]> git.ipfire.org Git - thirdparty/bash.git/blob - parse.y
Bash-4.2 patch 34
[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 (expanding_alias () && shell_input_line[shell_input_line_index+1] == '\0')
2397 {
2398 uc = 0;
2399 goto pop_alias;
2400 }
2401 else if (expanding_alias () && shell_input_line[shell_input_line_index+1] != '\0')
2402 {
2403 shell_input_line_index++; /* skip newline */
2404 goto next_alias_char; /* and get next character */
2405 }
2406 else
2407 goto restart_read;
2408 }
2409
2410 if (uc == 0 && shell_input_line_terminator == EOF)
2411 return ((shell_input_line_index != 0) ? '\n' : EOF);
2412
2413 return (uc);
2414 }
2415
2416 /* Put C back into the input for the shell. This might need changes for
2417 HANDLE_MULTIBYTE around EOLs. Since we (currently) never push back a
2418 character different than we read, shell_input_line_property doesn't need
2419 to change when manipulating shell_input_line. The define for
2420 last_shell_getc_is_singlebyte should take care of it, though. */
2421 static void
2422 shell_ungetc (c)
2423 int c;
2424 {
2425 if (shell_input_line && shell_input_line_index)
2426 shell_input_line[--shell_input_line_index] = c;
2427 else
2428 eol_ungetc_lookahead = c;
2429 }
2430
2431 #ifdef INCLUDE_UNUSED
2432 /* Back the input pointer up by one, effectively `ungetting' a character. */
2433 static void
2434 shell_ungetchar ()
2435 {
2436 if (shell_input_line && shell_input_line_index)
2437 shell_input_line_index--;
2438 }
2439 #endif
2440
2441 /* Discard input until CHARACTER is seen, then push that character back
2442 onto the input stream. */
2443 static void
2444 discard_until (character)
2445 int character;
2446 {
2447 int c;
2448
2449 while ((c = shell_getc (0)) != EOF && c != character)
2450 ;
2451
2452 if (c != EOF)
2453 shell_ungetc (c);
2454 }
2455
2456 void
2457 execute_variable_command (command, vname)
2458 char *command, *vname;
2459 {
2460 char *last_lastarg;
2461 sh_parser_state_t ps;
2462
2463 save_parser_state (&ps);
2464 last_lastarg = get_string_value ("_");
2465 if (last_lastarg)
2466 last_lastarg = savestring (last_lastarg);
2467
2468 parse_and_execute (savestring (command), vname, SEVAL_NONINT|SEVAL_NOHIST);
2469
2470 restore_parser_state (&ps);
2471 bind_variable ("_", last_lastarg, 0);
2472 FREE (last_lastarg);
2473
2474 if (token_to_read == '\n') /* reset_parser was called */
2475 token_to_read = 0;
2476 }
2477
2478 /* Place to remember the token. We try to keep the buffer
2479 at a reasonable size, but it can grow. */
2480 static char *token = (char *)NULL;
2481
2482 /* Current size of the token buffer. */
2483 static int token_buffer_size;
2484
2485 /* Command to read_token () explaining what we want it to do. */
2486 #define READ 0
2487 #define RESET 1
2488 #define prompt_is_ps1 \
2489 (!prompt_string_pointer || prompt_string_pointer == &ps1_prompt)
2490
2491 /* Function for yyparse to call. yylex keeps track of
2492 the last two tokens read, and calls read_token. */
2493 static int
2494 yylex ()
2495 {
2496 if (interactive && (current_token == 0 || current_token == '\n'))
2497 {
2498 /* Before we print a prompt, we might have to check mailboxes.
2499 We do this only if it is time to do so. Notice that only here
2500 is the mail alarm reset; nothing takes place in check_mail ()
2501 except the checking of mail. Please don't change this. */
2502 if (prompt_is_ps1 && parse_and_execute_level == 0 && time_to_check_mail ())
2503 {
2504 check_mail ();
2505 reset_mail_timer ();
2506 }
2507
2508 /* Avoid printing a prompt if we're not going to read anything, e.g.
2509 after resetting the parser with read_token (RESET). */
2510 if (token_to_read == 0 && SHOULD_PROMPT ())
2511 prompt_again ();
2512 }
2513
2514 two_tokens_ago = token_before_that;
2515 token_before_that = last_read_token;
2516 last_read_token = current_token;
2517 current_token = read_token (READ);
2518
2519 if ((parser_state & PST_EOFTOKEN) && current_token == shell_eof_token)
2520 {
2521 current_token = yacc_EOF;
2522 if (bash_input.type == st_string)
2523 rewind_input_string ();
2524 }
2525 parser_state &= ~PST_EOFTOKEN;
2526
2527 return (current_token);
2528 }
2529
2530 /* When non-zero, we have read the required tokens
2531 which allow ESAC to be the next one read. */
2532 static int esacs_needed_count;
2533
2534 void
2535 gather_here_documents ()
2536 {
2537 int r;
2538
2539 r = 0;
2540 while (need_here_doc)
2541 {
2542 parser_state |= PST_HEREDOC;
2543 make_here_document (redir_stack[r++], line_number);
2544 parser_state &= ~PST_HEREDOC;
2545 need_here_doc--;
2546 }
2547 }
2548
2549 /* When non-zero, an open-brace used to create a group is awaiting a close
2550 brace partner. */
2551 static int open_brace_count;
2552
2553 #define command_token_position(token) \
2554 (((token) == ASSIGNMENT_WORD) || (parser_state&PST_REDIRLIST) || \
2555 ((token) != SEMI_SEMI && (token) != SEMI_AND && (token) != SEMI_SEMI_AND && reserved_word_acceptable(token)))
2556
2557 #define assignment_acceptable(token) \
2558 (command_token_position(token) && ((parser_state & PST_CASEPAT) == 0))
2559
2560 /* Check to see if TOKEN is a reserved word and return the token
2561 value if it is. */
2562 #define CHECK_FOR_RESERVED_WORD(tok) \
2563 do { \
2564 if (!dollar_present && !quoted && \
2565 reserved_word_acceptable (last_read_token)) \
2566 { \
2567 int i; \
2568 for (i = 0; word_token_alist[i].word != (char *)NULL; i++) \
2569 if (STREQ (tok, word_token_alist[i].word)) \
2570 { \
2571 if ((parser_state & PST_CASEPAT) && (word_token_alist[i].token != ESAC)) \
2572 break; \
2573 if (word_token_alist[i].token == TIME && time_command_acceptable () == 0) \
2574 break; \
2575 if (word_token_alist[i].token == ESAC) \
2576 parser_state &= ~(PST_CASEPAT|PST_CASESTMT); \
2577 else if (word_token_alist[i].token == CASE) \
2578 parser_state |= PST_CASESTMT; \
2579 else if (word_token_alist[i].token == COND_END) \
2580 parser_state &= ~(PST_CONDCMD|PST_CONDEXPR); \
2581 else if (word_token_alist[i].token == COND_START) \
2582 parser_state |= PST_CONDCMD; \
2583 else if (word_token_alist[i].token == '{') \
2584 open_brace_count++; \
2585 else if (word_token_alist[i].token == '}' && open_brace_count) \
2586 open_brace_count--; \
2587 return (word_token_alist[i].token); \
2588 } \
2589 } \
2590 } while (0)
2591
2592 #if defined (ALIAS)
2593
2594 /* OK, we have a token. Let's try to alias expand it, if (and only if)
2595 it's eligible.
2596
2597 It is eligible for expansion if EXPAND_ALIASES is set, and
2598 the token is unquoted and the last token read was a command
2599 separator (or expand_next_token is set), and we are currently
2600 processing an alias (pushed_string_list is non-empty) and this
2601 token is not the same as the current or any previously
2602 processed alias.
2603
2604 Special cases that disqualify:
2605 In a pattern list in a case statement (parser_state & PST_CASEPAT). */
2606
2607 static char *
2608 mk_alexpansion (s)
2609 char *s;
2610 {
2611 int l;
2612 char *r;
2613
2614 l = strlen (s);
2615 r = xmalloc (l + 2);
2616 strcpy (r, s);
2617 /* If the last character in the alias is a newline, don't add a trailing
2618 space to the expansion. Works with shell_getc above. */
2619 if (r[l - 1] != ' ' && r[l - 1] != '\n')
2620 r[l++] = ' ';
2621 r[l] = '\0';
2622 return r;
2623 }
2624
2625 static int
2626 alias_expand_token (tokstr)
2627 char *tokstr;
2628 {
2629 char *expanded;
2630 alias_t *ap;
2631
2632 if (((parser_state & PST_ALEXPNEXT) || command_token_position (last_read_token)) &&
2633 (parser_state & PST_CASEPAT) == 0)
2634 {
2635 ap = find_alias (tokstr);
2636
2637 /* Currently expanding this token. */
2638 if (ap && (ap->flags & AL_BEINGEXPANDED))
2639 return (NO_EXPANSION);
2640
2641 /* mk_alexpansion puts an extra space on the end of the alias expansion,
2642 so the lookahead by the parser works right. If this gets changed,
2643 make sure the code in shell_getc that deals with reaching the end of
2644 an expanded alias is changed with it. */
2645 expanded = ap ? mk_alexpansion (ap->value) : (char *)NULL;
2646
2647 if (expanded)
2648 {
2649 push_string (expanded, ap->flags & AL_EXPANDNEXT, ap);
2650 return (RE_READ_TOKEN);
2651 }
2652 else
2653 /* This is an eligible token that does not have an expansion. */
2654 return (NO_EXPANSION);
2655 }
2656 return (NO_EXPANSION);
2657 }
2658 #endif /* ALIAS */
2659
2660 static int
2661 time_command_acceptable ()
2662 {
2663 #if defined (COMMAND_TIMING)
2664 int i;
2665
2666 if (posixly_correct && shell_compatibility_level > 41)
2667 {
2668 /* Quick check of the rest of the line to find the next token. If it
2669 begins with a `-', Posix says to not return `time' as the token.
2670 This was interp 267. */
2671 i = shell_input_line_index;
2672 while (i < shell_input_line_len && (shell_input_line[i] == ' ' || shell_input_line[i] == '\t'))
2673 i++;
2674 if (shell_input_line[i] == '-')
2675 return 0;
2676 }
2677
2678 switch (last_read_token)
2679 {
2680 case 0:
2681 case ';':
2682 case '\n':
2683 case AND_AND:
2684 case OR_OR:
2685 case '&':
2686 case DO:
2687 case THEN:
2688 case ELSE:
2689 case '{': /* } */
2690 case '(': /* ) */
2691 case BANG: /* ! time pipeline */
2692 case TIME: /* time time pipeline */
2693 case TIMEOPT: /* time -p time pipeline */
2694 case TIMEIGN: /* time -p -- ... */
2695 return 1;
2696 default:
2697 return 0;
2698 }
2699 #else
2700 return 0;
2701 #endif /* COMMAND_TIMING */
2702 }
2703
2704 /* Handle special cases of token recognition:
2705 IN is recognized if the last token was WORD and the token
2706 before that was FOR or CASE or SELECT.
2707
2708 DO is recognized if the last token was WORD and the token
2709 before that was FOR or SELECT.
2710
2711 ESAC is recognized if the last token caused `esacs_needed_count'
2712 to be set
2713
2714 `{' is recognized if the last token as WORD and the token
2715 before that was FUNCTION, or if we just parsed an arithmetic
2716 `for' command.
2717
2718 `}' is recognized if there is an unclosed `{' present.
2719
2720 `-p' is returned as TIMEOPT if the last read token was TIME.
2721 `--' is returned as TIMEIGN if the last read token was TIMEOPT.
2722
2723 ']]' is returned as COND_END if the parser is currently parsing
2724 a conditional expression ((parser_state & PST_CONDEXPR) != 0)
2725
2726 `time' is returned as TIME if and only if it is immediately
2727 preceded by one of `;', `\n', `||', `&&', or `&'.
2728 */
2729
2730 static int
2731 special_case_tokens (tokstr)
2732 char *tokstr;
2733 {
2734 if ((last_read_token == WORD) &&
2735 #if defined (SELECT_COMMAND)
2736 ((token_before_that == FOR) || (token_before_that == CASE) || (token_before_that == SELECT)) &&
2737 #else
2738 ((token_before_that == FOR) || (token_before_that == CASE)) &&
2739 #endif
2740 (tokstr[0] == 'i' && tokstr[1] == 'n' && tokstr[2] == 0))
2741 {
2742 if (token_before_that == CASE)
2743 {
2744 parser_state |= PST_CASEPAT;
2745 esacs_needed_count++;
2746 }
2747 return (IN);
2748 }
2749
2750 if (last_read_token == WORD &&
2751 #if defined (SELECT_COMMAND)
2752 (token_before_that == FOR || token_before_that == SELECT) &&
2753 #else
2754 (token_before_that == FOR) &&
2755 #endif
2756 (tokstr[0] == 'd' && tokstr[1] == 'o' && tokstr[2] == '\0'))
2757 return (DO);
2758
2759 /* Ditto for ESAC in the CASE case.
2760 Specifically, this handles "case word in esac", which is a legal
2761 construct, certainly because someone will pass an empty arg to the
2762 case construct, and we don't want it to barf. Of course, we should
2763 insist that the case construct has at least one pattern in it, but
2764 the designers disagree. */
2765 if (esacs_needed_count)
2766 {
2767 esacs_needed_count--;
2768 if (STREQ (tokstr, "esac"))
2769 {
2770 parser_state &= ~PST_CASEPAT;
2771 return (ESAC);
2772 }
2773 }
2774
2775 /* The start of a shell function definition. */
2776 if (parser_state & PST_ALLOWOPNBRC)
2777 {
2778 parser_state &= ~PST_ALLOWOPNBRC;
2779 if (tokstr[0] == '{' && tokstr[1] == '\0') /* } */
2780 {
2781 open_brace_count++;
2782 function_bstart = line_number;
2783 return ('{'); /* } */
2784 }
2785 }
2786
2787 /* We allow a `do' after a for ((...)) without an intervening
2788 list_terminator */
2789 if (last_read_token == ARITH_FOR_EXPRS && tokstr[0] == 'd' && tokstr[1] == 'o' && !tokstr[2])
2790 return (DO);
2791 if (last_read_token == ARITH_FOR_EXPRS && tokstr[0] == '{' && tokstr[1] == '\0') /* } */
2792 {
2793 open_brace_count++;
2794 return ('{'); /* } */
2795 }
2796
2797 if (open_brace_count && reserved_word_acceptable (last_read_token) && tokstr[0] == '}' && !tokstr[1])
2798 {
2799 open_brace_count--; /* { */
2800 return ('}');
2801 }
2802
2803 #if defined (COMMAND_TIMING)
2804 /* Handle -p after `time'. */
2805 if (last_read_token == TIME && tokstr[0] == '-' && tokstr[1] == 'p' && !tokstr[2])
2806 return (TIMEOPT);
2807 /* Handle -- after `time -p'. */
2808 if (last_read_token == TIMEOPT && tokstr[0] == '-' && tokstr[1] == '-' && !tokstr[2])
2809 return (TIMEIGN);
2810 #endif
2811
2812 #if defined (COND_COMMAND) /* [[ */
2813 if ((parser_state & PST_CONDEXPR) && tokstr[0] == ']' && tokstr[1] == ']' && tokstr[2] == '\0')
2814 return (COND_END);
2815 #endif
2816
2817 return (-1);
2818 }
2819
2820 /* Called from shell.c when Control-C is typed at top level. Or
2821 by the error rule at top level. */
2822 void
2823 reset_parser ()
2824 {
2825 dstack.delimiter_depth = 0; /* No delimiters found so far. */
2826 open_brace_count = 0;
2827
2828 #if defined (EXTENDED_GLOB)
2829 /* Reset to global value of extended glob */
2830 if (parser_state & PST_EXTPAT)
2831 extended_glob = global_extglob;
2832 #endif
2833
2834 parser_state = 0;
2835
2836 #if defined (ALIAS) || defined (DPAREN_ARITHMETIC)
2837 if (pushed_string_list)
2838 free_string_list ();
2839 #endif /* ALIAS || DPAREN_ARITHMETIC */
2840
2841 if (shell_input_line)
2842 {
2843 free (shell_input_line);
2844 shell_input_line = (char *)NULL;
2845 shell_input_line_size = shell_input_line_index = 0;
2846 }
2847
2848 FREE (word_desc_to_read);
2849 word_desc_to_read = (WORD_DESC *)NULL;
2850
2851 current_token = '\n'; /* XXX */
2852 last_read_token = '\n';
2853 token_to_read = '\n';
2854 }
2855
2856 /* Read the next token. Command can be READ (normal operation) or
2857 RESET (to normalize state). */
2858 static int
2859 read_token (command)
2860 int command;
2861 {
2862 int character; /* Current character. */
2863 int peek_char; /* Temporary look-ahead character. */
2864 int result; /* The thing to return. */
2865
2866 if (command == RESET)
2867 {
2868 reset_parser ();
2869 return ('\n');
2870 }
2871
2872 if (token_to_read)
2873 {
2874 result = token_to_read;
2875 if (token_to_read == WORD || token_to_read == ASSIGNMENT_WORD)
2876 {
2877 yylval.word = word_desc_to_read;
2878 word_desc_to_read = (WORD_DESC *)NULL;
2879 }
2880 token_to_read = 0;
2881 return (result);
2882 }
2883
2884 #if defined (COND_COMMAND)
2885 if ((parser_state & (PST_CONDCMD|PST_CONDEXPR)) == PST_CONDCMD)
2886 {
2887 cond_lineno = line_number;
2888 parser_state |= PST_CONDEXPR;
2889 yylval.command = parse_cond_command ();
2890 if (cond_token != COND_END)
2891 {
2892 cond_error ();
2893 return (-1);
2894 }
2895 token_to_read = COND_END;
2896 parser_state &= ~(PST_CONDEXPR|PST_CONDCMD);
2897 return (COND_CMD);
2898 }
2899 #endif
2900
2901 #if defined (ALIAS)
2902 /* This is a place to jump back to once we have successfully expanded a
2903 token with an alias and pushed the string with push_string () */
2904 re_read_token:
2905 #endif /* ALIAS */
2906
2907 /* Read a single word from input. Start by skipping blanks. */
2908 while ((character = shell_getc (1)) != EOF && shellblank (character))
2909 ;
2910
2911 if (character == EOF)
2912 {
2913 EOF_Reached = 1;
2914 return (yacc_EOF);
2915 }
2916
2917 if MBTEST(character == '#' && (!interactive || interactive_comments))
2918 {
2919 /* A comment. Discard until EOL or EOF, and then return a newline. */
2920 discard_until ('\n');
2921 shell_getc (0);
2922 character = '\n'; /* this will take the next if statement and return. */
2923 }
2924
2925 if (character == '\n')
2926 {
2927 /* If we're about to return an unquoted newline, we can go and collect
2928 the text of any pending here document. */
2929 if (need_here_doc)
2930 gather_here_documents ();
2931
2932 #if defined (ALIAS)
2933 parser_state &= ~PST_ALEXPNEXT;
2934 #endif /* ALIAS */
2935
2936 parser_state &= ~PST_ASSIGNOK;
2937
2938 return (character);
2939 }
2940
2941 if (parser_state & PST_REGEXP)
2942 goto tokword;
2943
2944 /* Shell meta-characters. */
2945 if MBTEST(shellmeta (character) && ((parser_state & PST_DBLPAREN) == 0))
2946 {
2947 #if defined (ALIAS)
2948 /* Turn off alias tokenization iff this character sequence would
2949 not leave us ready to read a command. */
2950 if (character == '<' || character == '>')
2951 parser_state &= ~PST_ALEXPNEXT;
2952 #endif /* ALIAS */
2953
2954 parser_state &= ~PST_ASSIGNOK;
2955
2956 peek_char = shell_getc (1);
2957 if (character == peek_char)
2958 {
2959 switch (character)
2960 {
2961 case '<':
2962 /* If '<' then we could be at "<<" or at "<<-". We have to
2963 look ahead one more character. */
2964 peek_char = shell_getc (1);
2965 if MBTEST(peek_char == '-')
2966 return (LESS_LESS_MINUS);
2967 else if MBTEST(peek_char == '<')
2968 return (LESS_LESS_LESS);
2969 else
2970 {
2971 shell_ungetc (peek_char);
2972 return (LESS_LESS);
2973 }
2974
2975 case '>':
2976 return (GREATER_GREATER);
2977
2978 case ';':
2979 parser_state |= PST_CASEPAT;
2980 #if defined (ALIAS)
2981 parser_state &= ~PST_ALEXPNEXT;
2982 #endif /* ALIAS */
2983
2984 peek_char = shell_getc (1);
2985 if MBTEST(peek_char == '&')
2986 return (SEMI_SEMI_AND);
2987 else
2988 {
2989 shell_ungetc (peek_char);
2990 return (SEMI_SEMI);
2991 }
2992
2993 case '&':
2994 return (AND_AND);
2995
2996 case '|':
2997 return (OR_OR);
2998
2999 #if defined (DPAREN_ARITHMETIC) || defined (ARITH_FOR_COMMAND)
3000 case '(': /* ) */
3001 result = parse_dparen (character);
3002 if (result == -2)
3003 break;
3004 else
3005 return result;
3006 #endif
3007 }
3008 }
3009 else if MBTEST(character == '<' && peek_char == '&')
3010 return (LESS_AND);
3011 else if MBTEST(character == '>' && peek_char == '&')
3012 return (GREATER_AND);
3013 else if MBTEST(character == '<' && peek_char == '>')
3014 return (LESS_GREATER);
3015 else if MBTEST(character == '>' && peek_char == '|')
3016 return (GREATER_BAR);
3017 else if MBTEST(character == '&' && peek_char == '>')
3018 {
3019 peek_char = shell_getc (1);
3020 if MBTEST(peek_char == '>')
3021 return (AND_GREATER_GREATER);
3022 else
3023 {
3024 shell_ungetc (peek_char);
3025 return (AND_GREATER);
3026 }
3027 }
3028 else if MBTEST(character == '|' && peek_char == '&')
3029 return (BAR_AND);
3030 else if MBTEST(character == ';' && peek_char == '&')
3031 {
3032 parser_state |= PST_CASEPAT;
3033 #if defined (ALIAS)
3034 parser_state &= ~PST_ALEXPNEXT;
3035 #endif /* ALIAS */
3036 return (SEMI_AND);
3037 }
3038
3039 shell_ungetc (peek_char);
3040
3041 /* If we look like we are reading the start of a function
3042 definition, then let the reader know about it so that
3043 we will do the right thing with `{'. */
3044 if MBTEST(character == ')' && last_read_token == '(' && token_before_that == WORD)
3045 {
3046 parser_state |= PST_ALLOWOPNBRC;
3047 #if defined (ALIAS)
3048 parser_state &= ~PST_ALEXPNEXT;
3049 #endif /* ALIAS */
3050 function_dstart = line_number;
3051 }
3052
3053 /* case pattern lists may be preceded by an optional left paren. If
3054 we're not trying to parse a case pattern list, the left paren
3055 indicates a subshell. */
3056 if MBTEST(character == '(' && (parser_state & PST_CASEPAT) == 0) /* ) */
3057 parser_state |= PST_SUBSHELL;
3058 /*(*/
3059 else if MBTEST((parser_state & PST_CASEPAT) && character == ')')
3060 parser_state &= ~PST_CASEPAT;
3061 /*(*/
3062 else if MBTEST((parser_state & PST_SUBSHELL) && character == ')')
3063 parser_state &= ~PST_SUBSHELL;
3064
3065 #if defined (PROCESS_SUBSTITUTION)
3066 /* Check for the constructs which introduce process substitution.
3067 Shells running in `posix mode' don't do process substitution. */
3068 if MBTEST(posixly_correct || ((character != '>' && character != '<') || peek_char != '(')) /*)*/
3069 #endif /* PROCESS_SUBSTITUTION */
3070 return (character);
3071 }
3072
3073 /* Hack <&- (close stdin) case. Also <&N- (dup and close). */
3074 if MBTEST(character == '-' && (last_read_token == LESS_AND || last_read_token == GREATER_AND))
3075 return (character);
3076
3077 tokword:
3078 /* Okay, if we got this far, we have to read a word. Read one,
3079 and then check it against the known ones. */
3080 result = read_token_word (character);
3081 #if defined (ALIAS)
3082 if (result == RE_READ_TOKEN)
3083 goto re_read_token;
3084 #endif
3085 return result;
3086 }
3087
3088 /*
3089 * Match a $(...) or other grouping construct. This has to handle embedded
3090 * quoted strings ('', ``, "") and nested constructs. It also must handle
3091 * reprompting the user, if necessary, after reading a newline, and returning
3092 * correct error values if it reads EOF.
3093 */
3094 #define P_FIRSTCLOSE 0x0001
3095 #define P_ALLOWESC 0x0002
3096 #define P_DQUOTE 0x0004
3097 #define P_COMMAND 0x0008 /* parsing a command, so look for comments */
3098 #define P_BACKQUOTE 0x0010 /* parsing a backquoted command substitution */
3099 #define P_ARRAYSUB 0x0020 /* parsing a [...] array subscript for assignment */
3100 #define P_DOLBRACE 0x0040 /* parsing a ${...} construct */
3101
3102 /* Lexical state while parsing a grouping construct or $(...). */
3103 #define LEX_WASDOL 0x001
3104 #define LEX_CKCOMMENT 0x002
3105 #define LEX_INCOMMENT 0x004
3106 #define LEX_PASSNEXT 0x008
3107 #define LEX_RESWDOK 0x010
3108 #define LEX_CKCASE 0x020
3109 #define LEX_INCASE 0x040
3110 #define LEX_INHEREDOC 0x080
3111 #define LEX_HEREDELIM 0x100 /* reading here-doc delimiter */
3112 #define LEX_STRIPDOC 0x200 /* <<- strip tabs from here doc delim */
3113 #define LEX_INWORD 0x400
3114
3115 #define COMSUB_META(ch) ((ch) == ';' || (ch) == '&' || (ch) == '|')
3116
3117 #define CHECK_NESTRET_ERROR() \
3118 do { \
3119 if (nestret == &matched_pair_error) \
3120 { \
3121 free (ret); \
3122 return &matched_pair_error; \
3123 } \
3124 } while (0)
3125
3126 #define APPEND_NESTRET() \
3127 do { \
3128 if (nestlen) \
3129 { \
3130 RESIZE_MALLOCED_BUFFER (ret, retind, nestlen, retsize, 64); \
3131 strcpy (ret + retind, nestret); \
3132 retind += nestlen; \
3133 } \
3134 } while (0)
3135
3136 static char matched_pair_error;
3137
3138 static char *
3139 parse_matched_pair (qc, open, close, lenp, flags)
3140 int qc; /* `"' if this construct is within double quotes */
3141 int open, close;
3142 int *lenp, flags;
3143 {
3144 int count, ch, tflags;
3145 int nestlen, ttranslen, start_lineno;
3146 char *ret, *nestret, *ttrans;
3147 int retind, retsize, rflags;
3148 int dolbrace_state;
3149
3150 dolbrace_state = (flags & P_DOLBRACE) ? DOLBRACE_PARAM : 0;
3151
3152 /*itrace("parse_matched_pair[%d]: open = %c close = %c flags = %d", line_number, open, close, flags);*/
3153 count = 1;
3154 tflags = 0;
3155
3156 if ((flags & P_COMMAND) && qc != '`' && qc != '\'' && qc != '"' && (flags & P_DQUOTE) == 0)
3157 tflags |= LEX_CKCOMMENT;
3158
3159 /* RFLAGS is the set of flags we want to pass to recursive calls. */
3160 rflags = (qc == '"') ? P_DQUOTE : (flags & P_DQUOTE);
3161
3162 ret = (char *)xmalloc (retsize = 64);
3163 retind = 0;
3164
3165 start_lineno = line_number;
3166 while (count)
3167 {
3168 ch = shell_getc (qc != '\'' && (tflags & (LEX_PASSNEXT)) == 0);
3169
3170 if (ch == EOF)
3171 {
3172 free (ret);
3173 parser_error (start_lineno, _("unexpected EOF while looking for matching `%c'"), close);
3174 EOF_Reached = 1; /* XXX */
3175 return (&matched_pair_error);
3176 }
3177
3178 /* Possible reprompting. */
3179 if (ch == '\n' && SHOULD_PROMPT ())
3180 prompt_again ();
3181
3182 /* Don't bother counting parens or doing anything else if in a comment
3183 or part of a case statement */
3184 if (tflags & LEX_INCOMMENT)
3185 {
3186 /* Add this character. */
3187 RESIZE_MALLOCED_BUFFER (ret, retind, 1, retsize, 64);
3188 ret[retind++] = ch;
3189
3190 if (ch == '\n')
3191 tflags &= ~LEX_INCOMMENT;
3192
3193 continue;
3194 }
3195
3196 /* Not exactly right yet, should handle shell metacharacters, too. If
3197 any changes are made to this test, make analogous changes to subst.c:
3198 extract_delimited_string(). */
3199 else if MBTEST((tflags & LEX_CKCOMMENT) && (tflags & LEX_INCOMMENT) == 0 && ch == '#' && (retind == 0 || ret[retind-1] == '\n' || shellblank (ret[retind - 1])))
3200 tflags |= LEX_INCOMMENT;
3201
3202 if (tflags & LEX_PASSNEXT) /* last char was backslash */
3203 {
3204 tflags &= ~LEX_PASSNEXT;
3205 if (qc != '\'' && ch == '\n') /* double-quoted \<newline> disappears. */
3206 {
3207 if (retind > 0)
3208 retind--; /* swallow previously-added backslash */
3209 continue;
3210 }
3211
3212 RESIZE_MALLOCED_BUFFER (ret, retind, 2, retsize, 64);
3213 if MBTEST(ch == CTLESC || ch == CTLNUL)
3214 ret[retind++] = CTLESC;
3215 ret[retind++] = ch;
3216 continue;
3217 }
3218 /* If we're reparsing the input (e.g., from parse_string_to_word_list),
3219 we've already prepended CTLESC to single-quoted results of $'...'.
3220 We may want to do this for other CTLESC-quoted characters in
3221 reparse, too. */
3222 else if MBTEST((parser_state & PST_REPARSE) && open == '\'' && (ch == CTLESC || ch == CTLNUL))
3223 {
3224 RESIZE_MALLOCED_BUFFER (ret, retind, 1, retsize, 64);
3225 ret[retind++] = ch;
3226 continue;
3227 }
3228 else if MBTEST(ch == CTLESC || ch == CTLNUL) /* special shell escapes */
3229 {
3230 RESIZE_MALLOCED_BUFFER (ret, retind, 2, retsize, 64);
3231 ret[retind++] = CTLESC;
3232 ret[retind++] = ch;
3233 continue;
3234 }
3235 else if MBTEST(ch == close) /* ending delimiter */
3236 count--;
3237 /* handle nested ${...} specially. */
3238 else if MBTEST(open != close && (tflags & LEX_WASDOL) && open == '{' && ch == open) /* } */
3239 count++;
3240 else if MBTEST(((flags & P_FIRSTCLOSE) == 0) && ch == open) /* nested begin */
3241 count++;
3242
3243 /* Add this character. */
3244 RESIZE_MALLOCED_BUFFER (ret, retind, 1, retsize, 64);
3245 ret[retind++] = ch;
3246
3247 /* If we just read the ending character, don't bother continuing. */
3248 if (count == 0)
3249 break;
3250
3251 if (open == '\'') /* '' inside grouping construct */
3252 {
3253 if MBTEST((flags & P_ALLOWESC) && ch == '\\')
3254 tflags |= LEX_PASSNEXT;
3255 continue;
3256 }
3257
3258 if MBTEST(ch == '\\') /* backslashes */
3259 tflags |= LEX_PASSNEXT;
3260
3261 /* Based on which dolstate is currently in (param, op, or word),
3262 decide what the op is. We're really only concerned if it's % or
3263 #, so we can turn on a flag that says whether or not we should
3264 treat single quotes as special when inside a double-quoted
3265 ${...}. This logic must agree with subst.c:extract_dollar_brace_string
3266 since they share the same defines. */
3267 if (flags & P_DOLBRACE)
3268 {
3269 /* ${param%[%]word} */
3270 if MBTEST(dolbrace_state == DOLBRACE_PARAM && ch == '%' && retind > 1)
3271 dolbrace_state = DOLBRACE_QUOTE;
3272 /* ${param#[#]word} */
3273 else if MBTEST(dolbrace_state == DOLBRACE_PARAM && ch == '#' && retind > 1)
3274 dolbrace_state = DOLBRACE_QUOTE;
3275 /* ${param/[/]pat/rep} */
3276 else if MBTEST(dolbrace_state == DOLBRACE_PARAM && ch == '/' && retind > 1)
3277 dolbrace_state = DOLBRACE_QUOTE;
3278 /* ${param^[^]pat} */
3279 else if MBTEST(dolbrace_state == DOLBRACE_PARAM && ch == '^' && retind > 1)
3280 dolbrace_state = DOLBRACE_QUOTE;
3281 /* ${param,[,]pat} */
3282 else if MBTEST(dolbrace_state == DOLBRACE_PARAM && ch == ',' && retind > 1)
3283 dolbrace_state = DOLBRACE_QUOTE;
3284 else if MBTEST(dolbrace_state == DOLBRACE_PARAM && strchr ("#%^,~:-=?+/", ch) != 0)
3285 dolbrace_state = DOLBRACE_OP;
3286 else if MBTEST(dolbrace_state == DOLBRACE_OP && strchr ("#%^,~:-=?+/", ch) == 0)
3287 dolbrace_state = DOLBRACE_WORD;
3288 }
3289
3290 /* The big hammer. Single quotes aren't special in double quotes. The
3291 problem is that Posix used to say the single quotes are semi-special:
3292 within a double-quoted ${...} construct "an even number of
3293 unescaped double-quotes or single-quotes, if any, shall occur." */
3294 /* This was changed in Austin Group Interp 221 */
3295 if MBTEST(posixly_correct && shell_compatibility_level > 41 && dolbrace_state != DOLBRACE_QUOTE && (flags & P_DQUOTE) && (flags & P_DOLBRACE) && ch == '\'')
3296 continue;
3297
3298 /* Could also check open == '`' if we want to parse grouping constructs
3299 inside old-style command substitution. */
3300 if (open != close) /* a grouping construct */
3301 {
3302 if MBTEST(shellquote (ch))
3303 {
3304 /* '', ``, or "" inside $(...) or other grouping construct. */
3305 push_delimiter (dstack, ch);
3306 if MBTEST((tflags & LEX_WASDOL) && ch == '\'') /* $'...' inside group */
3307 nestret = parse_matched_pair (ch, ch, ch, &nestlen, P_ALLOWESC|rflags);
3308 else
3309 nestret = parse_matched_pair (ch, ch, ch, &nestlen, rflags);
3310 pop_delimiter (dstack);
3311 CHECK_NESTRET_ERROR ();
3312
3313 if MBTEST((tflags & LEX_WASDOL) && ch == '\'' && (extended_quote || (rflags & P_DQUOTE) == 0))
3314 {
3315 /* Translate $'...' here. */
3316 ttrans = ansiexpand (nestret, 0, nestlen - 1, &ttranslen);
3317 xfree (nestret);
3318
3319 if ((rflags & P_DQUOTE) == 0)
3320 {
3321 nestret = sh_single_quote (ttrans);
3322 free (ttrans);
3323 nestlen = strlen (nestret);
3324 }
3325 else
3326 {
3327 nestret = ttrans;
3328 nestlen = ttranslen;
3329 }
3330 retind -= 2; /* back up before the $' */
3331 }
3332 else if MBTEST((tflags & LEX_WASDOL) && ch == '"' && (extended_quote || (rflags & P_DQUOTE) == 0))
3333 {
3334 /* Locale expand $"..." here. */
3335 ttrans = localeexpand (nestret, 0, nestlen - 1, start_lineno, &ttranslen);
3336 xfree (nestret);
3337
3338 nestret = sh_mkdoublequoted (ttrans, ttranslen, 0);
3339 free (ttrans);
3340 nestlen = ttranslen + 2;
3341 retind -= 2; /* back up before the $" */
3342 }
3343
3344 APPEND_NESTRET ();
3345 FREE (nestret);
3346 }
3347 else if ((flags & P_ARRAYSUB) && (tflags & LEX_WASDOL) && (ch == '(' || ch == '{' || ch == '[')) /* ) } ] */
3348 goto parse_dollar_word;
3349 }
3350 /* Parse an old-style command substitution within double quotes as a
3351 single word. */
3352 /* XXX - sh and ksh93 don't do this - XXX */
3353 else if MBTEST(open == '"' && ch == '`')
3354 {
3355 nestret = parse_matched_pair (0, '`', '`', &nestlen, rflags);
3356
3357 CHECK_NESTRET_ERROR ();
3358 APPEND_NESTRET ();
3359
3360 FREE (nestret);
3361 }
3362 else if MBTEST(open != '`' && (tflags & LEX_WASDOL) && (ch == '(' || ch == '{' || ch == '[')) /* ) } ] */
3363 /* check for $(), $[], or ${} inside quoted string. */
3364 {
3365 parse_dollar_word:
3366 if (open == ch) /* undo previous increment */
3367 count--;
3368 if (ch == '(') /* ) */
3369 nestret = parse_comsub (0, '(', ')', &nestlen, (rflags|P_COMMAND) & ~P_DQUOTE);
3370 else if (ch == '{') /* } */
3371 nestret = parse_matched_pair (0, '{', '}', &nestlen, P_FIRSTCLOSE|P_DOLBRACE|rflags);
3372 else if (ch == '[') /* ] */
3373 nestret = parse_matched_pair (0, '[', ']', &nestlen, rflags);
3374
3375 CHECK_NESTRET_ERROR ();
3376 APPEND_NESTRET ();
3377
3378 FREE (nestret);
3379 }
3380 if MBTEST(ch == '$')
3381 tflags |= LEX_WASDOL;
3382 else
3383 tflags &= ~LEX_WASDOL;
3384 }
3385
3386 ret[retind] = '\0';
3387 if (lenp)
3388 *lenp = retind;
3389 /*itrace("parse_matched_pair[%d]: returning %s", line_number, ret);*/
3390 return ret;
3391 }
3392
3393 /* Parse a $(...) command substitution. This is messier than I'd like, and
3394 reproduces a lot more of the token-reading code than I'd like. */
3395 static char *
3396 parse_comsub (qc, open, close, lenp, flags)
3397 int qc; /* `"' if this construct is within double quotes */
3398 int open, close;
3399 int *lenp, flags;
3400 {
3401 int count, ch, peekc, tflags, lex_rwlen, lex_wlen, lex_firstind;
3402 int nestlen, ttranslen, start_lineno;
3403 char *ret, *nestret, *ttrans, *heredelim;
3404 int retind, retsize, rflags, hdlen;
3405
3406 /* Posix interp 217 says arithmetic expressions have precedence, so
3407 assume $(( introduces arithmetic expansion and parse accordingly. */
3408 peekc = shell_getc (0);
3409 shell_ungetc (peekc);
3410 if (peekc == '(')
3411 return (parse_matched_pair (qc, open, close, lenp, 0));
3412
3413 /*itrace("parse_comsub: qc = `%c' open = %c close = %c", qc, open, close);*/
3414 count = 1;
3415 tflags = LEX_RESWDOK;
3416
3417 if ((flags & P_COMMAND) && qc != '\'' && qc != '"' && (flags & P_DQUOTE) == 0)
3418 tflags |= LEX_CKCASE;
3419 if ((tflags & LEX_CKCASE) && (interactive == 0 || interactive_comments))
3420 tflags |= LEX_CKCOMMENT;
3421
3422 /* RFLAGS is the set of flags we want to pass to recursive calls. */
3423 rflags = (flags & P_DQUOTE);
3424
3425 ret = (char *)xmalloc (retsize = 64);
3426 retind = 0;
3427
3428 start_lineno = line_number;
3429 lex_rwlen = lex_wlen = 0;
3430
3431 heredelim = 0;
3432 lex_firstind = -1;
3433
3434 while (count)
3435 {
3436 comsub_readchar:
3437 ch = shell_getc (qc != '\'' && (tflags & (LEX_INCOMMENT|LEX_PASSNEXT)) == 0);
3438
3439 if (ch == EOF)
3440 {
3441 eof_error:
3442 free (ret);
3443 FREE (heredelim);
3444 parser_error (start_lineno, _("unexpected EOF while looking for matching `%c'"), close);
3445 EOF_Reached = 1; /* XXX */
3446 return (&matched_pair_error);
3447 }
3448
3449 /* If we hit the end of a line and are reading the contents of a here
3450 document, and it's not the same line that the document starts on,
3451 check for this line being the here doc delimiter. Otherwise, if
3452 we're in a here document, mark the next character as the beginning
3453 of a line. */
3454 if (ch == '\n')
3455 {
3456 if ((tflags & LEX_HEREDELIM) && heredelim)
3457 {
3458 tflags &= ~LEX_HEREDELIM;
3459 tflags |= LEX_INHEREDOC;
3460 lex_firstind = retind + 1;
3461 }
3462 else if (tflags & LEX_INHEREDOC)
3463 {
3464 int tind;
3465 tind = lex_firstind;
3466 while ((tflags & LEX_STRIPDOC) && ret[tind] == '\t')
3467 tind++;
3468 if (STREQN (ret + tind, heredelim, hdlen))
3469 {
3470 tflags &= ~(LEX_STRIPDOC|LEX_INHEREDOC);
3471 /*itrace("parse_comsub:%d: found here doc end `%s'", line_number, ret + tind);*/
3472 free (heredelim);
3473 heredelim = 0;
3474 lex_firstind = -1;
3475 }
3476 else
3477 lex_firstind = retind + 1;
3478 }
3479 }
3480
3481 /* Possible reprompting. */
3482 if (ch == '\n' && SHOULD_PROMPT ())
3483 prompt_again ();
3484
3485 /* XXX -- possibly allow here doc to be delimited by ending right
3486 paren. */
3487 if ((tflags & LEX_INHEREDOC) && ch == close && count == 1)
3488 {
3489 int tind;
3490 /*itrace("parse_comsub: in here doc, ch == close, retind - firstind = %d hdlen = %d retind = %d", retind-lex_firstind, hdlen, retind);*/
3491 tind = lex_firstind;
3492 while ((tflags & LEX_STRIPDOC) && ret[tind] == '\t')
3493 tind++;
3494 if (retind-tind == hdlen && STREQN (ret + tind, heredelim, hdlen))
3495 {
3496 tflags &= ~(LEX_STRIPDOC|LEX_INHEREDOC);
3497 /*itrace("parse_comsub:%d: found here doc end `%s'", line_number, ret + tind);*/
3498 free (heredelim);
3499 heredelim = 0;
3500 lex_firstind = -1;
3501 }
3502 }
3503
3504 /* Don't bother counting parens or doing anything else if in a comment */
3505 if (tflags & (LEX_INCOMMENT|LEX_INHEREDOC))
3506 {
3507 /* Add this character. */
3508 RESIZE_MALLOCED_BUFFER (ret, retind, 1, retsize, 64);
3509 ret[retind++] = ch;
3510
3511 if ((tflags & LEX_INCOMMENT) && ch == '\n')
3512 {
3513 /*itrace("parse_comsub:%d: lex_incomment -> 0 ch = `%c'", line_number, ch);*/
3514 tflags &= ~LEX_INCOMMENT;
3515 }
3516
3517 continue;
3518 }
3519
3520 if (tflags & LEX_PASSNEXT) /* last char was backslash */
3521 {
3522 /*itrace("parse_comsub:%d: lex_passnext -> 0 ch = `%c' (%d)", line_number, ch, __LINE__);*/
3523 tflags &= ~LEX_PASSNEXT;
3524 if (qc != '\'' && ch == '\n') /* double-quoted \<newline> disappears. */
3525 {
3526 if (retind > 0)
3527 retind--; /* swallow previously-added backslash */
3528 continue;
3529 }
3530
3531 RESIZE_MALLOCED_BUFFER (ret, retind, 2, retsize, 64);
3532 if MBTEST(ch == CTLESC || ch == CTLNUL)
3533 ret[retind++] = CTLESC;
3534 ret[retind++] = ch;
3535 continue;
3536 }
3537
3538 /* If this is a shell break character, we are not in a word. If not,
3539 we either start or continue a word. */
3540 if MBTEST(shellbreak (ch))
3541 {
3542 tflags &= ~LEX_INWORD;
3543 /*itrace("parse_comsub:%d: lex_inword -> 0 ch = `%c' (%d)", line_number, ch, __LINE__);*/
3544 }
3545 else
3546 {
3547 if (tflags & LEX_INWORD)
3548 {
3549 lex_wlen++;
3550 /*itrace("parse_comsub:%d: lex_inword == 1 ch = `%c' lex_wlen = %d (%d)", line_number, ch, lex_wlen, __LINE__);*/
3551 }
3552 else
3553 {
3554 /*itrace("parse_comsub:%d: lex_inword -> 1 ch = `%c' (%d)", line_number, ch, __LINE__);*/
3555 tflags |= LEX_INWORD;
3556 lex_wlen = 0;
3557 }
3558 }
3559
3560 /* Skip whitespace */
3561 if MBTEST(shellblank (ch) && (tflags & LEX_HEREDELIM) == 0 && lex_rwlen == 0)
3562 {
3563 /* Add this character. */
3564 RESIZE_MALLOCED_BUFFER (ret, retind, 1, retsize, 64);
3565 ret[retind++] = ch;
3566 continue;
3567 }
3568
3569 /* Either we are looking for the start of the here-doc delimiter
3570 (lex_firstind == -1) or we are reading one (lex_firstind >= 0).
3571 If this character is a shell break character and we are reading
3572 the delimiter, save it and note that we are now reading a here
3573 document. If we've found the start of the delimiter, note it by
3574 setting lex_firstind. Backslashes can quote shell metacharacters
3575 in here-doc delimiters. */
3576 if (tflags & LEX_HEREDELIM)
3577 {
3578 if (lex_firstind == -1 && shellbreak (ch) == 0)
3579 lex_firstind = retind;
3580 #if 0
3581 else if (heredelim && (tflags & LEX_PASSNEXT) == 0 && ch == '\n')
3582 {
3583 tflags |= LEX_INHEREDOC;
3584 tflags &= ~LEX_HEREDELIM;
3585 lex_firstind = retind + 1;
3586 }
3587 #endif
3588 else if (lex_firstind >= 0 && (tflags & LEX_PASSNEXT) == 0 && shellbreak (ch))
3589 {
3590 if (heredelim == 0)
3591 {
3592 nestret = substring (ret, lex_firstind, retind);
3593 heredelim = string_quote_removal (nestret, 0);
3594 free (nestret);
3595 hdlen = STRLEN(heredelim);
3596 /*itrace("parse_comsub:%d: found here doc delimiter `%s' (%d)", line_number, heredelim, hdlen);*/
3597 }
3598 if (ch == '\n')
3599 {
3600 tflags |= LEX_INHEREDOC;
3601 tflags &= ~LEX_HEREDELIM;
3602 lex_firstind = retind + 1;
3603 }
3604 else
3605 lex_firstind = -1;
3606 }
3607 }
3608
3609 /* Meta-characters that can introduce a reserved word. Not perfect yet. */
3610 if MBTEST((tflags & LEX_RESWDOK) == 0 && (tflags & LEX_CKCASE) && (tflags & LEX_INCOMMENT) == 0 && (shellmeta(ch) || ch == '\n'))
3611 {
3612 /* Add this character. */
3613 RESIZE_MALLOCED_BUFFER (ret, retind, 1, retsize, 64);
3614 ret[retind++] = ch;
3615 peekc = shell_getc (1);
3616 if (ch == peekc && (ch == '&' || ch == '|' || ch == ';')) /* two-character tokens */
3617 {
3618 RESIZE_MALLOCED_BUFFER (ret, retind, 1, retsize, 64);
3619 ret[retind++] = peekc;
3620 /*itrace("parse_comsub:%d: set lex_reswordok = 1, ch = `%c'", line_number, ch);*/
3621 tflags |= LEX_RESWDOK;
3622 lex_rwlen = 0;
3623 continue;
3624 }
3625 else if (ch == '\n' || COMSUB_META(ch))
3626 {
3627 shell_ungetc (peekc);
3628 /*itrace("parse_comsub:%d: set lex_reswordok = 1, ch = `%c'", line_number, ch);*/
3629 tflags |= LEX_RESWDOK;
3630 lex_rwlen = 0;
3631 continue;
3632 }
3633 else if (ch == EOF)
3634 goto eof_error;
3635 else
3636 {
3637 /* `unget' the character we just added and fall through */
3638 retind--;
3639 shell_ungetc (peekc);
3640 }
3641 }
3642
3643 /* If we can read a reserved word, try to read one. */
3644 if (tflags & LEX_RESWDOK)
3645 {
3646 if MBTEST(islower (ch))
3647 {
3648 /* Add this character. */
3649 RESIZE_MALLOCED_BUFFER (ret, retind, 1, retsize, 64);
3650 ret[retind++] = ch;
3651 lex_rwlen++;
3652 continue;
3653 }
3654 else if MBTEST(lex_rwlen == 4 && shellbreak (ch))
3655 {
3656 if (STREQN (ret + retind - 4, "case", 4))
3657 {
3658 tflags |= LEX_INCASE;
3659 /*itrace("parse_comsub:%d: found `case', lex_incase -> 1 lex_reswdok -> 0", line_number);*/
3660 }
3661 else if (STREQN (ret + retind - 4, "esac", 4))
3662 {
3663 tflags &= ~LEX_INCASE;
3664 /*itrace("parse_comsub:%d: found `esac', lex_incase -> 0 lex_reswdok -> 0", line_number);*/
3665 }
3666 tflags &= ~LEX_RESWDOK;
3667 }
3668 else if MBTEST((tflags & LEX_CKCOMMENT) && ch == '#' && (lex_rwlen == 0 || ((tflags & LEX_INWORD) && lex_wlen == 0)))
3669 ; /* don't modify LEX_RESWDOK if we're starting a comment */
3670 else if MBTEST((tflags & LEX_INCASE) && ch != '\n')
3671 /* If we can read a reserved word and we're in case, we're at the
3672 point where we can read a new pattern list or an esac. We
3673 handle the esac case above. If we read a newline, we want to
3674 leave LEX_RESWDOK alone. If we read anything else, we want to
3675 turn off LEX_RESWDOK, since we're going to read a pattern list. */
3676 {
3677 tflags &= ~LEX_RESWDOK;
3678 /*itrace("parse_comsub:%d: lex_incase == 1 found `%c', lex_reswordok -> 0", line_number, ch);*/
3679 }
3680 else if MBTEST(shellbreak (ch) == 0)
3681 {
3682 tflags &= ~LEX_RESWDOK;
3683 /*itrace("parse_comsub:%d: found `%c', lex_reswordok -> 0", line_number, ch);*/
3684 }
3685 }
3686
3687 /* Might be the start of a here-doc delimiter */
3688 if MBTEST((tflags & LEX_INCOMMENT) == 0 && (tflags & LEX_CKCASE) && ch == '<')
3689 {
3690 /* Add this character. */
3691 RESIZE_MALLOCED_BUFFER (ret, retind, 1, retsize, 64);
3692 ret[retind++] = ch;
3693 peekc = shell_getc (1);
3694 if (peekc == EOF)
3695 goto eof_error;
3696 if (peekc == ch)
3697 {
3698 RESIZE_MALLOCED_BUFFER (ret, retind, 1, retsize, 64);
3699 ret[retind++] = peekc;
3700 peekc = shell_getc (1);
3701 if (peekc == EOF)
3702 goto eof_error;
3703 if (peekc == '-')
3704 {
3705 RESIZE_MALLOCED_BUFFER (ret, retind, 1, retsize, 64);
3706 ret[retind++] = peekc;
3707 tflags |= LEX_STRIPDOC;
3708 }
3709 else
3710 shell_ungetc (peekc);
3711 if (peekc != '<')
3712 {
3713 tflags |= LEX_HEREDELIM;
3714 lex_firstind = -1;
3715 }
3716 continue;
3717 }
3718 else
3719 ch = peekc; /* fall through and continue XXX */
3720 }
3721 else if MBTEST((tflags & LEX_CKCOMMENT) && (tflags & LEX_INCOMMENT) == 0 && ch == '#' && (((tflags & LEX_RESWDOK) && lex_rwlen == 0) || ((tflags & LEX_INWORD) && lex_wlen == 0)))
3722 {
3723 /*itrace("parse_comsub:%d: lex_incomment -> 1 (%d)", line_number, __LINE__);*/
3724 tflags |= LEX_INCOMMENT;
3725 }
3726
3727 if MBTEST(ch == CTLESC || ch == CTLNUL) /* special shell escapes */
3728 {
3729 RESIZE_MALLOCED_BUFFER (ret, retind, 2, retsize, 64);
3730 ret[retind++] = CTLESC;
3731 ret[retind++] = ch;
3732 continue;
3733 }
3734 #if 0
3735 else if MBTEST((tflags & LEX_INCASE) && ch == close && close == ')')
3736 tflags &= ~LEX_INCASE; /* XXX */
3737 #endif
3738 else if MBTEST(ch == close && (tflags & LEX_INCASE) == 0) /* ending delimiter */
3739 {
3740 count--;
3741 /*itrace("parse_comsub:%d: found close: count = %d", line_number, count);*/
3742 }
3743 else if MBTEST(((flags & P_FIRSTCLOSE) == 0) && (tflags & LEX_INCASE) == 0 && ch == open) /* nested begin */
3744 {
3745 count++;
3746 /*itrace("parse_comsub:%d: found open: count = %d", line_number, count);*/
3747 }
3748
3749 /* Add this character. */
3750 RESIZE_MALLOCED_BUFFER (ret, retind, 1, retsize, 64);
3751 ret[retind++] = ch;
3752
3753 /* If we just read the ending character, don't bother continuing. */
3754 if (count == 0)
3755 break;
3756
3757 if MBTEST(ch == '\\') /* backslashes */
3758 tflags |= LEX_PASSNEXT;
3759
3760 if MBTEST(shellquote (ch))
3761 {
3762 /* '', ``, or "" inside $(...). */
3763 push_delimiter (dstack, ch);
3764 if MBTEST((tflags & LEX_WASDOL) && ch == '\'') /* $'...' inside group */
3765 nestret = parse_matched_pair (ch, ch, ch, &nestlen, P_ALLOWESC|rflags);
3766 else
3767 nestret = parse_matched_pair (ch, ch, ch, &nestlen, rflags);
3768 pop_delimiter (dstack);
3769 CHECK_NESTRET_ERROR ();
3770
3771 if MBTEST((tflags & LEX_WASDOL) && ch == '\'' && (extended_quote || (rflags & P_DQUOTE) == 0))
3772 {
3773 /* Translate $'...' here. */
3774 ttrans = ansiexpand (nestret, 0, nestlen - 1, &ttranslen);
3775 xfree (nestret);
3776
3777 if ((rflags & P_DQUOTE) == 0)
3778 {
3779 nestret = sh_single_quote (ttrans);
3780 free (ttrans);
3781 nestlen = strlen (nestret);
3782 }
3783 else
3784 {
3785 nestret = ttrans;
3786 nestlen = ttranslen;
3787 }
3788 retind -= 2; /* back up before the $' */
3789 }
3790 else if MBTEST((tflags & LEX_WASDOL) && ch == '"' && (extended_quote || (rflags & P_DQUOTE) == 0))
3791 {
3792 /* Locale expand $"..." here. */
3793 ttrans = localeexpand (nestret, 0, nestlen - 1, start_lineno, &ttranslen);
3794 xfree (nestret);
3795
3796 nestret = sh_mkdoublequoted (ttrans, ttranslen, 0);
3797 free (ttrans);
3798 nestlen = ttranslen + 2;
3799 retind -= 2; /* back up before the $" */
3800 }
3801
3802 APPEND_NESTRET ();
3803 FREE (nestret);
3804 }
3805 else if MBTEST((tflags & LEX_WASDOL) && (ch == '(' || ch == '{' || ch == '[')) /* ) } ] */
3806 /* check for $(), $[], or ${} inside command substitution. */
3807 {
3808 if ((tflags & LEX_INCASE) == 0 && open == ch) /* undo previous increment */
3809 count--;
3810 if (ch == '(') /* ) */
3811 nestret = parse_comsub (0, '(', ')', &nestlen, (rflags|P_COMMAND) & ~P_DQUOTE);
3812 else if (ch == '{') /* } */
3813 nestret = parse_matched_pair (0, '{', '}', &nestlen, P_FIRSTCLOSE|P_DOLBRACE|rflags);
3814 else if (ch == '[') /* ] */
3815 nestret = parse_matched_pair (0, '[', ']', &nestlen, rflags);
3816
3817 CHECK_NESTRET_ERROR ();
3818 APPEND_NESTRET ();
3819
3820 FREE (nestret);
3821 }
3822 if MBTEST(ch == '$')
3823 tflags |= LEX_WASDOL;
3824 else
3825 tflags &= ~LEX_WASDOL;
3826 }
3827
3828 FREE (heredelim);
3829 ret[retind] = '\0';
3830 if (lenp)
3831 *lenp = retind;
3832 /*itrace("parse_comsub:%d: returning `%s'", line_number, ret);*/
3833 return ret;
3834 }
3835
3836 /* Recursively call the parser to parse a $(...) command substitution. */
3837 char *
3838 xparse_dolparen (base, string, indp, flags)
3839 char *base;
3840 char *string;
3841 int *indp;
3842 int flags;
3843 {
3844 sh_parser_state_t ps;
3845 sh_input_line_state_t ls;
3846 int orig_ind, nc, sflags;
3847 char *ret, *s, *ep, *ostring;
3848
3849 /*yydebug = 1;*/
3850 orig_ind = *indp;
3851 ostring = string;
3852
3853 /*itrace("xparse_dolparen: size = %d shell_input_line = `%s'", shell_input_line_size, shell_input_line);*/
3854 sflags = SEVAL_NONINT|SEVAL_NOHIST|SEVAL_NOFREE;
3855 if (flags & SX_NOLONGJMP)
3856 sflags |= SEVAL_NOLONGJMP;
3857 save_parser_state (&ps);
3858 save_input_line_state (&ls);
3859
3860 /*(*/
3861 parser_state |= PST_CMDSUBST|PST_EOFTOKEN; /* allow instant ')' */ /*(*/
3862 shell_eof_token = ')';
3863 parse_string (string, "command substitution", sflags, &ep);
3864
3865 restore_parser_state (&ps);
3866 reset_parser ();
3867 /* reset_parser clears shell_input_line and associated variables */
3868 restore_input_line_state (&ls);
3869 if (interactive)
3870 token_to_read = 0;
3871
3872 /* Need to find how many characters parse_and_execute consumed, update
3873 *indp, if flags != 0, copy the portion of the string parsed into RET
3874 and return it. If flags & 1 (EX_NOALLOC) we can return NULL. */
3875
3876 /*(*/
3877 if (ep[-1] != ')')
3878 {
3879 #if DEBUG
3880 if (ep[-1] != '\n')
3881 itrace("xparse_dolparen:%d: ep[-1] != RPAREN (%d), ep = `%s'", line_number, ep[-1], ep);
3882 #endif
3883 while (ep > ostring && ep[-1] == '\n') ep--;
3884 }
3885
3886 nc = ep - ostring;
3887 *indp = ep - base - 1;
3888
3889 /*(*/
3890 #if DEBUG
3891 if (base[*indp] != ')')
3892 itrace("xparse_dolparen:%d: base[%d] != RPAREN (%d), base = `%s'", line_number, *indp, base[*indp], base);
3893 #endif
3894
3895 if (flags & SX_NOALLOC)
3896 return (char *)NULL;
3897
3898 if (nc == 0)
3899 {
3900 ret = xmalloc (1);
3901 ret[0] = '\0';
3902 }
3903 else
3904 ret = substring (ostring, 0, nc - 1);
3905
3906 return ret;
3907 }
3908
3909 #if defined (DPAREN_ARITHMETIC) || defined (ARITH_FOR_COMMAND)
3910 /* Parse a double-paren construct. It can be either an arithmetic
3911 command, an arithmetic `for' command, or a nested subshell. Returns
3912 the parsed token, -1 on error, or -2 if we didn't do anything and
3913 should just go on. */
3914 static int
3915 parse_dparen (c)
3916 int c;
3917 {
3918 int cmdtyp, sline;
3919 char *wval;
3920 WORD_DESC *wd;
3921
3922 #if defined (ARITH_FOR_COMMAND)
3923 if (last_read_token == FOR)
3924 {
3925 arith_for_lineno = line_number;
3926 cmdtyp = parse_arith_cmd (&wval, 0);
3927 if (cmdtyp == 1)
3928 {
3929 wd = alloc_word_desc ();
3930 wd->word = wval;
3931 yylval.word_list = make_word_list (wd, (WORD_LIST *)NULL);
3932 return (ARITH_FOR_EXPRS);
3933 }
3934 else
3935 return -1; /* ERROR */
3936 }
3937 #endif
3938
3939 #if defined (DPAREN_ARITHMETIC)
3940 if (reserved_word_acceptable (last_read_token))
3941 {
3942 sline = line_number;
3943
3944 cmdtyp = parse_arith_cmd (&wval, 0);
3945 if (cmdtyp == 1) /* arithmetic command */
3946 {
3947 wd = alloc_word_desc ();
3948 wd->word = wval;
3949 wd->flags = W_QUOTED|W_NOSPLIT|W_NOGLOB|W_DQUOTE;
3950 yylval.word_list = make_word_list (wd, (WORD_LIST *)NULL);
3951 return (ARITH_CMD);
3952 }
3953 else if (cmdtyp == 0) /* nested subshell */
3954 {
3955 push_string (wval, 0, (alias_t *)NULL);
3956 if ((parser_state & PST_CASEPAT) == 0)
3957 parser_state |= PST_SUBSHELL;
3958 return (c);
3959 }
3960 else /* ERROR */
3961 return -1;
3962 }
3963 #endif
3964
3965 return -2; /* XXX */
3966 }
3967
3968 /* We've seen a `(('. Look for the matching `))'. If we get it, return 1.
3969 If not, assume it's a nested subshell for backwards compatibility and
3970 return 0. In any case, put the characters we've consumed into a locally-
3971 allocated buffer and make *ep point to that buffer. Return -1 on an
3972 error, for example EOF. */
3973 static int
3974 parse_arith_cmd (ep, adddq)
3975 char **ep;
3976 int adddq;
3977 {
3978 int exp_lineno, rval, c;
3979 char *ttok, *tokstr;
3980 int ttoklen;
3981
3982 exp_lineno = line_number;
3983 ttok = parse_matched_pair (0, '(', ')', &ttoklen, 0);
3984 rval = 1;
3985 if (ttok == &matched_pair_error)
3986 return -1;
3987 /* Check that the next character is the closing right paren. If
3988 not, this is a syntax error. ( */
3989 c = shell_getc (0);
3990 if MBTEST(c != ')')
3991 rval = 0;
3992
3993 tokstr = (char *)xmalloc (ttoklen + 4);
3994
3995 /* if ADDDQ != 0 then (( ... )) -> "..." */
3996 if (rval == 1 && adddq) /* arith cmd, add double quotes */
3997 {
3998 tokstr[0] = '"';
3999 strncpy (tokstr + 1, ttok, ttoklen - 1);
4000 tokstr[ttoklen] = '"';
4001 tokstr[ttoklen+1] = '\0';
4002 }
4003 else if (rval == 1) /* arith cmd, don't add double quotes */
4004 {
4005 strncpy (tokstr, ttok, ttoklen - 1);
4006 tokstr[ttoklen-1] = '\0';
4007 }
4008 else /* nested subshell */
4009 {
4010 tokstr[0] = '(';
4011 strncpy (tokstr + 1, ttok, ttoklen - 1);
4012 tokstr[ttoklen] = ')';
4013 tokstr[ttoklen+1] = c;
4014 tokstr[ttoklen+2] = '\0';
4015 }
4016
4017 *ep = tokstr;
4018 FREE (ttok);
4019 return rval;
4020 }
4021 #endif /* DPAREN_ARITHMETIC || ARITH_FOR_COMMAND */
4022
4023 #if defined (COND_COMMAND)
4024 static void
4025 cond_error ()
4026 {
4027 char *etext;
4028
4029 if (EOF_Reached && cond_token != COND_ERROR) /* [[ */
4030 parser_error (cond_lineno, _("unexpected EOF while looking for `]]'"));
4031 else if (cond_token != COND_ERROR)
4032 {
4033 if (etext = error_token_from_token (cond_token))
4034 {
4035 parser_error (cond_lineno, _("syntax error in conditional expression: unexpected token `%s'"), etext);
4036 free (etext);
4037 }
4038 else
4039 parser_error (cond_lineno, _("syntax error in conditional expression"));
4040 }
4041 }
4042
4043 static COND_COM *
4044 cond_expr ()
4045 {
4046 return (cond_or ());
4047 }
4048
4049 static COND_COM *
4050 cond_or ()
4051 {
4052 COND_COM *l, *r;
4053
4054 l = cond_and ();
4055 if (cond_token == OR_OR)
4056 {
4057 r = cond_or ();
4058 l = make_cond_node (COND_OR, (WORD_DESC *)NULL, l, r);
4059 }
4060 return l;
4061 }
4062
4063 static COND_COM *
4064 cond_and ()
4065 {
4066 COND_COM *l, *r;
4067
4068 l = cond_term ();
4069 if (cond_token == AND_AND)
4070 {
4071 r = cond_and ();
4072 l = make_cond_node (COND_AND, (WORD_DESC *)NULL, l, r);
4073 }
4074 return l;
4075 }
4076
4077 static int
4078 cond_skip_newlines ()
4079 {
4080 while ((cond_token = read_token (READ)) == '\n')
4081 {
4082 if (SHOULD_PROMPT ())
4083 prompt_again ();
4084 }
4085 return (cond_token);
4086 }
4087
4088 #define COND_RETURN_ERROR() \
4089 do { cond_token = COND_ERROR; return ((COND_COM *)NULL); } while (0)
4090
4091 static COND_COM *
4092 cond_term ()
4093 {
4094 WORD_DESC *op;
4095 COND_COM *term, *tleft, *tright;
4096 int tok, lineno;
4097 char *etext;
4098
4099 /* Read a token. It can be a left paren, a `!', a unary operator, or a
4100 word that should be the first argument of a binary operator. Start by
4101 skipping newlines, since this is a compound command. */
4102 tok = cond_skip_newlines ();
4103 lineno = line_number;
4104 if (tok == COND_END)
4105 {
4106 COND_RETURN_ERROR ();
4107 }
4108 else if (tok == '(')
4109 {
4110 term = cond_expr ();
4111 if (cond_token != ')')
4112 {
4113 if (term)
4114 dispose_cond_node (term); /* ( */
4115 if (etext = error_token_from_token (cond_token))
4116 {
4117 parser_error (lineno, _("unexpected token `%s', expected `)'"), etext);
4118 free (etext);
4119 }
4120 else
4121 parser_error (lineno, _("expected `)'"));
4122 COND_RETURN_ERROR ();
4123 }
4124 term = make_cond_node (COND_EXPR, (WORD_DESC *)NULL, term, (COND_COM *)NULL);
4125 (void)cond_skip_newlines ();
4126 }
4127 else if (tok == BANG || (tok == WORD && (yylval.word->word[0] == '!' && yylval.word->word[1] == '\0')))
4128 {
4129 if (tok == WORD)
4130 dispose_word (yylval.word); /* not needed */
4131 term = cond_term ();
4132 if (term)
4133 term->flags |= CMD_INVERT_RETURN;
4134 }
4135 else if (tok == WORD && yylval.word->word[0] == '-' && yylval.word->word[2] == 0 && test_unop (yylval.word->word))
4136 {
4137 op = yylval.word;
4138 tok = read_token (READ);
4139 if (tok == WORD)
4140 {
4141 tleft = make_cond_node (COND_TERM, yylval.word, (COND_COM *)NULL, (COND_COM *)NULL);
4142 term = make_cond_node (COND_UNARY, op, tleft, (COND_COM *)NULL);
4143 }
4144 else
4145 {
4146 dispose_word (op);
4147 if (etext = error_token_from_token (tok))
4148 {
4149 parser_error (line_number, _("unexpected argument `%s' to conditional unary operator"), etext);
4150 free (etext);
4151 }
4152 else
4153 parser_error (line_number, _("unexpected argument to conditional unary operator"));
4154 COND_RETURN_ERROR ();
4155 }
4156
4157 (void)cond_skip_newlines ();
4158 }
4159 else if (tok == WORD) /* left argument to binary operator */
4160 {
4161 /* lhs */
4162 tleft = make_cond_node (COND_TERM, yylval.word, (COND_COM *)NULL, (COND_COM *)NULL);
4163
4164 /* binop */
4165 tok = read_token (READ);
4166 if (tok == WORD && test_binop (yylval.word->word))
4167 {
4168 op = yylval.word;
4169 if (op->word[0] == '=' && (op->word[1] == '\0' || (op->word[1] == '=' && op->word[2] == '\0')))
4170 parser_state |= PST_EXTPAT;
4171 else if (op->word[0] == '!' && op->word[1] == '=' && op->word[2] == '\0')
4172 parser_state |= PST_EXTPAT;
4173 }
4174 #if defined (COND_REGEXP)
4175 else if (tok == WORD && STREQ (yylval.word->word, "=~"))
4176 {
4177 op = yylval.word;
4178 parser_state |= PST_REGEXP;
4179 }
4180 #endif
4181 else if (tok == '<' || tok == '>')
4182 op = make_word_from_token (tok); /* ( */
4183 /* There should be a check before blindly accepting the `)' that we have
4184 seen the opening `('. */
4185 else if (tok == COND_END || tok == AND_AND || tok == OR_OR || tok == ')')
4186 {
4187 /* Special case. [[ x ]] is equivalent to [[ -n x ]], just like
4188 the test command. Similarly for [[ x && expr ]] or
4189 [[ x || expr ]] or [[ (x) ]]. */
4190 op = make_word ("-n");
4191 term = make_cond_node (COND_UNARY, op, tleft, (COND_COM *)NULL);
4192 cond_token = tok;
4193 return (term);
4194 }
4195 else
4196 {
4197 if (etext = error_token_from_token (tok))
4198 {
4199 parser_error (line_number, _("unexpected token `%s', conditional binary operator expected"), etext);
4200 free (etext);
4201 }
4202 else
4203 parser_error (line_number, _("conditional binary operator expected"));
4204 dispose_cond_node (tleft);
4205 COND_RETURN_ERROR ();
4206 }
4207
4208 /* rhs */
4209 if (parser_state & PST_EXTPAT)
4210 extended_glob = 1;
4211 tok = read_token (READ);
4212 if (parser_state & PST_EXTPAT)
4213 extended_glob = global_extglob;
4214 parser_state &= ~(PST_REGEXP|PST_EXTPAT);
4215
4216 if (tok == WORD)
4217 {
4218 tright = make_cond_node (COND_TERM, yylval.word, (COND_COM *)NULL, (COND_COM *)NULL);
4219 term = make_cond_node (COND_BINARY, op, tleft, tright);
4220 }
4221 else
4222 {
4223 if (etext = error_token_from_token (tok))
4224 {
4225 parser_error (line_number, _("unexpected argument `%s' to conditional binary operator"), etext);
4226 free (etext);
4227 }
4228 else
4229 parser_error (line_number, _("unexpected argument to conditional binary operator"));
4230 dispose_cond_node (tleft);
4231 dispose_word (op);
4232 COND_RETURN_ERROR ();
4233 }
4234
4235 (void)cond_skip_newlines ();
4236 }
4237 else
4238 {
4239 if (tok < 256)
4240 parser_error (line_number, _("unexpected token `%c' in conditional command"), tok);
4241 else if (etext = error_token_from_token (tok))
4242 {
4243 parser_error (line_number, _("unexpected token `%s' in conditional command"), etext);
4244 free (etext);
4245 }
4246 else
4247 parser_error (line_number, _("unexpected token %d in conditional command"), tok);
4248 COND_RETURN_ERROR ();
4249 }
4250 return (term);
4251 }
4252
4253 /* This is kind of bogus -- we slip a mini recursive-descent parser in
4254 here to handle the conditional statement syntax. */
4255 static COMMAND *
4256 parse_cond_command ()
4257 {
4258 COND_COM *cexp;
4259
4260 global_extglob = extended_glob;
4261 cexp = cond_expr ();
4262 return (make_cond_command (cexp));
4263 }
4264 #endif
4265
4266 #if defined (ARRAY_VARS)
4267 /* When this is called, it's guaranteed that we don't care about anything
4268 in t beyond i. We do save and restore the chars, though. */
4269 static int
4270 token_is_assignment (t, i)
4271 char *t;
4272 int i;
4273 {
4274 unsigned char c, c1;
4275 int r;
4276
4277 c = t[i]; c1 = t[i+1];
4278 t[i] = '='; t[i+1] = '\0';
4279 r = assignment (t, (parser_state & PST_COMPASSIGN) != 0);
4280 t[i] = c; t[i+1] = c1;
4281 return r;
4282 }
4283
4284 /* XXX - possible changes here for `+=' */
4285 static int
4286 token_is_ident (t, i)
4287 char *t;
4288 int i;
4289 {
4290 unsigned char c;
4291 int r;
4292
4293 c = t[i];
4294 t[i] = '\0';
4295 r = legal_identifier (t);
4296 t[i] = c;
4297 return r;
4298 }
4299 #endif
4300
4301 static int
4302 read_token_word (character)
4303 int character;
4304 {
4305 /* The value for YYLVAL when a WORD is read. */
4306 WORD_DESC *the_word;
4307
4308 /* Index into the token that we are building. */
4309 int token_index;
4310
4311 /* ALL_DIGITS becomes zero when we see a non-digit. */
4312 int all_digit_token;
4313
4314 /* DOLLAR_PRESENT becomes non-zero if we see a `$'. */
4315 int dollar_present;
4316
4317 /* COMPOUND_ASSIGNMENT becomes non-zero if we are parsing a compound
4318 assignment. */
4319 int compound_assignment;
4320
4321 /* QUOTED becomes non-zero if we see one of ("), ('), (`), or (\). */
4322 int quoted;
4323
4324 /* Non-zero means to ignore the value of the next character, and just
4325 to add it no matter what. */
4326 int pass_next_character;
4327
4328 /* The current delimiting character. */
4329 int cd;
4330 int result, peek_char;
4331 char *ttok, *ttrans;
4332 int ttoklen, ttranslen;
4333 intmax_t lvalue;
4334
4335 if (token_buffer_size < TOKEN_DEFAULT_INITIAL_SIZE)
4336 token = (char *)xrealloc (token, token_buffer_size = TOKEN_DEFAULT_INITIAL_SIZE);
4337
4338 token_index = 0;
4339 all_digit_token = DIGIT (character);
4340 dollar_present = quoted = pass_next_character = compound_assignment = 0;
4341
4342 for (;;)
4343 {
4344 if (character == EOF)
4345 goto got_token;
4346
4347 if (pass_next_character)
4348 {
4349 pass_next_character = 0;
4350 goto got_escaped_character;
4351 }
4352
4353 cd = current_delimiter (dstack);
4354
4355 /* Handle backslashes. Quote lots of things when not inside of
4356 double-quotes, quote some things inside of double-quotes. */
4357 if MBTEST(character == '\\')
4358 {
4359 peek_char = shell_getc (0);
4360
4361 /* Backslash-newline is ignored in all cases except
4362 when quoted with single quotes. */
4363 if (peek_char == '\n')
4364 {
4365 character = '\n';
4366 goto next_character;
4367 }
4368 else
4369 {
4370 shell_ungetc (peek_char);
4371
4372 /* If the next character is to be quoted, note it now. */
4373 if (cd == 0 || cd == '`' ||
4374 (cd == '"' && peek_char >= 0 && (sh_syntaxtab[peek_char] & CBSDQUOTE)))
4375 pass_next_character++;
4376
4377 quoted = 1;
4378 goto got_character;
4379 }
4380 }
4381
4382 /* Parse a matched pair of quote characters. */
4383 if MBTEST(shellquote (character))
4384 {
4385 push_delimiter (dstack, character);
4386 ttok = parse_matched_pair (character, character, character, &ttoklen, (character == '`') ? P_COMMAND : 0);
4387 pop_delimiter (dstack);
4388 if (ttok == &matched_pair_error)
4389 return -1; /* Bail immediately. */
4390 RESIZE_MALLOCED_BUFFER (token, token_index, ttoklen + 2,
4391 token_buffer_size, TOKEN_DEFAULT_GROW_SIZE);
4392 token[token_index++] = character;
4393 strcpy (token + token_index, ttok);
4394 token_index += ttoklen;
4395 all_digit_token = 0;
4396 quoted = 1;
4397 dollar_present |= (character == '"' && strchr (ttok, '$') != 0);
4398 FREE (ttok);
4399 goto next_character;
4400 }
4401
4402 #ifdef COND_REGEXP
4403 /* When parsing a regexp as a single word inside a conditional command,
4404 we need to special-case characters special to both the shell and
4405 regular expressions. Right now, that is only '(' and '|'. */ /*)*/
4406 if MBTEST((parser_state & PST_REGEXP) && (character == '(' || character == '|')) /*)*/
4407 {
4408 if (character == '|')
4409 goto got_character;
4410
4411 push_delimiter (dstack, character);
4412 ttok = parse_matched_pair (cd, '(', ')', &ttoklen, 0);
4413 pop_delimiter (dstack);
4414 if (ttok == &matched_pair_error)
4415 return -1; /* Bail immediately. */
4416 RESIZE_MALLOCED_BUFFER (token, token_index, ttoklen + 2,
4417 token_buffer_size, TOKEN_DEFAULT_GROW_SIZE);
4418 token[token_index++] = character;
4419 strcpy (token + token_index, ttok);
4420 token_index += ttoklen;
4421 FREE (ttok);
4422 dollar_present = all_digit_token = 0;
4423 goto next_character;
4424 }
4425 #endif /* COND_REGEXP */
4426
4427 #ifdef EXTENDED_GLOB
4428 /* Parse a ksh-style extended pattern matching specification. */
4429 if MBTEST(extended_glob && PATTERN_CHAR (character))
4430 {
4431 peek_char = shell_getc (1);
4432 if MBTEST(peek_char == '(') /* ) */
4433 {
4434 push_delimiter (dstack, peek_char);
4435 ttok = parse_matched_pair (cd, '(', ')', &ttoklen, 0);
4436 pop_delimiter (dstack);
4437 if (ttok == &matched_pair_error)
4438 return -1; /* Bail immediately. */
4439 RESIZE_MALLOCED_BUFFER (token, token_index, ttoklen + 2,
4440 token_buffer_size,
4441 TOKEN_DEFAULT_GROW_SIZE);
4442 token[token_index++] = character;
4443 token[token_index++] = peek_char;
4444 strcpy (token + token_index, ttok);
4445 token_index += ttoklen;
4446 FREE (ttok);
4447 dollar_present = all_digit_token = 0;
4448 goto next_character;
4449 }
4450 else
4451 shell_ungetc (peek_char);
4452 }
4453 #endif /* EXTENDED_GLOB */
4454
4455 /* If the delimiter character is not single quote, parse some of
4456 the shell expansions that must be read as a single word. */
4457 if (shellexp (character))
4458 {
4459 peek_char = shell_getc (1);
4460 /* $(...), <(...), >(...), $((...)), ${...}, and $[...] constructs */
4461 if MBTEST(peek_char == '(' || \
4462 ((peek_char == '{' || peek_char == '[') && character == '$')) /* ) ] } */
4463 {
4464 if (peek_char == '{') /* } */
4465 ttok = parse_matched_pair (cd, '{', '}', &ttoklen, P_FIRSTCLOSE|P_DOLBRACE);
4466 else if (peek_char == '(') /* ) */
4467 {
4468 /* XXX - push and pop the `(' as a delimiter for use by
4469 the command-oriented-history code. This way newlines
4470 appearing in the $(...) string get added to the
4471 history literally rather than causing a possibly-
4472 incorrect `;' to be added. ) */
4473 push_delimiter (dstack, peek_char);
4474 ttok = parse_comsub (cd, '(', ')', &ttoklen, P_COMMAND);
4475 pop_delimiter (dstack);
4476 }
4477 else
4478 ttok = parse_matched_pair (cd, '[', ']', &ttoklen, 0);
4479 if (ttok == &matched_pair_error)
4480 return -1; /* Bail immediately. */
4481 RESIZE_MALLOCED_BUFFER (token, token_index, ttoklen + 2,
4482 token_buffer_size,
4483 TOKEN_DEFAULT_GROW_SIZE);
4484 token[token_index++] = character;
4485 token[token_index++] = peek_char;
4486 strcpy (token + token_index, ttok);
4487 token_index += ttoklen;
4488 FREE (ttok);
4489 dollar_present = 1;
4490 all_digit_token = 0;
4491 goto next_character;
4492 }
4493 /* This handles $'...' and $"..." new-style quoted strings. */
4494 else if MBTEST(character == '$' && (peek_char == '\'' || peek_char == '"'))
4495 {
4496 int first_line;
4497
4498 first_line = line_number;
4499 push_delimiter (dstack, peek_char);
4500 ttok = parse_matched_pair (peek_char, peek_char, peek_char,
4501 &ttoklen,
4502 (peek_char == '\'') ? P_ALLOWESC : 0);
4503 pop_delimiter (dstack);
4504 if (ttok == &matched_pair_error)
4505 return -1;
4506 if (peek_char == '\'')
4507 {
4508 ttrans = ansiexpand (ttok, 0, ttoklen - 1, &ttranslen);
4509 free (ttok);
4510
4511 /* Insert the single quotes and correctly quote any
4512 embedded single quotes (allowed because P_ALLOWESC was
4513 passed to parse_matched_pair). */
4514 ttok = sh_single_quote (ttrans);
4515 free (ttrans);
4516 ttranslen = strlen (ttok);
4517 ttrans = ttok;
4518 }
4519 else
4520 {
4521 /* Try to locale-expand the converted string. */
4522 ttrans = localeexpand (ttok, 0, ttoklen - 1, first_line, &ttranslen);
4523 free (ttok);
4524
4525 /* Add the double quotes back */
4526 ttok = sh_mkdoublequoted (ttrans, ttranslen, 0);
4527 free (ttrans);
4528 ttranslen += 2;
4529 ttrans = ttok;
4530 }
4531
4532 RESIZE_MALLOCED_BUFFER (token, token_index, ttranslen + 2,
4533 token_buffer_size,
4534 TOKEN_DEFAULT_GROW_SIZE);
4535 strcpy (token + token_index, ttrans);
4536 token_index += ttranslen;
4537 FREE (ttrans);
4538 quoted = 1;
4539 all_digit_token = 0;
4540 goto next_character;
4541 }
4542 /* This could eventually be extended to recognize all of the
4543 shell's single-character parameter expansions, and set flags.*/
4544 else if MBTEST(character == '$' && peek_char == '$')
4545 {
4546 ttok = (char *)xmalloc (3);
4547 ttok[0] = ttok[1] = '$';
4548 ttok[2] = '\0';
4549 RESIZE_MALLOCED_BUFFER (token, token_index, 3,
4550 token_buffer_size,
4551 TOKEN_DEFAULT_GROW_SIZE);
4552 strcpy (token + token_index, ttok);
4553 token_index += 2;
4554 dollar_present = 1;
4555 all_digit_token = 0;
4556 FREE (ttok);
4557 goto next_character;
4558 }
4559 else
4560 shell_ungetc (peek_char);
4561 }
4562
4563 #if defined (ARRAY_VARS)
4564 /* Identify possible array subscript assignment; match [...]. If
4565 parser_state&PST_COMPASSIGN, we need to parse [sub]=words treating
4566 `sub' as if it were enclosed in double quotes. */
4567 else if MBTEST(character == '[' && /* ] */
4568 ((token_index > 0 && assignment_acceptable (last_read_token) && token_is_ident (token, token_index)) ||
4569 (token_index == 0 && (parser_state&PST_COMPASSIGN))))
4570 {
4571 ttok = parse_matched_pair (cd, '[', ']', &ttoklen, P_ARRAYSUB);
4572 if (ttok == &matched_pair_error)
4573 return -1; /* Bail immediately. */
4574 RESIZE_MALLOCED_BUFFER (token, token_index, ttoklen + 2,
4575 token_buffer_size,
4576 TOKEN_DEFAULT_GROW_SIZE);
4577 token[token_index++] = character;
4578 strcpy (token + token_index, ttok);
4579 token_index += ttoklen;
4580 FREE (ttok);
4581 all_digit_token = 0;
4582 goto next_character;
4583 }
4584 /* Identify possible compound array variable assignment. */
4585 else if MBTEST(character == '=' && token_index > 0 && (assignment_acceptable (last_read_token) || (parser_state & PST_ASSIGNOK)) && token_is_assignment (token, token_index))
4586 {
4587 peek_char = shell_getc (1);
4588 if MBTEST(peek_char == '(') /* ) */
4589 {
4590 ttok = parse_compound_assignment (&ttoklen);
4591
4592 RESIZE_MALLOCED_BUFFER (token, token_index, ttoklen + 4,
4593 token_buffer_size,
4594 TOKEN_DEFAULT_GROW_SIZE);
4595
4596 token[token_index++] = '=';
4597 token[token_index++] = '(';
4598 if (ttok)
4599 {
4600 strcpy (token + token_index, ttok);
4601 token_index += ttoklen;
4602 }
4603 token[token_index++] = ')';
4604 FREE (ttok);
4605 all_digit_token = 0;
4606 compound_assignment = 1;
4607 #if 1
4608 goto next_character;
4609 #else
4610 goto got_token; /* ksh93 seems to do this */
4611 #endif
4612 }
4613 else
4614 shell_ungetc (peek_char);
4615 }
4616 #endif
4617
4618 /* When not parsing a multi-character word construct, shell meta-
4619 characters break words. */
4620 if MBTEST(shellbreak (character))
4621 {
4622 shell_ungetc (character);
4623 goto got_token;
4624 }
4625
4626 got_character:
4627
4628 if (character == CTLESC || character == CTLNUL)
4629 token[token_index++] = CTLESC;
4630
4631 got_escaped_character:
4632
4633 all_digit_token &= DIGIT (character);
4634 dollar_present |= character == '$';
4635
4636 token[token_index++] = character;
4637
4638 RESIZE_MALLOCED_BUFFER (token, token_index, 1, token_buffer_size,
4639 TOKEN_DEFAULT_GROW_SIZE);
4640
4641 next_character:
4642 if (character == '\n' && SHOULD_PROMPT ())
4643 prompt_again ();
4644
4645 /* We want to remove quoted newlines (that is, a \<newline> pair)
4646 unless we are within single quotes or pass_next_character is
4647 set (the shell equivalent of literal-next). */
4648 cd = current_delimiter (dstack);
4649 character = shell_getc (cd != '\'' && pass_next_character == 0);
4650 } /* end for (;;) */
4651
4652 got_token:
4653
4654 token[token_index] = '\0';
4655
4656 /* Check to see what thing we should return. If the last_read_token
4657 is a `<', or a `&', or the character which ended this token is
4658 a '>' or '<', then, and ONLY then, is this input token a NUMBER.
4659 Otherwise, it is just a word, and should be returned as such. */
4660 if MBTEST(all_digit_token && (character == '<' || character == '>' || \
4661 last_read_token == LESS_AND || \
4662 last_read_token == GREATER_AND))
4663 {
4664 if (legal_number (token, &lvalue) && (int)lvalue == lvalue)
4665 yylval.number = lvalue;
4666 else
4667 yylval.number = -1;
4668 return (NUMBER);
4669 }
4670
4671 /* Check for special case tokens. */
4672 result = (last_shell_getc_is_singlebyte) ? special_case_tokens (token) : -1;
4673 if (result >= 0)
4674 return result;
4675
4676 #if defined (ALIAS)
4677 /* Posix.2 does not allow reserved words to be aliased, so check for all
4678 of them, including special cases, before expanding the current token
4679 as an alias. */
4680 if MBTEST(posixly_correct)
4681 CHECK_FOR_RESERVED_WORD (token);
4682
4683 /* Aliases are expanded iff EXPAND_ALIASES is non-zero, and quoting
4684 inhibits alias expansion. */
4685 if (expand_aliases && quoted == 0)
4686 {
4687 result = alias_expand_token (token);
4688 if (result == RE_READ_TOKEN)
4689 return (RE_READ_TOKEN);
4690 else if (result == NO_EXPANSION)
4691 parser_state &= ~PST_ALEXPNEXT;
4692 }
4693
4694 /* If not in Posix.2 mode, check for reserved words after alias
4695 expansion. */
4696 if MBTEST(posixly_correct == 0)
4697 #endif
4698 CHECK_FOR_RESERVED_WORD (token);
4699
4700 the_word = (WORD_DESC *)xmalloc (sizeof (WORD_DESC));
4701 the_word->word = (char *)xmalloc (1 + token_index);
4702 the_word->flags = 0;
4703 strcpy (the_word->word, token);
4704 if (dollar_present)
4705 the_word->flags |= W_HASDOLLAR;
4706 if (quoted)
4707 the_word->flags |= W_QUOTED; /*(*/
4708 if (compound_assignment && token[token_index-1] == ')')
4709 the_word->flags |= W_COMPASSIGN;
4710 /* A word is an assignment if it appears at the beginning of a
4711 simple command, or after another assignment word. This is
4712 context-dependent, so it cannot be handled in the grammar. */
4713 if (assignment (token, (parser_state & PST_COMPASSIGN) != 0))
4714 {
4715 the_word->flags |= W_ASSIGNMENT;
4716 /* Don't perform word splitting on assignment statements. */
4717 if (assignment_acceptable (last_read_token) || (parser_state & PST_COMPASSIGN) != 0)
4718 the_word->flags |= W_NOSPLIT;
4719 }
4720
4721 if (command_token_position (last_read_token))
4722 {
4723 struct builtin *b;
4724 b = builtin_address_internal (token, 0);
4725 if (b && (b->flags & ASSIGNMENT_BUILTIN))
4726 parser_state |= PST_ASSIGNOK;
4727 else if (STREQ (token, "eval") || STREQ (token, "let"))
4728 parser_state |= PST_ASSIGNOK;
4729 }
4730
4731 yylval.word = the_word;
4732
4733 if (token[0] == '{' && token[token_index-1] == '}' &&
4734 (character == '<' || character == '>'))
4735 {
4736 /* can use token; already copied to the_word */
4737 token[token_index-1] = '\0';
4738 if (legal_identifier (token+1))
4739 {
4740 strcpy (the_word->word, token+1);
4741 /*itrace("read_token_word: returning REDIR_WORD for %s", the_word->word);*/
4742 return (REDIR_WORD);
4743 }
4744 }
4745
4746 result = ((the_word->flags & (W_ASSIGNMENT|W_NOSPLIT)) == (W_ASSIGNMENT|W_NOSPLIT))
4747 ? ASSIGNMENT_WORD : WORD;
4748
4749 switch (last_read_token)
4750 {
4751 case FUNCTION:
4752 parser_state |= PST_ALLOWOPNBRC;
4753 function_dstart = line_number;
4754 break;
4755 case CASE:
4756 case SELECT:
4757 case FOR:
4758 if (word_top < MAX_CASE_NEST)
4759 word_top++;
4760 word_lineno[word_top] = line_number;
4761 break;
4762 }
4763
4764 return (result);
4765 }
4766
4767 /* Return 1 if TOKSYM is a token that after being read would allow
4768 a reserved word to be seen, else 0. */
4769 static int
4770 reserved_word_acceptable (toksym)
4771 int toksym;
4772 {
4773 switch (toksym)
4774 {
4775 case '\n':
4776 case ';':
4777 case '(':
4778 case ')':
4779 case '|':
4780 case '&':
4781 case '{':
4782 case '}': /* XXX */
4783 case AND_AND:
4784 case BANG:
4785 case BAR_AND:
4786 case DO:
4787 case DONE:
4788 case ELIF:
4789 case ELSE:
4790 case ESAC:
4791 case FI:
4792 case IF:
4793 case OR_OR:
4794 case SEMI_SEMI:
4795 case SEMI_AND:
4796 case SEMI_SEMI_AND:
4797 case THEN:
4798 case TIME:
4799 case TIMEOPT:
4800 case TIMEIGN:
4801 case COPROC:
4802 case UNTIL:
4803 case WHILE:
4804 case 0:
4805 return 1;
4806 default:
4807 #if defined (COPROCESS_SUPPORT)
4808 if (last_read_token == WORD && token_before_that == COPROC)
4809 return 1;
4810 #endif
4811 if (last_read_token == WORD && token_before_that == FUNCTION)
4812 return 1;
4813 return 0;
4814 }
4815 }
4816
4817 /* Return the index of TOKEN in the alist of reserved words, or -1 if
4818 TOKEN is not a shell reserved word. */
4819 int
4820 find_reserved_word (tokstr)
4821 char *tokstr;
4822 {
4823 int i;
4824 for (i = 0; word_token_alist[i].word; i++)
4825 if (STREQ (tokstr, word_token_alist[i].word))
4826 return i;
4827 return -1;
4828 }
4829
4830 #if 0
4831 #if defined (READLINE)
4832 /* Called after each time readline is called. This insures that whatever
4833 the new prompt string is gets propagated to readline's local prompt
4834 variable. */
4835 static void
4836 reset_readline_prompt ()
4837 {
4838 char *temp_prompt;
4839
4840 if (prompt_string_pointer)
4841 {
4842 temp_prompt = (*prompt_string_pointer)
4843 ? decode_prompt_string (*prompt_string_pointer)
4844 : (char *)NULL;
4845
4846 if (temp_prompt == 0)
4847 {
4848 temp_prompt = (char *)xmalloc (1);
4849 temp_prompt[0] = '\0';
4850 }
4851
4852 FREE (current_readline_prompt);
4853 current_readline_prompt = temp_prompt;
4854 }
4855 }
4856 #endif /* READLINE */
4857 #endif /* 0 */
4858
4859 #if defined (HISTORY)
4860 /* A list of tokens which can be followed by newlines, but not by
4861 semi-colons. When concatenating multiple lines of history, the
4862 newline separator for such tokens is replaced with a space. */
4863 static const int no_semi_successors[] = {
4864 '\n', '{', '(', ')', ';', '&', '|',
4865 CASE, DO, ELSE, IF, SEMI_SEMI, SEMI_AND, SEMI_SEMI_AND, THEN, UNTIL,
4866 WHILE, AND_AND, OR_OR, IN,
4867 0
4868 };
4869
4870 /* If we are not within a delimited expression, try to be smart
4871 about which separators can be semi-colons and which must be
4872 newlines. Returns the string that should be added into the
4873 history entry. LINE is the line we're about to add; it helps
4874 make some more intelligent decisions in certain cases. */
4875 char *
4876 history_delimiting_chars (line)
4877 const char *line;
4878 {
4879 static int last_was_heredoc = 0; /* was the last entry the start of a here document? */
4880 register int i;
4881
4882 if ((parser_state & PST_HEREDOC) == 0)
4883 last_was_heredoc = 0;
4884
4885 if (dstack.delimiter_depth != 0)
4886 return ("\n");
4887
4888 /* We look for current_command_line_count == 2 because we are looking to
4889 add the first line of the body of the here document (the second line
4890 of the command). We also keep LAST_WAS_HEREDOC as a private sentinel
4891 variable to note when we think we added the first line of a here doc
4892 (the one with a "<<" somewhere in it) */
4893 if (parser_state & PST_HEREDOC)
4894 {
4895 if (last_was_heredoc)
4896 {
4897 last_was_heredoc = 0;
4898 return "\n";
4899 }
4900 return (current_command_line_count == 2 ? "\n" : "");
4901 }
4902
4903 if (parser_state & PST_COMPASSIGN)
4904 return (" ");
4905
4906 /* First, handle some special cases. */
4907 /*(*/
4908 /* If we just read `()', assume it's a function definition, and don't
4909 add a semicolon. If the token before the `)' was not `(', and we're
4910 not in the midst of parsing a case statement, assume it's a
4911 parenthesized command and add the semicolon. */
4912 /*)(*/
4913 if (token_before_that == ')')
4914 {
4915 if (two_tokens_ago == '(') /*)*/ /* function def */
4916 return " ";
4917 /* This does not work for subshells inside case statement
4918 command lists. It's a suboptimal solution. */
4919 else if (parser_state & PST_CASESTMT) /* case statement pattern */
4920 return " ";
4921 else
4922 return "; "; /* (...) subshell */
4923 }
4924 else if (token_before_that == WORD && two_tokens_ago == FUNCTION)
4925 return " "; /* function def using `function name' without `()' */
4926
4927 /* If we're not in a here document, but we think we're about to parse one,
4928 and we would otherwise return a `;', return a newline to delimit the
4929 line with the here-doc delimiter */
4930 else if ((parser_state & PST_HEREDOC) == 0 && current_command_line_count > 1 && last_read_token == '\n' && strstr (line, "<<"))
4931 {
4932 last_was_heredoc = 1;
4933 return "\n";
4934 }
4935
4936 else if (token_before_that == WORD && two_tokens_ago == FOR)
4937 {
4938 /* Tricky. `for i\nin ...' should not have a semicolon, but
4939 `for i\ndo ...' should. We do what we can. */
4940 for (i = shell_input_line_index; whitespace (shell_input_line[i]); i++)
4941 ;
4942 if (shell_input_line[i] && shell_input_line[i] == 'i' && shell_input_line[i+1] == 'n')
4943 return " ";
4944 return ";";
4945 }
4946 else if (two_tokens_ago == CASE && token_before_that == WORD && (parser_state & PST_CASESTMT))
4947 return " ";
4948
4949 for (i = 0; no_semi_successors[i]; i++)
4950 {
4951 if (token_before_that == no_semi_successors[i])
4952 return (" ");
4953 }
4954
4955 return ("; ");
4956 }
4957 #endif /* HISTORY */
4958
4959 /* Issue a prompt, or prepare to issue a prompt when the next character
4960 is read. */
4961 static void
4962 prompt_again ()
4963 {
4964 char *temp_prompt;
4965
4966 if (interactive == 0 || expanding_alias ()) /* XXX */
4967 return;
4968
4969 ps1_prompt = get_string_value ("PS1");
4970 ps2_prompt = get_string_value ("PS2");
4971
4972 if (!prompt_string_pointer)
4973 prompt_string_pointer = &ps1_prompt;
4974
4975 temp_prompt = *prompt_string_pointer
4976 ? decode_prompt_string (*prompt_string_pointer)
4977 : (char *)NULL;
4978
4979 if (temp_prompt == 0)
4980 {
4981 temp_prompt = (char *)xmalloc (1);
4982 temp_prompt[0] = '\0';
4983 }
4984
4985 current_prompt_string = *prompt_string_pointer;
4986 prompt_string_pointer = &ps2_prompt;
4987
4988 #if defined (READLINE)
4989 if (!no_line_editing)
4990 {
4991 FREE (current_readline_prompt);
4992 current_readline_prompt = temp_prompt;
4993 }
4994 else
4995 #endif /* READLINE */
4996 {
4997 FREE (current_decoded_prompt);
4998 current_decoded_prompt = temp_prompt;
4999 }
5000 }
5001
5002 int
5003 get_current_prompt_level ()
5004 {
5005 return ((current_prompt_string && current_prompt_string == ps2_prompt) ? 2 : 1);
5006 }
5007
5008 void
5009 set_current_prompt_level (x)
5010 int x;
5011 {
5012 prompt_string_pointer = (x == 2) ? &ps2_prompt : &ps1_prompt;
5013 current_prompt_string = *prompt_string_pointer;
5014 }
5015
5016 static void
5017 print_prompt ()
5018 {
5019 fprintf (stderr, "%s", current_decoded_prompt);
5020 fflush (stderr);
5021 }
5022
5023 /* Return a string which will be printed as a prompt. The string
5024 may contain special characters which are decoded as follows:
5025
5026 \a bell (ascii 07)
5027 \d the date in Day Mon Date format
5028 \e escape (ascii 033)
5029 \h the hostname up to the first `.'
5030 \H the hostname
5031 \j the number of active jobs
5032 \l the basename of the shell's tty device name
5033 \n CRLF
5034 \r CR
5035 \s the name of the shell
5036 \t the time in 24-hour hh:mm:ss format
5037 \T the time in 12-hour hh:mm:ss format
5038 \@ the time in 12-hour hh:mm am/pm format
5039 \A the time in 24-hour hh:mm format
5040 \D{fmt} the result of passing FMT to strftime(3)
5041 \u your username
5042 \v the version of bash (e.g., 2.00)
5043 \V the release of bash, version + patchlevel (e.g., 2.00.0)
5044 \w the current working directory
5045 \W the last element of $PWD
5046 \! the history number of this command
5047 \# the command number of this command
5048 \$ a $ or a # if you are root
5049 \nnn character code nnn in octal
5050 \\ a backslash
5051 \[ begin a sequence of non-printing chars
5052 \] end a sequence of non-printing chars
5053 */
5054 #define PROMPT_GROWTH 48
5055 char *
5056 decode_prompt_string (string)
5057 char *string;
5058 {
5059 WORD_LIST *list;
5060 char *result, *t;
5061 struct dstack save_dstack;
5062 int last_exit_value, last_comsub_pid;
5063 #if defined (PROMPT_STRING_DECODE)
5064 int result_size, result_index;
5065 int c, n, i;
5066 char *temp, octal_string[4];
5067 struct tm *tm;
5068 time_t the_time;
5069 char timebuf[128];
5070 char *timefmt;
5071
5072 result = (char *)xmalloc (result_size = PROMPT_GROWTH);
5073 result[result_index = 0] = 0;
5074 temp = (char *)NULL;
5075
5076 while (c = *string++)
5077 {
5078 if (posixly_correct && c == '!')
5079 {
5080 if (*string == '!')
5081 {
5082 temp = savestring ("!");
5083 goto add_string;
5084 }
5085 else
5086 {
5087 #if !defined (HISTORY)
5088 temp = savestring ("1");
5089 #else /* HISTORY */
5090 temp = itos (history_number ());
5091 #endif /* HISTORY */
5092 string--; /* add_string increments string again. */
5093 goto add_string;
5094 }
5095 }
5096 if (c == '\\')
5097 {
5098 c = *string;
5099
5100 switch (c)
5101 {
5102 case '0':
5103 case '1':
5104 case '2':
5105 case '3':
5106 case '4':
5107 case '5':
5108 case '6':
5109 case '7':
5110 strncpy (octal_string, string, 3);
5111 octal_string[3] = '\0';
5112
5113 n = read_octal (octal_string);
5114 temp = (char *)xmalloc (3);
5115
5116 if (n == CTLESC || n == CTLNUL)
5117 {
5118 temp[0] = CTLESC;
5119 temp[1] = n;
5120 temp[2] = '\0';
5121 }
5122 else if (n == -1)
5123 {
5124 temp[0] = '\\';
5125 temp[1] = '\0';
5126 }
5127 else
5128 {
5129 temp[0] = n;
5130 temp[1] = '\0';
5131 }
5132
5133 for (c = 0; n != -1 && c < 3 && ISOCTAL (*string); c++)
5134 string++;
5135
5136 c = 0; /* tested at add_string: */
5137 goto add_string;
5138
5139 case 'd':
5140 case 't':
5141 case 'T':
5142 case '@':
5143 case 'A':
5144 /* Make the current time/date into a string. */
5145 (void) time (&the_time);
5146 #if defined (HAVE_TZSET)
5147 sv_tz ("TZ"); /* XXX -- just make sure */
5148 #endif
5149 tm = localtime (&the_time);
5150
5151 if (c == 'd')
5152 n = strftime (timebuf, sizeof (timebuf), "%a %b %d", tm);
5153 else if (c == 't')
5154 n = strftime (timebuf, sizeof (timebuf), "%H:%M:%S", tm);
5155 else if (c == 'T')
5156 n = strftime (timebuf, sizeof (timebuf), "%I:%M:%S", tm);
5157 else if (c == '@')
5158 n = strftime (timebuf, sizeof (timebuf), "%I:%M %p", tm);
5159 else if (c == 'A')
5160 n = strftime (timebuf, sizeof (timebuf), "%H:%M", tm);
5161
5162 if (n == 0)
5163 timebuf[0] = '\0';
5164 else
5165 timebuf[sizeof(timebuf) - 1] = '\0';
5166
5167 temp = savestring (timebuf);
5168 goto add_string;
5169
5170 case 'D': /* strftime format */
5171 if (string[1] != '{') /* } */
5172 goto not_escape;
5173
5174 (void) time (&the_time);
5175 tm = localtime (&the_time);
5176 string += 2; /* skip { */
5177 timefmt = xmalloc (strlen (string) + 3);
5178 for (t = timefmt; *string && *string != '}'; )
5179 *t++ = *string++;
5180 *t = '\0';
5181 c = *string; /* tested at add_string */
5182 if (timefmt[0] == '\0')
5183 {
5184 timefmt[0] = '%';
5185 timefmt[1] = 'X'; /* locale-specific current time */
5186 timefmt[2] = '\0';
5187 }
5188 n = strftime (timebuf, sizeof (timebuf), timefmt, tm);
5189 free (timefmt);
5190
5191 if (n == 0)
5192 timebuf[0] = '\0';
5193 else
5194 timebuf[sizeof(timebuf) - 1] = '\0';
5195
5196 if (promptvars || posixly_correct)
5197 /* Make sure that expand_prompt_string is called with a
5198 second argument of Q_DOUBLE_QUOTES if we use this
5199 function here. */
5200 temp = sh_backslash_quote_for_double_quotes (timebuf);
5201 else
5202 temp = savestring (timebuf);
5203 goto add_string;
5204
5205 case 'n':
5206 temp = (char *)xmalloc (3);
5207 temp[0] = no_line_editing ? '\n' : '\r';
5208 temp[1] = no_line_editing ? '\0' : '\n';
5209 temp[2] = '\0';
5210 goto add_string;
5211
5212 case 's':
5213 temp = base_pathname (shell_name);
5214 temp = savestring (temp);
5215 goto add_string;
5216
5217 case 'v':
5218 case 'V':
5219 temp = (char *)xmalloc (16);
5220 if (c == 'v')
5221 strcpy (temp, dist_version);
5222 else
5223 sprintf (temp, "%s.%d", dist_version, patch_level);
5224 goto add_string;
5225
5226 case 'w':
5227 case 'W':
5228 {
5229 /* Use the value of PWD because it is much more efficient. */
5230 char t_string[PATH_MAX];
5231 int tlen;
5232
5233 temp = get_string_value ("PWD");
5234
5235 if (temp == 0)
5236 {
5237 if (getcwd (t_string, sizeof(t_string)) == 0)
5238 {
5239 t_string[0] = '.';
5240 tlen = 1;
5241 }
5242 else
5243 tlen = strlen (t_string);
5244 }
5245 else
5246 {
5247 tlen = sizeof (t_string) - 1;
5248 strncpy (t_string, temp, tlen);
5249 }
5250 t_string[tlen] = '\0';
5251
5252 #if defined (MACOSX)
5253 /* Convert from "fs" format to "input" format */
5254 temp = fnx_fromfs (t_string, strlen (t_string));
5255 if (temp != t_string)
5256 strcpy (t_string, temp);
5257 #endif
5258
5259 #define ROOT_PATH(x) ((x)[0] == '/' && (x)[1] == 0)
5260 #define DOUBLE_SLASH_ROOT(x) ((x)[0] == '/' && (x)[1] == '/' && (x)[2] == 0)
5261 /* Abbreviate \W as ~ if $PWD == $HOME */
5262 if (c == 'W' && (((t = get_string_value ("HOME")) == 0) || STREQ (t, t_string) == 0))
5263 {
5264 if (ROOT_PATH (t_string) == 0 && DOUBLE_SLASH_ROOT (t_string) == 0)
5265 {
5266 t = strrchr (t_string, '/');
5267 if (t)
5268 memmove (t_string, t + 1, strlen (t)); /* strlen(t) to copy NULL */
5269 }
5270 }
5271 #undef ROOT_PATH
5272 #undef DOUBLE_SLASH_ROOT
5273 else
5274 /* polite_directory_format is guaranteed to return a string
5275 no longer than PATH_MAX - 1 characters. */
5276 strcpy (t_string, polite_directory_format (t_string));
5277
5278 temp = trim_pathname (t_string, PATH_MAX - 1);
5279 /* If we're going to be expanding the prompt string later,
5280 quote the directory name. */
5281 if (promptvars || posixly_correct)
5282 /* Make sure that expand_prompt_string is called with a
5283 second argument of Q_DOUBLE_QUOTES if we use this
5284 function here. */
5285 temp = sh_backslash_quote_for_double_quotes (t_string);
5286 else
5287 temp = savestring (t_string);
5288
5289 goto add_string;
5290 }
5291
5292 case 'u':
5293 if (current_user.user_name == 0)
5294 get_current_user_info ();
5295 temp = savestring (current_user.user_name);
5296 goto add_string;
5297
5298 case 'h':
5299 case 'H':
5300 temp = savestring (current_host_name);
5301 if (c == 'h' && (t = (char *)strchr (temp, '.')))
5302 *t = '\0';
5303 goto add_string;
5304
5305 case '#':
5306 temp = itos (current_command_number);
5307 goto add_string;
5308
5309 case '!':
5310 #if !defined (HISTORY)
5311 temp = savestring ("1");
5312 #else /* HISTORY */
5313 temp = itos (history_number ());
5314 #endif /* HISTORY */
5315 goto add_string;
5316
5317 case '$':
5318 t = temp = (char *)xmalloc (3);
5319 if ((promptvars || posixly_correct) && (current_user.euid != 0))
5320 *t++ = '\\';
5321 *t++ = current_user.euid == 0 ? '#' : '$';
5322 *t = '\0';
5323 goto add_string;
5324
5325 case 'j':
5326 temp = itos (count_all_jobs ());
5327 goto add_string;
5328
5329 case 'l':
5330 #if defined (HAVE_TTYNAME)
5331 temp = (char *)ttyname (fileno (stdin));
5332 t = temp ? base_pathname (temp) : "tty";
5333 temp = savestring (t);
5334 #else
5335 temp = savestring ("tty");
5336 #endif /* !HAVE_TTYNAME */
5337 goto add_string;
5338
5339 #if defined (READLINE)
5340 case '[':
5341 case ']':
5342 if (no_line_editing)
5343 {
5344 string++;
5345 break;
5346 }
5347 temp = (char *)xmalloc (3);
5348 n = (c == '[') ? RL_PROMPT_START_IGNORE : RL_PROMPT_END_IGNORE;
5349 i = 0;
5350 if (n == CTLESC || n == CTLNUL)
5351 temp[i++] = CTLESC;
5352 temp[i++] = n;
5353 temp[i] = '\0';
5354 goto add_string;
5355 #endif /* READLINE */
5356
5357 case '\\':
5358 case 'a':
5359 case 'e':
5360 case 'r':
5361 temp = (char *)xmalloc (2);
5362 if (c == 'a')
5363 temp[0] = '\07';
5364 else if (c == 'e')
5365 temp[0] = '\033';
5366 else if (c == 'r')
5367 temp[0] = '\r';
5368 else /* (c == '\\') */
5369 temp[0] = c;
5370 temp[1] = '\0';
5371 goto add_string;
5372
5373 default:
5374 not_escape:
5375 temp = (char *)xmalloc (3);
5376 temp[0] = '\\';
5377 temp[1] = c;
5378 temp[2] = '\0';
5379
5380 add_string:
5381 if (c)
5382 string++;
5383 result =
5384 sub_append_string (temp, result, &result_index, &result_size);
5385 temp = (char *)NULL; /* Freed in sub_append_string (). */
5386 result[result_index] = '\0';
5387 break;
5388 }
5389 }
5390 else
5391 {
5392 RESIZE_MALLOCED_BUFFER (result, result_index, 3, result_size, PROMPT_GROWTH);
5393 result[result_index++] = c;
5394 result[result_index] = '\0';
5395 }
5396 }
5397 #else /* !PROMPT_STRING_DECODE */
5398 result = savestring (string);
5399 #endif /* !PROMPT_STRING_DECODE */
5400
5401 /* Save the delimiter stack and point `dstack' to temp space so any
5402 command substitutions in the prompt string won't result in screwing
5403 up the parser's quoting state. */
5404 save_dstack = dstack;
5405 dstack = temp_dstack;
5406 dstack.delimiter_depth = 0;
5407
5408 /* Perform variable and parameter expansion and command substitution on
5409 the prompt string. */
5410 if (promptvars || posixly_correct)
5411 {
5412 last_exit_value = last_command_exit_value;
5413 last_comsub_pid = last_command_subst_pid;
5414 list = expand_prompt_string (result, Q_DOUBLE_QUOTES, 0);
5415 free (result);
5416 result = string_list (list);
5417 dispose_words (list);
5418 last_command_exit_value = last_exit_value;
5419 last_command_subst_pid = last_comsub_pid;
5420 }
5421 else
5422 {
5423 t = dequote_string (result);
5424 free (result);
5425 result = t;
5426 }
5427
5428 dstack = save_dstack;
5429
5430 return (result);
5431 }
5432
5433 /************************************************
5434 * *
5435 * ERROR HANDLING *
5436 * *
5437 ************************************************/
5438
5439 /* Report a syntax error, and restart the parser. Call here for fatal
5440 errors. */
5441 int
5442 yyerror (msg)
5443 const char *msg;
5444 {
5445 report_syntax_error ((char *)NULL);
5446 reset_parser ();
5447 return (0);
5448 }
5449
5450 static char *
5451 error_token_from_token (tok)
5452 int tok;
5453 {
5454 char *t;
5455
5456 if (t = find_token_in_alist (tok, word_token_alist, 0))
5457 return t;
5458
5459 if (t = find_token_in_alist (tok, other_token_alist, 0))
5460 return t;
5461
5462 t = (char *)NULL;
5463 /* This stuff is dicy and needs closer inspection */
5464 switch (current_token)
5465 {
5466 case WORD:
5467 case ASSIGNMENT_WORD:
5468 if (yylval.word)
5469 t = savestring (yylval.word->word);
5470 break;
5471 case NUMBER:
5472 t = itos (yylval.number);
5473 break;
5474 case ARITH_CMD:
5475 if (yylval.word_list)
5476 t = string_list (yylval.word_list);
5477 break;
5478 case ARITH_FOR_EXPRS:
5479 if (yylval.word_list)
5480 t = string_list_internal (yylval.word_list, " ; ");
5481 break;
5482 case COND_CMD:
5483 t = (char *)NULL; /* punt */
5484 break;
5485 }
5486
5487 return t;
5488 }
5489
5490 static char *
5491 error_token_from_text ()
5492 {
5493 char *msg, *t;
5494 int token_end, i;
5495
5496 t = shell_input_line;
5497 i = shell_input_line_index;
5498 token_end = 0;
5499 msg = (char *)NULL;
5500
5501 if (i && t[i] == '\0')
5502 i--;
5503
5504 while (i && (whitespace (t[i]) || t[i] == '\n'))
5505 i--;
5506
5507 if (i)
5508 token_end = i + 1;
5509
5510 while (i && (member (t[i], " \n\t;|&") == 0))
5511 i--;
5512
5513 while (i != token_end && (whitespace (t[i]) || t[i] == '\n'))
5514 i++;
5515
5516 /* Return our idea of the offending token. */
5517 if (token_end || (i == 0 && token_end == 0))
5518 {
5519 if (token_end)
5520 msg = substring (t, i, token_end);
5521 else /* one-character token */
5522 {
5523 msg = (char *)xmalloc (2);
5524 msg[0] = t[i];
5525 msg[1] = '\0';
5526 }
5527 }
5528
5529 return (msg);
5530 }
5531
5532 static void
5533 print_offending_line ()
5534 {
5535 char *msg;
5536 int token_end;
5537
5538 msg = savestring (shell_input_line);
5539 token_end = strlen (msg);
5540 while (token_end && msg[token_end - 1] == '\n')
5541 msg[--token_end] = '\0';
5542
5543 parser_error (line_number, "`%s'", msg);
5544 free (msg);
5545 }
5546
5547 /* Report a syntax error with line numbers, etc.
5548 Call here for recoverable errors. If you have a message to print,
5549 then place it in MESSAGE, otherwise pass NULL and this will figure
5550 out an appropriate message for you. */
5551 static void
5552 report_syntax_error (message)
5553 char *message;
5554 {
5555 char *msg, *p;
5556
5557 if (message)
5558 {
5559 parser_error (line_number, "%s", message);
5560 if (interactive && EOF_Reached)
5561 EOF_Reached = 0;
5562 last_command_exit_value = parse_and_execute_level ? EX_BADSYNTAX : EX_BADUSAGE;
5563 return;
5564 }
5565
5566 /* If the line of input we're reading is not null, try to find the
5567 objectionable token. First, try to figure out what token the
5568 parser's complaining about by looking at current_token. */
5569 if (current_token != 0 && EOF_Reached == 0 && (msg = error_token_from_token (current_token)))
5570 {
5571 if (ansic_shouldquote (msg))
5572 {
5573 p = ansic_quote (msg, 0, NULL);
5574 free (msg);
5575 msg = p;
5576 }
5577 parser_error (line_number, _("syntax error near unexpected token `%s'"), msg);
5578 free (msg);
5579
5580 if (interactive == 0)
5581 print_offending_line ();
5582
5583 last_command_exit_value = parse_and_execute_level ? EX_BADSYNTAX : EX_BADUSAGE;
5584 return;
5585 }
5586
5587 /* If looking at the current token doesn't prove fruitful, try to find the
5588 offending token by analyzing the text of the input line near the current
5589 input line index and report what we find. */
5590 if (shell_input_line && *shell_input_line)
5591 {
5592 msg = error_token_from_text ();
5593 if (msg)
5594 {
5595 parser_error (line_number, _("syntax error near `%s'"), msg);
5596 free (msg);
5597 }
5598
5599 /* If not interactive, print the line containing the error. */
5600 if (interactive == 0)
5601 print_offending_line ();
5602 }
5603 else
5604 {
5605 msg = EOF_Reached ? _("syntax error: unexpected end of file") : _("syntax error");
5606 parser_error (line_number, "%s", msg);
5607 /* When the shell is interactive, this file uses EOF_Reached
5608 only for error reporting. Other mechanisms are used to
5609 decide whether or not to exit. */
5610 if (interactive && EOF_Reached)
5611 EOF_Reached = 0;
5612 }
5613
5614 last_command_exit_value = parse_and_execute_level ? EX_BADSYNTAX : EX_BADUSAGE;
5615 }
5616
5617 /* ??? Needed function. ??? We have to be able to discard the constructs
5618 created during parsing. In the case of error, we want to return
5619 allocated objects to the memory pool. In the case of no error, we want
5620 to throw away the information about where the allocated objects live.
5621 (dispose_command () will actually free the command.) */
5622 static void
5623 discard_parser_constructs (error_p)
5624 int error_p;
5625 {
5626 }
5627
5628 /************************************************
5629 * *
5630 * EOF HANDLING *
5631 * *
5632 ************************************************/
5633
5634 /* Do that silly `type "bye" to exit' stuff. You know, "ignoreeof". */
5635
5636 /* A flag denoting whether or not ignoreeof is set. */
5637 int ignoreeof = 0;
5638
5639 /* The number of times that we have encountered an EOF character without
5640 another character intervening. When this gets above the limit, the
5641 shell terminates. */
5642 int eof_encountered = 0;
5643
5644 /* The limit for eof_encountered. */
5645 int eof_encountered_limit = 10;
5646
5647 /* If we have EOF as the only input unit, this user wants to leave
5648 the shell. If the shell is not interactive, then just leave.
5649 Otherwise, if ignoreeof is set, and we haven't done this the
5650 required number of times in a row, print a message. */
5651 static void
5652 handle_eof_input_unit ()
5653 {
5654 if (interactive)
5655 {
5656 /* shell.c may use this to decide whether or not to write out the
5657 history, among other things. We use it only for error reporting
5658 in this file. */
5659 if (EOF_Reached)
5660 EOF_Reached = 0;
5661
5662 /* If the user wants to "ignore" eof, then let her do so, kind of. */
5663 if (ignoreeof)
5664 {
5665 if (eof_encountered < eof_encountered_limit)
5666 {
5667 fprintf (stderr, _("Use \"%s\" to leave the shell.\n"),
5668 login_shell ? "logout" : "exit");
5669 eof_encountered++;
5670 /* Reset the parsing state. */
5671 last_read_token = current_token = '\n';
5672 /* Reset the prompt string to be $PS1. */
5673 prompt_string_pointer = (char **)NULL;
5674 prompt_again ();
5675 return;
5676 }
5677 }
5678
5679 /* In this case EOF should exit the shell. Do it now. */
5680 reset_parser ();
5681 exit_builtin ((WORD_LIST *)NULL);
5682 }
5683 else
5684 {
5685 /* We don't write history files, etc., for non-interactive shells. */
5686 EOF_Reached = 1;
5687 }
5688 }
5689
5690 /************************************************
5691 * *
5692 * STRING PARSING FUNCTIONS *
5693 * *
5694 ************************************************/
5695
5696 /* It's very important that these two functions treat the characters
5697 between ( and ) identically. */
5698
5699 static WORD_LIST parse_string_error;
5700
5701 /* Take a string and run it through the shell parser, returning the
5702 resultant word list. Used by compound array assignment. */
5703 WORD_LIST *
5704 parse_string_to_word_list (s, flags, whom)
5705 char *s;
5706 int flags;
5707 const char *whom;
5708 {
5709 WORD_LIST *wl;
5710 int tok, orig_current_token, orig_line_number, orig_input_terminator;
5711 int orig_line_count;
5712 int old_echo_input, old_expand_aliases;
5713 #if defined (HISTORY)
5714 int old_remember_on_history, old_history_expansion_inhibited;
5715 #endif
5716
5717 #if defined (HISTORY)
5718 old_remember_on_history = remember_on_history;
5719 # if defined (BANG_HISTORY)
5720 old_history_expansion_inhibited = history_expansion_inhibited;
5721 # endif
5722 bash_history_disable ();
5723 #endif
5724
5725 orig_line_number = line_number;
5726 orig_line_count = current_command_line_count;
5727 orig_input_terminator = shell_input_line_terminator;
5728 old_echo_input = echo_input_at_read;
5729 old_expand_aliases = expand_aliases;
5730
5731 push_stream (1);
5732 last_read_token = WORD; /* WORD to allow reserved words here */
5733 current_command_line_count = 0;
5734 echo_input_at_read = expand_aliases = 0;
5735
5736 with_input_from_string (s, whom);
5737 wl = (WORD_LIST *)NULL;
5738
5739 if (flags & 1)
5740 parser_state |= PST_COMPASSIGN|PST_REPARSE;
5741
5742 while ((tok = read_token (READ)) != yacc_EOF)
5743 {
5744 if (tok == '\n' && *bash_input.location.string == '\0')
5745 break;
5746 if (tok == '\n') /* Allow newlines in compound assignments */
5747 continue;
5748 if (tok != WORD && tok != ASSIGNMENT_WORD)
5749 {
5750 line_number = orig_line_number + line_number - 1;
5751 orig_current_token = current_token;
5752 current_token = tok;
5753 yyerror (NULL); /* does the right thing */
5754 current_token = orig_current_token;
5755 if (wl)
5756 dispose_words (wl);
5757 wl = &parse_string_error;
5758 break;
5759 }
5760 wl = make_word_list (yylval.word, wl);
5761 }
5762
5763 last_read_token = '\n';
5764 pop_stream ();
5765
5766 #if defined (HISTORY)
5767 remember_on_history = old_remember_on_history;
5768 # if defined (BANG_HISTORY)
5769 history_expansion_inhibited = old_history_expansion_inhibited;
5770 # endif /* BANG_HISTORY */
5771 #endif /* HISTORY */
5772
5773 echo_input_at_read = old_echo_input;
5774 expand_aliases = old_expand_aliases;
5775
5776 current_command_line_count = orig_line_count;
5777 shell_input_line_terminator = orig_input_terminator;
5778
5779 if (flags & 1)
5780 parser_state &= ~(PST_COMPASSIGN|PST_REPARSE);
5781
5782 if (wl == &parse_string_error)
5783 {
5784 last_command_exit_value = EXECUTION_FAILURE;
5785 if (interactive_shell == 0 && posixly_correct)
5786 jump_to_top_level (FORCE_EOF);
5787 else
5788 jump_to_top_level (DISCARD);
5789 }
5790
5791 return (REVERSE_LIST (wl, WORD_LIST *));
5792 }
5793
5794 static char *
5795 parse_compound_assignment (retlenp)
5796 int *retlenp;
5797 {
5798 WORD_LIST *wl, *rl;
5799 int tok, orig_line_number, orig_token_size, orig_last_token, assignok;
5800 char *saved_token, *ret;
5801
5802 saved_token = token;
5803 orig_token_size = token_buffer_size;
5804 orig_line_number = line_number;
5805 orig_last_token = last_read_token;
5806
5807 last_read_token = WORD; /* WORD to allow reserved words here */
5808
5809 token = (char *)NULL;
5810 token_buffer_size = 0;
5811
5812 assignok = parser_state&PST_ASSIGNOK; /* XXX */
5813
5814 wl = (WORD_LIST *)NULL; /* ( */
5815 parser_state |= PST_COMPASSIGN;
5816
5817 while ((tok = read_token (READ)) != ')')
5818 {
5819 if (tok == '\n') /* Allow newlines in compound assignments */
5820 {
5821 if (SHOULD_PROMPT ())
5822 prompt_again ();
5823 continue;
5824 }
5825 if (tok != WORD && tok != ASSIGNMENT_WORD)
5826 {
5827 current_token = tok; /* for error reporting */
5828 if (tok == yacc_EOF) /* ( */
5829 parser_error (orig_line_number, _("unexpected EOF while looking for matching `)'"));
5830 else
5831 yyerror(NULL); /* does the right thing */
5832 if (wl)
5833 dispose_words (wl);
5834 wl = &parse_string_error;
5835 break;
5836 }
5837 wl = make_word_list (yylval.word, wl);
5838 }
5839
5840 FREE (token);
5841 token = saved_token;
5842 token_buffer_size = orig_token_size;
5843
5844 parser_state &= ~PST_COMPASSIGN;
5845
5846 if (wl == &parse_string_error)
5847 {
5848 last_command_exit_value = EXECUTION_FAILURE;
5849 last_read_token = '\n'; /* XXX */
5850 if (interactive_shell == 0 && posixly_correct)
5851 jump_to_top_level (FORCE_EOF);
5852 else
5853 jump_to_top_level (DISCARD);
5854 }
5855
5856 last_read_token = orig_last_token; /* XXX - was WORD? */
5857
5858 if (wl)
5859 {
5860 rl = REVERSE_LIST (wl, WORD_LIST *);
5861 ret = string_list (rl);
5862 dispose_words (rl);
5863 }
5864 else
5865 ret = (char *)NULL;
5866
5867 if (retlenp)
5868 *retlenp = (ret && *ret) ? strlen (ret) : 0;
5869
5870 if (assignok)
5871 parser_state |= PST_ASSIGNOK;
5872
5873 return ret;
5874 }
5875
5876 /************************************************
5877 * *
5878 * SAVING AND RESTORING PARTIAL PARSE STATE *
5879 * *
5880 ************************************************/
5881
5882 sh_parser_state_t *
5883 save_parser_state (ps)
5884 sh_parser_state_t *ps;
5885 {
5886 if (ps == 0)
5887 ps = (sh_parser_state_t *)xmalloc (sizeof (sh_parser_state_t));
5888 if (ps == 0)
5889 return ((sh_parser_state_t *)NULL);
5890
5891 ps->parser_state = parser_state;
5892 ps->token_state = save_token_state ();
5893
5894 ps->input_line_terminator = shell_input_line_terminator;
5895 ps->eof_encountered = eof_encountered;
5896
5897 ps->prompt_string_pointer = prompt_string_pointer;
5898
5899 ps->current_command_line_count = current_command_line_count;
5900
5901 #if defined (HISTORY)
5902 ps->remember_on_history = remember_on_history;
5903 # if defined (BANG_HISTORY)
5904 ps->history_expansion_inhibited = history_expansion_inhibited;
5905 # endif
5906 #endif
5907
5908 ps->last_command_exit_value = last_command_exit_value;
5909 #if defined (ARRAY_VARS)
5910 ps->pipestatus = save_pipestatus_array ();
5911 #endif
5912
5913 ps->last_shell_builtin = last_shell_builtin;
5914 ps->this_shell_builtin = this_shell_builtin;
5915
5916 ps->expand_aliases = expand_aliases;
5917 ps->echo_input_at_read = echo_input_at_read;
5918
5919 ps->token = token;
5920 ps->token_buffer_size = token_buffer_size;
5921 /* Force reallocation on next call to read_token_word */
5922 token = 0;
5923 token_buffer_size = 0;
5924
5925 return (ps);
5926 }
5927
5928 void
5929 restore_parser_state (ps)
5930 sh_parser_state_t *ps;
5931 {
5932 if (ps == 0)
5933 return;
5934
5935 parser_state = ps->parser_state;
5936 if (ps->token_state)
5937 {
5938 restore_token_state (ps->token_state);
5939 free (ps->token_state);
5940 }
5941
5942 shell_input_line_terminator = ps->input_line_terminator;
5943 eof_encountered = ps->eof_encountered;
5944
5945 prompt_string_pointer = ps->prompt_string_pointer;
5946
5947 current_command_line_count = ps->current_command_line_count;
5948
5949 #if defined (HISTORY)
5950 remember_on_history = ps->remember_on_history;
5951 # if defined (BANG_HISTORY)
5952 history_expansion_inhibited = ps->history_expansion_inhibited;
5953 # endif
5954 #endif
5955
5956 last_command_exit_value = ps->last_command_exit_value;
5957 #if defined (ARRAY_VARS)
5958 restore_pipestatus_array (ps->pipestatus);
5959 #endif
5960
5961 last_shell_builtin = ps->last_shell_builtin;
5962 this_shell_builtin = ps->this_shell_builtin;
5963
5964 expand_aliases = ps->expand_aliases;
5965 echo_input_at_read = ps->echo_input_at_read;
5966
5967 FREE (token);
5968 token = ps->token;
5969 token_buffer_size = ps->token_buffer_size;
5970 }
5971
5972 sh_input_line_state_t *
5973 save_input_line_state (ls)
5974 sh_input_line_state_t *ls;
5975 {
5976 if (ls == 0)
5977 ls = (sh_input_line_state_t *)xmalloc (sizeof (sh_input_line_state_t));
5978 if (ls == 0)
5979 return ((sh_input_line_state_t *)NULL);
5980
5981 ls->input_line = shell_input_line;
5982 ls->input_line_size = shell_input_line_size;
5983 ls->input_line_len = shell_input_line_len;
5984 ls->input_line_index = shell_input_line_index;
5985
5986 /* force reallocation */
5987 shell_input_line = 0;
5988 shell_input_line_size = shell_input_line_len = shell_input_line_index = 0;
5989 }
5990
5991 void
5992 restore_input_line_state (ls)
5993 sh_input_line_state_t *ls;
5994 {
5995 FREE (shell_input_line);
5996 shell_input_line = ls->input_line;
5997 shell_input_line_size = ls->input_line_size;
5998 shell_input_line_len = ls->input_line_len;
5999 shell_input_line_index = ls->input_line_index;
6000
6001 set_line_mbstate ();
6002 }
6003
6004 /************************************************
6005 * *
6006 * MULTIBYTE CHARACTER HANDLING *
6007 * *
6008 ************************************************/
6009
6010 #if defined (HANDLE_MULTIBYTE)
6011 static void
6012 set_line_mbstate ()
6013 {
6014 int i, previ, len, c;
6015 mbstate_t mbs, prevs;
6016 size_t mbclen;
6017
6018 if (shell_input_line == NULL)
6019 return;
6020 len = strlen (shell_input_line); /* XXX - shell_input_line_len ? */
6021 FREE (shell_input_line_property);
6022 shell_input_line_property = (char *)xmalloc (len + 1);
6023
6024 memset (&prevs, '\0', sizeof (mbstate_t));
6025 for (i = previ = 0; i < len; i++)
6026 {
6027 mbs = prevs;
6028
6029 c = shell_input_line[i];
6030 if (c == EOF)
6031 {
6032 int j;
6033 for (j = i; j < len; j++)
6034 shell_input_line_property[j] = 1;
6035 break;
6036 }
6037
6038 mbclen = mbrlen (shell_input_line + previ, i - previ + 1, &mbs);
6039 if (mbclen == 1 || mbclen == (size_t)-1)
6040 {
6041 mbclen = 1;
6042 previ = i + 1;
6043 }
6044 else if (mbclen == (size_t)-2)
6045 mbclen = 0;
6046 else if (mbclen > 1)
6047 {
6048 mbclen = 0;
6049 previ = i + 1;
6050 prevs = mbs;
6051 }
6052 else
6053 {
6054 /* XXX - what to do if mbrlen returns 0? (null wide character) */
6055 int j;
6056 for (j = i; j < len; j++)
6057 shell_input_line_property[j] = 1;
6058 break;
6059 }
6060
6061 shell_input_line_property[i] = mbclen;
6062 }
6063 }
6064 #endif /* HANDLE_MULTIBYTE */