xref: /freebsd/contrib/bmake/str.c (revision 5ca8e32633c4ffbbcd6762e5888b6a4ba0708c6c)
1 /*	$NetBSD: str.c,v 1.102 2024/01/05 23:22:06 rillig Exp $	*/
2 
3 /*
4  * Copyright (c) 1988, 1989, 1990, 1993
5  *	The Regents of the University of California.  All rights reserved.
6  *
7  * This code is derived from software contributed to Berkeley by
8  * Adam de Boor.
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 
35 /*
36  * Copyright (c) 1989 by Berkeley Softworks
37  * All rights reserved.
38  *
39  * This code is derived from software contributed to Berkeley by
40  * Adam de Boor.
41  *
42  * Redistribution and use in source and binary forms, with or without
43  * modification, are permitted provided that the following conditions
44  * are met:
45  * 1. Redistributions of source code must retain the above copyright
46  *    notice, this list of conditions and the following disclaimer.
47  * 2. Redistributions in binary form must reproduce the above copyright
48  *    notice, this list of conditions and the following disclaimer in the
49  *    documentation and/or other materials provided with the distribution.
50  * 3. All advertising materials mentioning features or use of this software
51  *    must display the following acknowledgement:
52  *	This product includes software developed by the University of
53  *	California, Berkeley and its contributors.
54  * 4. Neither the name of the University nor the names of its contributors
55  *    may be used to endorse or promote products derived from this software
56  *    without specific prior written permission.
57  *
58  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
59  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
60  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
61  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
62  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
63  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
64  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
65  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
66  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
67  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
68  * SUCH DAMAGE.
69  */
70 
71 #include "make.h"
72 
73 /*	"@(#)str.c	5.8 (Berkeley) 6/1/90"	*/
74 MAKE_RCSID("$NetBSD: str.c,v 1.102 2024/01/05 23:22:06 rillig Exp $");
75 
76 
77 static HashTable interned_strings;
78 
79 
80 /* Return the concatenation of s1 and s2, freshly allocated. */
81 char *
82 str_concat2(const char *s1, const char *s2)
83 {
84 	size_t len1 = strlen(s1);
85 	size_t len2 = strlen(s2);
86 	char *result = bmake_malloc(len1 + len2 + 1);
87 	memcpy(result, s1, len1);
88 	memcpy(result + len1, s2, len2 + 1);
89 	return result;
90 }
91 
92 /* Return the concatenation of s1, s2 and s3, freshly allocated. */
93 char *
94 str_concat3(const char *s1, const char *s2, const char *s3)
95 {
96 	size_t len1 = strlen(s1);
97 	size_t len2 = strlen(s2);
98 	size_t len3 = strlen(s3);
99 	char *result = bmake_malloc(len1 + len2 + len3 + 1);
100 	memcpy(result, s1, len1);
101 	memcpy(result + len1, s2, len2);
102 	memcpy(result + len1 + len2, s3, len3 + 1);
103 	return result;
104 }
105 
106 /*
107  * Fracture a string into an array of words (as delineated by tabs or spaces)
108  * taking quotation marks into account.
109  *
110  * A string that is empty or only contains whitespace nevertheless results in
111  * a single word.  This is unexpected in many places, and the caller needs to
112  * correct for this edge case.
113  *
114  * If expand is true, quotes are removed and escape sequences such as \r, \t,
115  * etc... are expanded. In this case, return NULL on parse errors.
116  *
117  * Returns the fractured words, which must be freed later using Words_Free,
118  * unless the returned Words.words was NULL.
119  */
120 SubstringWords
121 Substring_Words(const char *str, bool expand)
122 {
123 	size_t str_len;
124 	char *words_buf;
125 	size_t words_cap;
126 	Substring *words;
127 	size_t words_len;
128 	char inquote;
129 	char *word_start;
130 	char *word_end;
131 	const char *str_p;
132 
133 	/* XXX: why only hspace, not whitespace? */
134 	cpp_skip_hspace(&str);	/* skip leading space chars. */
135 
136 	/* words_buf holds the words, separated by '\0'. */
137 	str_len = strlen(str);
138 	words_buf = bmake_malloc(str_len + 1);
139 
140 	words_cap = str_len / 5 > 50 ? str_len / 5 : 50;
141 	words = bmake_malloc((words_cap + 1) * sizeof(words[0]));
142 
143 	/*
144 	 * copy the string; at the same time, parse backslashes,
145 	 * quotes and build the word list.
146 	 */
147 	words_len = 0;
148 	inquote = '\0';
149 	word_start = words_buf;
150 	word_end = words_buf;
151 	for (str_p = str;; str_p++) {
152 		char ch = *str_p;
153 		switch (ch) {
154 		case '"':
155 		case '\'':
156 			if (inquote != '\0') {
157 				if (inquote == ch)
158 					inquote = '\0';
159 				else
160 					break;
161 			} else {
162 				inquote = ch;
163 				/* Don't miss "" or '' */
164 				if (word_start == NULL && str_p[1] == inquote) {
165 					if (!expand) {
166 						word_start = word_end;
167 						*word_end++ = ch;
168 					} else
169 						word_start = word_end + 1;
170 					str_p++;
171 					inquote = '\0';
172 					break;
173 				}
174 			}
175 			if (!expand) {
176 				if (word_start == NULL)
177 					word_start = word_end;
178 				*word_end++ = ch;
179 			}
180 			continue;
181 		case ' ':
182 		case '\t':
183 		case '\n':
184 			if (inquote != '\0')
185 				break;
186 			if (word_start == NULL)
187 				continue;
188 			/* FALLTHROUGH */
189 		case '\0':
190 			/*
191 			 * end of a token -- make sure there's enough words
192 			 * space and save off a pointer.
193 			 */
194 			if (word_start == NULL)
195 				goto done;
196 
197 			*word_end++ = '\0';
198 			if (words_len == words_cap) {
199 				words_cap *= 2;
200 				words = bmake_realloc(words,
201 				    (words_cap + 1) * sizeof(words[0]));
202 			}
203 			words[words_len++] =
204 			    Substring_Init(word_start, word_end - 1);
205 			word_start = NULL;
206 			if (ch == '\n' || ch == '\0') {
207 				if (expand && inquote != '\0') {
208 					SubstringWords res;
209 
210 					free(words);
211 					free(words_buf);
212 
213 					res.words = NULL;
214 					res.len = 0;
215 					res.freeIt = NULL;
216 					return res;
217 				}
218 				goto done;
219 			}
220 			continue;
221 		case '\\':
222 			if (!expand) {
223 				if (word_start == NULL)
224 					word_start = word_end;
225 				*word_end++ = '\\';
226 				/* catch lonely '\' at end of string */
227 				if (str_p[1] == '\0')
228 					continue;
229 				ch = *++str_p;
230 				break;
231 			}
232 
233 			switch (ch = *++str_p) {
234 			case '\0':
235 			case '\n':
236 				/* hmmm; fix it up as best we can */
237 				ch = '\\';
238 				str_p--;
239 				break;
240 			case 'b':
241 				ch = '\b';
242 				break;
243 			case 'f':
244 				ch = '\f';
245 				break;
246 			case 'n':
247 				ch = '\n';
248 				break;
249 			case 'r':
250 				ch = '\r';
251 				break;
252 			case 't':
253 				ch = '\t';
254 				break;
255 			}
256 			break;
257 		}
258 		if (word_start == NULL)
259 			word_start = word_end;
260 		*word_end++ = ch;
261 	}
262 done:
263 	words[words_len] = Substring_Init(NULL, NULL);	/* useful for argv */
264 
265 	{
266 		SubstringWords result;
267 
268 		result.words = words;
269 		result.len = words_len;
270 		result.freeIt = words_buf;
271 		return result;
272 	}
273 }
274 
275 Words
276 Str_Words(const char *str, bool expand)
277 {
278 	SubstringWords swords;
279 	Words words;
280 	size_t i;
281 
282 	swords = Substring_Words(str, expand);
283 	if (swords.words == NULL) {
284 		words.words = NULL;
285 		words.len = 0;
286 		words.freeIt = NULL;
287 		return words;
288 	}
289 
290 	words.words = bmake_malloc((swords.len + 1) * sizeof(words.words[0]));
291 	words.len = swords.len;
292 	words.freeIt = swords.freeIt;
293 	for (i = 0; i < swords.len + 1; i++)
294 		words.words[i] = UNCONST(swords.words[i].start);
295 	free(swords.words);
296 	return words;
297 }
298 
299 /*
300  * XXX: In the extreme edge case that one of the characters is from the basic
301  * execution character set and the other isn't, the result of the comparison
302  * differs depending on whether plain char is signed or unsigned.
303  *
304  * An example is the character range from \xE4 to 'a', where \xE4 may come
305  * from U+00E4 'Latin small letter A with diaeresis'.
306  *
307  * If char is signed, \xE4 evaluates to -28, the first half of the condition
308  * becomes -28 <= '0' && '0' <= 'a', which evaluates to true.
309  *
310  * If char is unsigned, \xE4 evaluates to 228, the second half of the
311  * condition becomes 'a' <= '0' && '0' <= 228, which evaluates to false.
312  */
313 static bool
314 in_range(char e1, char c, char e2)
315 {
316 	return (e1 <= c && c <= e2) || (e2 <= c && c <= e1);
317 }
318 
319 /*
320  * Test if a string matches a pattern like "*.[ch]". The pattern matching
321  * characters are '*', '?' and '[]', as in fnmatch(3).
322  *
323  * See varmod-match.mk for examples and edge cases.
324  */
325 StrMatchResult
326 Str_Match(const char *str, const char *pat)
327 {
328 	StrMatchResult res = { NULL, false };
329 	bool asterisk = false;
330 	const char *fixed_str = str;
331 	const char *fixed_pat = pat;
332 
333 match_fixed_length:
334 	str = fixed_str;
335 	pat = fixed_pat;
336 	for (; *pat != '\0' && *pat != '*'; str++, pat++) {
337 		if (*str == '\0')
338 			return res;
339 
340 		if (*pat == '?')	/* match any single character */
341 			continue;
342 
343 		if (*pat == '[') {	/* match a character from a list */
344 			bool neg = pat[1] == '^';
345 			pat += neg ? 2 : 1;
346 
347 		next_char_in_list:
348 			if (*pat == '\0')
349 				res.error = "Unfinished character list";
350 			if (*pat == ']' || *pat == '\0') {
351 				if (neg)
352 					goto end_of_char_list;
353 				goto no_match;
354 			}
355 			if (*pat == *str)
356 				goto end_of_char_list;
357 			if (pat[1] == '-' && pat[2] == '\0') {
358 				res.error = "Unfinished character range";
359 				res.matched = neg;
360 				return res;
361 			}
362 			if (pat[1] == '-') {
363 				if (in_range(pat[0], *str, pat[2]))
364 					goto end_of_char_list;
365 				pat += 2;
366 			}
367 			pat++;
368 			goto next_char_in_list;
369 
370 		end_of_char_list:
371 			if (neg && *pat != ']' && *pat != '\0')
372 				goto no_match;
373 			while (*pat != ']' && *pat != '\0')
374 				pat++;
375 			if (*pat == '\0')
376 				pat--;
377 			continue;
378 		}
379 
380 		if (*pat == '\\')	/* match the next character exactly */
381 			pat++;
382 		if (*pat != *str) {
383 			if (asterisk && str == fixed_str) {
384 				while (*str != '\0' && *str != *pat)
385 					str++;
386 				fixed_str = str;
387 				goto match_fixed_length;
388 			}
389 			goto no_match;
390 		}
391 	}
392 
393 	if (*pat == '*') {
394 		asterisk = true;
395 		while (*pat == '*')
396 			pat++;
397 		if (*pat == '\0') {
398 			res.matched = true;
399 			return res;
400 		}
401 		fixed_str = str;
402 		fixed_pat = pat;
403 		goto match_fixed_length;
404 	}
405 	if (asterisk && *str != '\0') {
406 		fixed_str += strlen(str);
407 		goto match_fixed_length;
408 	}
409 	res.matched = *str == '\0';
410 	return res;
411 
412 no_match:
413 	if (!asterisk)
414 		return res;
415 	fixed_str++;
416 	goto match_fixed_length;
417 }
418 
419 void
420 Str_Intern_Init(void)
421 {
422 	HashTable_Init(&interned_strings);
423 }
424 
425 void
426 Str_Intern_End(void)
427 {
428 #ifdef CLEANUP
429 	HashTable_Done(&interned_strings);
430 #endif
431 }
432 
433 /* Return a canonical instance of str, with unlimited lifetime. */
434 const char *
435 Str_Intern(const char *str)
436 {
437 	return HashTable_CreateEntry(&interned_strings, str, NULL)->key;
438 }
439