xref: /freebsd/contrib/tcsh/tc.str.c (revision 2e1417489338b971e5fd599ff48b5f65df9e8d3b) 
1 /* $Header: /p/tcsh/cvsroot/tcsh/tc.str.c,v 3.30 2009/06/25 21:27:38 christos Exp $ */
2 /*
3  * tc.str.c: Short string package
4  * 	     This has been a lesson of how to write buggy code!
5  */
6 /*-
7  * Copyright (c) 1980, 1991 The Regents of the University of California.
8  * All rights reserved.
9  *
10  * Redistribution and use in source and binary forms, with or without
11  * modification, are permitted provided that the following conditions
12  * are met:
13  * 1. Redistributions of source code must retain the above copyright
14  *    notice, this list of conditions and the following disclaimer.
15  * 2. Redistributions in binary form must reproduce the above copyright
16  *    notice, this list of conditions and the following disclaimer in the
17  *    documentation and/or other materials provided with the distribution.
18  * 3. Neither the name of the University nor the names of its contributors
19  *    may be used to endorse or promote products derived from this software
20  *    without specific prior written permission.
21  *
22  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
23  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
24  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
25  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
26  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
27  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
28  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
29  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
30  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
31  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
32  * SUCH DAMAGE.
33  */
34 #include "sh.h"
35 
36 #include <limits.h>
37 
38 RCSID("$tcsh: tc.str.c,v 3.30 2009/06/25 21:27:38 christos Exp $")
39 
40 #define MALLOC_INCR	128
41 #ifdef WIDE_STRINGS
42 #define MALLOC_SURPLUS	MB_LEN_MAX /* Space for one multibyte character */
43 #else
44 #define MALLOC_SURPLUS	0
45 #endif
46 
47 #ifdef WIDE_STRINGS
48 size_t
49 one_mbtowc(wchar_t *pwc, const char *s, size_t n)
50 {
51     int len;
52 
53     len = rt_mbtowc(pwc, s, n);
54     if (len == -1) {
55         reset_mbtowc();
56 	*pwc = (unsigned char)*s | INVALID_BYTE;
57     }
58     if (len <= 0)
59 	len = 1;
60     return len;
61 }
62 
63 size_t
64 one_wctomb(char *s, wchar_t wchar)
65 {
66     int len;
67 
68     if (wchar & INVALID_BYTE) {
69 	s[0] = wchar & 0xFF;
70 	len = 1;
71     } else {
72 	len = wctomb(s, wchar);
73 	if (len == -1)
74 	    s[0] = wchar;
75 	if (len <= 0)
76 	    len = 1;
77     }
78     return len;
79 }
80 
81 int
82 rt_mbtowc(wchar_t *pwc, const char *s, size_t n)
83 {
84     int ret;
85     char back[MB_LEN_MAX];
86 
87     ret = mbtowc(pwc, s, n);
88     if (ret > 0 && (wctomb(back, *pwc) != ret || memcmp(s, back, ret) != 0))
89 	ret = -1;
90     return ret;
91 }
92 #endif
93 
94 #ifdef SHORT_STRINGS
95 Char  **
96 blk2short(char **src)
97 {
98     size_t     n;
99     Char **sdst, **dst;
100 
101     /*
102      * Count
103      */
104     for (n = 0; src[n] != NULL; n++)
105 	continue;
106     sdst = dst = xmalloc((n + 1) * sizeof(Char *));
107 
108     for (; *src != NULL; src++)
109 	*dst++ = SAVE(*src);
110     *dst = NULL;
111     return (sdst);
112 }
113 
114 char  **
115 short2blk(Char **src)
116 {
117     size_t     n;
118     char **sdst, **dst;
119 
120     /*
121      * Count
122      */
123     for (n = 0; src[n] != NULL; n++)
124 	continue;
125     sdst = dst = xmalloc((n + 1) * sizeof(char *));
126 
127     for (; *src != NULL; src++)
128 	*dst++ = strsave(short2str(*src));
129     *dst = NULL;
130     return (sdst);
131 }
132 
133 Char   *
134 str2short(const char *src)
135 {
136     static struct Strbuf buf; /* = Strbuf_INIT; */
137 
138     if (src == NULL)
139 	return (NULL);
140 
141     buf.len = 0;
142     while (*src) {
143 	Char wc;
144 
145 	src += one_mbtowc(&wc, src, MB_LEN_MAX);
146 	Strbuf_append1(&buf, wc);
147     }
148     Strbuf_terminate(&buf);
149     return buf.s;
150 }
151 
152 char   *
153 short2str(const Char *src)
154 {
155     static char *sdst = NULL;
156     static size_t dstsize = 0;
157     char *dst, *edst;
158 
159     if (src == NULL)
160 	return (NULL);
161 
162     if (sdst == NULL) {
163 	dstsize = MALLOC_INCR;
164 	sdst = xmalloc((dstsize + MALLOC_SURPLUS) * sizeof(char));
165     }
166     dst = sdst;
167     edst = &dst[dstsize];
168     while (*src) {
169 	dst += one_wctomb(dst, *src & CHAR);
170 	src++;
171 	if (dst >= edst) {
172 	    char *wdst = dst;
173 	    char *wedst = edst;
174 
175 	    dstsize += MALLOC_INCR;
176 	    sdst = xrealloc(sdst, (dstsize + MALLOC_SURPLUS) * sizeof(char));
177 	    edst = &sdst[dstsize];
178 	    dst = &edst[-MALLOC_INCR];
179 	    while (wdst > wedst) {
180 		dst++;
181 		wdst--;
182 	    }
183 	}
184     }
185     *dst = 0;
186     return (sdst);
187 }
188 
189 #ifndef WIDE_STRINGS
190 Char   *
191 s_strcpy(Char *dst, const Char *src)
192 {
193     Char *sdst;
194 
195     sdst = dst;
196     while ((*dst++ = *src++) != '\0')
197 	continue;
198     return (sdst);
199 }
200 
201 Char   *
202 s_strncpy(Char *dst, const Char *src, size_t n)
203 {
204     Char *sdst;
205 
206     if (n == 0)
207 	return(dst);
208 
209     sdst = dst;
210     do
211 	if ((*dst++ = *src++) == '\0') {
212 	    while (--n != 0)
213 		*dst++ = '\0';
214 	    return(sdst);
215 	}
216     while (--n != 0);
217     return (sdst);
218 }
219 
220 Char   *
221 s_strcat(Char *dst, const Char *src)
222 {
223     Strcpy(Strend(dst), src);
224     return dst;
225 }
226 
227 #ifdef NOTUSED
228 Char   *
229 s_strncat(Char *dst, const Char *src, size_t n)
230 {
231     Char *sdst;
232 
233     if (n == 0)
234 	return (dst);
235 
236     sdst = dst;
237 
238     while (*dst)
239 	dst++;
240 
241     do
242 	if ((*dst++ = *src++) == '\0')
243 	    return(sdst);
244     while (--n != 0)
245 	continue;
246 
247     *dst = '\0';
248     return (sdst);
249 }
250 
251 #endif
252 
253 Char   *
254 s_strchr(const Char *str, int ch)
255 {
256     do
257 	if (*str == ch)
258 	    return ((Char *)(intptr_t)str);
259     while (*str++);
260     return (NULL);
261 }
262 
263 Char   *
264 s_strrchr(const Char *str, int ch)
265 {
266     const Char *rstr;
267 
268     rstr = NULL;
269     do
270 	if (*str == ch)
271 	    rstr = str;
272     while (*str++);
273     return ((Char *)(intptr_t)rstr);
274 }
275 
276 size_t
277 s_strlen(const Char *str)
278 {
279     size_t n;
280 
281     for (n = 0; *str++; n++)
282 	continue;
283     return (n);
284 }
285 
286 int
287 s_strcmp(const Char *str1, const Char *str2)
288 {
289     for (; *str1 && *str1 == *str2; str1++, str2++)
290 	continue;
291     /*
292      * The following case analysis is necessary so that characters which look
293      * negative collate low against normal characters but high against the
294      * end-of-string NUL.
295      */
296     if (*str1 == '\0' && *str2 == '\0')
297 	return (0);
298     else if (*str1 == '\0')
299 	return (-1);
300     else if (*str2 == '\0')
301 	return (1);
302     else
303 	return (*str1 - *str2);
304 }
305 
306 int
307 s_strncmp(const Char *str1, const Char *str2, size_t n)
308 {
309     if (n == 0)
310 	return (0);
311     do {
312 	if (*str1 != *str2) {
313 	    /*
314 	     * The following case analysis is necessary so that characters
315 	     * which look negative collate low against normal characters
316 	     * but high against the end-of-string NUL.
317 	     */
318 	    if (*str1 == '\0')
319 		return (-1);
320 	    else if (*str2 == '\0')
321 		return (1);
322 	    else
323 		return (*str1 - *str2);
324 	}
325         if (*str1 == '\0')
326 	    return(0);
327 	str1++, str2++;
328     } while (--n != 0);
329     return(0);
330 }
331 #endif /* not WIDE_STRINGS */
332 
333 int
334 s_strcasecmp(const Char *str1, const Char *str2)
335 {
336 #ifdef WIDE_STRINGS
337     wchar_t l1 = 0, l2 = 0;
338     for (; *str1 && ((*str1 == *str2 && (l1 = l2 = 0) == 0) ||
339 	(l1 = towlower(*str1)) == (l2 = towlower(*str2))); str1++, str2++)
340 	continue;
341 
342 #else
343     unsigned char c1, c2, l1 = 0, l2 = 0;
344     for (; *str1 && ((*str1 == *str2 && (l1 = l2 = 0) == 0) ||
345 	((c1 = (unsigned char)*str1) == *str1 &&
346 	 (c2 = (unsigned char)*str2) == *str2 &&
347 	(l1 = tolower(c1)) == (l2 = tolower(c2)))); str1++, str2++)
348 	continue;
349 #endif
350     /*
351      * The following case analysis is necessary so that characters which look
352      * negative collate low against normal characters but high against the
353      * end-of-string NUL.
354      */
355     if (*str1 == '\0' && *str2 == '\0')
356 	return (0);
357     else if (*str1 == '\0')
358 	return (-1);
359     else if (*str2 == '\0')
360 	return (1);
361     else if (l1 == l2)	/* They are zero when they are equal */
362 	return (*str1 - *str2);
363     else
364 	return (l1 - l2);
365 }
366 
367 Char   *
368 s_strnsave(const Char *s, size_t len)
369 {
370     Char *n;
371 
372     n = xmalloc((len + 1) * sizeof (*n));
373     memcpy(n, s, len * sizeof (*n));
374     n[len] = '\0';
375     return n;
376 }
377 
378 Char   *
379 s_strsave(const Char *s)
380 {
381     Char   *n;
382     size_t size;
383 
384     if (s == NULL)
385 	s = STRNULL;
386     size = (Strlen(s) + 1) * sizeof(*n);
387     n = xmalloc(size);
388     memcpy(n, s, size);
389     return (n);
390 }
391 
392 Char   *
393 s_strspl(const Char *cp, const Char *dp)
394 {
395     Char *res, *ep;
396     const Char *p, *q;
397 
398     if (!cp)
399 	cp = STRNULL;
400     if (!dp)
401 	dp = STRNULL;
402     for (p = cp; *p++;)
403 	continue;
404     for (q = dp; *q++;)
405 	continue;
406     res = xmalloc(((p - cp) + (q - dp) - 1) * sizeof(Char));
407     for (ep = res, q = cp; (*ep++ = *q++) != '\0';)
408 	continue;
409     for (ep--, q = dp; (*ep++ = *q++) != '\0';)
410 	continue;
411     return (res);
412 }
413 
414 Char   *
415 s_strend(const Char *cp)
416 {
417     if (!cp)
418 	return ((Char *)(intptr_t) cp);
419     while (*cp)
420 	cp++;
421     return ((Char *)(intptr_t) cp);
422 }
423 
424 Char   *
425 s_strstr(const Char *s, const Char *t)
426 {
427     do {
428 	const Char *ss = s;
429 	const Char *tt = t;
430 
431 	do
432 	    if (*tt == '\0')
433 		return ((Char *)(intptr_t) s);
434 	while (*ss++ == *tt++);
435     } while (*s++ != '\0');
436     return (NULL);
437 }
438 
439 #else /* !SHORT_STRINGS */
440 char *
441 caching_strip(const char *s)
442 {
443     static char *buf = NULL;
444     static size_t buf_size = 0;
445     size_t size;
446 
447     if (s == NULL)
448       return NULL;
449     size = strlen(s) + 1;
450     if (buf_size < size) {
451 	buf = xrealloc(buf, size);
452 	buf_size = size;
453     }
454     memcpy(buf, s, size);
455     strip(buf);
456     return buf;
457 }
458 #endif
459 
460 char   *
461 short2qstr(const Char *src)
462 {
463     static char *sdst = NULL;
464     static size_t dstsize = 0;
465     char *dst, *edst;
466 
467     if (src == NULL)
468 	return (NULL);
469 
470     if (sdst == NULL) {
471 	dstsize = MALLOC_INCR;
472 	sdst = xmalloc((dstsize + MALLOC_SURPLUS) * sizeof(char));
473     }
474     dst = sdst;
475     edst = &dst[dstsize];
476     while (*src) {
477 	if (*src & QUOTE) {
478 	    *dst++ = '\\';
479 	    if (dst == edst) {
480 		dstsize += MALLOC_INCR;
481 		sdst = xrealloc(sdst,
482 				(dstsize + MALLOC_SURPLUS) * sizeof(char));
483 		edst = &sdst[dstsize];
484 		dst = &edst[-MALLOC_INCR];
485 	    }
486 	}
487 	dst += one_wctomb(dst, *src & CHAR);
488 	src++;
489 	if (dst >= edst) {
490 	    ptrdiff_t i = dst - edst;
491 	    dstsize += MALLOC_INCR;
492 	    sdst = xrealloc(sdst, (dstsize + MALLOC_SURPLUS) * sizeof(char));
493 	    edst = &sdst[dstsize];
494 	    dst = &edst[-MALLOC_INCR + i];
495 	}
496     }
497     *dst = 0;
498     return (sdst);
499 }
500 
501 struct blk_buf *
502 bb_alloc()
503 {
504     return xcalloc(1, sizeof(struct blk_buf));
505 }
506 
507 static void
508 bb_store(struct blk_buf *bb, Char *str)
509 {
510     if (bb->len == bb->size) { /* Keep space for terminating NULL */
511 	if (bb->size == 0)
512 	    bb->size = 16; /* Arbitrary */
513 	else
514 	    bb->size *= 2;
515 	bb->vec = xrealloc(bb->vec, bb->size * sizeof (*bb->vec));
516     }
517     bb->vec[bb->len] = str;
518 }
519 
520 void
521 bb_append(struct blk_buf *bb, Char *str)
522 {
523     bb_store(bb, str);
524     bb->len++;
525 }
526 
527 void
528 bb_cleanup(void *xbb)
529 {
530     struct blk_buf *bb;
531     size_t i;
532 
533     bb = xbb;
534     for (i = 0; i < bb->len; i++)
535 	xfree(bb->vec[i]);
536     xfree(bb->vec);
537 }
538 
539 void
540 bb_free(void *bb)
541 {
542     bb_cleanup(bb);
543     xfree(bb);
544 }
545 
546 Char **
547 bb_finish(struct blk_buf *bb)
548 {
549     bb_store(bb, NULL);
550     return xrealloc(bb->vec, (bb->len + 1) * sizeof (*bb->vec));
551 }
552 
553 #define DO_STRBUF(STRBUF, CHAR, STRLEN)				\
554 								\
555 struct STRBUF *							\
556 STRBUF##_alloc(void)						\
557 {								\
558     return xcalloc(1, sizeof(struct STRBUF));			\
559 }								\
560 								\
561 static void							\
562 STRBUF##_store1(struct STRBUF *buf, CHAR c)			\
563 {								\
564     if (buf->size == buf->len) {				\
565 	if (buf->size == 0)					\
566 	    buf->size = 64; /* Arbitrary */			\
567 	else							\
568 	    buf->size *= 2;					\
569 	buf->s = xrealloc(buf->s, buf->size * sizeof(*buf->s));	\
570     }								\
571     buf->s[buf->len] = c;					\
572 }								\
573 								\
574 /* Like strbuf_append1(buf, '\0'), but don't advance len */	\
575 void								\
576 STRBUF##_terminate(struct STRBUF *buf)				\
577 {								\
578     STRBUF##_store1(buf, '\0');					\
579 }								\
580 								\
581 void								\
582 STRBUF##_append1(struct STRBUF *buf, CHAR c)			\
583 {								\
584     STRBUF##_store1(buf, c);					\
585     buf->len++;							\
586 }								\
587 								\
588 void								\
589 STRBUF##_appendn(struct STRBUF *buf, const CHAR *s, size_t len)	\
590 {								\
591     if (buf->size < buf->len + len) {				\
592 	if (buf->size == 0)					\
593 	    buf->size = 64; /* Arbitrary */			\
594 	while (buf->size < buf->len + len)			\
595 	    buf->size *= 2;					\
596 	buf->s = xrealloc(buf->s, buf->size * sizeof(*buf->s));	\
597     }								\
598     memcpy(buf->s + buf->len, s, len * sizeof(*buf->s));	\
599     buf->len += len;						\
600 }								\
601 								\
602 void								\
603 STRBUF##_append(struct STRBUF *buf, const CHAR *s)		\
604 {								\
605     STRBUF##_appendn(buf, s, STRLEN(s));			\
606 }								\
607 								\
608 CHAR *								\
609 STRBUF##_finish(struct STRBUF *buf)				\
610 {								\
611     STRBUF##_append1(buf, 0);					\
612     return xrealloc(buf->s, buf->len * sizeof(*buf->s));	\
613 }								\
614 								\
615 void								\
616 STRBUF##_cleanup(void *xbuf)					\
617 {								\
618     struct STRBUF *buf;						\
619 								\
620     buf = xbuf;							\
621     xfree(buf->s);						\
622 }								\
623 								\
624 void								\
625 STRBUF##_free(void *xbuf)					\
626 {								\
627     STRBUF##_cleanup(xbuf);					\
628     xfree(xbuf);						\
629 }								\
630 								\
631 const struct STRBUF STRBUF##_init /* = STRBUF##_INIT; */
632 
633 DO_STRBUF(strbuf, char, strlen);
634 DO_STRBUF(Strbuf, Char, Strlen);
635