1/*-
2 * Copyright (c) 1992 Diomidis Spinellis.
3 * Copyright (c) 1992, 1993
4 * The Regents of the University of California. All rights reserved.
5 *
6 * This code is derived from software contributed to Berkeley by
7 * Diomidis Spinellis of Imperial College, University of London.
8 *
9 * Redistribution and use in source and binary forms, with or without
10 * modification, are permitted provided that the following conditions
11 * are met:
12 * 1. Redistributions of source code must retain the above copyright
13 * notice, this list of conditions and the following disclaimer.
14 * 2. Redistributions in binary form must reproduce the above copyright
15 * notice, this list of conditions and the following disclaimer in the
16 * documentation and/or other materials provided with the distribution.
17 * 4. Neither the name of the University nor the names of its contributors
18 * may be used to endorse or promote products derived from this software
19 * without specific prior written permission.
20 *
21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31 * SUCH DAMAGE.
32 */
33
34#include <sys/cdefs.h>
| 1/*-
2 * Copyright (c) 1992 Diomidis Spinellis.
3 * Copyright (c) 1992, 1993
4 * The Regents of the University of California. All rights reserved.
5 *
6 * This code is derived from software contributed to Berkeley by
7 * Diomidis Spinellis of Imperial College, University of London.
8 *
9 * Redistribution and use in source and binary forms, with or without
10 * modification, are permitted provided that the following conditions
11 * are met:
12 * 1. Redistributions of source code must retain the above copyright
13 * notice, this list of conditions and the following disclaimer.
14 * 2. Redistributions in binary form must reproduce the above copyright
15 * notice, this list of conditions and the following disclaimer in the
16 * documentation and/or other materials provided with the distribution.
17 * 4. Neither the name of the University nor the names of its contributors
18 * may be used to endorse or promote products derived from this software
19 * without specific prior written permission.
20 *
21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31 * SUCH DAMAGE.
32 */
33
34#include <sys/cdefs.h>
|
35__FBSDID("$FreeBSD: head/usr.bin/sed/compile.c 197361 2009-09-20 15:17:40Z dds $");
| 35__FBSDID("$FreeBSD: head/usr.bin/sed/compile.c 197362 2009-09-20 15:47:31Z dds $");
|
36
37#ifndef lint
38static const char sccsid[] = "@(#)compile.c 8.1 (Berkeley) 6/6/93";
39#endif
40
41#include <sys/types.h>
42#include <sys/stat.h>
43
44#include <ctype.h>
45#include <err.h>
46#include <errno.h>
47#include <fcntl.h>
48#include <limits.h>
49#include <regex.h>
50#include <stdio.h>
51#include <stdlib.h>
52#include <string.h>
53#include <wchar.h>
54
55#include "defs.h"
56#include "extern.h"
57
58#define LHSZ 128
59#define LHMASK (LHSZ - 1)
60static struct labhash {
61 struct labhash *lh_next;
62 u_int lh_hash;
63 struct s_command *lh_cmd;
64 int lh_ref;
65} *labels[LHSZ];
66
67static char *compile_addr(char *, struct s_addr *);
68static char *compile_ccl(char **, char *);
69static char *compile_delimited(char *, char *, int);
70static char *compile_flags(char *, struct s_subst *);
71static regex_t *compile_re(char *, int);
72static char *compile_subst(char *, struct s_subst *);
73static char *compile_text(void);
74static char *compile_tr(char *, struct s_tr **);
75static struct s_command
76 **compile_stream(struct s_command **);
77static char *duptoeol(char *, const char *);
78static void enterlabel(struct s_command *);
79static struct s_command
80 *findlabel(char *);
81static void fixuplabel(struct s_command *, struct s_command *);
82static void uselabel(void);
83
84/*
85 * Command specification. This is used to drive the command parser.
86 */
87struct s_format {
88 char code; /* Command code */
89 int naddr; /* Number of address args */
90 enum e_args args; /* Argument type */
91};
92
93static struct s_format cmd_fmts[] = {
94 {'{', 2, GROUP},
95 {'}', 0, ENDGROUP},
96 {'a', 1, TEXT},
97 {'b', 2, BRANCH},
98 {'c', 2, TEXT},
99 {'d', 2, EMPTY},
100 {'D', 2, EMPTY},
101 {'g', 2, EMPTY},
102 {'G', 2, EMPTY},
103 {'h', 2, EMPTY},
104 {'H', 2, EMPTY},
105 {'i', 1, TEXT},
106 {'l', 2, EMPTY},
107 {'n', 2, EMPTY},
108 {'N', 2, EMPTY},
109 {'p', 2, EMPTY},
110 {'P', 2, EMPTY},
111 {'q', 1, EMPTY},
112 {'r', 1, RFILE},
113 {'s', 2, SUBST},
114 {'t', 2, BRANCH},
115 {'w', 2, WFILE},
116 {'x', 2, EMPTY},
117 {'y', 2, TR},
118 {'!', 2, NONSEL},
119 {':', 0, LABEL},
120 {'#', 0, COMMENT},
121 {'=', 1, EMPTY},
122 {'\0', 0, COMMENT},
123};
124
125/* The compiled program. */
126struct s_command *prog;
127
128/*
129 * Compile the program into prog.
130 * Initialise appends.
131 */
132void
133compile(void)
134{
135 *compile_stream(&prog) = NULL;
136 fixuplabel(prog, NULL);
137 uselabel();
138 if (appendnum == 0)
139 appends = NULL;
140 else if ((appends = malloc(sizeof(struct s_appends) * appendnum)) ==
141 NULL)
142 err(1, "malloc");
143 if ((match = malloc((maxnsub + 1) * sizeof(regmatch_t))) == NULL)
144 err(1, "malloc");
145}
146
147#define EATSPACE() do { \
148 if (p) \
149 while (*p && isspace((unsigned char)*p)) \
150 p++; \
151 } while (0)
152
153static struct s_command **
154compile_stream(struct s_command **link)
155{
156 char *p;
157 static char lbuf[_POSIX2_LINE_MAX + 1]; /* To save stack */
158 struct s_command *cmd, *cmd2, *stack;
159 struct s_format *fp;
160 char re[_POSIX2_LINE_MAX + 1];
161 int naddr; /* Number of addresses */
162
163 stack = 0;
164 for (;;) {
165 if ((p = cu_fgets(lbuf, sizeof(lbuf), NULL)) == NULL) {
166 if (stack != 0)
167 errx(1, "%lu: %s: unexpected EOF (pending }'s)",
168 linenum, fname);
169 return (link);
170 }
171
172semicolon: EATSPACE();
173 if (p) {
174 if (*p == '#' || *p == '\0')
175 continue;
176 else if (*p == ';') {
177 p++;
178 goto semicolon;
179 }
180 }
181 if ((*link = cmd = malloc(sizeof(struct s_command))) == NULL)
182 err(1, "malloc");
183 link = &cmd->next;
184 cmd->startline = cmd->nonsel = 0;
185 /* First parse the addresses */
186 naddr = 0;
187
188/* Valid characters to start an address */
189#define addrchar(c) (strchr("0123456789/\\$", (c)))
190 if (addrchar(*p)) {
191 naddr++;
192 if ((cmd->a1 = malloc(sizeof(struct s_addr))) == NULL)
193 err(1, "malloc");
194 p = compile_addr(p, cmd->a1);
195 EATSPACE(); /* EXTENSION */
196 if (*p == ',') {
197 p++;
198 EATSPACE(); /* EXTENSION */
199 naddr++;
200 if ((cmd->a2 = malloc(sizeof(struct s_addr)))
201 == NULL)
202 err(1, "malloc");
203 p = compile_addr(p, cmd->a2);
204 EATSPACE();
205 } else
206 cmd->a2 = 0;
207 } else
208 cmd->a1 = cmd->a2 = 0;
209
210nonsel: /* Now parse the command */
211 if (!*p)
212 errx(1, "%lu: %s: command expected", linenum, fname);
213 cmd->code = *p;
214 for (fp = cmd_fmts; fp->code; fp++)
215 if (fp->code == *p)
216 break;
217 if (!fp->code)
218 errx(1, "%lu: %s: invalid command code %c", linenum, fname, *p);
219 if (naddr > fp->naddr)
220 errx(1,
221 "%lu: %s: command %c expects up to %d address(es), found %d",
222 linenum, fname, *p, fp->naddr, naddr);
223 switch (fp->args) {
224 case NONSEL: /* ! */
225 p++;
226 EATSPACE();
227 cmd->nonsel = 1;
228 goto nonsel;
229 case GROUP: /* { */
230 p++;
231 EATSPACE();
232 cmd->next = stack;
233 stack = cmd;
234 link = &cmd->u.c;
235 if (*p)
236 goto semicolon;
237 break;
238 case ENDGROUP:
239 /*
240 * Short-circuit command processing, since end of
241 * group is really just a noop.
242 */
243 cmd->nonsel = 1;
244 if (stack == 0)
245 errx(1, "%lu: %s: unexpected }", linenum, fname);
246 cmd2 = stack;
247 stack = cmd2->next;
248 cmd2->next = cmd;
249 /*FALLTHROUGH*/
250 case EMPTY: /* d D g G h H l n N p P q x = \0 */
251 p++;
252 EATSPACE();
253 if (*p == ';') {
254 p++;
255 link = &cmd->next;
256 goto semicolon;
257 }
258 if (*p)
259 errx(1, "%lu: %s: extra characters at the end of %c command",
260 linenum, fname, cmd->code);
261 break;
262 case TEXT: /* a c i */
263 p++;
264 EATSPACE();
265 if (*p != '\\')
266 errx(1,
267"%lu: %s: command %c expects \\ followed by text", linenum, fname, cmd->code);
268 p++;
269 EATSPACE();
270 if (*p)
271 errx(1,
272 "%lu: %s: extra characters after \\ at the end of %c command",
273 linenum, fname, cmd->code);
274 cmd->t = compile_text();
275 break;
276 case COMMENT: /* \0 # */
277 break;
278 case WFILE: /* w */
279 p++;
280 EATSPACE();
281 if (*p == '\0')
282 errx(1, "%lu: %s: filename expected", linenum, fname);
283 cmd->t = duptoeol(p, "w command");
284 if (aflag)
285 cmd->u.fd = -1;
286 else if ((cmd->u.fd = open(p,
287 O_WRONLY|O_APPEND|O_CREAT|O_TRUNC,
288 DEFFILEMODE)) == -1)
289 err(1, "%s", p);
290 break;
291 case RFILE: /* r */
292 p++;
293 EATSPACE();
294 if (*p == '\0')
295 errx(1, "%lu: %s: filename expected", linenum, fname);
296 else
297 cmd->t = duptoeol(p, "read command");
298 break;
299 case BRANCH: /* b t */
300 p++;
301 EATSPACE();
302 if (*p == '\0')
303 cmd->t = NULL;
304 else
305 cmd->t = duptoeol(p, "branch");
306 break;
307 case LABEL: /* : */
308 p++;
309 EATSPACE();
310 cmd->t = duptoeol(p, "label");
311 if (strlen(p) == 0)
312 errx(1, "%lu: %s: empty label", linenum, fname);
313 enterlabel(cmd);
314 break;
315 case SUBST: /* s */
316 p++;
317 if (*p == '\0' || *p == '\\')
318 errx(1,
319"%lu: %s: substitute pattern can not be delimited by newline or backslash",
320 linenum, fname);
321 if ((cmd->u.s = calloc(1, sizeof(struct s_subst))) == NULL)
322 err(1, "malloc");
323 p = compile_delimited(p, re, 0);
324 if (p == NULL)
325 errx(1,
326 "%lu: %s: unterminated substitute pattern", linenum, fname);
327
328 /* Compile RE with no case sensitivity temporarily */
329 if (*re == '\0')
330 cmd->u.s->re = NULL;
331 else
332 cmd->u.s->re = compile_re(re, 0);
333 --p;
334 p = compile_subst(p, cmd->u.s);
335 p = compile_flags(p, cmd->u.s);
336
337 /* Recompile RE with case sensitivity from "I" flag if any */
338 if (*re == '\0')
339 cmd->u.s->re = NULL;
340 else
341 cmd->u.s->re = compile_re(re, cmd->u.s->icase);
342 EATSPACE();
343 if (*p == ';') {
344 p++;
345 link = &cmd->next;
346 goto semicolon;
347 }
348 break;
349 case TR: /* y */
350 p++;
351 p = compile_tr(p, &cmd->u.y);
352 EATSPACE();
353 if (*p == ';') {
354 p++;
355 link = &cmd->next;
356 goto semicolon;
357 }
358 if (*p)
359 errx(1,
360"%lu: %s: extra text at the end of a transform command", linenum, fname);
361 break;
362 }
363 }
364}
365
366/*
367 * Get a delimited string. P points to the delimeter of the string; d points
368 * to a buffer area. Newline and delimiter escapes are processed; other
369 * escapes are ignored.
370 *
371 * Returns a pointer to the first character after the final delimiter or NULL
372 * in the case of a non-terminated string. The character array d is filled
373 * with the processed string.
374 */
375static char *
376compile_delimited(char *p, char *d, int is_tr)
377{
378 char c;
379
380 c = *p++;
381 if (c == '\0')
382 return (NULL);
383 else if (c == '\\')
384 errx(1, "%lu: %s: \\ can not be used as a string delimiter",
385 linenum, fname);
386 else if (c == '\n')
387 errx(1, "%lu: %s: newline can not be used as a string delimiter",
388 linenum, fname);
389 while (*p) {
390 if (*p == '[' && *p != c) {
391 if ((d = compile_ccl(&p, d)) == NULL)
392 errx(1, "%lu: %s: unbalanced brackets ([])", linenum, fname);
393 continue;
394 } else if (*p == '\\' && p[1] == '[') {
395 *d++ = *p++;
396 } else if (*p == '\\' && p[1] == c)
397 p++;
398 else if (*p == '\\' && p[1] == 'n') {
399 *d++ = '\n';
400 p += 2;
401 continue;
402 } else if (*p == '\\' && p[1] == '\\') {
403 if (is_tr)
404 p++;
405 else
406 *d++ = *p++;
407 } else if (*p == c) {
408 *d = '\0';
409 return (p + 1);
410 }
411 *d++ = *p++;
412 }
413 return (NULL);
414}
415
416
417/* compile_ccl: expand a POSIX character class */
418static char *
419compile_ccl(char **sp, char *t)
420{
421 int c, d;
422 char *s = *sp;
423
424 *t++ = *s++;
425 if (*s == '^')
426 *t++ = *s++;
427 if (*s == ']')
428 *t++ = *s++;
429 for (; *s && (*t = *s) != ']'; s++, t++)
430 if (*s == '[' && ((d = *(s+1)) == '.' || d == ':' || d == '=')) {
431 *++t = *++s, t++, s++;
432 for (c = *s; (*t = *s) != ']' || c != d; s++, t++)
433 if ((c = *s) == '\0')
434 return NULL;
| 36
37#ifndef lint
38static const char sccsid[] = "@(#)compile.c 8.1 (Berkeley) 6/6/93";
39#endif
40
41#include <sys/types.h>
42#include <sys/stat.h>
43
44#include <ctype.h>
45#include <err.h>
46#include <errno.h>
47#include <fcntl.h>
48#include <limits.h>
49#include <regex.h>
50#include <stdio.h>
51#include <stdlib.h>
52#include <string.h>
53#include <wchar.h>
54
55#include "defs.h"
56#include "extern.h"
57
58#define LHSZ 128
59#define LHMASK (LHSZ - 1)
60static struct labhash {
61 struct labhash *lh_next;
62 u_int lh_hash;
63 struct s_command *lh_cmd;
64 int lh_ref;
65} *labels[LHSZ];
66
67static char *compile_addr(char *, struct s_addr *);
68static char *compile_ccl(char **, char *);
69static char *compile_delimited(char *, char *, int);
70static char *compile_flags(char *, struct s_subst *);
71static regex_t *compile_re(char *, int);
72static char *compile_subst(char *, struct s_subst *);
73static char *compile_text(void);
74static char *compile_tr(char *, struct s_tr **);
75static struct s_command
76 **compile_stream(struct s_command **);
77static char *duptoeol(char *, const char *);
78static void enterlabel(struct s_command *);
79static struct s_command
80 *findlabel(char *);
81static void fixuplabel(struct s_command *, struct s_command *);
82static void uselabel(void);
83
84/*
85 * Command specification. This is used to drive the command parser.
86 */
87struct s_format {
88 char code; /* Command code */
89 int naddr; /* Number of address args */
90 enum e_args args; /* Argument type */
91};
92
93static struct s_format cmd_fmts[] = {
94 {'{', 2, GROUP},
95 {'}', 0, ENDGROUP},
96 {'a', 1, TEXT},
97 {'b', 2, BRANCH},
98 {'c', 2, TEXT},
99 {'d', 2, EMPTY},
100 {'D', 2, EMPTY},
101 {'g', 2, EMPTY},
102 {'G', 2, EMPTY},
103 {'h', 2, EMPTY},
104 {'H', 2, EMPTY},
105 {'i', 1, TEXT},
106 {'l', 2, EMPTY},
107 {'n', 2, EMPTY},
108 {'N', 2, EMPTY},
109 {'p', 2, EMPTY},
110 {'P', 2, EMPTY},
111 {'q', 1, EMPTY},
112 {'r', 1, RFILE},
113 {'s', 2, SUBST},
114 {'t', 2, BRANCH},
115 {'w', 2, WFILE},
116 {'x', 2, EMPTY},
117 {'y', 2, TR},
118 {'!', 2, NONSEL},
119 {':', 0, LABEL},
120 {'#', 0, COMMENT},
121 {'=', 1, EMPTY},
122 {'\0', 0, COMMENT},
123};
124
125/* The compiled program. */
126struct s_command *prog;
127
128/*
129 * Compile the program into prog.
130 * Initialise appends.
131 */
132void
133compile(void)
134{
135 *compile_stream(&prog) = NULL;
136 fixuplabel(prog, NULL);
137 uselabel();
138 if (appendnum == 0)
139 appends = NULL;
140 else if ((appends = malloc(sizeof(struct s_appends) * appendnum)) ==
141 NULL)
142 err(1, "malloc");
143 if ((match = malloc((maxnsub + 1) * sizeof(regmatch_t))) == NULL)
144 err(1, "malloc");
145}
146
147#define EATSPACE() do { \
148 if (p) \
149 while (*p && isspace((unsigned char)*p)) \
150 p++; \
151 } while (0)
152
153static struct s_command **
154compile_stream(struct s_command **link)
155{
156 char *p;
157 static char lbuf[_POSIX2_LINE_MAX + 1]; /* To save stack */
158 struct s_command *cmd, *cmd2, *stack;
159 struct s_format *fp;
160 char re[_POSIX2_LINE_MAX + 1];
161 int naddr; /* Number of addresses */
162
163 stack = 0;
164 for (;;) {
165 if ((p = cu_fgets(lbuf, sizeof(lbuf), NULL)) == NULL) {
166 if (stack != 0)
167 errx(1, "%lu: %s: unexpected EOF (pending }'s)",
168 linenum, fname);
169 return (link);
170 }
171
172semicolon: EATSPACE();
173 if (p) {
174 if (*p == '#' || *p == '\0')
175 continue;
176 else if (*p == ';') {
177 p++;
178 goto semicolon;
179 }
180 }
181 if ((*link = cmd = malloc(sizeof(struct s_command))) == NULL)
182 err(1, "malloc");
183 link = &cmd->next;
184 cmd->startline = cmd->nonsel = 0;
185 /* First parse the addresses */
186 naddr = 0;
187
188/* Valid characters to start an address */
189#define addrchar(c) (strchr("0123456789/\\$", (c)))
190 if (addrchar(*p)) {
191 naddr++;
192 if ((cmd->a1 = malloc(sizeof(struct s_addr))) == NULL)
193 err(1, "malloc");
194 p = compile_addr(p, cmd->a1);
195 EATSPACE(); /* EXTENSION */
196 if (*p == ',') {
197 p++;
198 EATSPACE(); /* EXTENSION */
199 naddr++;
200 if ((cmd->a2 = malloc(sizeof(struct s_addr)))
201 == NULL)
202 err(1, "malloc");
203 p = compile_addr(p, cmd->a2);
204 EATSPACE();
205 } else
206 cmd->a2 = 0;
207 } else
208 cmd->a1 = cmd->a2 = 0;
209
210nonsel: /* Now parse the command */
211 if (!*p)
212 errx(1, "%lu: %s: command expected", linenum, fname);
213 cmd->code = *p;
214 for (fp = cmd_fmts; fp->code; fp++)
215 if (fp->code == *p)
216 break;
217 if (!fp->code)
218 errx(1, "%lu: %s: invalid command code %c", linenum, fname, *p);
219 if (naddr > fp->naddr)
220 errx(1,
221 "%lu: %s: command %c expects up to %d address(es), found %d",
222 linenum, fname, *p, fp->naddr, naddr);
223 switch (fp->args) {
224 case NONSEL: /* ! */
225 p++;
226 EATSPACE();
227 cmd->nonsel = 1;
228 goto nonsel;
229 case GROUP: /* { */
230 p++;
231 EATSPACE();
232 cmd->next = stack;
233 stack = cmd;
234 link = &cmd->u.c;
235 if (*p)
236 goto semicolon;
237 break;
238 case ENDGROUP:
239 /*
240 * Short-circuit command processing, since end of
241 * group is really just a noop.
242 */
243 cmd->nonsel = 1;
244 if (stack == 0)
245 errx(1, "%lu: %s: unexpected }", linenum, fname);
246 cmd2 = stack;
247 stack = cmd2->next;
248 cmd2->next = cmd;
249 /*FALLTHROUGH*/
250 case EMPTY: /* d D g G h H l n N p P q x = \0 */
251 p++;
252 EATSPACE();
253 if (*p == ';') {
254 p++;
255 link = &cmd->next;
256 goto semicolon;
257 }
258 if (*p)
259 errx(1, "%lu: %s: extra characters at the end of %c command",
260 linenum, fname, cmd->code);
261 break;
262 case TEXT: /* a c i */
263 p++;
264 EATSPACE();
265 if (*p != '\\')
266 errx(1,
267"%lu: %s: command %c expects \\ followed by text", linenum, fname, cmd->code);
268 p++;
269 EATSPACE();
270 if (*p)
271 errx(1,
272 "%lu: %s: extra characters after \\ at the end of %c command",
273 linenum, fname, cmd->code);
274 cmd->t = compile_text();
275 break;
276 case COMMENT: /* \0 # */
277 break;
278 case WFILE: /* w */
279 p++;
280 EATSPACE();
281 if (*p == '\0')
282 errx(1, "%lu: %s: filename expected", linenum, fname);
283 cmd->t = duptoeol(p, "w command");
284 if (aflag)
285 cmd->u.fd = -1;
286 else if ((cmd->u.fd = open(p,
287 O_WRONLY|O_APPEND|O_CREAT|O_TRUNC,
288 DEFFILEMODE)) == -1)
289 err(1, "%s", p);
290 break;
291 case RFILE: /* r */
292 p++;
293 EATSPACE();
294 if (*p == '\0')
295 errx(1, "%lu: %s: filename expected", linenum, fname);
296 else
297 cmd->t = duptoeol(p, "read command");
298 break;
299 case BRANCH: /* b t */
300 p++;
301 EATSPACE();
302 if (*p == '\0')
303 cmd->t = NULL;
304 else
305 cmd->t = duptoeol(p, "branch");
306 break;
307 case LABEL: /* : */
308 p++;
309 EATSPACE();
310 cmd->t = duptoeol(p, "label");
311 if (strlen(p) == 0)
312 errx(1, "%lu: %s: empty label", linenum, fname);
313 enterlabel(cmd);
314 break;
315 case SUBST: /* s */
316 p++;
317 if (*p == '\0' || *p == '\\')
318 errx(1,
319"%lu: %s: substitute pattern can not be delimited by newline or backslash",
320 linenum, fname);
321 if ((cmd->u.s = calloc(1, sizeof(struct s_subst))) == NULL)
322 err(1, "malloc");
323 p = compile_delimited(p, re, 0);
324 if (p == NULL)
325 errx(1,
326 "%lu: %s: unterminated substitute pattern", linenum, fname);
327
328 /* Compile RE with no case sensitivity temporarily */
329 if (*re == '\0')
330 cmd->u.s->re = NULL;
331 else
332 cmd->u.s->re = compile_re(re, 0);
333 --p;
334 p = compile_subst(p, cmd->u.s);
335 p = compile_flags(p, cmd->u.s);
336
337 /* Recompile RE with case sensitivity from "I" flag if any */
338 if (*re == '\0')
339 cmd->u.s->re = NULL;
340 else
341 cmd->u.s->re = compile_re(re, cmd->u.s->icase);
342 EATSPACE();
343 if (*p == ';') {
344 p++;
345 link = &cmd->next;
346 goto semicolon;
347 }
348 break;
349 case TR: /* y */
350 p++;
351 p = compile_tr(p, &cmd->u.y);
352 EATSPACE();
353 if (*p == ';') {
354 p++;
355 link = &cmd->next;
356 goto semicolon;
357 }
358 if (*p)
359 errx(1,
360"%lu: %s: extra text at the end of a transform command", linenum, fname);
361 break;
362 }
363 }
364}
365
366/*
367 * Get a delimited string. P points to the delimeter of the string; d points
368 * to a buffer area. Newline and delimiter escapes are processed; other
369 * escapes are ignored.
370 *
371 * Returns a pointer to the first character after the final delimiter or NULL
372 * in the case of a non-terminated string. The character array d is filled
373 * with the processed string.
374 */
375static char *
376compile_delimited(char *p, char *d, int is_tr)
377{
378 char c;
379
380 c = *p++;
381 if (c == '\0')
382 return (NULL);
383 else if (c == '\\')
384 errx(1, "%lu: %s: \\ can not be used as a string delimiter",
385 linenum, fname);
386 else if (c == '\n')
387 errx(1, "%lu: %s: newline can not be used as a string delimiter",
388 linenum, fname);
389 while (*p) {
390 if (*p == '[' && *p != c) {
391 if ((d = compile_ccl(&p, d)) == NULL)
392 errx(1, "%lu: %s: unbalanced brackets ([])", linenum, fname);
393 continue;
394 } else if (*p == '\\' && p[1] == '[') {
395 *d++ = *p++;
396 } else if (*p == '\\' && p[1] == c)
397 p++;
398 else if (*p == '\\' && p[1] == 'n') {
399 *d++ = '\n';
400 p += 2;
401 continue;
402 } else if (*p == '\\' && p[1] == '\\') {
403 if (is_tr)
404 p++;
405 else
406 *d++ = *p++;
407 } else if (*p == c) {
408 *d = '\0';
409 return (p + 1);
410 }
411 *d++ = *p++;
412 }
413 return (NULL);
414}
415
416
417/* compile_ccl: expand a POSIX character class */
418static char *
419compile_ccl(char **sp, char *t)
420{
421 int c, d;
422 char *s = *sp;
423
424 *t++ = *s++;
425 if (*s == '^')
426 *t++ = *s++;
427 if (*s == ']')
428 *t++ = *s++;
429 for (; *s && (*t = *s) != ']'; s++, t++)
430 if (*s == '[' && ((d = *(s+1)) == '.' || d == ':' || d == '=')) {
431 *++t = *++s, t++, s++;
432 for (c = *s; (*t = *s) != ']' || c != d; s++, t++)
433 if ((c = *s) == '\0')
434 return NULL;
|
435 } else if (*s == '\\' && s[1] == 'n')
436 *t = '\n', s++;
| 435 }
|
437 return (*s == ']') ? *sp = ++s, ++t : NULL;
438}
439
440/*
441 * Compiles the regular expression in RE and returns a pointer to the compiled
442 * regular expression.
443 * Cflags are passed to regcomp.
444 */
445static regex_t *
446compile_re(char *re, int case_insensitive)
447{
448 regex_t *rep;
449 int eval, flags;
450
451
452 flags = rflags;
453 if (case_insensitive)
454 flags |= REG_ICASE;
455 if ((rep = malloc(sizeof(regex_t))) == NULL)
456 err(1, "malloc");
457 if ((eval = regcomp(rep, re, flags)) != 0)
458 errx(1, "%lu: %s: RE error: %s",
459 linenum, fname, strregerror(eval, rep));
460 if (maxnsub < rep->re_nsub)
461 maxnsub = rep->re_nsub;
462 return (rep);
463}
464
465/*
466 * Compile the substitution string of a regular expression and set res to
467 * point to a saved copy of it. Nsub is the number of parenthesized regular
468 * expressions.
469 */
470static char *
471compile_subst(char *p, struct s_subst *s)
472{
473 static char lbuf[_POSIX2_LINE_MAX + 1];
474 int asize, size;
475 u_char ref;
476 char c, *text, *op, *sp;
477 int more = 1, sawesc = 0;
478
479 c = *p++; /* Terminator character */
480 if (c == '\0')
481 return (NULL);
482
483 s->maxbref = 0;
484 s->linenum = linenum;
485 asize = 2 * _POSIX2_LINE_MAX + 1;
486 if ((text = malloc(asize)) == NULL)
487 err(1, "malloc");
488 size = 0;
489 do {
490 op = sp = text + size;
491 for (; *p; p++) {
492 if (*p == '\\' || sawesc) {
493 /*
494 * If this is a continuation from the last
495 * buffer, we won't have a character to
496 * skip over.
497 */
498 if (sawesc)
499 sawesc = 0;
500 else
501 p++;
502
503 if (*p == '\0') {
504 /*
505 * This escaped character is continued
506 * in the next part of the line. Note
507 * this fact, then cause the loop to
508 * exit w/ normal EOL case and reenter
509 * above with the new buffer.
510 */
511 sawesc = 1;
512 p--;
513 continue;
514 } else if (strchr("123456789", *p) != NULL) {
515 *sp++ = '\\';
516 ref = *p - '0';
517 if (s->re != NULL &&
518 ref > s->re->re_nsub)
519 errx(1, "%lu: %s: \\%c not defined in the RE",
520 linenum, fname, *p);
521 if (s->maxbref < ref)
522 s->maxbref = ref;
523 } else if (*p == '&' || *p == '\\')
524 *sp++ = '\\';
525 } else if (*p == c) {
526 if (*++p == '\0' && more) {
527 if (cu_fgets(lbuf, sizeof(lbuf), &more))
528 p = lbuf;
529 }
530 *sp++ = '\0';
531 size += sp - op;
532 if ((s->new = realloc(text, size)) == NULL)
533 err(1, "realloc");
534 return (p);
535 } else if (*p == '\n') {
536 errx(1,
537"%lu: %s: unescaped newline inside substitute pattern", linenum, fname);
538 /* NOTREACHED */
539 }
540 *sp++ = *p;
541 }
542 size += sp - op;
543 if (asize - size < _POSIX2_LINE_MAX + 1) {
544 asize *= 2;
545 if ((text = realloc(text, asize)) == NULL)
546 err(1, "realloc");
547 }
548 } while (cu_fgets(p = lbuf, sizeof(lbuf), &more));
549 errx(1, "%lu: %s: unterminated substitute in regular expression",
550 linenum, fname);
551 /* NOTREACHED */
552}
553
554/*
555 * Compile the flags of the s command
556 */
557static char *
558compile_flags(char *p, struct s_subst *s)
559{
560 int gn; /* True if we have seen g or n */
561 unsigned long nval;
562 char wfile[_POSIX2_LINE_MAX + 1], *q;
563
564 s->n = 1; /* Default */
565 s->p = 0;
566 s->wfile = NULL;
567 s->wfd = -1;
568 s->icase = 0;
569 for (gn = 0;;) {
570 EATSPACE(); /* EXTENSION */
571 switch (*p) {
572 case 'g':
573 if (gn)
574 errx(1,
575"%lu: %s: more than one number or 'g' in substitute flags", linenum, fname);
576 gn = 1;
577 s->n = 0;
578 break;
579 case '\0':
580 case '\n':
581 case ';':
582 return (p);
583 case 'p':
584 s->p = 1;
585 break;
586 case 'I':
587 s->icase = 1;
588 break;
589 case '1': case '2': case '3':
590 case '4': case '5': case '6':
591 case '7': case '8': case '9':
592 if (gn)
593 errx(1,
594"%lu: %s: more than one number or 'g' in substitute flags", linenum, fname);
595 gn = 1;
596 errno = 0;
597 nval = strtol(p, &p, 10);
598 if (errno == ERANGE || nval > INT_MAX)
599 errx(1,
600"%lu: %s: overflow in the 'N' substitute flag", linenum, fname);
601 s->n = nval;
602 p--;
603 break;
604 case 'w':
605 p++;
606#ifdef HISTORIC_PRACTICE
607 if (*p != ' ') {
608 warnx("%lu: %s: space missing before w wfile", linenum, fname);
609 return (p);
610 }
611#endif
612 EATSPACE();
613 q = wfile;
614 while (*p) {
615 if (*p == '\n')
616 break;
617 *q++ = *p++;
618 }
619 *q = '\0';
620 if (q == wfile)
621 errx(1, "%lu: %s: no wfile specified", linenum, fname);
622 s->wfile = strdup(wfile);
623 if (!aflag && (s->wfd = open(wfile,
624 O_WRONLY|O_APPEND|O_CREAT|O_TRUNC,
625 DEFFILEMODE)) == -1)
626 err(1, "%s", wfile);
627 return (p);
628 default:
629 errx(1, "%lu: %s: bad flag in substitute command: '%c'",
630 linenum, fname, *p);
631 break;
632 }
633 p++;
634 }
635}
636
637/*
638 * Compile a translation set of strings into a lookup table.
639 */
640static char *
641compile_tr(char *p, struct s_tr **py)
642{
643 struct s_tr *y;
644 int i;
645 const char *op, *np;
646 char old[_POSIX2_LINE_MAX + 1];
647 char new[_POSIX2_LINE_MAX + 1];
648 size_t oclen, oldlen, nclen, newlen;
649 mbstate_t mbs1, mbs2;
650
651 if ((*py = y = malloc(sizeof(*y))) == NULL)
652 err(1, NULL);
653 y->multis = NULL;
654 y->nmultis = 0;
655
656 if (*p == '\0' || *p == '\\')
657 errx(1,
658 "%lu: %s: transform pattern can not be delimited by newline or backslash",
659 linenum, fname);
660 p = compile_delimited(p, old, 1);
661 if (p == NULL)
662 errx(1, "%lu: %s: unterminated transform source string",
663 linenum, fname);
664 p = compile_delimited(p - 1, new, 1);
665 if (p == NULL)
666 errx(1, "%lu: %s: unterminated transform target string",
667 linenum, fname);
668 EATSPACE();
669 op = old;
670 oldlen = mbsrtowcs(NULL, &op, 0, NULL);
671 if (oldlen == (size_t)-1)
672 err(1, NULL);
673 np = new;
674 newlen = mbsrtowcs(NULL, &np, 0, NULL);
675 if (newlen == (size_t)-1)
676 err(1, NULL);
677 if (newlen != oldlen)
678 errx(1, "%lu: %s: transform strings are not the same length",
679 linenum, fname);
680 if (MB_CUR_MAX == 1) {
681 /*
682 * The single-byte encoding case is easy: generate a
683 * lookup table.
684 */
685 for (i = 0; i <= UCHAR_MAX; i++)
686 y->bytetab[i] = (char)i;
687 for (; *op; op++, np++)
688 y->bytetab[(u_char)*op] = *np;
689 } else {
690 /*
691 * Multi-byte encoding case: generate a lookup table as
692 * above, but only for single-byte characters. The first
693 * bytes of multi-byte characters have their lookup table
694 * entries set to 0, which causes do_tr() to search through
695 * an auxiliary vector of multi-byte mappings.
696 */
697 memset(&mbs1, 0, sizeof(mbs1));
698 memset(&mbs2, 0, sizeof(mbs2));
699 for (i = 0; i <= UCHAR_MAX; i++)
700 y->bytetab[i] = (btowc(i) != WEOF) ? i : 0;
701 while (*op != '\0') {
702 oclen = mbrlen(op, MB_LEN_MAX, &mbs1);
703 if (oclen == (size_t)-1 || oclen == (size_t)-2)
704 errc(1, EILSEQ, NULL);
705 nclen = mbrlen(np, MB_LEN_MAX, &mbs2);
706 if (nclen == (size_t)-1 || nclen == (size_t)-2)
707 errc(1, EILSEQ, NULL);
708 if (oclen == 1 && nclen == 1)
709 y->bytetab[(u_char)*op] = *np;
710 else {
711 y->bytetab[(u_char)*op] = 0;
712 y->multis = realloc(y->multis,
713 (y->nmultis + 1) * sizeof(*y->multis));
714 if (y->multis == NULL)
715 err(1, NULL);
716 i = y->nmultis++;
717 y->multis[i].fromlen = oclen;
718 memcpy(y->multis[i].from, op, oclen);
719 y->multis[i].tolen = nclen;
720 memcpy(y->multis[i].to, np, nclen);
721 }
722 op += oclen;
723 np += nclen;
724 }
725 }
726 return (p);
727}
728
729/*
730 * Compile the text following an a or i command.
731 */
732static char *
733compile_text(void)
734{
735 int asize, esc_nl, size;
736 char *text, *p, *op, *s;
737 char lbuf[_POSIX2_LINE_MAX + 1];
738
739 asize = 2 * _POSIX2_LINE_MAX + 1;
740 if ((text = malloc(asize)) == NULL)
741 err(1, "malloc");
742 size = 0;
743 while (cu_fgets(lbuf, sizeof(lbuf), NULL)) {
744 op = s = text + size;
745 p = lbuf;
746 EATSPACE();
747 for (esc_nl = 0; *p != '\0'; p++) {
748 if (*p == '\\' && p[1] != '\0' && *++p == '\n')
749 esc_nl = 1;
750 *s++ = *p;
751 }
752 size += s - op;
753 if (!esc_nl) {
754 *s = '\0';
755 break;
756 }
757 if (asize - size < _POSIX2_LINE_MAX + 1) {
758 asize *= 2;
759 if ((text = realloc(text, asize)) == NULL)
760 err(1, "realloc");
761 }
762 }
763 text[size] = '\0';
764 if ((p = realloc(text, size + 1)) == NULL)
765 err(1, "realloc");
766 return (p);
767}
768
769/*
770 * Get an address and return a pointer to the first character after
771 * it. Fill the structure pointed to according to the address.
772 */
773static char *
774compile_addr(char *p, struct s_addr *a)
775{
776 char *end, re[_POSIX2_LINE_MAX + 1];
777 int icase;
778
779 icase = 0;
780
781 a->type = 0;
782 switch (*p) {
783 case '\\': /* Context address */
784 ++p;
785 /* FALLTHROUGH */
786 case '/': /* Context address */
787 p = compile_delimited(p, re, 0);
788 if (p == NULL)
789 errx(1, "%lu: %s: unterminated regular expression", linenum, fname);
790 /* Check for case insensitive regexp flag */
791 if (*p == 'I') {
792 icase = 1;
793 p++;
794 }
795 if (*re == '\0')
796 a->u.r = NULL;
797 else
798 a->u.r = compile_re(re, icase);
799 a->type = AT_RE;
800 return (p);
801
802 case '$': /* Last line */
803 a->type = AT_LAST;
804 return (p + 1);
805
806 case '+': /* Relative line number */
807 a->type = AT_RELLINE;
808 p++;
809 /* FALLTHROUGH */
810 /* Line number */
811 case '0': case '1': case '2': case '3': case '4':
812 case '5': case '6': case '7': case '8': case '9':
813 if (a->type == 0)
814 a->type = AT_LINE;
815 a->u.l = strtol(p, &end, 10);
816 return (end);
817 default:
818 errx(1, "%lu: %s: expected context address", linenum, fname);
819 return (NULL);
820 }
821}
822
823/*
824 * duptoeol --
825 * Return a copy of all the characters up to \n or \0.
826 */
827static char *
828duptoeol(char *s, const char *ctype)
829{
830 size_t len;
831 int ws;
832 char *p, *start;
833
834 ws = 0;
835 for (start = s; *s != '\0' && *s != '\n'; ++s)
836 ws = isspace((unsigned char)*s);
837 *s = '\0';
838 if (ws)
839 warnx("%lu: %s: whitespace after %s", linenum, fname, ctype);
840 len = s - start + 1;
841 if ((p = malloc(len)) == NULL)
842 err(1, "malloc");
843 return (memmove(p, start, len));
844}
845
846/*
847 * Convert goto label names to addresses, and count a and r commands, in
848 * the given subset of the script. Free the memory used by labels in b
849 * and t commands (but not by :).
850 *
851 * TODO: Remove } nodes
852 */
853static void
854fixuplabel(struct s_command *cp, struct s_command *end)
855{
856
857 for (; cp != end; cp = cp->next)
858 switch (cp->code) {
859 case 'a':
860 case 'r':
861 appendnum++;
862 break;
863 case 'b':
864 case 't':
865 /* Resolve branch target. */
866 if (cp->t == NULL) {
867 cp->u.c = NULL;
868 break;
869 }
870 if ((cp->u.c = findlabel(cp->t)) == NULL)
871 errx(1, "%lu: %s: undefined label '%s'", linenum, fname, cp->t);
872 free(cp->t);
873 break;
874 case '{':
875 /* Do interior commands. */
876 fixuplabel(cp->u.c, cp->next);
877 break;
878 }
879}
880
881/*
882 * Associate the given command label for later lookup.
883 */
884static void
885enterlabel(struct s_command *cp)
886{
887 struct labhash **lhp, *lh;
888 u_char *p;
889 u_int h, c;
890
891 for (h = 0, p = (u_char *)cp->t; (c = *p) != 0; p++)
892 h = (h << 5) + h + c;
893 lhp = &labels[h & LHMASK];
894 for (lh = *lhp; lh != NULL; lh = lh->lh_next)
895 if (lh->lh_hash == h && strcmp(cp->t, lh->lh_cmd->t) == 0)
896 errx(1, "%lu: %s: duplicate label '%s'", linenum, fname, cp->t);
897 if ((lh = malloc(sizeof *lh)) == NULL)
898 err(1, "malloc");
899 lh->lh_next = *lhp;
900 lh->lh_hash = h;
901 lh->lh_cmd = cp;
902 lh->lh_ref = 0;
903 *lhp = lh;
904}
905
906/*
907 * Find the label contained in the command l in the command linked
908 * list cp. L is excluded from the search. Return NULL if not found.
909 */
910static struct s_command *
911findlabel(char *name)
912{
913 struct labhash *lh;
914 u_char *p;
915 u_int h, c;
916
917 for (h = 0, p = (u_char *)name; (c = *p) != 0; p++)
918 h = (h << 5) + h + c;
919 for (lh = labels[h & LHMASK]; lh != NULL; lh = lh->lh_next) {
920 if (lh->lh_hash == h && strcmp(name, lh->lh_cmd->t) == 0) {
921 lh->lh_ref = 1;
922 return (lh->lh_cmd);
923 }
924 }
925 return (NULL);
926}
927
928/*
929 * Warn about any unused labels. As a side effect, release the label hash
930 * table space.
931 */
932static void
933uselabel(void)
934{
935 struct labhash *lh, *next;
936 int i;
937
938 for (i = 0; i < LHSZ; i++) {
939 for (lh = labels[i]; lh != NULL; lh = next) {
940 next = lh->lh_next;
941 if (!lh->lh_ref)
942 warnx("%lu: %s: unused label '%s'",
943 linenum, fname, lh->lh_cmd->t);
944 free(lh);
945 }
946 }
947}
| 436 return (*s == ']') ? *sp = ++s, ++t : NULL;
437}
438
439/*
440 * Compiles the regular expression in RE and returns a pointer to the compiled
441 * regular expression.
442 * Cflags are passed to regcomp.
443 */
444static regex_t *
445compile_re(char *re, int case_insensitive)
446{
447 regex_t *rep;
448 int eval, flags;
449
450
451 flags = rflags;
452 if (case_insensitive)
453 flags |= REG_ICASE;
454 if ((rep = malloc(sizeof(regex_t))) == NULL)
455 err(1, "malloc");
456 if ((eval = regcomp(rep, re, flags)) != 0)
457 errx(1, "%lu: %s: RE error: %s",
458 linenum, fname, strregerror(eval, rep));
459 if (maxnsub < rep->re_nsub)
460 maxnsub = rep->re_nsub;
461 return (rep);
462}
463
464/*
465 * Compile the substitution string of a regular expression and set res to
466 * point to a saved copy of it. Nsub is the number of parenthesized regular
467 * expressions.
468 */
469static char *
470compile_subst(char *p, struct s_subst *s)
471{
472 static char lbuf[_POSIX2_LINE_MAX + 1];
473 int asize, size;
474 u_char ref;
475 char c, *text, *op, *sp;
476 int more = 1, sawesc = 0;
477
478 c = *p++; /* Terminator character */
479 if (c == '\0')
480 return (NULL);
481
482 s->maxbref = 0;
483 s->linenum = linenum;
484 asize = 2 * _POSIX2_LINE_MAX + 1;
485 if ((text = malloc(asize)) == NULL)
486 err(1, "malloc");
487 size = 0;
488 do {
489 op = sp = text + size;
490 for (; *p; p++) {
491 if (*p == '\\' || sawesc) {
492 /*
493 * If this is a continuation from the last
494 * buffer, we won't have a character to
495 * skip over.
496 */
497 if (sawesc)
498 sawesc = 0;
499 else
500 p++;
501
502 if (*p == '\0') {
503 /*
504 * This escaped character is continued
505 * in the next part of the line. Note
506 * this fact, then cause the loop to
507 * exit w/ normal EOL case and reenter
508 * above with the new buffer.
509 */
510 sawesc = 1;
511 p--;
512 continue;
513 } else if (strchr("123456789", *p) != NULL) {
514 *sp++ = '\\';
515 ref = *p - '0';
516 if (s->re != NULL &&
517 ref > s->re->re_nsub)
518 errx(1, "%lu: %s: \\%c not defined in the RE",
519 linenum, fname, *p);
520 if (s->maxbref < ref)
521 s->maxbref = ref;
522 } else if (*p == '&' || *p == '\\')
523 *sp++ = '\\';
524 } else if (*p == c) {
525 if (*++p == '\0' && more) {
526 if (cu_fgets(lbuf, sizeof(lbuf), &more))
527 p = lbuf;
528 }
529 *sp++ = '\0';
530 size += sp - op;
531 if ((s->new = realloc(text, size)) == NULL)
532 err(1, "realloc");
533 return (p);
534 } else if (*p == '\n') {
535 errx(1,
536"%lu: %s: unescaped newline inside substitute pattern", linenum, fname);
537 /* NOTREACHED */
538 }
539 *sp++ = *p;
540 }
541 size += sp - op;
542 if (asize - size < _POSIX2_LINE_MAX + 1) {
543 asize *= 2;
544 if ((text = realloc(text, asize)) == NULL)
545 err(1, "realloc");
546 }
547 } while (cu_fgets(p = lbuf, sizeof(lbuf), &more));
548 errx(1, "%lu: %s: unterminated substitute in regular expression",
549 linenum, fname);
550 /* NOTREACHED */
551}
552
553/*
554 * Compile the flags of the s command
555 */
556static char *
557compile_flags(char *p, struct s_subst *s)
558{
559 int gn; /* True if we have seen g or n */
560 unsigned long nval;
561 char wfile[_POSIX2_LINE_MAX + 1], *q;
562
563 s->n = 1; /* Default */
564 s->p = 0;
565 s->wfile = NULL;
566 s->wfd = -1;
567 s->icase = 0;
568 for (gn = 0;;) {
569 EATSPACE(); /* EXTENSION */
570 switch (*p) {
571 case 'g':
572 if (gn)
573 errx(1,
574"%lu: %s: more than one number or 'g' in substitute flags", linenum, fname);
575 gn = 1;
576 s->n = 0;
577 break;
578 case '\0':
579 case '\n':
580 case ';':
581 return (p);
582 case 'p':
583 s->p = 1;
584 break;
585 case 'I':
586 s->icase = 1;
587 break;
588 case '1': case '2': case '3':
589 case '4': case '5': case '6':
590 case '7': case '8': case '9':
591 if (gn)
592 errx(1,
593"%lu: %s: more than one number or 'g' in substitute flags", linenum, fname);
594 gn = 1;
595 errno = 0;
596 nval = strtol(p, &p, 10);
597 if (errno == ERANGE || nval > INT_MAX)
598 errx(1,
599"%lu: %s: overflow in the 'N' substitute flag", linenum, fname);
600 s->n = nval;
601 p--;
602 break;
603 case 'w':
604 p++;
605#ifdef HISTORIC_PRACTICE
606 if (*p != ' ') {
607 warnx("%lu: %s: space missing before w wfile", linenum, fname);
608 return (p);
609 }
610#endif
611 EATSPACE();
612 q = wfile;
613 while (*p) {
614 if (*p == '\n')
615 break;
616 *q++ = *p++;
617 }
618 *q = '\0';
619 if (q == wfile)
620 errx(1, "%lu: %s: no wfile specified", linenum, fname);
621 s->wfile = strdup(wfile);
622 if (!aflag && (s->wfd = open(wfile,
623 O_WRONLY|O_APPEND|O_CREAT|O_TRUNC,
624 DEFFILEMODE)) == -1)
625 err(1, "%s", wfile);
626 return (p);
627 default:
628 errx(1, "%lu: %s: bad flag in substitute command: '%c'",
629 linenum, fname, *p);
630 break;
631 }
632 p++;
633 }
634}
635
636/*
637 * Compile a translation set of strings into a lookup table.
638 */
639static char *
640compile_tr(char *p, struct s_tr **py)
641{
642 struct s_tr *y;
643 int i;
644 const char *op, *np;
645 char old[_POSIX2_LINE_MAX + 1];
646 char new[_POSIX2_LINE_MAX + 1];
647 size_t oclen, oldlen, nclen, newlen;
648 mbstate_t mbs1, mbs2;
649
650 if ((*py = y = malloc(sizeof(*y))) == NULL)
651 err(1, NULL);
652 y->multis = NULL;
653 y->nmultis = 0;
654
655 if (*p == '\0' || *p == '\\')
656 errx(1,
657 "%lu: %s: transform pattern can not be delimited by newline or backslash",
658 linenum, fname);
659 p = compile_delimited(p, old, 1);
660 if (p == NULL)
661 errx(1, "%lu: %s: unterminated transform source string",
662 linenum, fname);
663 p = compile_delimited(p - 1, new, 1);
664 if (p == NULL)
665 errx(1, "%lu: %s: unterminated transform target string",
666 linenum, fname);
667 EATSPACE();
668 op = old;
669 oldlen = mbsrtowcs(NULL, &op, 0, NULL);
670 if (oldlen == (size_t)-1)
671 err(1, NULL);
672 np = new;
673 newlen = mbsrtowcs(NULL, &np, 0, NULL);
674 if (newlen == (size_t)-1)
675 err(1, NULL);
676 if (newlen != oldlen)
677 errx(1, "%lu: %s: transform strings are not the same length",
678 linenum, fname);
679 if (MB_CUR_MAX == 1) {
680 /*
681 * The single-byte encoding case is easy: generate a
682 * lookup table.
683 */
684 for (i = 0; i <= UCHAR_MAX; i++)
685 y->bytetab[i] = (char)i;
686 for (; *op; op++, np++)
687 y->bytetab[(u_char)*op] = *np;
688 } else {
689 /*
690 * Multi-byte encoding case: generate a lookup table as
691 * above, but only for single-byte characters. The first
692 * bytes of multi-byte characters have their lookup table
693 * entries set to 0, which causes do_tr() to search through
694 * an auxiliary vector of multi-byte mappings.
695 */
696 memset(&mbs1, 0, sizeof(mbs1));
697 memset(&mbs2, 0, sizeof(mbs2));
698 for (i = 0; i <= UCHAR_MAX; i++)
699 y->bytetab[i] = (btowc(i) != WEOF) ? i : 0;
700 while (*op != '\0') {
701 oclen = mbrlen(op, MB_LEN_MAX, &mbs1);
702 if (oclen == (size_t)-1 || oclen == (size_t)-2)
703 errc(1, EILSEQ, NULL);
704 nclen = mbrlen(np, MB_LEN_MAX, &mbs2);
705 if (nclen == (size_t)-1 || nclen == (size_t)-2)
706 errc(1, EILSEQ, NULL);
707 if (oclen == 1 && nclen == 1)
708 y->bytetab[(u_char)*op] = *np;
709 else {
710 y->bytetab[(u_char)*op] = 0;
711 y->multis = realloc(y->multis,
712 (y->nmultis + 1) * sizeof(*y->multis));
713 if (y->multis == NULL)
714 err(1, NULL);
715 i = y->nmultis++;
716 y->multis[i].fromlen = oclen;
717 memcpy(y->multis[i].from, op, oclen);
718 y->multis[i].tolen = nclen;
719 memcpy(y->multis[i].to, np, nclen);
720 }
721 op += oclen;
722 np += nclen;
723 }
724 }
725 return (p);
726}
727
728/*
729 * Compile the text following an a or i command.
730 */
731static char *
732compile_text(void)
733{
734 int asize, esc_nl, size;
735 char *text, *p, *op, *s;
736 char lbuf[_POSIX2_LINE_MAX + 1];
737
738 asize = 2 * _POSIX2_LINE_MAX + 1;
739 if ((text = malloc(asize)) == NULL)
740 err(1, "malloc");
741 size = 0;
742 while (cu_fgets(lbuf, sizeof(lbuf), NULL)) {
743 op = s = text + size;
744 p = lbuf;
745 EATSPACE();
746 for (esc_nl = 0; *p != '\0'; p++) {
747 if (*p == '\\' && p[1] != '\0' && *++p == '\n')
748 esc_nl = 1;
749 *s++ = *p;
750 }
751 size += s - op;
752 if (!esc_nl) {
753 *s = '\0';
754 break;
755 }
756 if (asize - size < _POSIX2_LINE_MAX + 1) {
757 asize *= 2;
758 if ((text = realloc(text, asize)) == NULL)
759 err(1, "realloc");
760 }
761 }
762 text[size] = '\0';
763 if ((p = realloc(text, size + 1)) == NULL)
764 err(1, "realloc");
765 return (p);
766}
767
768/*
769 * Get an address and return a pointer to the first character after
770 * it. Fill the structure pointed to according to the address.
771 */
772static char *
773compile_addr(char *p, struct s_addr *a)
774{
775 char *end, re[_POSIX2_LINE_MAX + 1];
776 int icase;
777
778 icase = 0;
779
780 a->type = 0;
781 switch (*p) {
782 case '\\': /* Context address */
783 ++p;
784 /* FALLTHROUGH */
785 case '/': /* Context address */
786 p = compile_delimited(p, re, 0);
787 if (p == NULL)
788 errx(1, "%lu: %s: unterminated regular expression", linenum, fname);
789 /* Check for case insensitive regexp flag */
790 if (*p == 'I') {
791 icase = 1;
792 p++;
793 }
794 if (*re == '\0')
795 a->u.r = NULL;
796 else
797 a->u.r = compile_re(re, icase);
798 a->type = AT_RE;
799 return (p);
800
801 case '$': /* Last line */
802 a->type = AT_LAST;
803 return (p + 1);
804
805 case '+': /* Relative line number */
806 a->type = AT_RELLINE;
807 p++;
808 /* FALLTHROUGH */
809 /* Line number */
810 case '0': case '1': case '2': case '3': case '4':
811 case '5': case '6': case '7': case '8': case '9':
812 if (a->type == 0)
813 a->type = AT_LINE;
814 a->u.l = strtol(p, &end, 10);
815 return (end);
816 default:
817 errx(1, "%lu: %s: expected context address", linenum, fname);
818 return (NULL);
819 }
820}
821
822/*
823 * duptoeol --
824 * Return a copy of all the characters up to \n or \0.
825 */
826static char *
827duptoeol(char *s, const char *ctype)
828{
829 size_t len;
830 int ws;
831 char *p, *start;
832
833 ws = 0;
834 for (start = s; *s != '\0' && *s != '\n'; ++s)
835 ws = isspace((unsigned char)*s);
836 *s = '\0';
837 if (ws)
838 warnx("%lu: %s: whitespace after %s", linenum, fname, ctype);
839 len = s - start + 1;
840 if ((p = malloc(len)) == NULL)
841 err(1, "malloc");
842 return (memmove(p, start, len));
843}
844
845/*
846 * Convert goto label names to addresses, and count a and r commands, in
847 * the given subset of the script. Free the memory used by labels in b
848 * and t commands (but not by :).
849 *
850 * TODO: Remove } nodes
851 */
852static void
853fixuplabel(struct s_command *cp, struct s_command *end)
854{
855
856 for (; cp != end; cp = cp->next)
857 switch (cp->code) {
858 case 'a':
859 case 'r':
860 appendnum++;
861 break;
862 case 'b':
863 case 't':
864 /* Resolve branch target. */
865 if (cp->t == NULL) {
866 cp->u.c = NULL;
867 break;
868 }
869 if ((cp->u.c = findlabel(cp->t)) == NULL)
870 errx(1, "%lu: %s: undefined label '%s'", linenum, fname, cp->t);
871 free(cp->t);
872 break;
873 case '{':
874 /* Do interior commands. */
875 fixuplabel(cp->u.c, cp->next);
876 break;
877 }
878}
879
880/*
881 * Associate the given command label for later lookup.
882 */
883static void
884enterlabel(struct s_command *cp)
885{
886 struct labhash **lhp, *lh;
887 u_char *p;
888 u_int h, c;
889
890 for (h = 0, p = (u_char *)cp->t; (c = *p) != 0; p++)
891 h = (h << 5) + h + c;
892 lhp = &labels[h & LHMASK];
893 for (lh = *lhp; lh != NULL; lh = lh->lh_next)
894 if (lh->lh_hash == h && strcmp(cp->t, lh->lh_cmd->t) == 0)
895 errx(1, "%lu: %s: duplicate label '%s'", linenum, fname, cp->t);
896 if ((lh = malloc(sizeof *lh)) == NULL)
897 err(1, "malloc");
898 lh->lh_next = *lhp;
899 lh->lh_hash = h;
900 lh->lh_cmd = cp;
901 lh->lh_ref = 0;
902 *lhp = lh;
903}
904
905/*
906 * Find the label contained in the command l in the command linked
907 * list cp. L is excluded from the search. Return NULL if not found.
908 */
909static struct s_command *
910findlabel(char *name)
911{
912 struct labhash *lh;
913 u_char *p;
914 u_int h, c;
915
916 for (h = 0, p = (u_char *)name; (c = *p) != 0; p++)
917 h = (h << 5) + h + c;
918 for (lh = labels[h & LHMASK]; lh != NULL; lh = lh->lh_next) {
919 if (lh->lh_hash == h && strcmp(name, lh->lh_cmd->t) == 0) {
920 lh->lh_ref = 1;
921 return (lh->lh_cmd);
922 }
923 }
924 return (NULL);
925}
926
927/*
928 * Warn about any unused labels. As a side effect, release the label hash
929 * table space.
930 */
931static void
932uselabel(void)
933{
934 struct labhash *lh, *next;
935 int i;
936
937 for (i = 0; i < LHSZ; i++) {
938 for (lh = labels[i]; lh != NULL; lh = next) {
939 next = lh->lh_next;
940 if (!lh->lh_ref)
941 warnx("%lu: %s: unused label '%s'",
942 linenum, fname, lh->lh_cmd->t);
943 free(lh);
944 }
945 }
946}
|