST_CASEPAT|PST_CASESTMT); \
2579 else if (word_token_alist[i].token == CASE) \
2580 parser_state |= PST_CASESTMT; \
2581 else if (word_token_alist[i].token == COND_END) \
2582 parser_state &= ~(PST_CONDCMD|PST_CONDEXPR); \
2583 else if (word_token_alist[i].token == COND_START) \
2584 parser_state |= PST_CONDCMD; \
2585 else if (word_token_alist[i].token == '{') \
2586 open_brace_count++; \
2587 else if (word_token_alist[i].token == '}' && open_brace_count) \
2588 open_brace_count--; \
2589 return (word_token_alist[i].token); \
2590 } \
2591 } \
2592 } while (0)
2593
2594 #if defined (ALIAS)
2595
2596 /* OK, we have a token. Let's try to alias expand it, if (and only if)
2597 it's eligible.
2598
2599 It is eligible for expansion if EXPAND_ALIASES is set, and
2600 the token is unquoted and the last token read was a command
2601 separator (or expand_next_token is set), and we are currently
2602 processing an alias (pushed_string_list is non-empty) and this
2603 token is not the same as the current or any previously
2604 processed alias.
2605
2606 Special cases that disqualify:
2607 In a pattern list in a case statement (parser_state & PST_CASEPAT). */
2608
2609 static char *
2610 mk_alexpansion (s)
2611 char *s;
2612 {
2613 int l;
2614 char *r;
2615
2616 l = strlen (s);
2617 r = xmalloc (l + 2);
2618 strcpy (r, s);
2619 /* If the last character in the alias is a newline, don't add a trailing
2620 space to the expansion. Works with shell_getc above. */
2621 if (r[l - 1] != ' ' && r[l - 1] != '\n')
2622 r[l++] = ' ';
2623 r[l] = '\0';
2624 return r;
2625 }
2626
2627 static int
2628 alias_expand_token (tokstr)
2629 char *tokstr;
2630 {
2631 char *expanded;
2632 alias_t *ap;
2633
2634 if (((parser_state & PST_ALEXPNEXT) || command_token_position (last_read_token)) &&
2635 (parser_state & PST_CASEPAT) == 0)
2636 {
2637 ap = find_alias (tokstr);
2638
2639 /* Currently expanding this token. */
2640 if (ap && (ap->flags & AL_BEINGEXPANDED))
2641 return (NO_EXPANSION);
2642
2643 /* mk_alexpansion puts an extra space on the end of the alias expansion,
2644 so the lookahead by the parser works right. If this gets changed,
2645 make sure the code in shell_getc that deals with reaching the end of
2646 an expanded alias is changed with it. */
2647 expanded = ap ? mk_alexpansion (ap->value) : (char *)NULL;
2648
2649 if (expanded)
2650 {
2651 push_string (expanded, ap->flags & AL_EXPANDNEXT, ap);
2652 return (RE_READ_TOKEN);
2653 }
2654 else
2655 /* This is an eligible token that does not have an expansion. */
2656 return (NO_EXPANSION);
2657 }
2658 return (NO_EXPANSION);
2659 }
2660 #endif /* ALIAS */
2661
2662 static int
2663 time_command_acceptable ()
2664 {
2665 #if defined (COMMAND_TIMING)
2666 int i;
2667
2668 if (posixly_correct && shell_compatibility_level > 41)
2669 {
2670 /* Quick check of the rest of the line to find the next token. If it
2671 begins with a `-', Posix says to not return `time' as the token.
2672 This was interp 267. */
2673 i = shell_input_line_index;
2674 while (i < shell_input_line_len && (shell_input_line[i] == ' ' || shell_input_line[i] == '\t'))
2675 i++;
2676 if (shell_input_line[i] == '-')
2677 return 0;
2678 }
2679
2680 switch (last_read_token)
2681 {
2682 case 0:
2683 case ';':
2684 case '\n':
2685 case AND_AND:
2686 case OR_OR:
2687 case '&':
2688 case DO:
2689 case THEN:
2690 case ELSE:
2691 case '{': /* } */
2692 case '(': /* ) */
2693 case BANG: /* ! time pipeline */
2694 case TIME: /* time time pipeline */
2695 case TIMEOPT: /* time -p time pipeline */
2696 case TIMEIGN: /* time -p -- ... */
2697 return 1;
2698 default:
2699 return 0;
2700 }
2701 #else
2702 return 0;
2703 #endif /* COMMAND_TIMING */
2704 }
2705
2706 /* Handle special cases of token recognition:
2707 IN is recognized if the last token was WORD and the token
2708 before that was FOR or CASE or SELECT.
2709
2710 DO is recognized if the last token was WORD and the token
2711 before that was FOR or SELECT.
2712
2713 ESAC is recognized if the last token caused `esacs_needed_count'
2714 to be set
2715
2716 `{' is recognized if the last token as WORD and the token
2717 before that was FUNCTION, or if we just parsed an arithmetic
2718 `for' command.
2719
2720 `}' is recognized if there is an unclosed `{' present.
2721
2722 `-p' is returned as TIMEOPT if the last read token was TIME.
2723 `--' is returned as TIMEIGN if the last read token was TIMEOPT.
2724
2725 ']]' is returned as COND_END if the parser is currently parsing
2726 a conditional expression ((parser_state & PST_CONDEXPR) != 0)
2727
2728 `time' is returned as TIME if and only if it is immediately
2729 preceded by one of `;', `\n', `||', `&&', or `&'.
2730 */
2731
2732 static int
2733 special_case_tokens (tokstr)
2734 char *tokstr;
2735 {
2736 if ((last_read_token == WORD) &&
2737 #if defined (SELECT_COMMAND)
2738 ((token_before_that == FOR) || (token_before_that == CASE) || (token_before_that == SELECT)) &&
2739 #else
2740 ((token_before_that == FOR) || (token_before_that == CASE)) &&
2741 #endif
2742 (tokstr[0] == 'i' && tokstr[1] == 'n' && tokstr[2] == 0))
2743 {
2744 if (token_before_that == CASE)
2745 {
2746 parser_state |= PST_CASEPAT;
2747 esacs_needed_count++;
2748 }
2749 return (IN);
2750 }
2751
2752 if (last_read_token == WORD &&
2753 #if defined (SELECT_COMMAND)
2754 (token_before_that == FOR || token_before_that == SELECT) &&
2755 #else
2756 (token_before_that == FOR) &&
2757 #endif
2758 (tokstr[0] == 'd' && tokstr[1] == 'o' && tokstr[2] == '\0'))
2759 return (DO);
2760
2761 /* Ditto for ESAC in the CASE case.
2762 Specifically, this handles "case word in esac", which is a legal
2763 construct, certainly because someone will pass an empty arg to the
2764 case construct, and we don't want it to barf. Of course, we should
2765 insist that the case construct has at least one pattern in it, but
2766 the designers disagree. */
2767 if (esacs_needed_count)
2768 {
2769 esacs_needed_count--;
2770 if (STREQ (tokstr, "esac"))
2771 {
2772 parser_state &= ~PST_CASEPAT;
2773 return (ESAC);
2774 }
2775 }
2776
2777 /* The start of a shell function definition. */
2778 if (parser_state & PST_ALLOWOPNBRC)
2779 {
2780 parser_state &= ~PST_ALLOWOPNBRC;
2781 if (tokstr[0] == '{' && tokstr[1] == '\0') /* } */
2782 {
2783 open_brace_count++;
2784 function_bstart = line_number;
2785 return ('{'); /* } */
2786 }
2787 }
2788
2789 /* We allow a `do' after a for ((...)) without an intervening
2790 list_terminator */
2791 if (last_read_token == ARITH_FOR_EXPRS && tokstr[0] == 'd' && tokstr[1] == 'o' && !tokstr[2])
2792 return (DO);
2793 if (last_read_token == ARITH_FOR_EXPRS && tokstr[0] == '{' && tokstr[1] == '\0') /* } */
2794 {
2795 open_brace_count++;
2796 return ('{'); /* } */
2797 }
2798
2799 if (open_brace_count && reserved_word_acceptable (last_read_token) && tokstr[0] == '}' && !tokstr[1])
2800 {
2801 open_brace_count--; /* { */
2802 return ('}');
2803 }
2804
2805 #if defined (COMMAND_TIMING)
2806 /* Handle -p after `time'. */
2807 if (last_read_token == TIME && tokstr[0] == '-' && tokstr[1] == 'p' && !tokstr[2])
2808 return (TIMEOPT);
2809 /* Handle -- after `time -p'. */
2810 if (last_read_token == TIMEOPT && tokstr[0] == '-' && tokstr[1] == '-' && !tokstr[2])
2811 return (TIMEIGN);
2812 #endif
2813
2814 #if defined (COND_COMMAND) /* [[ */
2815 if ((parser_state & PST_CONDEXPR) && tokstr[0] == ']' && tokstr[1] == ']' && tokstr[2] == '\0')
2816 return (COND_END);
2817 #endif
2818
2819 return (-1);
2820 }
2821
2822 /* Called from shell.c when Control-C is typed at top level. Or
2823 by the error rule at top level. */
2824 void
2825 reset_parser ()
2826 {
2827 dstack.delimiter_depth = 0; /* No delimiters found so far. */
2828 open_brace_count = 0;
2829
2830 #if defined (EXTENDED_GLOB)
2831 /* Reset to global value of extended glob */
2832 if (parser_state & PST_EXTPAT)
2833 extended_glob = global_extglob;
2834 #endif
2835
2836 parser_state = 0;
2837
2838 #if defined (ALIAS) || defined (DPAREN_ARITHMETIC)
2839 if (pushed_string_list)
2840 free_string_list ();
2841 #endif /* ALIAS || DPAREN_ARITHMETIC */
2842
2843 if (shell_input_line)
2844 {
2845 free (shell_input_line);
2846 shell_input_line = (char *)NULL;
2847 shell_input_line_size = shell_input_line_index = 0;
2848 }
2849
2850 FREE (word_desc_to_read);
2851 word_desc_to_read = (WORD_DESC *)NULL;
2852
2853 current_token = '\n'; /* XXX */
2854 last_read_token = '\n';
2855 token_to_read = '\n';
2856 }
2857
2858 /* Read the next token. Command can be READ (normal operation) or
2859 RESET (to normalize state). */
2860 static int
2861 read_token (command)
2862 int command;
2863 {
2864 int character; /* Current character. */
2865 int peek_char; /* Temporary look-ahead character. */
2866 int result; /* The thing to return. */
2867
2868 if (command == RESET)
2869 {
2870 reset_parser ();
2871 return ('\n');
2872 }
2873
2874 if (token_to_read)
2875 {
2876 result = token_to_read;
2877 if (token_to_read == WORD || token_to_read == ASSIGNMENT_WORD)
2878 {
2879 yylval.word = word_desc_to_read;
2880 word_desc_to_read = (WORD_DESC *)NULL;
2881 }
2882 token_to_read = 0;
2883 return (result);
2884 }
2885
2886 #if defined (COND_COMMAND)
2887 if ((parser_state & (PST_CONDCMD|PST_CONDEXPR)) == PST_CONDCMD)
2888 {
2889 cond_lineno = line_number;
2890 parser_state |= PST_CONDEXPR;
2891 yylval.command = parse_cond_command ();
2892 if (cond_token != COND_END)
2893 {
2894 cond_error ();
2895 return (-1);
2896 }
2897 token_to_read = COND_END;
2898 parser_state &= ~(PST_CONDEXPR|PST_CONDCMD);
2899 return (COND_CMD);
2900 }
2901 #endif
2902
2903 #if defined (ALIAS)
2904 /* This is a place to jump back to once we have successfully expanded a
2905 token with an alias and pushed the string with push_string () */
2906 re_read_token:
2907 #endif /* ALIAS */
2908
2909 /* Read a single word from input. Start by skipping blanks. */
2910 while ((character = shell_getc (1)) != EOF && shellblank (character))
2911 ;
2912
2913 if (character == EOF)
2914 {
2915 EOF_Reached = 1;
2916 return (yacc_EOF);
2917 }
2918
2919 if MBTEST(character == '#' && (!interactive || interactive_comments))
2920 {
2921 /* A comment. Discard until EOL or EOF, and then return a newline. */
2922 discard_until ('\n');
2923 shell_getc (0);
2924 character = '\n'; /* this will take the next if statement and return. */
2925 }
2926
2927 if (character == '\n')
2928 {
2929 /* If we're about to return an unquoted newline, we can go and collect
2930 the text of any pending here document. */
2931 if (need_here_doc)
2932 gather_here_documents ();
2933
2934 #if defined (ALIAS)
2935 parser_state &= ~PST_ALEXPNEXT;
2936 #endif /* ALIAS */
2937
2938 parser_state &= ~PST_ASSIGNOK;
2939
2940 return (character);
2941 }
2942
2943 if (parser_state & PST_REGEXP)
2944 goto tokword;
2945
2946 /* Shell meta-characters. */
2947 if MBTEST(shellmeta (character) && ((parser_state & PST_DBLPAREN) == 0))
2948 {
2949 #if defined (ALIAS)
2950 /* Turn off alias tokenization iff this character sequence would
2951 not leave us ready to read a command. */
2952 if (character == '<' || character == '>')
2953 parser_state &= ~PST_ALEXPNEXT;
2954 #endif /* ALIAS */
2955
2956 parser_state &= ~PST_ASSIGNOK;
2957
2958 peek_char = shell_getc (1);
2959 if (character == peek_char)
2960 {
2961 switch (character)
2962 {
2963 case '<':
2964 /* If '<' then we could be at "<<" or at "<<-". We have to
2965 look ahead one more character. */
2966 peek_char = shell_getc (1);
2967 if MBTEST(peek_char == '-')
2968 return (LESS_LESS_MINUS);
2969 else if MBTEST(peek_char == '<')
2970 return (LESS_LESS_LESS);
2971 else
2972 {
2973 shell_ungetc (peek_char);
2974 return (LESS_LESS);
2975 }
2976
2977 case '>':
2978 return (GREATER_GREATER);
2979
2980 case ';':
2981 parser_state |= PST_CASEPAT;
2982 #if defined (ALIAS)
2983 parser_state &= ~PST_ALEXPNEXT;
2984 #endif /* ALIAS */
2985
2986 peek_char = shell_getc (1);
2987 if MBTEST(peek_char == '&')
2988 return (SEMI_SEMI_AND);
2989 else
2990 {
2991 shell_ungetc (peek_char);
2992 return (SEMI_SEMI);
2993 }
2994
2995 case '&':
2996 return (AND_AND);
2997
2998 case '|':
2999 return (OR_OR);
3000
3001 #if defined (DPAREN_ARITHMETIC) || defined (ARITH_FOR_COMMAND)
3002 case '(': /* ) */
3003 result = parse_dparen (character);
3004 if (result == -2)
3005 break;
3006 else
3007 return result;
3008 #endif
3009 }
3010 }
3011 else if MBTEST(character == '<' && peek_char == '&')
3012 return (LESS_AND);
3013 else if MBTEST(character == '>' && peek_char == '&')
3014 return (GREATER_AND);
3015 else if MBTEST(character == '<' && peek_char == '>')
3016 return (LESS_GREATER);
3017 else if MBTEST(character == '>' && peek_char == '|')
3018 return (GREATER_BAR);
3019 else if MBTEST(character == '&' && peek_char == '>')
3020 {
3021 peek_char = shell_getc (1);
3022 if MBTEST(peek_char == '>')
3023 return (AND_GREATER_GREATER);
3024 else
3025 {
3026 shell_ungetc (peek_char);
3027 return (AND_GREATER);
3028 }
3029 }
3030 else if MBTEST(character == '|' && peek_char == '&')
3031 return (BAR_AND);
3032 else if MBTEST(character == ';' && peek_char == '&')
3033 {
3034 parser_state |= PST_CASEPAT;
3035 #if defined (ALIAS)
3036 parser_state &= ~PST_ALEXPNEXT;
3037 #endif /* ALIAS */
3038 return (SEMI_AND);
3039 }
3040
3041 shell_ungetc (peek_char);
3042
3043 /* If we look like we are reading the start of a function
3044 definition, then let the reader know about it so that
3045 we will do the right thing with `{'. */
3046 if MBTEST(character == ')' && last_read_token == '(' && token_before_that == WORD)
3047 {
3048 parser_state |= PST_ALLOWOPNBRC;
3049 #if defined (ALIAS)
3050 parser_state &= ~PST_ALEXPNEXT;
3051 #endif /* ALIAS */
3052 function_dstart = line_number;
3053 }
3054
3055 /* case pattern lists may be preceded by an optional left paren. If
3056 we're not trying to parse a case pattern list, the left paren
3057 indicates a subshell. */
3058 if MBTEST(character == '(' && (parser_state & PST_CASEPAT) == 0) /* ) */
3059 parser_state |= PST_SUBSHELL;
3060 /*(*/
3061 else if MBTEST((parser_state & PST_CASEPAT) && character == ')')
3062 parser_state &= ~PST_CASEPAT;
3063 /*(*/
3064 else if MBTEST((parser_state & PST_SUBSHELL) && character == ')')
3065 parser_state &= ~PST_SUBSHELL;
3066
3067 #if defined (PROCESS_SUBSTITUTION)
3068 /* Check for the constructs which introduce process substitution.
3069 Shells running in `posix mode' don't do process substitution. */
3070 if MBTEST(posixly_correct || ((character != '>' && character != '<') || peek_char != '(')) /*)*/
3071 #endif /* PROCESS_SUBSTITUTION */
3072 return (character);
3073 }
3074
3075 /* Hack <&- (close stdin) case. Also <&N- (dup and close). */
3076 if MBTEST(character == '-' && (last_read_token == LESS_AND || last_read_token == GREATER_AND))
3077 return (character);
3078
3079 tokword:
3080 /* Okay, if we got this far, we have to read a word. Read one,
3081 and then check it against the known ones. */
3082 result = read_token_word (character);
3083 #if defined (ALIAS)
3084 if (result == RE_READ_TOKEN)
3085 goto re_read_token;
3086 #endif
3087 return result;
3088 }
3089
3090 /*
3091 * Match a $(...) or other grouping construct. This has to handle embedded
3092 * quoted strings ('', ``, "") and nested constructs. It also must handle
3093 * reprompting the user, if necessary, after reading a newline, and returning
3094 * correct error values if it reads EOF.
3095 */
3096 #define P_FIRSTCLOSE 0x0001
3097 #define P_ALLOWESC 0x0002
3098 #define P_DQUOTE 0x0004
3099 #define P_COMMAND 0x0008 /* parsing a command, so look for comments */
3100 #define P_BACKQUOTE 0x0010 /* parsing a backquoted command substitution */
3101 #define P_ARRAYSUB 0x0020 /* parsing a [...] array subscript for assignment */
3102 #define P_DOLBRACE 0x0040 /* parsing a ${...} construct */
3103
3104 /* Lexical state while parsing a grouping construct or $(...). */
3105 #define LEX_WASDOL 0x001
3106 #define LEX_CKCOMMENT 0x002
3107 #define LEX_INCOMMENT 0x004
3108 #define LEX_PASSNEXT 0x008
3109 #define LEX_RESWDOK 0x010
3110 #define LEX_CKCASE 0x020
3111 #define LEX_INCASE 0x040
3112 #define LEX_INHEREDOC 0x080
3113 #define LEX_HEREDELIM 0x100 /* reading here-doc delimiter */
3114 #define LEX_STRIPDOC 0x200 /* <<- strip tabs from here doc delim */
3115 #define LEX_INWORD 0x400
3116
3117 #define COMSUB_META(ch) ((ch) == ';' || (ch) == '&' || (ch) == '|')
3118
3119 #define CHECK_NESTRET_ERROR() \
3120 do { \
3121 if (nestret == &matched_pair_error) \
3122 { \
3123 free (ret); \
3124 return &matched_pair_error; \
3125 } \
3126 } while (0)
3127
3128 #define APPEND_NESTRET() \
3129 do { \
3130 if (nestlen) \
3131 { \
3132 RESIZE_MALLOCED_BUFFER (ret, retind, nestlen, retsize, 64); \
3133 strcpy (ret + retind, nestret); \
3134 retind += nestlen; \
3135 } \
3136 } while (0)
3137
3138 static char matched_pair_error;
3139
3140 static char *
3141 parse_matched_pair (qc, open, close, lenp, flags)
3142 int qc; /* `"' if this construct is within double quotes */
3143 int open, close;
3144 int *lenp, flags;
3145 {
3146 int count, ch, tflags;
3147 int nestlen, ttranslen, start_lineno;
3148 char *ret, *nestret, *ttrans;
3149 int retind, retsize, rflags;
3150 int dolbrace_state;
3151
3152 dolbrace_state = (flags & P_DOLBRACE) ? DOLBRACE_PARAM : 0;
3153
3154 /*itrace("parse_matched_pair[%d]: open = %c close = %c flags = %d", line_number, open, close, flags);*/
3155 count = 1;
3156 tflags = 0;
3157
3158 if ((flags & P_COMMAND) && qc != '`' && qc != '\'' && qc != '"' && (flags & P_DQUOTE) == 0)
3159 tflags |= LEX_CKCOMMENT;
3160
3161 /* RFLAGS is the set of flags we want to pass to recursive calls. */
3162 rflags = (qc == '"') ? P_DQUOTE : (flags & P_DQUOTE);
3163
3164 ret = (char *)xmalloc (retsize = 64);
3165 retind = 0;
3166
3167 start_lineno = line_number;
3168 while (count)
3169 {
3170 ch = shell_getc (qc != '\'' && (tflags & (LEX_PASSNEXT)) == 0);
3171
3172 if (ch == EOF)
3173 {
3174 free (ret);
3175 parser_error (start_lineno, _("unexpected EOF while looking for matching `%c'"), close);
3176 EOF_Reached = 1; /* XXX */
3177 return (&matched_pair_error);
3178 }
3179
3180 /* Possible reprompting. */
3181 if (ch == '\n' && SHOULD_PROMPT ())
3182 prompt_again ();
3183
3184 /* Don't bother counting parens or doing anything else if in a comment
3185 or part of a case statement */
3186 if (tflags & LEX_INCOMMENT)
3187 {
3188 /* Add this character. */
3189 RESIZE_MALLOCED_BUFFER (ret, retind, 1, retsize, 64);
3190 ret[retind++] = ch;
3191
3192 if (ch == '\n')
3193 tflags &= ~LEX_INCOMMENT;
3194
3195 continue;
3196 }
3197
3198 /* Not exactly right yet, should handle shell metacharacters, too. If
3199 any changes are made to this test, make analogous changes to subst.c:
3200 extract_delimited_string(). */
3201 else if MBTEST((tflags & LEX_CKCOMMENT) && (tflags & LEX_INCOMMENT) == 0 && ch == '#' && (retind == 0 || ret[retind-1] == '\n' || shellblank (ret[retind - 1])))
3202 tflags |= LEX_INCOMMENT;
3203
3204 if (tflags & LEX_PASSNEXT) /* last char was backslash */
3205 {
3206 tflags &= ~LEX_PASSNEXT;
3207 if (qc != '\'' && ch == '\n') /* double-quoted \<newline> disappears. */
3208 {
3209 if (retind > 0)
3210 retind--; /* swallow previously-added backslash */
3211 continue;
3212 }
3213
3214 RESIZE_MALLOCED_BUFFER (ret, retind, 2, retsize, 64);
3215 if MBTEST(ch == CTLESC || ch == CTLNUL)
3216 ret[retind++] = CTLESC;
3217 ret[retind++] = ch;
3218 continue;
3219 }
3220 /* If we're reparsing the input (e.g., from parse_string_to_word_list),
3221 we've already prepended CTLESC to single-quoted results of $'...'.
3222 We may want to do this for other CTLESC-quoted characters in
3223 reparse, too. */
3224 else if MBTEST((parser_state & PST_REPARSE) && open == '\'' && (ch == CTLESC || ch == CTLNUL))
3225 {
3226 RESIZE_MALLOCED_BUFFER (ret, retind, 1, retsize, 64);
3227 ret[retind++] = ch;
3228 continue;
3229 }
3230 else if MBTEST(ch == CTLESC || ch == CTLNUL) /* special shell escapes */
3231 {
3232 RESIZE_MALLOCED_BUFFER (ret, retind, 2, retsize, 64);
3233 ret[retind++] = CTLESC;
3234 ret[retind++] = ch;
3235 continue;
3236 }
3237 else if MBTEST(ch == close) /* ending delimiter */
3238 count--;
3239 /* handle nested ${...} specially. */
3240 else if MBTEST(open != close && (tflags & LEX_WASDOL) && open == '{' && ch == open) /* } */
3241 count++;
3242 else if MBTEST(((flags & P_FIRSTCLOSE) == 0) && ch == open) /* nested begin */
3243 count++;
3244
3245 /* Add this character. */
3246 RESIZE_MALLOCED_BUFFER (ret, retind, 1, retsize, 64);
3247 ret[retind++] = ch;
3248
3249 /* If we just read the ending character, don't bother continuing. */
3250 if (count == 0)
3251 break;
3252
3253 if (open == '\'') /* '' inside grouping construct */
3254 {
3255 if MBTEST((flags & P_ALLOWESC) && ch == '\\')
3256 tflags |= LEX_PASSNEXT;
3257 continue;
3258 }
3259
3260 if MBTEST(ch == '\\') /* backslashes */
3261 tflags |= LEX_PASSNEXT;
3262
3263 /* Based on which dolstate is currently in (param, op, or word),
3264 decide what the op is. We're really only concerned if it's % or
3265 #, so we can turn on a flag that says whether or not we should
3266 treat single quotes as special when inside a double-quoted
3267 ${...}. This logic must agree with subst.c:extract_dollar_brace_string
3268 since they share the same defines. */
3269 if (flags & P_DOLBRACE)
3270 {
3271 /* ${param%[%]word} */
3272 if MBTEST(dolbrace_state == DOLBRACE_PARAM && ch == '%' && retind > 1)
3273 dolbrace_state = DOLBRACE_QUOTE;
3274 /* ${param#[#]word} */
3275 else if MBTEST(dolbrace_state == DOLBRACE_PARAM && ch == '#' && retind > 1)
3276 dolbrace_state = DOLBRACE_QUOTE;
3277 /* ${param/[/]pat/rep} */
3278 else if MBTEST(dolbrace_state == DOLBRACE_PARAM && ch == '/' && retind > 1)
3279 dolbrace_state = DOLBRACE_QUOTE;
3280 /* ${param^[^]pat} */
3281 else if MBTEST(dolbrace_state == DOLBRACE_PARAM && ch == '^' && retind > 1)
3282 dolbrace_state = DOLBRACE_QUOTE;
3283 /* ${param,[,]pat} */
3284 else if MBTEST(dolbrace_state == DOLBRACE_PARAM && ch == ',' && retind > 1)
3285 dolbrace_state = DOLBRACE_QUOTE;
3286 else if MBTEST(dolbrace_state == DOLBRACE_PARAM && strchr ("#%^,~:-=?+/", ch) != 0)
3287 dolbrace_state = DOLBRACE_OP;
3288 else if MBTEST(dolbrace_state == DOLBRACE_OP && strchr ("#%^,~:-=?+/", ch) == 0)
3289 dolbrace_state = DOLBRACE_WORD;
3290 }
3291
3292 /* The big hammer. Single quotes aren't special in double quotes. The
3293 problem is that Posix used to say the single quotes are semi-special:
3294 within a double-quoted ${...} construct "an even number of
3295 unescaped double-quotes or single-quotes, if any, shall occur." */
3296 /* This was changed in Austin Group Interp 221 */
3297 if MBTEST(posixly_correct && shell_compatibility_level > 41 && dolbrace_state != DOLBRACE_QUOTE && (flags & P_DQUOTE) && (flags & P_DOLBRACE) && ch == '\'')
3298 continue;
3299
3300 /* Could also check open == '`' if we want to parse grouping constructs
3301 inside old-style command substitution. */
3302 if (open != close) /* a grouping construct */
3303 {
3304 if MBTEST(shellquote (ch))
3305 {
3306 /* '', ``, or "" inside $(...) or other grouping construct. */
3307 push_delimiter (dstack, ch);
3308 if MBTEST((tflags & LEX_WASDOL) && ch == '\'') /* $'...' inside group */
3309 nestret = parse_matched_pair (ch, ch, ch, &nestlen, P_ALLOWESC|rflags);
3310 else
3311 nestret = parse_matched_pair (ch, ch, ch, &nestlen, rflags);
3312 pop_delimiter (dstack);
3313 CHECK_NESTRET_ERROR ();
3314
3315 if MBTEST((tflags & LEX_WASDOL) && ch == '\'' && (extended_quote || (rflags & P_DQUOTE) == 0))
3316 {
3317 /* Translate $'...' here. */
3318 ttrans = ansiexpand (nestret, 0, nestlen - 1, &ttranslen);
3319 xfree (nestret);
3320
3321 if ((rflags & P_DQUOTE) == 0)
3322 {
3323 nestret = sh_single_quote (ttrans);
3324 free (ttrans);
3325 nestlen = strlen (nestret);
3326 }
3327 else
3328 {
3329 nestret = ttrans;
3330 nestlen = ttranslen;
3331 }
3332 retind -= 2; /* back up before the $' */
3333 }
3334 else if MBTEST((tflags & LEX_WASDOL) && ch == '"' && (extended_quote || (rflags & P_DQUOTE) == 0))
3335 {
3336 /* Locale expand $"..." here. */
3337 ttrans = localeexpand (nestret, 0, nestlen - 1, start_lineno, &ttranslen);
3338 xfree (nestret);
3339
3340 nestret = sh_mkdoublequoted (ttrans, ttranslen, 0);
3341 free (ttrans);
3342 nestlen = ttranslen + 2;
3343 retind -= 2; /* back up before the $" */
3344 }
3345
3346 APPEND_NESTRET ();
3347 FREE (nestret);
3348 }
3349 else if ((flags & P_ARRAYSUB) && (tflags & LEX_WASDOL) && (ch == '(' || ch == '{' || ch == '[')) /* ) } ] */
3350 goto parse_dollar_word;
3351 }
3352 /* Parse an old-style command substitution within double quotes as a
3353 single word. */
3354 /* XXX - sh and ksh93 don't do this - XXX */
3355 else if MBTEST(open == '"' && ch == '`')
3356 {
3357 nestret = parse_matched_pair (0, '`', '`', &nestlen, rflags);
3358
3359 CHECK_NESTRET_ERROR ();
3360 APPEND_NESTRET ();
3361
3362 FREE (nestret);
3363 }
3364 else if MBTEST(open != '`' && (tflags & LEX_WASDOL) && (ch == '(' || ch == '{' || ch == '[')) /* ) } ] */
3365 /* check for $(), $[], or ${} inside quoted string. */
3366 {
3367 parse_dollar_word:
3368 if (open == ch) /* undo previous increment */
3369 count--;
3370 if (ch == '(') /* ) */
3371 nestret = parse_comsub (0, '(', ')', &nestlen, (rflags|P_COMMAND) & ~P_DQUOTE);
3372 else if (ch == '{') /* } */
3373 nestret = parse_matched_pair (0, '{', '}', &nestlen, P_FIRSTCLOSE|P_DOLBRACE|rflags);
3374 else if (ch == '[') /* ] */
3375 nestret = parse_matched_pair (0, '[', ']', &nestlen, rflags);
3376
3377 CHECK_NESTRET_ERROR ();
3378 APPEND_NESTRET ();
3379
3380 FREE (nestret);
3381 }
3382 if MBTEST(ch == '$')
3383 tflags |= LEX_WASDOL;
3384 else
3385 tflags &= ~LEX_WASDOL;
3386 }
3387
3388 ret[retind] = '\0';
3389 if (lenp)
3390 *lenp = retind;
3391 /*itrace("parse_matched_pair[%d]: returning %s", line_number, ret);*/
3392 return ret;
3393 }
3394
3395 /* Parse a $(...) command substitution. This is messier than I'd like, and
3396 reproduces a lot more of the token-reading code than I'd like. */
3397 static char *
3398 parse_comsub (qc, open, close, lenp, flags)
3399 int qc; /* `"' if this construct is within double quotes */
3400 int open, close;
3401 int *lenp, flags;
3402 {
3403 int count, ch, peekc, tflags, lex_rwlen, lex_wlen, lex_firstind;
3404 int nestlen, ttranslen, start_lineno;
3405 char *ret, *nestret, *ttrans, *heredelim;
3406 int retind, retsize, rflags, hdlen;
3407
3408 /* Posix interp 217 says arithmetic expressions have precedence, so
3409 assume $(( introduces arithmetic expansion and parse accordingly. */
3410 peekc = shell_getc (0);
3411 shell_ungetc (peekc);
3412 if (peekc == '(')
3413 return (parse_matched_pair (qc, open, close, lenp, 0));
3414
3415 /*itrace("parse_comsub: qc = `%c' open = %c close = %c", qc, open, close);*/
3416 count = 1;
3417 tflags = LEX_RESWDOK;
3418
3419 if ((flags & P_COMMAND) && qc != '\'' && qc != '"' && (flags & P_DQUOTE) == 0)
3420 tflags |= LEX_CKCASE;
3421 if ((tflags & LEX_CKCASE) && (interactive == 0 || interactive_comments))
3422 tflags |= LEX_CKCOMMENT;
3423
3424 /* RFLAGS is the set of flags we want to pass to recursive calls. */
3425 rflags = (flags & P_DQUOTE);
3426
3427 ret = (char *)xmalloc (retsize = 64);
3428 retind = 0;
3429
3430 start_lineno = line_number;
3431 lex_rwlen = lex_wlen = 0;
3432
3433 heredelim = 0;
3434 lex_firstind = -1;
3435
3436 while (count)
3437 {
3438 comsub_readchar:
3439 ch = shell_getc (qc != '\'' && (tflags & (LEX_INCOMMENT|LEX_PASSNEXT)) == 0);
3440
3441 if (ch == EOF)
3442 {
3443 eof_error:
3444 free (ret);
3445 FREE (heredelim);
3446 parser_error (start_lineno, _("unexpected EOF while looking for matching `%c'"), close);
3447 EOF_Reached = 1; /* XXX */
3448 return (&matched_pair_error);
3449 }
3450
3451 /* If we hit the end of a line and are reading the contents of a here
3452 document, and it's not the same line that the document starts on,
3453 check for this line being the here doc delimiter. Otherwise, if
3454 we're in a here document, mark the next character as the beginning
3455 of a line. */
3456 if (ch == '\n')
3457 {
3458 if ((tflags & LEX_HEREDELIM) && heredelim)
3459 {
3460 tflags &= ~LEX_HEREDELIM;
3461 tflags |= LEX_INHEREDOC;
3462 lex_firstind = retind + 1;
3463 }
3464 else if (tflags & LEX_INHEREDOC)
3465 {
3466 int tind;
3467 tind = lex_firstind;
3468 while ((tflags & LEX_STRIPDOC) && ret[tind] == '\t')
3469 tind++;
3470 if (STREQN (ret + tind, heredelim, hdlen))
3471 {
3472 tflags &= ~(LEX_STRIPDOC|LEX_INHEREDOC);
3473 /*itrace("parse_comsub:%d: found here doc end `%s'", line_number, ret + tind);*/
3474 free (heredelim);
3475 heredelim = 0;
3476 lex_firstind = -1;
3477 }
3478 else
3479 lex_firstind = retind + 1;
3480 }
3481 }
3482
3483 /* Possible reprompting. */
3484 if (ch == '\n' && SHOULD_PROMPT ())
3485 prompt_again ();
3486
3487 /* XXX -- possibly allow here doc to be delimited by ending right
3488 paren. */
3489 if ((tflags & LEX_INHEREDOC) && ch == close && count == 1)
3490 {
3491 int tind;
3492 /*itrace("parse_comsub: in here doc, ch == close, retind - firstind = %d hdlen = %d retind = %d", retind-lex_firstind, hdlen, retind);*/
3493 tind = lex_firstind;
3494 while ((tflags & LEX_STRIPDOC) && ret[tind] == '\t')
3495 tind++;
3496 if (retind-tind == hdlen && STREQN (ret + tind, heredelim, hdlen))
3497 {
3498 tflags &= ~(LEX_STRIPDOC|LEX_INHEREDOC);
3499 /*itrace("parse_comsub:%d: found here doc end `%s'", line_number, ret + tind);*/
3500 free (heredelim);
3501 heredelim = 0;
3502 lex_firstind = -1;
3503 }
3504 }
3505
3506 /* Don't bother counting parens or doing anything else if in a comment */
3507 if (tflags & (LEX_INCOMMENT|LEX_INHEREDOC))
3508 {
3509 /* Add this character. */
3510 RESIZE_MALLOCED_BUFFER (ret, retind, 1, retsize, 64);
3511 ret[retind++] = ch;
3512
3513 if ((tflags & LEX_INCOMMENT) && ch == '\n')
3514 {
3515 /*itrace("parse_comsub:%d: lex_incomment -> 0 ch = `%c'", line_number, ch);*/
3516 tflags &= ~LEX_INCOMMENT;
3517 }
3518
3519 continue;
3520 }
3521
3522 if (tflags & LEX_PASSNEXT) /* last char was backslash */
3523 {
3524 /*itrace("parse_comsub:%d: lex_passnext -> 0 ch = `%c' (%d)", line_number, ch, __LINE__);*/
3525 tflags &= ~LEX_PASSNEXT;
3526 if (qc != '\'' && ch == '\n') /* double-quoted \<newline> disappears. */
3527 {
3528 if (retind > 0)
3529 retind--; /* swallow previously-added backslash */
3530 continue;
3531 }
3532
3533 RESIZE_MALLOCED_BUFFER (ret, retind, 2, retsize, 64);
3534 if MBTEST(ch == CTLESC || ch == CTLNUL)
3535 ret[retind++] = CTLESC;
3536 ret[retind++] = ch;
3537 continue;
3538 }
3539
3540 /* If this is a shell break character, we are not in a word. If not,
3541 we either start or continue a word. */
3542 if MBTEST(shellbreak (ch))
3543 {
3544 tflags &= ~LEX_INWORD;
3545 /*itrace("parse_comsub:%d: lex_inword -> 0 ch = `%c' (%d)", line_number, ch, __LINE__);*/
3546 }
3547 else
3548 {
3549 if (tflags & LEX_INWORD)
3550 {
3551 lex_wlen++;
3552 /*itrace("parse_comsub:%d: lex_inword == 1 ch = `%c' lex_wlen = %d (%d)", line_number, ch, lex_wlen, __LINE__);*/
3553 }
3554 else
3555 {
3556 /*itrace("parse_comsub:%d: lex_inword -> 1 ch = `%c' (%d)", line_number, ch, __LINE__);*/
3557 tflags |= LEX_INWORD;
3558 lex_wlen = 0;
3559 }
3560 }
3561
3562 /* Skip whitespace */
3563 if MBTEST(shellblank (ch) && (tflags & LEX_HEREDELIM) == 0 && lex_rwlen == 0)
3564 {
3565 /* Add this character. */
3566 RESIZE_MALLOCED_BUFFER (ret, retind, 1, retsize, 64);
3567 ret[retind++] = ch;
3568 continue;
3569 }
3570
3571 /* Either we are looking for the start of the here-doc delimiter
3572 (lex_firstind == -1) or we are reading one (lex_firstind >= 0).
3573 If this character is a shell break character and we are reading
3574 the delimiter, save it and note that we are now reading a here
3575 document. If we've found the start of the delimiter, note it by
3576 setting lex_firstind. Backslashes can quote shell metacharacters
3577 in here-doc delimiters. */
3578 if (tflags & LEX_HEREDELIM)
3579 {
3580 if (lex_firstind == -1 && shellbreak (ch) == 0)
3581 lex_firstind = retind;
3582 #if 0
3583 else if (heredelim && (tflags & LEX_PASSNEXT) == 0 && ch == '\n')
3584 {
3585 tflags |= LEX_INHEREDOC;
3586 tflags &= ~LEX_HEREDELIM;
3587 lex_firstind = retind + 1;
3588 }
3589 #endif
3590 else if (lex_firstind >= 0 && (tflags & LEX_PASSNEXT) == 0 && shellbreak (ch))
3591 {
3592 if (heredelim == 0)
3593 {
3594 nestret = substring (ret, lex_firstind, retind);
3595 heredelim = string_quote_removal (nestret, 0);
3596 free (nestret);
3597 hdlen = STRLEN(heredelim);
3598 /*itrace("parse_comsub:%d: found here doc delimiter `%s' (%d)", line_number, heredelim, hdlen);*/
3599 }
3600 if (ch == '\n')
3601 {
3602 tflags |= LEX_INHEREDOC;
3603 tflags &= ~LEX_HEREDELIM;
3604 lex_firstind = retind + 1;
3605 }
3606 else
3607 lex_firstind = -1;
3608 }
3609 }
3610
3611 /* Meta-characters that can introduce a reserved word. Not perfect yet. */
3612 if MBTEST((tflags & LEX_RESWDOK) == 0 && (tflags & LEX_CKCASE) && (tflags & LEX_INCOMMENT) == 0 && (shellmeta(ch) || ch == '\n'))
3613 {
3614 /* Add this character. */
3615 RESIZE_MALLOCED_BUFFER (ret, retind, 1, retsize, 64);
3616 ret[retind++] = ch;
3617 peekc = shell_getc (1);
3618 if (ch == peekc && (ch == '&' || ch == '|' || ch == ';')) /* two-character tokens */
3619 {
3620 RESIZE_MALLOCED_BUFFER (ret, retind, 1, retsize, 64);
3621 ret[retind++] = peekc;
3622 /*itrace("parse_comsub:%d: set lex_reswordok = 1, ch = `%c'", line_number, ch);*/
3623 tflags |= LEX_RESWDOK;
3624 lex_rwlen = 0;
3625 continue;
3626 }
3627 else if (ch == '\n' || COMSUB_META(ch))
3628 {
3629 shell_ungetc (peekc);
3630 /*itrace("parse_comsub:%d: set lex_reswordok = 1, ch = `%c'", line_number, ch);*/
3631 tflags |= LEX_RESWDOK;
3632 lex_rwlen = 0;
3633 continue;
3634 }
3635 else if (ch == EOF)
3636 goto eof_error;
3637 else
3638 {
3639 /* `unget' the character we just added and fall through */
3640 retind--;
3641 shell_ungetc (peekc);
3642 }
3643 }
3644
3645 /* If we can read a reserved word, try to read one. */
3646 if (tflags & LEX_RESWDOK)
3647 {
3648 if MBTEST(islower (ch))
3649 {
3650 /* Add this character. */
3651 RESIZE_MALLOCED_BUFFER (ret, retind, 1, retsize, 64);
3652 ret[retind++] = ch;
3653 lex_rwlen++;
3654 continue;
3655 }
3656 else if MBTEST(lex_rwlen == 4 && shellbreak (ch))
3657 {
3658 if (STREQN (ret + retind - 4, "case", 4))
3659 {
3660 tflags |= LEX_INCASE;
3661 /*itrace("parse_comsub:%d: found `case', lex_incase -> 1 lex_reswdok -> 0", line_number);*/
3662 }
3663 else if (STREQN (ret + retind - 4, "esac", 4))
3664 {
3665 tflags &= ~LEX_INCASE;
3666 /*itrace("parse_comsub:%d: found `esac', lex_incase -> 0 lex_reswdok -> 0", line_number);*/
3667 }
3668 tflags &= ~LEX_RESWDOK;
3669 }
3670 else if MBTEST((tflags & LEX_CKCOMMENT) && ch == '#' && (lex_rwlen == 0 || ((tflags & LEX_INWORD) && lex_wlen == 0)))
3671 ; /* don't modify LEX_RESWDOK if we're starting a comment */
3672 else if MBTEST((tflags & LEX_INCASE) && ch != '\n')
3673 /* If we can read a reserved word and we're in case, we're at the
3674 point where we can read a new pattern list or an esac. We
3675 handle the esac case above. If we read a newline, we want to
3676 leave LEX_RESWDOK alone. If we read anything else, we want to
3677 turn off LEX_RESWDOK, since we're going to read a pattern list. */
3678 {
3679 tflags &= ~LEX_RESWDOK;
3680 /*itrace("parse_comsub:%d: lex_incase == 1 found `%c', lex_reswordok -> 0", line_number, ch);*/
3681 }
3682 else if MBTEST(shellbreak (ch) == 0)
3683 {
3684 tflags &= ~LEX_RESWDOK;
3685 /*itrace("parse_comsub:%d: found `%c', lex_reswordok -> 0", line_number, ch);*/
3686 }
3687 }
3688
3689 /* Might be the start of a here-doc delimiter */
3690 if MBTEST((tflags & LEX_INCOMMENT) == 0 && (tflags & LEX_CKCASE) && ch == '<')
3691 {
3692 /* Add this character. */
3693 RESIZE_MALLOCED_BUFFER (ret, retind, 1, retsize, 64);
3694 ret[retind++] = ch;
3695 peekc = shell_getc (1);
3696 if (peekc == EOF)
3697 goto eof_error;
3698 if (peekc == ch)
3699 {
3700 RESIZE_MALLOCED_BUFFER (ret, retind, 1, retsize, 64);
3701 ret[retind++] = peekc;
3702 peekc = shell_getc (1);
3703 if (peekc == EOF)
3704 goto eof_error;
3705 if (peekc == '-')
3706 {
3707 RESIZE_MALLOCED_BUFFER (ret, retind, 1, retsize, 64);
3708 ret[retind++] = peekc;
3709 tflags |= LEX_STRIPDOC;
3710 }
3711 else
3712 shell_ungetc (peekc);
3713 if (peekc != '<')
3714 {
3715 tflags |= LEX_HEREDELIM;
3716 lex_firstind = -1;
3717 }
3718 continue;
3719 }
3720 else
3721 ch = peekc; /* fall through and continue XXX */
3722 }
3723 else if MBTEST((tflags & LEX_CKCOMMENT) && (tflags & LEX_INCOMMENT) == 0 && ch == '#' && (((tflags & LEX_RESWDOK) && lex_rwlen == 0) || ((tflags & LEX_INWORD) && lex_wlen == 0)))
3724 {
3725 /*itrace("parse_comsub:%d: lex_incomment -> 1 (%d)", line_number, __LINE__);*/
3726 tflags |= LEX_INCOMMENT;
3727 }
3728
3729 if MBTEST(ch == CTLESC || ch == CTLNUL) /* special shell escapes */
3730 {
3731 RESIZE_MALLOCED_BUFFER (ret, retind, 2, retsize, 64);
3732 ret[retind++] = CTLESC;
3733 ret[retind++] = ch;
3734 continue;
3735 }
3736 #if 0
3737 else if MBTEST((tflags & LEX_INCASE) && ch == close && close == ')')
3738 tflags &= ~LEX_INCASE; /* XXX */
3739 #endif
3740 else if MBTEST(ch == close && (tflags & LEX_INCASE) == 0) /* ending delimiter */
3741 {
3742 count--;
3743 /*itrace("parse_comsub:%d: found close: count = %d", line_number, count);*/
3744 }
3745 else if MBTEST(((flags & P_FIRSTCLOSE) == 0) && (tflags & LEX_INCASE) == 0 && ch == open) /* nested begin */
3746 {
3747 count++;
3748 /*itrace("parse_comsub:%d: found open: count = %d", line_number, count);*/
3749 }
3750
3751 /* Add this character. */
3752 RESIZE_MALLOCED_BUFFER (ret, retind, 1, retsize, 64);
3753 ret[retind++] = ch;
3754
3755 /* If we just read the ending character, don't bother continuing. */
3756 if (count == 0)
3757 break;
3758
3759 if MBTEST(ch == '\\') /* backslashes */
3760 tflags |= LEX_PASSNEXT;
3761
3762 if MBTEST(shellquote (ch))
3763 {
3764 /* '', ``, or "" inside $(...). */
3765 push_delimiter (dstack, ch);
3766 if MBTEST((tflags & LEX_WASDOL) && ch == '\'') /* $'...' inside group */
3767 nestret = parse_matched_pair (ch, ch, ch, &nestlen, P_ALLOWESC|rflags);
3768 else
3769 nestret = parse_matched_pair (ch, ch, ch, &nestlen, rflags);
3770 pop_delimiter (dstack);
3771 CHECK_NESTRET_ERROR ();
3772
3773 if MBTEST((tflags & LEX_WASDOL) && ch == '\'' && (extended_quote || (rflags & P_DQUOTE) == 0))
3774 {
3775 /* Translate $'...' here. */
3776 ttrans = ansiexpand (nestret, 0, nestlen - 1, &ttranslen);
3777 xfree (nestret);
3778
3779 if ((rflags & P_DQUOTE) == 0)
3780 {
3781 nestret = sh_single_quote (ttrans);
3782 free (ttrans);
3783 nestlen = strlen (nestret);
3784 }
3785 else
3786 {
3787 nestret = ttrans;
3788 nestlen = ttranslen;
3789 }
3790 retind -= 2; /* back up before the $' */
3791 }
3792 else if MBTEST((tflags & LEX_WASDOL) && ch == '"' && (extended_quote || (rflags & P_DQUOTE) == 0))
3793 {
3794 /* Locale expand $"..." here. */
3795 ttrans = localeexpand (nestret, 0, nestlen - 1, start_lineno, &ttranslen);
3796 xfree (nestret);
3797
3798 nestret = sh_mkdoublequoted (ttrans, ttranslen, 0);
3799 free (ttrans);
3800 nestlen = ttranslen + 2;
3801 retind -= 2; /* back up before the $" */
3802 }
3803
3804 APPEND_NESTRET ();
3805 FREE (nestret);
3806 }
3807 else if MBTEST((tflags & LEX_WASDOL) && (ch == '(' || ch == '{' || ch == '[')) /* ) } ] */
3808 /* check for $(), $[], or ${} inside command substitution. */
3809 {
3810 if ((tflags & LEX_INCASE) == 0 && open == ch) /* undo previous increment */
3811 count--;
3812 if (ch == '(') /* ) */
3813 nestret = parse_comsub (0, '(', ')', &nestlen, (rflags|P_COMMAND) & ~P_DQUOTE);
3814 else if (ch == '{') /* } */
3815 nestret = parse_matched_pair (0, '{', '}', &nestlen, P_FIRSTCLOSE|P_DOLBRACE|rflags);
3816 else if (ch == '[') /* ] */
3817 nestret = parse_matched_pair (0, '[', ']', &nestlen, rflags);
3818
3819 CHECK_NESTRET_ERROR ();
3820 APPEND_NESTRET ();
3821
3822 FREE (nestret);
3823 }
3824 if MBTEST(ch == '$')
3825 tflags |= LEX_WASDOL;
3826 else
3827 tflags &= ~LEX_WASDOL;
3828 }
3829
3830 FREE (heredelim);
3831 ret[retind] = '\0';
3832 if (lenp)
3833 *lenp = retind;
3834 /*itrace("parse_comsub:%d: returning `%s'", line_number, ret);*/
3835 return ret;
3836 }
3837
3838 /* Recursively call the parser to parse a $(...) command substitution. */
3839 char *
3840 xparse_dolparen (base, string, indp, flags)
3841 char *base;
3842 char *string;
3843 int *indp;
3844 int flags;
3845 {
3846 sh_parser_state_t ps;
3847 sh_input_line_state_t ls;
3848 int orig_ind, nc, sflags;
3849 char *ret, *s, *ep, *ostring;
3850
3851 /*yydebug = 1;*/
3852 orig_ind = *indp;
3853 ostring = string;
3854
3855 /*itrace("xparse_dolparen: size = %d shell_input_line = `%s'", shell_input_line_size, shell_input_line);*/
3856 sflags = SEVAL_NONINT|SEVAL_NOHIST|SEVAL_NOFREE;
3857 if (flags & SX_NOLONGJMP)
3858 sflags |= SEVAL_NOLONGJMP;
3859 save_parser_state (&ps);
3860 save_input_line_state (&ls);
3861
3862 /*(*/
3863 parser_state |= PST_CMDSUBST|PST_EOFTOKEN; /* allow instant ')' */ /*(*/
3864 shell_eof_token = ')';
3865 parse_string (string, "command substitution", sflags, &ep);
3866
3867 restore_parser_state (&ps);
3868 reset_parser ();
3869 /* reset_parser clears shell_input_line and associated variables */
3870 restore_input_line_state (&ls);
3871 if (interactive)
3872 token_to_read = 0;
3873
3874 /* Need to find how many characters parse_and_execute consumed, update
3875 *indp, if flags != 0, copy the portion of the string parsed into RET
3876 and return it. If flags & 1 (EX_NOALLOC) we can return NULL. */
3877
3878 /*(*/
3879 if (ep[-1] != ')')
3880 {
3881 #if DEBUG
3882 if (ep[-1] != '\n')
3883 itrace("xparse_dolparen:%d: ep[-1] != RPAREN (%d), ep = `%s'", line_number, ep[-1], ep);
3884 #endif
3885 while (ep > ostring && ep[-1] == '\n') ep--;
3886 }
3887
3888 nc = ep - ostring;
3889 *indp = ep - base - 1;
3890
3891 /*(*/
3892 #if DEBUG
3893 if (base[*indp] != ')')
3894 itrace("xparse_dolparen:%d: base[%d] != RPAREN (%d), base = `%s'", line_number, *indp, base[*indp], base);
3895 #endif
3896
3897 if (flags & SX_NOALLOC)
3898 return (char *)NULL;
3899
3900 if (nc == 0)
3901 {
3902 ret = xmalloc (1);
3903 ret[0] = '\0';
3904 }
3905 else
3906 ret = substring (ostring, 0, nc - 1);
3907
3908 return ret;
3909 }
3910
3911 #if defined (DPAREN_ARITHMETIC) || defined (ARITH_FOR_COMMAND)
3912 /* Parse a double-paren construct. It can be either an arithmetic
3913 command, an arithmetic `for' command, or a nested subshell. Returns
3914 the parsed token, -1 on error, or -2 if we didn't do anything and
3915 should just go on. */
3916 static int
3917 parse_dparen (c)
3918 int c;
3919 {
3920 int cmdtyp, sline;
3921 char *wval;
3922 WORD_DESC *wd;
3923
3924 #if defined (ARITH_FOR_COMMAND)
3925 if (last_read_token == FOR)
3926 {
3927 arith_for_lineno = line_number;
3928 cmdtyp = parse_arith_cmd (&wval, 0);
3929 if (cmdtyp == 1)
3930 {
3931 wd = alloc_word_desc ();
3932 wd->word = wval;
3933 yylval.word_list = make_word_list (wd, (WORD_LIST *)NULL);
3934 return (ARITH_FOR_EXPRS);
3935 }
3936 else
3937 return -1; /* ERROR */
3938 }
3939 #endif
3940
3941 #if defined (DPAREN_ARITHMETIC)
3942 if (reserved_word_acceptable (last_read_token))
3943 {
3944 sline = line_number;
3945
3946 cmdtyp = parse_arith_cmd (&wval, 0);
3947 if (cmdtyp == 1) /* arithmetic command */
3948 {
3949 wd = alloc_word_desc ();
3950 wd->word = wval;
3951 wd->flags = W_QUOTED|W_NOSPLIT|W_NOGLOB|W_DQUOTE;
3952 yylval.word_list = make_word_list (wd, (WORD_LIST *)NULL);
3953 return (ARITH_CMD);
3954 }
3955 else if (cmdtyp == 0) /* nested subshell */
3956 {
3957 push_string (wval, 0, (alias_t *)NULL);
3958 if ((parser_state & PST_CASEPAT) == 0)
3959 parser_state |= PST_SUBSHELL;
3960 return (c);
3961 }
3962 else /* ERROR */
3963 return -1;
3964 }
3965 #endif
3966
3967 return -2; /* XXX */
3968 }
3969
3970 /* We've seen a `(('. Look for the matching `))'. If we get it, return 1.
3971 If not, assume it's a nested subshell for backwards compatibility and
3972 return 0. In any case, put the characters we've consumed into a locally-
3973 allocated buffer and make *ep point to that buffer. Return -1 on an
3974 error, for example EOF. */
3975 static int
3976 parse_arith_cmd (ep, adddq)
3977 char **ep;
3978 int adddq;
3979 {
3980 int exp_lineno, rval, c;
3981 char *ttok, *tokstr;
3982 int ttoklen;
3983
3984 exp_lineno = line_number;
3985 ttok = parse_matched_pair (0, '(', ')', &ttoklen, 0);
3986 rval = 1;
3987 if (ttok == &matched_pair_error)
3988 return -1;
3989 /* Check that the next character is the closing right paren. If
3990 not, this is a syntax error. ( */
3991 c = shell_getc (0);
3992 if MBTEST(c != ')')
3993 rval = 0;
3994
3995 tokstr = (char *)xmalloc (ttoklen + 4);
3996
3997 /* if ADDDQ != 0 then (( ... )) -> "..." */
3998 if (rval == 1 && adddq) /* arith cmd, add double quotes */
3999 {
4000 tokstr[0] = '"';
4001 strncpy (tokstr + 1, ttok, ttoklen - 1);
4002 tokstr[ttoklen] = '"';
4003 tokstr[ttoklen+1] = '\0';
4004 }
4005 else if (rval == 1) /* arith cmd, don't add double quotes */
4006 {
4007 strncpy (tokstr, ttok, ttoklen - 1);
4008 tokstr[ttoklen-1] = '\0';
4009 }
4010 else /* nested subshell */
4011 {
4012 tokstr[0] = '(';
4013 strncpy (tokstr + 1, ttok, ttoklen - 1);
4014 tokstr[ttoklen] = ')';
4015 tokstr[ttoklen+1] = c;
4016 tokstr[ttoklen+2] = '\0';
4017 }
4018
4019 *ep = tokstr;
4020 FREE (ttok);
4021 return rval;
4022 }
4023 #endif /* DPAREN_ARITHMETIC || ARITH_FOR_COMMAND */
4024
4025 #if defined (COND_COMMAND)
4026 static void
4027 cond_error ()
4028 {
4029 char *etext;
4030
4031 if (EOF_Reached && cond_token != COND_ERROR) /* [[ */
4032 parser_error (cond_lineno, _("unexpected EOF while looking for `]]'"));
4033 else if (cond_token != COND_ERROR)
4034 {
4035 if (etext = error_token_from_token (cond_token))
4036 {
4037 parser_error (cond_lineno, _("syntax error in conditional expression: unexpected token `%s'"), etext);
4038 free (etext);
4039 }
4040 else
4041 parser_error (cond_lineno, _("syntax error in conditional expression"));
4042 }
4043 }
4044
4045 static COND_COM *
4046 cond_expr ()
4047 {
4048 return (cond_or ());
4049 }
4050
4051 static COND_COM *
4052 cond_or ()
4053 {
4054 COND_COM *l, *r;
4055
4056 l = cond_and ();
4057 if (cond_token == OR_OR)
4058 {
4059 r = cond_or ();
4060 l = make_cond_node (COND_OR, (WORD_DESC *)NULL, l, r);
4061 }
4062 return l;
4063 }
4064
4065 static COND_COM *
4066 cond_and ()
4067 {
4068 COND_COM *l, *r;
4069
4070 l = cond_term ();
4071 if (cond_token == AND_AND)
4072 {
4073 r = cond_and ();
4074 l = make_cond_node (COND_AND, (WORD_DESC *)NULL, l, r);
4075 }
4076 return l;
4077 }
4078
4079 static int
4080 cond_skip_newlines ()
4081 {
4082 while ((cond_token = read_token (READ)) == '\n')
4083 {
4084 if (SHOULD_PROMPT ())
4085 prompt_again ();
4086 }
4087 return (cond_token);
4088 }
4089
4090 #define COND_RETURN_ERROR() \
4091 do { cond_token = COND_ERROR; return ((COND_COM *)NULL); } while (0)
4092
4093 static COND_COM *
4094 cond_term ()
4095 {
4096 WORD_DESC *op;
4097 COND_COM *term, *tleft, *tright;
4098 int tok, lineno;
4099 char *etext;
4100
4101 /* Read a token. It can be a left paren, a `!', a unary operator, or a
4102 word that should be the first argument of a binary operator. Start by
4103 skipping newlines, since this is a compound command. */
4104 tok = cond_skip_newlines ();
4105 lineno = line_number;
4106 if (tok == COND_END)
4107 {
4108 COND_RETURN_ERROR ();
4109 }
4110 else if (tok == '(')
4111 {
4112 term = cond_expr ();
4113 if (cond_token != ')')
4114 {
4115 if (term)
4116 dispose_cond_node (term); /* ( */
4117 if (etext = error_token_from_token (cond_token))
4118 {
4119 parser_error (lineno, _("unexpected token `%s', expected `)'"), etext);
4120 free (etext);
4121 }
4122 else
4123 parser_error (lineno, _("expected `)'"));
4124 COND_RETURN_ERROR ();
4125 }
4126 term = make_cond_node (COND_EXPR, (WORD_DESC *)NULL, term, (COND_COM *)NULL);
4127 (void)cond_skip_newlines ();
4128 }
4129 else if (tok == BANG || (tok == WORD && (yylval.word->word[0] == '!' && yylval.word->word[1] == '\0')))
4130 {
4131 if (tok == WORD)
4132 dispose_word (yylval.word); /* not needed */
4133 term = cond_term ();
4134 if (term)
4135 term->flags |= CMD_INVERT_RETURN;
4136 }
4137 else if (tok == WORD && yylval.word->word[0] == '-' && yylval.word->word[2] == 0 && test_unop (yylval.word->word))
4138 {
4139 op = yylval.word;
4140 tok = read_token (READ);
4141 if (tok == WORD)
4142 {
4143 tleft = make_cond_node (COND_TERM, yylval.word, (COND_COM *)NULL, (COND_COM *)NULL);
4144 term = make_cond_node (COND_UNARY, op, tleft, (COND_COM *)NULL);
4145 }
4146 else
4147 {
4148 dispose_word (op);
4149 if (etext = error_token_from_token (tok))
4150 {
4151 parser_error (line_number, _("unexpected argument `%s' to conditional unary operator"), etext);
4152 free (etext);
4153 }
4154 else
4155 parser_error (line_number, _("unexpected argument to conditional unary operator"));
4156 COND_RETURN_ERROR ();
4157 }
4158
4159 (void)cond_skip_newlines ();
4160 }
4161 else if (tok == WORD) /* left argument to binary operator */
4162 {
4163 /* lhs */
4164 tleft = make_cond_node (COND_TERM, yylval.word, (COND_COM *)NULL, (COND_COM *)NULL);
4165
4166 /* binop */
4167 tok = read_token (READ);
4168 if (tok == WORD && test_binop (yylval.word->word))
4169 {
4170 op = yylval.word;
4171 if (op->word[0] == '=' && (op->word[1] == '\0' || (op->word[1] == '=' && op->word[2] == '\0')))
4172 parser_state |= PST_EXTPAT;
4173 else if (op->word[0] == '!' && op->word[1] == '=' && op->word[2] == '\0')
4174 parser_state |= PST_EXTPAT;
4175 }
4176 #if defined (COND_REGEXP)
4177 else if (tok == WORD && STREQ (yylval.word->word, "=~"))
4178 {
4179 op = yylval.word;
4180 parser_state |= PST_REGEXP;
4181 }
4182 #endif
4183 else if (tok == '<' || tok == '>')
4184 op = make_word_from_token (tok); /* ( */
4185 /* There should be a check before blindly accepting the `)' that we have
4186 seen the opening `('. */
4187 else if (tok == COND_END || tok == AND_AND || tok == OR_OR || tok == ')')
4188 {
4189 /* Special case. [[ x ]] is equivalent to [[ -n x ]], just like
4190 the test command. Similarly for [[ x && expr ]] or
4191 [[ x || expr ]] or [[ (x) ]]. */
4192 op = make_word ("-n");
4193 term = make_cond_node (COND_UNARY, op, tleft, (COND_COM *)NULL);
4194 cond_token = tok;
4195 return (term);
4196 }
4197 else
4198 {
4199 if (etext = error_token_from_token (tok))
4200 {
4201 parser_error (line_number, _("unexpected token `%s', conditional binary operator expected"), etext);
4202 free (etext);
4203 }
4204 else
4205 parser_error (line_number, _("conditional binary operator expected"));
4206 dispose_cond_node (tleft);
4207 COND_RETURN_ERROR ();
4208 }
4209
4210 /* rhs */
4211 if (parser_state & PST_EXTPAT)
4212 extended_glob = 1;
4213 tok = read_token (READ);
4214 if (parser_state & PST_EXTPAT)
4215 extended_glob = global_extglob;
4216 parser_state &= ~(PST_REGEXP|PST_EXTPAT);
4217
4218 if (tok == WORD)
4219 {
4220 tright = make_cond_node (COND_TERM, yylval.word, (COND_COM *)NULL, (COND_COM *)NULL);
4221 term = make_cond_node (COND_BINARY, op, tleft, tright);
4222 }
4223 else
4224 {
4225 if (etext = error_token_from_token (tok))
4226 {
4227 parser_error (line_number, _("unexpected argument `%s' to conditional binary operator"), etext);
4228 free (etext);
4229 }
4230 else
4231 parser_error (line_number, _("unexpected argument to conditional binary operator"));
4232 dispose_cond_node (tleft);
4233 dispose_word (op);
4234 COND_RETURN_ERROR ();
4235 }
4236
4237 (void)cond_skip_newlines ();
4238 }
4239 else
4240 {
4241 if (tok < 256)
4242 parser_error (line_number, _("unexpected token `%c' in conditional command"), tok);
4243 else if (etext = error_token_from_token (tok))
4244 {
4245 parser_error (line_number, _("unexpected token `%s' in conditional command"), etext);
4246 free (etext);
4247 }
4248 else
4249 parser_error (line_number, _("unexpected token %d in conditional command"), tok);
4250 COND_RETURN_ERROR ();
4251 }
4252 return (term);
4253 }
4254
4255 /* This is kind of bogus -- we slip a mini recursive-descent parser in
4256 here to handle the conditional statement syntax. */
4257 static COMMAND *
4258 parse_cond_command ()
4259 {
4260 COND_COM *cexp;
4261
4262 global_extglob = extended_glob;
4263 cexp = cond_expr ();
4264 return (make_cond_command (cexp));
4265 }
4266 #endif
4267
4268 #if defined (ARRAY_VARS)
4269 /* When this is called, it's guaranteed that we don't care about anything
4270 in t beyond i. We do save and restore the chars, though. */
4271 static int
4272 token_is_assignment (t, i)
4273 char *t;
4274 int i;
4275 {
4276 unsigned char c, c1;
4277 int r;
4278
4279 c = t[i]; c1 = t[i+1];
4280 t[i] = '='; t[i+1] = '\0';
4281 r = assignment (t, (parser_state & PST_COMPASSIGN) != 0);
4282 t[i] = c; t[i+1] = c1;
4283 return r;
4284 }
4285
4286 /* XXX - possible changes here for `+=' */
4287 static int
4288 token_is_ident (t, i)
4289 char *t;
4290 int i;
4291 {
4292 unsigned char c;
4293 int r;
4294
4295 c = t[i];
4296 t[i] = '\0';
4297 r = legal_identifier (t);
4298 t[i] = c;
4299 return r;
4300 }
4301 #endif
4302
4303 static int
4304 read_token_word (character)
4305 int character;
4306 {
4307 /* The value for YYLVAL when a WORD is read. */
4308 WORD_DESC *the_word;
4309
4310 /* Index into the token that we are building. */
4311 int token_index;
4312
4313 /* ALL_DIGITS becomes zero when we see a non-digit. */
4314 int all_digit_token;
4315
4316 /* DOLLAR_PRESENT becomes non-zero if we see a `$'. */
4317 int dollar_present;
4318
4319 /* COMPOUND_ASSIGNMENT becomes non-zero if we are parsing a compound
4320 assignment. */
4321 int compound_assignment;
4322
4323 /* QUOTED becomes non-zero if we see one of ("), ('), (`), or (\). */
4324 int quoted;
4325
4326 /* Non-zero means to ignore the value of the next character, and just
4327 to add it no matter what. */
4328 int pass_next_character;
4329
4330 /* The current delimiting character. */
4331 int cd;
4332 int result, peek_char;
4333 char *ttok, *ttrans;
4334 int ttoklen, ttranslen;
4335 intmax_t lvalue;
4336
4337 if (token_buffer_size < TOKEN_DEFAULT_INITIAL_SIZE)
4338 token = (char *)xrealloc (token, token_buffer_size = TOKEN_DEFAULT_INITIAL_SIZE);
4339
4340 token_index = 0;
4341 all_digit_token = DIGIT (character);
4342 dollar_present = quoted = pass_next_character = compound_assignment = 0;
4343
4344 for (;;)
4345 {
4346 if (character == EOF)
4347 goto got_token;
4348
4349 if (pass_next_character)
4350 {
4351 pass_next_character = 0;
4352 goto got_escaped_character;
4353 }
4354
4355 cd = current_delimiter (dstack);
4356
4357 /* Handle backslashes. Quote lots of things when not inside of
4358 double-quotes, quote some things inside of double-quotes. */
4359 if MBTEST(character == '\\')
4360 {
4361 peek_char = shell_getc (0);
4362
4363 /* Backslash-newline is ignored in all cases except
4364 when quoted with single quotes. */
4365 if (peek_char == '\n')
4366 {
4367 character = '\n';
4368 goto next_character;
4369 }
4370 else
4371 {
4372 shell_ungetc (peek_char);
4373
4374 /* If the next character is to be quoted, note it now. */
4375 if (cd == 0 || cd == '`' ||
4376 (cd == '"' && peek_char >= 0 && (sh_syntaxtab[peek_char] & CBSDQUOTE)))
4377 pass_next_character++;
4378
4379 quoted = 1;
4380 goto got_character;
4381 }
4382 }
4383
4384 /* Parse a matched pair of quote characters. */
4385 if MBTEST(shellquote (character))
4386 {
4387 push_delimiter (dstack, character);
4388 ttok = parse_matched_pair (character, character, character, &ttoklen, (character == '`') ? P_COMMAND : 0);
4389 pop_delimiter (dstack);
4390 if (ttok == &matched_pair_error)
4391 return -1; /* Bail immediately. */
4392 RESIZE_MALLOCED_BUFFER (token, token_index, ttoklen + 2,
4393 token_buffer_size, TOKEN_DEFAULT_GROW_SIZE);
4394 token[token_index++] = character;
4395 strcpy (token + token_index, ttok);
4396 token_index += ttoklen;
4397 all_digit_token = 0;
4398 quoted = 1;
4399 dollar_present |= (character == '"' && strchr (ttok, '$') != 0);
4400 FREE (ttok);
4401 goto next_character;
4402 }
4403
4404 #ifdef COND_REGEXP
4405 /* When parsing a regexp as a single word inside a conditional command,
4406 we need to special-case characters special to both the shell and
4407 regular expressions. Right now, that is only '(' and '|'. */ /*)*/
4408 if MBTEST((parser_state & PST_REGEXP) && (character == '(' || character == '|')) /*)*/
4409 {
4410 if (character == '|')
4411 goto got_character;
4412
4413 push_delimiter (dstack, character);
4414 ttok = parse_matched_pair (cd, '(', ')', &ttoklen, 0);
4415 pop_delimiter (dstack);
4416 if (ttok == &matched_pair_error)
4417 return -1; /* Bail immediately. */
4418 RESIZE_MALLOCED_BUFFER (token, token_index, ttoklen + 2,
4419 token_buffer_size, TOKEN_DEFAULT_GROW_SIZE);
4420 token[token_index++] = character;
4421 strcpy (token + token_index, ttok);
4422 token_index += ttoklen;
4423 FREE (ttok);
4424 dollar_present = all_digit_token = 0;
4425 goto next_character;
4426 }
4427 #endif /* COND_REGEXP */
4428
4429 #ifdef EXTENDED_GLOB
4430 /* Parse a ksh-style extended pattern matching specification. */
4431 if MBTEST(extended_glob && PATTERN_CHAR (character))
4432 {
4433 peek_char = shell_getc (1);
4434 if MBTEST(peek_char == '(') /* ) */
4435 {
4436 push_delimiter (dstack, peek_char);
4437 ttok = parse_matched_pair (cd, '(', ')', &ttoklen, 0);
4438 pop_delimiter (dstack);
4439 if (ttok == &matched_pair_error)
4440 return -1; /* Bail immediately. */
4441 RESIZE_MALLOCED_BUFFER (token, token_index, ttoklen + 2,
4442 token_buffer_size,
4443 TOKEN_DEFAULT_GROW_SIZE);
4444 token[token_index++] = character;
4445 token[token_index++] = peek_char;
4446 strcpy (token + token_index, ttok);
4447 token_index += ttoklen;
4448 FREE (ttok);
4449 dollar_present = all_digit_token = 0;
4450 goto next_character;
4451 }
4452 else
4453 shell_ungetc (peek_char);
4454 }
4455 #endif /* EXTENDED_GLOB */
4456
4457 /* If the delimiter character is not single quote, parse some of
4458 the shell expansions that must be read as a single word. */
4459 if (shellexp (character))
4460 {
4461 peek_char = shell_getc (1);
4462 /* $(...), <(...), >(...), $((...)), ${...}, and $[...] constructs */
4463 if MBTEST(peek_char == '(' || \
4464 ((peek_char == '{' || peek_char == '[') && character == '$')) /* ) ] } */
4465 {
4466 if (peek_char == '{') /* } */
4467 ttok = parse_matched_pair (cd, '{', '}', &ttoklen, P_FIRSTCLOSE|P_DOLBRACE);
4468 else if (peek_char == '(') /* ) */
4469 {
4470 /* XXX - push and pop the `(' as a delimiter for use by
4471 the command-oriented-history code. This way newlines
4472 appearing in the $(...) string get added to the
4473 history literally rather than causing a possibly-
4474 incorrect `;' to be added. ) */
4475 push_delimiter (dstack, peek_char);
4476 ttok = parse_comsub (cd, '(', ')', &ttoklen, P_COMMAND);
4477 pop_delimiter (dstack);
4478 }
4479 else
4480 ttok = parse_matched_pair (cd, '[', ']', &ttoklen, 0);
4481 if (ttok == &matched_pair_error)
4482 return -1; /* Bail immediately. */
4483 RESIZE_MALLOCED_BUFFER (token, token_index, ttoklen + 2,
4484 token_buffer_size,
4485 TOKEN_DEFAULT_GROW_SIZE);
4486 token[token_index++] = character;
4487 token[token_index++] = peek_char;
4488 strcpy (token + token_index, ttok);
4489 token_index += ttoklen;
4490 FREE (ttok);
4491 dollar_present = 1;
4492 all_digit_token = 0;
4493 goto next_character;
4494 }
4495 /* This handles $'...' and $"..." new-style quoted strings. */
4496 else if MBTEST(character == '$' && (peek_char == '\'' || peek_char == '"'))
4497 {
4498 int first_line;
4499
4500 first_line = line_number;
4501 push_delimiter (dstack, peek_char);
4502 ttok = parse_matched_pair (peek_char, peek_char, peek_char,
4503 &ttoklen,
4504 (peek_char == '\'') ? P_ALLOWESC : 0);
4505 pop_delimiter (dstack);
4506 if (ttok == &matched_pair_error)
4507 return -1;
4508 if (peek_char == '\'')
4509 {
4510 ttrans = ansiexpand (ttok, 0, ttoklen - 1, &ttranslen);
4511 free (ttok);
4512
4513 /* Insert the single quotes and correctly quote any
4514 embedded single quotes (allowed because P_ALLOWESC was
4515 passed to parse_matched_pair). */
4516 ttok = sh_single_quote (ttrans);
4517 free (ttrans);
4518 ttranslen = strlen (ttok);
4519 ttrans = ttok;
4520 }
4521 else
4522 {
4523 /* Try to locale-expand the converted string. */
4524 ttrans = localeexpand (ttok, 0, ttoklen - 1, first_line, &ttranslen);
4525 free (ttok);
4526
4527 /* Add the double quotes back */
4528 ttok = sh_mkdoublequoted (ttrans, ttranslen, 0);
4529 free (ttrans);
4530 ttranslen += 2;
4531 ttrans = ttok;
4532 }
4533
4534 RESIZE_MALLOCED_BUFFER (token, token_index, ttranslen + 2,
4535 token_buffer_size,
4536 TOKEN_DEFAULT_GROW_SIZE);
4537 strcpy (token + token_index, ttrans);
4538 token_index += ttranslen;
4539 FREE (ttrans);
4540 quoted = 1;
4541 all_digit_token = 0;
4542 goto next_character;
4543 }
4544 /* This could eventually be extended to recognize all of the
4545 shell's single-character parameter expansions, and set flags.*/
4546 else if MBTEST(character == '$' && peek_char == '$')
4547 {
4548 ttok = (char *)xmalloc (3);
4549 ttok[0] = ttok[1] = '$';
4550 ttok[2] = '\0';
4551 RESIZE_MALLOCED_BUFFER (token, token_index, 3,
4552 token_buffer_size,
4553 TOKEN_DEFAULT_GROW_SIZE);
4554 strcpy (token + token_index, ttok);
4555 token_index += 2;
4556 dollar_present = 1;
4557 all_digit_token = 0;
4558 FREE (ttok);
4559 goto next_character;
4560 }
4561 else
4562 shell_ungetc (peek_char);
4563 }
4564
4565 #if defined (ARRAY_VARS)
4566 /* Identify possible array subscript assignment; match [...]. If
4567 parser_state&PST_COMPASSIGN, we need to parse [sub]=words treating
4568 `sub' as if it were enclosed in double quotes. */
4569 else if MBTEST(character == '[' && /* ] */
4570 ((token_index > 0 && assignment_acceptable (last_read_token) && token_is_ident (token, token_index)) ||
4571 (token_index == 0 && (parser_state&PST_COMPASSIGN))))
4572 {
4573 ttok = parse_matched_pair (cd, '[', ']', &ttoklen, P_ARRAYSUB);
4574 if (ttok == &matched_pair_error)
4575 return -1; /* Bail immediately. */
4576 RESIZE_MALLOCED_BUFFER (token, token_index, ttoklen + 2,
4577 token_buffer_size,
4578 TOKEN_DEFAULT_GROW_SIZE);
4579 token[token_index++] = character;
4580 strcpy (token + token_index, ttok);
4581 token_index += ttoklen;
4582 FREE (ttok);
4583 all_digit_token = 0;
4584 goto next_character;
4585 }
4586 /* Identify possible compound array variable assignment. */
4587 else if MBTEST(character == '=' && token_index > 0 && (assignment_acceptable (last_read_token) || (parser_state & PST_ASSIGNOK)) && token_is_assignment (token, token_index))
4588 {
4589 peek_char = shell_getc (1);
4590 if MBTEST(peek_char == '(') /* ) */
4591 {
4592 ttok = parse_compound_assignment (&ttoklen);
4593
4594 RESIZE_MALLOCED_BUFFER (token, token_index, ttoklen + 4,
4595 token_buffer_size,
4596 TOKEN_DEFAULT_GROW_SIZE);
4597
4598 token[token_index++] = '=';
4599 token[token_index++] = '(';
4600 if (ttok)
4601 {
4602 strcpy (token + token_index, ttok);
4603 token_index += ttoklen;
4604 }
4605 token[token_index++] = ')';
4606 FREE (ttok);
4607 all_digit_token = 0;
4608 compound_assignment = 1;
4609 #if 1
4610 goto next_character;
4611 #else
4612 goto got_token; /* ksh93 seems to do this */
4613 #endif
4614 }
4615 else
4616 shell_ungetc (peek_char);
4617 }
4618 #endif
4619
4620 /* When not parsing a multi-character word construct, shell meta-
4621 characters break words. */
4622 if MBTEST(shellbreak (character))
4623 {
4624 shell_ungetc (character);
4625 goto got_token;
4626 }
4627
4628 got_character:
4629
4630 if (character == CTLESC || character == CTLNUL)
4631 token[token_index++] = CTLESC;
4632
4633 got_escaped_character:
4634
4635 all_digit_token &= DIGIT (character);
4636 dollar_present |= character == '$';
4637
4638 token[token_index++] = character;
4639
4640 RESIZE_MALLOCED_BUFFER (token, token_index, 1, token_buffer_size,
4641 TOKEN_DEFAULT_GROW_SIZE);
4642
4643 next_character:
4644 if (character == '\n' && SHOULD_PROMPT ())
4645 prompt_again ();
4646
4647 /* We want to remove quoted newlines (that is, a \<newline> pair)
4648 unless we are within single quotes or pass_next_character is
4649 set (the shell equivalent of literal-next). */
4650 cd = current_delimiter (dstack);
4651 character = shell_getc (cd != '\'' && pass_next_character == 0);
4652 } /* end for (;;) */
4653
4654 got_token:
4655
4656 token[token_index] = '\0';
4657
4658 /* Check to see what thing we should return. If the last_read_token
4659 is a `<', or a `&', or the character which ended this token is
4660 a '>' or '<', then, and ONLY then, is this input token a NUMBER.
4661 Otherwise, it is just a word, and should be returned as such. */
4662 if MBTEST(all_digit_token && (character == '<' || character == '>' || \
4663 last_read_token == LESS_AND || \
4664 last_read_token == GREATER_AND))
4665 {
4666 if (legal_number (token, &lvalue) && (int)lvalue == lvalue)
4667 yylval.number = lvalue;
4668 else
4669 yylval.number = -1;
4670 return (NUMBER);
4671 }
4672
4673 /* Check for special case tokens. */
4674 result = (last_shell_getc_is_singlebyte) ? special_case_tokens (token) : -1;
4675 if (result >= 0)
4676 return result;
4677
4678 #if defined (ALIAS)
4679 /* Posix.2 does not allow reserved words to be aliased, so check for all
4680 of them, including special cases, before expanding the current token
4681 as an alias. */
4682 if MBTEST(posixly_correct)
4683 CHECK_FOR_RESERVED_WORD (token);
4684
4685 /* Aliases are expanded iff EXPAND_ALIASES is non-zero, and quoting
4686 inhibits alias expansion. */
4687 if (expand_aliases && quoted == 0)
4688 {
4689 result = alias_expand_token (token);
4690 if (result == RE_READ_TOKEN)
4691 return (RE_READ_TOKEN);
4692 else if (result == NO_EXPANSION)
4693 parser_state &= ~PST_ALEXPNEXT;
4694 }
4695
4696 /* If not in Posix.2 mode, check for reserved words after alias
4697 expansion. */
4698 if MBTEST(posixly_correct == 0)
4699 #endif
4700 CHECK_FOR_RESERVED_WORD (token);
4701
4702 the_word = (WORD_DESC *)xmalloc (sizeof (WORD_DESC));
4703 the_word->word = (char *)xmalloc (1 + token_index);
4704 the_word->flags = 0;
4705 strcpy (the_word->word, token);
4706 if (dollar_present)
4707 the_word->flags |= W_HASDOLLAR;
4708 if (quoted)
4709 the_word->flags |= W_QUOTED; /*(*/
4710 if (compound_assignment && token[token_index-1] == ')')
4711 the_word->flags |= W_COMPASSIGN;
4712 /* A word is an assignment if it appears at the beginning of a
4713 simple command, or after another assignment word. This is
4714 context-dependent, so it cannot be handled in the grammar. */
4715 if (assignment (token, (parser_state & PST_COMPASSIGN) != 0))
4716 {
4717 the_word->flags |= W_ASSIGNMENT;
4718 /* Don't perform word splitting on assignment statements. */
4719 if (assignment_acceptable (last_read_token) || (parser_state & PST_COMPASSIGN) != 0)
4720 the_word->flags |= W_NOSPLIT;
4721 }
4722
4723 if (command_token_position (last_read_token))
4724 {
4725 struct builtin *b;
4726 b = builtin_address_internal (token, 0);
4727 if (b && (b->flags & ASSIGNMENT_BUILTIN))
4728 parser_state |= PST_ASSIGNOK;
4729 else if (STREQ (token, "eval") || STREQ (token, "let"))
4730 parser_state |= PST_ASSIGNOK;
4731 }
4732
4733 yylval.word = the_word;
4734
4735 if (token[0] == '{' && token[token_index-1] == '}' &&
4736 (character == '<' || character == '>'))
4737 {
4738 /* can use token; already copied to the_word */
4739 token[token_index-1] = '\0';
4740 if (legal_identifier (token+1))
4741 {
4742 strcpy (the_word->word, token+1);
4743 /*itrace("read_token_word: returning REDIR_WORD for %s", the_word->word);*/
4744 return (REDIR_WORD);
4745 }
4746 }
4747
4748 result = ((the_word->flags & (W_ASSIGNMENT|W_NOSPLIT)) == (W_ASSIGNMENT|W_NOSPLIT))
4749 ? ASSIGNMENT_WORD : WORD;
4750
4751 switch (last_read_token)
4752 {
4753 case FUNCTION:
4754 parser_state |= PST_ALLOWOPNBRC;
4755 function_dstart = line_number;
4756 break;
4757 case CASE:
4758 case SELECT:
4759 case FOR:
4760 if (word_top < MAX_CASE_NEST)
4761 word_top++;
4762 word_lineno[word_top] = line_number;
4763 break;
4764 }
4765
4766 return (result);
4767 }
4768
4769 /* Return 1 if TOKSYM is a token that after being read would allow
4770 a reserved word to be seen, else 0. */
4771 static int
4772 reserved_word_acceptable (toksym)
4773 int toksym;
4774 {
4775 switch (toksym)
4776 {
4777 case '\n':
4778 case ';':
4779 case '(':
4780 case ')':
4781 case '|':
4782 case '&':
4783 case '{':
4784 case '}': /* XXX */
4785 case AND_AND:
4786 case BANG:
4787 case BAR_AND:
4788 case DO:
4789 case DONE:
4790 case ELIF:
4791 case ELSE:
4792 case ESAC:
4793 case FI:
4794 case IF:
4795 case OR_OR:
4796 case SEMI_SEMI:
4797 case SEMI_AND:
4798 case SEMI_SEMI_AND:
4799 case THEN:
4800 case TIME:
4801 case TIMEOPT:
4802 case TIMEIGN:
4803 case COPROC:
4804 case UNTIL:
4805 case WHILE:
4806 case 0:
4807 return 1;
4808 default:
4809 #if defined (COPROCESS_SUPPORT)
4810 if (last_read_token == WORD && token_before_that == COPROC)
4811 return 1;
4812 #endif
4813 if (last_read_token == WORD && token_before_that == FUNCTION)
4814 return 1;
4815 return 0;
4816 }
4817 }
4818
4819 /* Return the index of TOKEN in the alist of reserved words, or -1 if
4820 TOKEN is not a shell reserved word. */
4821 int
4822 find_reserved_word (tokstr)
4823 char *tokstr;
4824 {
4825 int i;
4826 for (i = 0; word_token_alist[i].word; i++)
4827 if (STREQ (tokstr, word_token_alist[i].word))
4828 return i;
4829 return -1;
4830 }
4831
4832 #if 0
4833 #if defined (READLINE)
4834 /* Called after each time readline is called. This insures that whatever
4835 the new prompt string is gets propagated to readline's local prompt
4836 variable. */
4837 static void
4838 reset_readline_prompt ()
4839 {
4840 char *temp_prompt;
4841
4842 if (prompt_string_pointer)
4843 {
4844 temp_prompt = (*prompt_string_pointer)
4845 ? decode_prompt_string (*prompt_string_pointer)
4846 : (char *)NULL;
4847
4848 if (temp_prompt == 0)
4849 {
4850 temp_prompt = (char *)xmalloc (1);
4851 temp_prompt[0] = '\0';
4852 }
4853
4854 FREE (current_readline_prompt);
4855 current_readline_prompt = temp_prompt;
4856 }
4857 }
4858 #endif /* READLINE */
4859 #endif /* 0 */
4860
4861 #if defined (HISTORY)
4862 /* A list of tokens which can be followed by newlines, but not by
4863 semi-colons. When concatenating multiple lines of history, the
4864 newline separator for such tokens is replaced with a space. */
4865 static const int no_semi_successors[] = {
4866 '\n', '{', '(', ')', ';', '&', '|',
4867 CASE, DO, ELSE, IF, SEMI_SEMI, SEMI_AND, SEMI_SEMI_AND, THEN, UNTIL,
4868 WHILE, AND_AND, OR_OR, IN,
4869 0
4870 };
4871
4872 /* If we are not within a delimited expression, try to be smart
4873 about which separators can be semi-colons and which must be
4874 newlines. Returns the string that should be added into the
4875 history entry. LINE is the line we're about to add; it helps
4876 make some more intelligent decisions in certain cases. */
4877 char *
4878 history_delimiting_chars (line)
4879 const char *line;
4880 {
4881 static int last_was_heredoc = 0; /* was the last entry the start of a here document? */
4882 register int i;
4883
4884 if ((parser_state & PST_HEREDOC) == 0)
4885 last_was_heredoc = 0;
4886
4887 if (dstack.delimiter_depth != 0)
4888 return ("\n");
4889
4890 /* We look for current_command_line_count == 2 because we are looking to
4891 add the first line of the body of the here document (the second line
4892 of the command). We also keep LAST_WAS_HEREDOC as a private sentinel
4893 variable to note when we think we added the first line of a here doc
4894 (the one with a "<<" somewhere in it) */
4895 if (parser_state & PST_HEREDOC)
4896 {
4897 if (last_was_heredoc)
4898 {
4899 last_was_heredoc = 0;
4900 return "\n";
4901 }
4902 return (current_command_line_count == 2 ? "\n" : "");
4903 }
4904
4905 if (parser_state & PST_COMPASSIGN)
4906 return (" ");
4907
4908 /* First, handle some special cases. */
4909 /*(*/
4910 /* If we just read `()', assume it's a function definition, and don't
4911 add a semicolon. If the token before the `)' was not `(', and we're
4912 not in the midst of parsing a case statement, assume it's a
4913 parenthesized command and add the semicolon. */
4914 /*)(*/
4915 if (token_before_that == ')')
4916 {
4917 if (two_tokens_ago == '(') /*)*/ /* function def */
4918 return " ";
4919 /* This does not work for subshells inside case statement
4920 command lists. It's a suboptimal solution. */
4921 else if (parser_state & PST_CASESTMT) /* case statement pattern */
4922 return " ";
4923 else
4924 return "; "; /* (...) subshell */
4925 }
4926 else if (token_before_that == WORD && two_tokens_ago == FUNCTION)
4927 return " "; /* function def using `function name' without `()' */
4928
4929 /* If we're not in a here document, but we think we're about to parse one,
4930 and we would otherwise return a `;', return a newline to delimit the
4931 line with the here-doc delimiter */
4932 else if ((parser_state & PST_HEREDOC) == 0 && current_command_line_count > 1 && last_read_token == '\n' && strstr (line, "<<"))
4933 {
4934 last_was_heredoc = 1;
4935 return "\n";
4936 }
4937
4938 else if (token_before_that == WORD && two_tokens_ago == FOR)
4939 {
4940 /* Tricky. `for i\nin ...' should not have a semicolon, but
4941 `for i\ndo ...' should. We do what we can. */
4942 for (i = shell_input_line_index; whitespace (shell_input_line[i]); i++)
4943 ;
4944 if (shell_input_line[i] && shell_input_line[i] == 'i' && shell_input_line[i+1] == 'n')
4945 return " ";
4946 return ";";
4947 }
4948 else if (two_tokens_ago == CASE && token_before_that == WORD && (parser_state & PST_CASESTMT))
4949 return " ";
4950
4951 for (i = 0; no_semi_successors[i]; i++)
4952 {
4953 if (token_before_that == no_semi_successors[i])
4954 return (" ");
4955 }
4956
4957 return ("; ");
4958 }
4959 #endif /* HISTORY */
4960
4961 /* Issue a prompt, or prepare to issue a prompt when the next character
4962 is read. */
4963 static void
4964 prompt_again ()
4965 {
4966 char *temp_prompt;
4967
4968 if (interactive == 0 || expanding_alias ()) /* XXX */
4969 return;
4970
4971 ps1_prompt = get_string_value ("PS1");
4972 ps2_prompt = get_string_value ("PS2");
4973
4974 if (!prompt_string_pointer)
4975 prompt_string_pointer = &ps1_prompt;
4976
4977 temp_prompt = *prompt_string_pointer
4978 ? decode_prompt_string (*prompt_string_pointer)
4979 : (char *)NULL;
4980
4981 if (temp_prompt == 0)
4982 {
4983 temp_prompt = (char *)xmalloc (1);
4984 temp_prompt[0] = '\0';
4985 }
4986
4987 current_prompt_string = *prompt_string_pointer;
4988 prompt_string_pointer = &ps2_prompt;
4989
4990 #if defined (READLINE)
4991 if (!no_line_editing)
4992 {
4993 FREE (current_readline_prompt);
4994 current_readline_prompt = temp_prompt;
4995 }
4996 else
4997 #endif /* READLINE */
4998 {
4999 FREE (current_decoded_prompt);
5000 current_decoded_prompt = temp_prompt;
5001 }
5002 }
5003
5004 int
5005 get_current_prompt_level ()
5006 {
5007 return ((current_prompt_string && current_prompt_string == ps2_prompt) ? 2 : 1);
5008 }
5009
5010 void
5011 set_current_prompt_level (x)
5012 int x;
5013 {
5014 prompt_string_pointer = (x == 2) ? &ps2_prompt : &ps1_prompt;
5015 current_prompt_string = *prompt_string_pointer;
5016 }
5017
5018 static void
5019 print_prompt ()
5020 {
5021 fprintf (stderr, "%s", current_decoded_prompt);
5022 fflush (stderr);
5023 }
5024
5025 /* Return a string which will be printed as a prompt. The string
5026 may contain special characters which are decoded as follows:
5027
5028 \a bell (ascii 07)
5029 \d the date in Day Mon Date format
5030 \e escape (ascii 033)
5031 \h the hostname up to the first `.'
5032 \H the hostname
5033 \j the number of active jobs
5034 \l the basename of the shell's tty device name
5035 \n CRLF
5036 \r CR
5037 \s the name of the shell
5038 \t the time in 24-hour hh:mm:ss format
5039 \T the time in 12-hour hh:mm:ss format
5040 \@ the time in 12-hour hh:mm am/pm format
5041 \A the time in 24-hour hh:mm format
5042 \D{fmt} the result of passing FMT to strftime(3)
5043 \u your username
5044 \v the version of bash (e.g., 2.00)
5045 \V the release of bash, version + patchlevel (e.g., 2.00.0)
5046 \w the current working directory
5047 \W the last element of $PWD
5048 \! the history number of this command
5049 \# the command number of this command
5050 \$ a $ or a # if you are root
5051 \nnn character code nnn in octal
5052 \\ a backslash
5053 \[ begin a sequence of non-printing chars
5054 \] end a sequence of non-printing chars
5055 */
5056 #define PROMPT_GROWTH 48
5057 char *
5058 decode_prompt_string (string)
5059 char *string;
5060 {
5061 WORD_LIST *list;
5062 char *result, *t;
5063 struct dstack save_dstack;
5064 int last_exit_value, last_comsub_pid;
5065 #if defined (PROMPT_STRING_DECODE)
5066 int result_size, result_index;
5067 int c, n, i;
5068 char *temp, octal_string[4];
5069 struct tm *tm;
5070 time_t the_time;
5071 char timebuf[128];
5072 char *timefmt;
5073
5074 result = (char *)xmalloc (result_size = PROMPT_GROWTH);
5075 result[result_index = 0] = 0;
5076 temp = (char *)NULL;
5077
5078 while (c = *string++)
5079 {
5080 if (posixly_correct && c == '!')
5081 {
5082 if (*string == '!')
5083 {
5084 temp = savestring ("!");
5085 goto add_string;
5086 }
5087 else
5088 {
5089 #if !defined (HISTORY)
5090 temp = savestring ("1");
5091 #else /* HISTORY */
5092 temp = itos (history_number ());
5093 #endif /* HISTORY */
5094 string--; /* add_string increments string again. */
5095 goto add_string;
5096 }
5097 }
5098 if (c == '\\')
5099 {
5100 c = *string;
5101
5102 switch (c)
5103 {
5104 case '0':
5105 case '1':
5106 case '2':
5107 case '3':
5108 case '4':
5109 case '5':
5110 case '6':
5111 case '7':
5112 strncpy (octal_string, string, 3);
5113 octal_string[3] = '\0';
5114
5115 n = read_octal (octal_string);
5116 temp = (char *)xmalloc (3);
5117
5118 if (n == CTLESC || n == CTLNUL)
5119 {
5120 temp[0] = CTLESC;
5121 temp[1] = n;
5122 temp[2] = '\0';
5123 }
5124 else if (n == -1)
5125 {
5126 temp[0] = '\\';
5127 temp[1] = '\0';
5128 }
5129 else
5130 {
5131 temp[0] = n;
5132 temp[1] = '\0';
5133 }
5134
5135 for (c = 0; n != -1 && c < 3 && ISOCTAL (*string); c++)
5136 string++;
5137
5138 c = 0; /* tested at add_string: */
5139 goto add_string;
5140
5141 case 'd':
5142 case 't':
5143 case 'T':
5144 case '@':
5145 case 'A':
5146 /* Make the current time/date into a string. */
5147 (void) time (&the_time);
5148 #if defined (HAVE_TZSET)
5149 sv_tz ("TZ"); /* XXX -- just make sure */
5150 #endif
5151 tm = localtime (&the_time);
5152
5153 if (c == 'd')
5154 n = strftime (timebuf, sizeof (timebuf), "%a %b %d", tm);
5155 else if (c == 't')
5156 n = strftime (timebuf, sizeof (timebuf), "%H:%M:%S", tm);
5157 else if (c == 'T')
5158 n = strftime (timebuf, sizeof (timebuf), "%I:%M:%S", tm);
5159 else if (c == '@')
5160 n = strftime (timebuf, sizeof (timebuf), "%I:%M %p", tm);
5161 else if (c == 'A')
5162 n = strftime (timebuf, sizeof (timebuf), "%H:%M", tm);
5163
5164 if (n == 0)
5165 timebuf[0] = '\0';
5166 else
5167 timebuf[sizeof(timebuf) - 1] = '\0';
5168
5169 temp = savestring (timebuf);
5170 goto add_string;
5171
5172 case 'D': /* strftime format */
5173 if (string[1] != '{') /* } */
5174 goto not_escape;
5175
5176 (void) time (&the_time);
5177 tm = localtime (&the_time);
5178 string += 2; /* skip { */
5179 timefmt = xmalloc (strlen (string) + 3);
5180 for (t = timefmt; *string && *string != '}'; )
5181 *t++ = *string++;
5182 *t = '\0';
5183 c = *string; /* tested at add_string */
5184 if (timefmt[0] == '\0')
5185 {
5186 timefmt[0] = '%';
5187 timefmt[1] = 'X'; /* locale-specific current time */
5188 timefmt[2] = '\0';
5189 }
5190 n = strftime (timebuf, sizeof (timebuf), timefmt, tm);
5191 free (timefmt);
5192
5193 if (n == 0)
5194 timebuf[0] = '\0';
5195 else
5196 timebuf[sizeof(timebuf) - 1] = '\0';
5197
5198 if (promptvars || posixly_correct)
5199 /* Make sure that expand_prompt_string is called with a
5200 second argument of Q_DOUBLE_QUOTES if we use this
5201 function here. */
5202 temp = sh_backslash_quote_for_double_quotes (timebuf);
5203 else
5204 temp = savestring (timebuf);
5205 goto add_string;
5206
5207 case 'n':
5208 temp = (char *)xmalloc (3);
5209 temp[0] = no_line_editing ? '\n' : '\r';
5210 temp[1] = no_line_editing ? '\0' : '\n';
5211 temp[2] = '\0';
5212 goto add_string;
5213
5214 case 's':
5215 temp = base_pathname (shell_name);
5216 temp = savestring (temp);
5217 goto add_string;
5218
5219 case 'v':
5220 case 'V':
5221 temp = (char *)xmalloc (16);
5222 if (c == 'v')
5223 strcpy (temp, dist_version);
5224 else
5225 sprintf (temp, "%s.%d", dist_version, patch_level);
5226 goto add_string;
5227
5228 case 'w':
5229 case 'W':
5230 {
5231 /* Use the value of PWD because it is much more efficient. */
5232 char t_string[PATH_MAX];
5233 int tlen;
5234
5235 temp = get_string_value ("PWD");
5236
5237 if (temp == 0)
5238 {
5239 if (getcwd (t_string, sizeof(t_string)) == 0)
5240 {
5241 t_string[0] = '.';
5242 tlen = 1;
5243 }
5244 else
5245 tlen = strlen (t_string);
5246 }
5247 else
5248 {
5249 tlen = sizeof (t_string) - 1;
5250 strncpy (t_string, temp, tlen);
5251 }
5252 t_string[tlen] = '\0';
5253
5254 #if defined (MACOSX)
5255 /* Convert from "fs" format to "input" format */
5256 temp = fnx_fromfs (t_string, strlen (t_string));
5257 if (temp != t_string)
5258 strcpy (t_string, temp);
5259 #endif
5260
5261 #define ROOT_PATH(x) ((x)[0] == '/' && (x)[1] == 0)
5262 #define DOUBLE_SLASH_ROOT(x) ((x)[0] == '/' && (x)[1] == '/' && (x)[2] == 0)
5263 /* Abbreviate \W as ~ if $PWD == $HOME */
5264 if (c == 'W' && (((t = get_string_value ("HOME")) == 0) || STREQ (t, t_string) == 0))
5265 {
5266 if (ROOT_PATH (t_string) == 0 && DOUBLE_SLASH_ROOT (t_string) == 0)
5267 {
5268 t = strrchr (t_string, '/');
5269 if (t)
5270 memmove (t_string, t + 1, strlen (t)); /* strlen(t) to copy NULL */
5271 }
5272 }
5273 #undef ROOT_PATH
5274 #undef DOUBLE_SLASH_ROOT
5275 else
5276 /* polite_directory_format is guaranteed to return a string
5277 no longer than PATH_MAX - 1 characters. */
5278 strcpy (t_string, polite_directory_format (t_string));
5279
5280 temp = trim_pathname (t_string, PATH_MAX - 1);
5281 /* If we're going to be expanding the prompt string later,
5282 quote the directory name. */
5283 if (promptvars || posixly_correct)
5284 /* Make sure that expand_prompt_string is called with a
5285 second argument of Q_DOUBLE_QUOTES if we use this
5286 function here. */
5287 temp = sh_backslash_quote_for_double_quotes (t_string);
5288 else
5289 temp = savestring (t_string);
5290
5291 goto add_string;
5292 }
5293
5294 case 'u':
5295 if (current_user.user_name == 0)
5296 get_current_user_info ();
5297 temp = savestring (current_user.user_name);
5298 goto add_string;
5299
5300 case 'h':
5301 case 'H':
5302 temp = savestring (current_host_name);
5303 if (c == 'h' && (t = (char *)strchr (temp, '.')))
5304 *t = '\0';
5305 goto add_string;
5306
5307 case '#':
5308 temp = itos (current_command_number);
5309 goto add_string;
5310
5311 case '!':
5312 #if !defined (HISTORY)
5313 temp = savestring ("1");
5314 #else /* HISTORY */
5315 temp = itos (history_number ());
5316 #endif /* HISTORY */
5317 goto add_string;
5318
5319 case '$':
5320 t = temp = (char *)xmalloc (3);
5321 if ((promptvars || posixly_correct) && (current_user.euid != 0))
5322 *t++ = '\\';
5323 *t++ = current_user.euid == 0 ? '#' : '$';
5324 *t = '\0';
5325 goto add_string;
5326
5327 case 'j':
5328 temp = itos (count_all_jobs ());
5329 goto add_string;
5330
5331 case 'l':
5332 #if defined (HAVE_TTYNAME)
5333 temp = (char *)ttyname (fileno (stdin));
5334 t = temp ? base_pathname (temp) : "tty";
5335 temp = savestring (t);
5336 #else
5337 temp = savestring ("tty");
5338 #endif /* !HAVE_TTYNAME */
5339 goto add_string;
5340
5341 #if defined (READLINE)
5342 case '[':
5343 case ']':
5344 if (no_line_editing)
5345 {
5346 string++;
5347 break;
5348 }
5349 temp = (char *)xmalloc (3);
5350 n = (c == '[') ? RL_PROMPT_START_IGNORE : RL_PROMPT_END_IGNORE;
5351 i = 0;
5352 if (n == CTLESC || n == CTLNUL)
5353 temp[i++] = CTLESC;
5354 temp[i++] = n;
5355 temp[i] = '\0';
5356 goto add_string;
5357 #endif /* READLINE */
5358
5359 case '\\':
5360 case 'a':
5361 case 'e':
5362 case 'r':
5363 temp = (char *)xmalloc (2);
5364 if (c == 'a')
5365 temp[0] = '\07';
5366 else if (c == 'e')
5367 temp[0] = '\033';
5368 else if (c == 'r')
5369 temp[0] = '\r';
5370 else /* (c == '\\') */
5371 temp[0] = c;
5372 temp[1] = '\0';
5373 goto add_string;
5374
5375 default:
5376 not_escape:
5377 temp = (char *)xmalloc (3);
5378 temp[0] = '\\';
5379 temp[1] = c;
5380 temp[2] = '\0';
5381
5382 add_string:
5383 if (c)
5384 string++;
5385 result =
5386 sub_append_string (temp, result, &result_index, &result_size);
5387 temp = (char *)NULL; /* Freed in sub_append_string (). */
5388 result[result_index] = '\0';
5389 break;
5390 }
5391 }
5392 else
5393 {
5394 RESIZE_MALLOCED_BUFFER (result, result_index, 3, result_size, PROMPT_GROWTH);
5395 result[result_index++] = c;
5396 result[result_index] = '\0';
5397 }
5398 }
5399 #else /* !PROMPT_STRING_DECODE */
5400 result = savestring (string);
5401 #endif /* !PROMPT_STRING_DECODE */
5402
5403 /* Save the delimiter stack and point `dstack' to temp space so any
5404 command substitutions in the prompt string won't result in screwing
5405 up the parser's quoting state. */
5406 save_dstack = dstack;
5407 dstack = temp_dstack;
5408 dstack.delimiter_depth = 0;
5409
5410 /* Perform variable and parameter expansion and command substitution on
5411 the prompt string. */
5412 if (promptvars || posixly_correct)
5413 {
5414 last_exit_value = last_command_exit_value;
5415 last_comsub_pid = last_command_subst_pid;
5416 list = expand_prompt_string (result, Q_DOUBLE_QUOTES, 0);
5417 free (result);
5418 result = string_list (list);
5419 dispose_words (list);
5420 last_command_exit_value = last_exit_value;
5421 last_command_subst_pid = last_comsub_pid;
5422 }
5423 else
5424 {
5425 t = dequote_string (result);
5426 free (result);
5427 result = t;
5428 }
5429
5430 dstack = save_dstack;
5431
5432 return (result);
5433 }
5434
5435 /************************************************
5436 * *
5437 * ERROR HANDLING *
5438 * *
5439 ************************************************/
5440
5441 /* Report a syntax error, and restart the parser. Call here for fatal
5442 errors. */
5443 int
5444 yyerror (msg)
5445 const char *msg;
5446 {
5447 report_syntax_error ((char *)NULL);
5448 reset_parser ();
5449 return (0);
5450 }
5451
5452 static char *
5453 error_token_from_token (tok)
5454 int tok;
5455 {
5456 char *t;
5457
5458 if (t = find_token_in_alist (tok, word_token_alist, 0))
5459 return t;
5460
5461 if (t = find_token_in_alist (tok, other_token_alist, 0))
5462 return t;
5463
5464 t = (char *)NULL;
5465 /* This stuff is dicy and needs closer inspection */
5466 switch (current_token)
5467 {
5468 case WORD:
5469 case ASSIGNMENT_WORD:
5470 if (yylval.word)
5471 t = savestring (yylval.word->word);
5472 break;
5473 case NUMBER:
5474 t = itos (yylval.number);
5475 break;
5476 case ARITH_CMD:
5477 if (yylval.word_list)
5478 t = string_list (yylval.word_list);
5479 break;
5480 case ARITH_FOR_EXPRS:
5481 if (yylval.word_list)
5482 t = string_list_internal (yylval.word_list, " ; ");
5483 break;
5484 case COND_CMD:
5485 t = (char *)NULL; /* punt */
5486 break;
5487 }
5488
5489 return t;
5490 }
5491
5492 static char *
5493 error_token_from_text ()
5494 {
5495 char *msg, *t;
5496 int token_end, i;
5497
5498 t = shell_input_line;
5499 i = shell_input_line_index;
5500 token_end = 0;
5501 msg = (char *)NULL;
5502
5503 if (i && t[i] == '\0')
5504 i--;
5505
5506 while (i && (whitespace (t[i]) || t[i] == '\n'))
5507 i--;
5508
5509 if (i)
5510 token_end = i + 1;
5511
5512 while (i && (member (t[i], " \n\t;|&") == 0))
5513 i--;
5514
5515 while (i != token_end && (whitespace (t[i]) || t[i] == '\n'))
5516 i++;
5517
5518 /* Return our idea of the offending token. */
5519 if (token_end || (i == 0 && token_end == 0))
5520 {
5521 if (token_end)
5522 msg = substring (t, i, token_end);
5523 else /* one-character token */
5524 {
5525 msg = (char *)xmalloc (2);
5526 msg[0] = t[i];
5527 msg[1] = '\0';
5528 }
5529 }
5530
5531 return (msg);
5532 }
5533
5534 static void
5535 print_offending_line ()
5536 {
5537 char *msg;
5538 int token_end;
5539
5540 msg = savestring (shell_input_line);
5541 token_end = strlen (msg);
5542 while (token_end && msg[token_end - 1] == '\n')
5543 msg[--token_end] = '\0';
5544
5545 parser_error (line_number, "`%s'", msg);
5546 free (msg);
5547 }
5548
5549 /* Report a syntax error with line numbers, etc.
5550 Call here for recoverable errors. If you have a message to print,
5551 then place it in MESSAGE, otherwise pass NULL and this will figure
5552 out an appropriate message for you. */
5553 static void
5554 report_syntax_error (message)
5555 char *message;
5556 {
5557 char *msg, *p;
5558
5559 if (message)
5560 {
5561 parser_error (line_number, "%s", message);
5562 if (interactive && EOF_Reached)
5563 EOF_Reached = 0;
5564 last_command_exit_value = parse_and_execute_level ? EX_BADSYNTAX : EX_BADUSAGE;
5565 return;
5566 }
5567
5568 /* If the line of input we're reading is not null, try to find the
5569 objectionable token. First, try to figure out what token the
5570 parser's complaining about by looking at current_token. */
5571 if (current_token != 0 && EOF_Reached == 0 && (msg = error_token_from_token (current_token)))
5572 {
5573 if (ansic_shouldquote (msg))
5574 {
5575 p = ansic_quote (msg, 0, NULL);
5576 free (msg);
5577 msg = p;
5578 }
5579 parser_error (line_number, _("syntax error near unexpected token `%s'"), msg);
5580 free (msg);
5581
5582 if (interactive == 0)
5583 print_offending_line ();
5584
5585 last_command_exit_value = parse_and_execute_level ? EX_BADSYNTAX : EX_BADUSAGE;
5586 return;
5587 }
5588
5589 /* If looking at the current token doesn't prove fruitful, try to find the
5590 offending token by analyzing the text of the input line near the current
5591 input line index and report what we find. */
5592 if (shell_input_line && *shell_input_line)
5593 {
5594 msg = error_token_from_text ();
5595 if (msg)
5596 {
5597 parser_error (line_number, _("syntax error near `%s'"), msg);
5598 free (msg);
5599 }
5600
5601 /* If not interactive, print the line containing the error. */
5602 if (interactive == 0)
5603 print_offending_line ();
5604 }
5605 else
5606 {
5607 msg = EOF_Reached ? _("syntax error: unexpected end of file") : _("syntax error");
5608 parser_error (line_number, "%s", msg);
5609 /* When the shell is interactive, this file uses EOF_Reached
5610 only for error reporting. Other mechanisms are used to
5611 decide whether or not to exit. */
5612 if (interactive && EOF_Reached)
5613 EOF_Reached = 0;
5614 }
5615
5616 last_command_exit_value = parse_and_execute_level ? EX_BADSYNTAX : EX_BADUSAGE;
5617 }
5618
5619 /* ??? Needed function. ??? We have to be able to discard the constructs
5620 created during parsing. In the case of error, we want to return
5621 allocated objects to the memory pool. In the case of no error, we want
5622 to throw away the information about where the allocated objects live.
5623 (dispose_command () will actually free the command.) */
5624 static void
5625 discard_parser_constructs (error_p)
5626 int error_p;
5627 {
5628 }
5629
5630 /************************************************
5631 * *
5632 * EOF HANDLING *
5633 * *
5634 ************************************************/
5635
5636 /* Do that silly `type "bye" to exit' stuff. You know, "ignoreeof". */
5637
5638 /* A flag denoting whether or not ignoreeof is set. */
5639 int ignoreeof = 0;
5640
5641 /* The number of times that we have encountered an EOF character without
5642 another character intervening. When this gets above the limit, the
5643 shell terminates. */
5644 int eof_encountered = 0;
5645
5646 /* The limit for eof_encountered. */
5647 int eof_encountered_limit = 10;
5648
5649 /* If we have EOF as the only input unit, this user wants to leave
5650 the shell. If the shell is not interactive, then just leave.
5651 Otherwise, if ignoreeof is set, and we haven't done this the
5652 required number of times in a row, print a message. */
5653 static void
5654 handle_eof_input_unit ()
5655 {
5656 if (interactive)
5657 {
5658 /* shell.c may use this to decide whether or not to write out the
5659 history, among other things. We use it only for error reporting
5660 in this file. */
5661 if (EOF_Reached)
5662 EOF_Reached = 0;
5663
5664 /* If the user wants to "ignore" eof, then let her do so, kind of. */
5665 if (ignoreeof)
5666 {
5667 if (eof_encountered < eof_encountered_limit)
5668 {
5669 fprintf (stderr, _("Use \"%s\" to leave the shell.\n"),
5670 login_shell ? "logout" : "exit");
5671 eof_encountered++;
5672 /* Reset the parsing state. */
5673 last_read_token = current_token = '\n';
5674 /* Reset the prompt string to be $PS1. */
5675 prompt_string_pointer = (char **)NULL;
5676 prompt_again ();
5677 return;
5678 }
5679 }
5680
5681 /* In this case EOF should exit the shell. Do it now. */
5682 reset_parser ();
5683 exit_builtin ((WORD_LIST *)NULL);
5684 }
5685 else
5686 {
5687 /* We don't write history files, etc., for non-interactive shells. */
5688 EOF_Reached = 1;
5689 }
5690 }
5691
5692 /************************************************
5693 * *
5694 * STRING PARSING FUNCTIONS *
5695 * *
5696 ************************************************/
5697
5698 /* It's very important that these two functions treat the characters
5699 between ( and ) identically. */
5700
5701 static WORD_LIST parse_string_error;
5702
5703 /* Take a string and run it through the shell parser, returning the
5704 resultant word list. Used by compound array assignment. */
5705 WORD_LIST *
5706 parse_string_to_word_list (s, flags, whom)
5707 char *s;
5708 int flags;
5709 const char *whom;
5710 {
5711 WORD_LIST *wl;
5712 int tok, orig_current_token, orig_line_number, orig_input_terminator;
5713 int orig_line_count;
5714 int old_echo_input, old_expand_aliases;
5715 #if defined (HISTORY)
5716 int old_remember_on_history, old_history_expansion_inhibited;
5717 #endif
5718
5719 #if defined (HISTORY)
5720 old_remember_on_history = remember_on_history;
5721 # if defined (BANG_HISTORY)
5722 old_history_expansion_inhibited = history_expansion_inhibited;
5723 # endif
5724 bash_history_disable ();
5725 #endif
5726
5727 orig_line_number = line_number;
5728 orig_line_count = current_command_line_count;
5729 orig_input_terminator = shell_input_line_terminator;
5730 old_echo_input = echo_input_at_read;
5731 old_expand_aliases = expand_aliases;
5732
5733 push_stream (1);
5734 last_read_token = WORD; /* WORD to allow reserved words here */
5735 current_command_line_count = 0;
5736 echo_input_at_read = expand_aliases = 0;
5737
5738 with_input_from_string (s, whom);
5739 wl = (WORD_LIST *)NULL;
5740
5741 if (flags & 1)
5742 parser_state |= PST_COMPASSIGN|PST_REPARSE;
5743
5744 while ((tok = read_token (READ)) != yacc_EOF)
5745 {
5746 if (tok == '\n' && *bash_input.location.string == '\0')
5747 break;
5748 if (tok == '\n') /* Allow newlines in compound assignments */
5749 continue;
5750 if (tok != WORD && tok != ASSIGNMENT_WORD)
5751 {
5752 line_number = orig_line_number + line_number - 1;
5753 orig_current_token = current_token;
5754 current_token = tok;
5755 yyerror (NULL); /* does the right thing */
5756 current_token = orig_current_token;
5757 if (wl)
5758 dispose_words (wl);
5759 wl = &parse_string_error;
5760 break;
5761 }
5762 wl = make_word_list (yylval.word, wl);
5763 }
5764
5765 last_read_token = '\n';
5766 pop_stream ();
5767
5768 #if defined (HISTORY)
5769 remember_on_history = old_remember_on_history;
5770 # if defined (BANG_HISTORY)
5771 history_expansion_inhibited = old_history_expansion_inhibited;
5772 # endif /* BANG_HISTORY */
5773 #endif /* HISTORY */
5774
5775 echo_input_at_read = old_echo_input;
5776 expand_aliases = old_expand_aliases;
5777
5778 current_command_line_count = orig_line_count;
5779 shell_input_line_terminator = orig_input_terminator;
5780
5781 if (flags & 1)
5782 parser_state &= ~(PST_COMPASSIGN|PST_REPARSE);
5783
5784 if (wl == &parse_string_error)
5785 {
5786 last_command_exit_value = EXECUTION_FAILURE;
5787 if (interactive_shell == 0 && posixly_correct)
5788 jump_to_top_level (FORCE_EOF);
5789 else
5790 jump_to_top_level (DISCARD);
5791 }
5792
5793 return (REVERSE_LIST (wl, WORD_LIST *));
5794 }
5795
5796 static char *
5797 parse_compound_assignment (retlenp)
5798 int *retlenp;
5799 {
5800 WORD_LIST *wl, *rl;
5801 int tok, orig_line_number, orig_token_size, orig_last_token, assignok;
5802 char *saved_token, *ret;
5803
5804 saved_token = token;
5805 orig_token_size = token_buffer_size;
5806 orig_line_number = line_number;
5807 orig_last_token = last_read_token;
5808
5809 last_read_token = WORD; /* WORD to allow reserved words here */
5810
5811 token = (char *)NULL;
5812 token_buffer_size = 0;
5813
5814 assignok = parser_state&PST_ASSIGNOK; /* XXX */
5815
5816 wl = (WORD_LIST *)NULL; /* ( */
5817 parser_state |= PST_COMPASSIGN;
5818
5819 while ((tok = read_token (READ)) != ')')
5820 {
5821 if (tok == '\n') /* Allow newlines in compound assignments */
5822 {
5823 if (SHOULD_PROMPT ())
5824 prompt_again ();
5825 continue;
5826 }
5827 if (tok != WORD && tok != ASSIGNMENT_WORD)
5828 {
5829 current_token = tok; /* for error reporting */
5830 if (tok == yacc_EOF) /* ( */
5831 parser_error (orig_line_number, _("unexpected EOF while looking for matching `)'"));
5832 else
5833 yyerror(NULL); /* does the right thing */
5834 if (wl)
5835 dispose_words (wl);
5836 wl = &parse_string_error;
5837 break;
5838 }
5839 wl = make_word_list (yylval.word, wl);
5840 }
5841
5842 FREE (token);
5843 token = saved_token;
5844 token_buffer_size = orig_token_size;
5845
5846 parser_state &= ~PST_COMPASSIGN;
5847
5848 if (wl == &parse_string_error)
5849 {
5850 last_command_exit_value = EXECUTION_FAILURE;
5851 last_read_token = '\n'; /* XXX */
5852 if (interactive_shell == 0 && posixly_correct)
5853 jump_to_top_level (FORCE_EOF);
5854 else
5855 jump_to_top_level (DISCARD);
5856 }
5857
5858 last_read_token = orig_last_token; /* XXX - was WORD? */
5859
5860 if (wl)
5861 {
5862 rl = REVERSE_LIST (wl, WORD_LIST *);
5863 ret = string_list (rl);
5864 dispose_words (rl);
5865 }
5866 else
5867 ret = (char *)NULL;
5868
5869 if (retlenp)
5870 *retlenp = (ret && *ret) ? strlen (ret) : 0;
5871
5872 if (assignok)
5873 parser_state |= PST_ASSIGNOK;
5874
5875 return ret;
5876 }
5877
5878 /************************************************
5879 * *
5880 * SAVING AND RESTORING PARTIAL PARSE STATE *
5881 * *
5882 ************************************************/
5883
5884 sh_parser_state_t *
5885 save_parser_state (ps)
5886 sh_parser_state_t *ps;
5887 {
5888 if (ps == 0)
5889 ps = (sh_parser_state_t *)xmalloc (sizeof (sh_parser_state_t));
5890 if (ps == 0)
5891 return ((sh_parser_state_t *)NULL);
5892
5893 ps->parser_state = parser_state;
5894 ps->token_state = save_token_state ();
5895
5896 ps->input_line_terminator = shell_input_line_terminator;
5897 ps->eof_encountered = eof_encountered;
5898
5899 ps->prompt_string_pointer = prompt_string_pointer;
5900
5901 ps->current_command_line_count = current_command_line_count;
5902
5903 #if defined (HISTORY)
5904 ps->remember_on_history = remember_on_history;
5905 # if defined (BANG_HISTORY)
5906 ps->history_expansion_inhibited = history_expansion_inhibited;
5907 # endif
5908 #endif
5909
5910 ps->last_command_exit_value = last_command_exit_value;
5911 #if defined (ARRAY_VARS)
5912 ps->pipestatus = save_pipestatus_array ();
5913 #endif
5914
5915 ps->last_shell_builtin = last_shell_builtin;
5916 ps->this_shell_builtin = this_shell_builtin;
5917
5918 ps->expand_aliases = expand_aliases;
5919 ps->echo_input_at_read = echo_input_at_read;
5920
5921 ps->token = token;
5922 ps->token_buffer_size = token_buffer_size;
5923 /* Force reallocation on next call to read_token_word */
5924 token = 0;
5925 token_buffer_size = 0;
5926
5927 return (ps);
5928 }
5929
5930 void
5931 restore_parser_state (ps)
5932 sh_parser_state_t *ps;
5933 {
5934 if (ps == 0)
5935 return;
5936
5937 parser_state = ps->parser_state;
5938 if (ps->token_state)
5939 {
5940 restore_token_state (ps->token_state);
5941 free (ps->token_state);
5942 }
5943
5944 shell_input_line_terminator = ps->input_line_terminator;
5945 eof_encountered = ps->eof_encountered;
5946
5947 prompt_string_pointer = ps->prompt_string_pointer;
5948
5949 current_command_line_count = ps->current_command_line_count;
5950
5951 #if defined (HISTORY)
5952 remember_on_history = ps->remember_on_history;
5953 # if defined (BANG_HISTORY)
5954 history_expansion_inhibited = ps->history_expansion_inhibited;
5955 # endif
5956 #endif
5957
5958 last_command_exit_value = ps->last_command_exit_value;
5959 #if defined (ARRAY_VARS)
5960 restore_pipestatus_array (ps->pipestatus);
5961 #endif
5962
5963 last_shell_builtin = ps->last_shell_builtin;
5964 this_shell_builtin = ps->this_shell_builtin;
5965
5966 expand_aliases = ps->expand_aliases;
5967 echo_input_at_read = ps->echo_input_at_read;
5968
5969 FREE (token);
5970 token = ps->token;
5971 token_buffer_size = ps->token_buffer_size;
5972 }
5973
5974 sh_input_line_state_t *
5975 save_input_line_state (ls)
5976 sh_input_line_state_t *ls;
5977 {
5978 if (ls == 0)
5979 ls = (sh_input_line_state_t *)xmalloc (sizeof (sh_input_line_state_t));
5980 if (ls == 0)
5981 return ((sh_input_line_state_t *)NULL);
5982
5983 ls->input_line = shell_input_line;
5984 ls->input_line_size = shell_input_line_size;
5985 ls->input_line_len = shell_input_line_len;
5986 ls->input_line_index = shell_input_line_index;
5987
5988 /* force reallocation */
5989 shell_input_line = 0;
5990 shell_input_line_size = shell_input_line_len = shell_input_line_index = 0;
5991 }
5992
5993 void
5994 restore_input_line_state (ls)
5995 sh_input_line_state_t *ls;
5996 {
5997 FREE (shell_input_line);
5998 shell_input_line = ls->input_line;
5999 shell_input_line_size = ls->input_line_size;
6000 shell_input_line_len = ls->input_line_len;
6001 shell_input_line_index = ls->input_line_index;
6002
6003 set_line_mbstate ();
6004 }
6005
6006 /************************************************
6007 * *
6008 * MULTIBYTE CHARACTER HANDLING *
6009 * *
6010 ************************************************/
6011
6012 #if defined (HANDLE_MULTIBYTE)
6013 static void
6014 set_line_mbstate ()
6015 {
6016 int i, previ, len, c;
6017 mbstate_t mbs, prevs;
6018 size_t mbclen;
6019
6020 if (shell_input_line == NULL)
6021 return;
6022 len = strlen (shell_input_line); /* XXX - shell_input_line_len ? */
6023 FREE (shell_input_line_property);
6024 shell_input_line_property = (char *)xmalloc (len + 1);
6025
6026 memset (&prevs, '\0', sizeof (mbstate_t));
6027 for (i = previ = 0; i < len; i++)
6028 {
6029 mbs = prevs;
6030
6031 c = shell_input_line[i];
6032 if (c == EOF)
6033 {
6034 int j;
6035 for (j = i; j < len; j++)
6036 shell_input_line_property[j] = 1;
6037 break;
6038 }
6039
6040 mbclen = mbrlen (shell_input_line + previ, i - previ + 1, &mbs);
6041 if (mbclen == 1 || mbclen == (size_t)-1)
6042 {
6043 mbclen = 1;
6044 previ = i + 1;
6045 }
6046 else if (mbclen == (size_t)-2)
6047 mbclen = 0;
6048 else if (mbclen > 1)
6049 {
6050 mbclen = 0;
6051 previ = i + 1;
6052 prevs = mbs;
6053 }
6054 else
6055 {
6056 /* XXX - what to do if mbrlen returns 0? (null wide character) */
6057 int j;
6058 for (j = i; j < len; j++)
6059 shell_input_line_property[j] = 1;
6060 break;
6061 }
6062
6063 shell_input_line_property[i] = mbclen;
6064 }
6065 }
6066 #endif /* HANDLE_MULTIBYTE */