Commit | Line | Data |
---|---|---|
9c55c241 | 1 | /* |
ab5be9fa | 2 | * Copyright (C) 2017 Philippe Proulx <pproulx@efficios.com> |
9c55c241 | 3 | * |
ab5be9fa | 4 | * SPDX-License-Identifier: GPL-2.0-only |
9c55c241 | 5 | * |
9c55c241 PP |
6 | */ |
7 | ||
8 | #define _LGPL_SOURCE | |
9 | #include <stdlib.h> | |
10 | #include <string.h> | |
11 | #include <stdbool.h> | |
9c55c241 PP |
12 | |
13 | #include "string-utils.h" | |
14 | #include "../macros.h" | |
15 | ||
16 | enum star_glob_pattern_type_flags { | |
17 | STAR_GLOB_PATTERN_TYPE_FLAG_NONE = 0, | |
18 | STAR_GLOB_PATTERN_TYPE_FLAG_PATTERN = 1, | |
19 | STAR_GLOB_PATTERN_TYPE_FLAG_END_ONLY = 2, | |
20 | }; | |
21 | ||
22 | /* | |
23 | * Normalizes the star-only globbing pattern `pattern`, that is, crushes | |
24 | * consecutive `*` characters into a single `*`, avoiding `\*`. | |
25 | */ | |
9c55c241 PP |
26 | void strutils_normalize_star_glob_pattern(char *pattern) |
27 | { | |
28 | const char *p; | |
29 | char *np; | |
30 | bool got_star = false; | |
31 | ||
a0377dfe | 32 | LTTNG_ASSERT(pattern); |
9c55c241 PP |
33 | |
34 | for (p = pattern, np = pattern; *p != '\0'; p++) { | |
35 | switch (*p) { | |
36 | case '*': | |
37 | if (got_star) { | |
38 | /* Avoid consecutive stars. */ | |
39 | continue; | |
40 | } | |
41 | ||
42 | got_star = true; | |
43 | break; | |
44 | case '\\': | |
45 | /* Copy backslash character. */ | |
46 | *np = *p; | |
47 | np++; | |
48 | p++; | |
49 | ||
50 | if (*p == '\0') { | |
51 | goto end; | |
52 | } | |
53 | ||
54 | /* Fall through default case. */ | |
55 | default: | |
56 | got_star = false; | |
57 | break; | |
58 | } | |
59 | ||
60 | /* Copy single character. */ | |
61 | *np = *p; | |
62 | np++; | |
63 | } | |
64 | ||
65 | end: | |
66 | *np = '\0'; | |
67 | } | |
68 | ||
69 | static | |
70 | enum star_glob_pattern_type_flags strutils_test_glob_pattern(const char *pattern) | |
71 | { | |
72 | enum star_glob_pattern_type_flags ret = | |
73 | STAR_GLOB_PATTERN_TYPE_FLAG_NONE; | |
74 | const char *p; | |
75 | ||
a0377dfe | 76 | LTTNG_ASSERT(pattern); |
9c55c241 PP |
77 | |
78 | for (p = pattern; *p != '\0'; p++) { | |
79 | switch (*p) { | |
80 | case '*': | |
81 | ret = STAR_GLOB_PATTERN_TYPE_FLAG_PATTERN; | |
82 | ||
83 | if (p[1] == '\0') { | |
84 | ret |= STAR_GLOB_PATTERN_TYPE_FLAG_END_ONLY; | |
85 | } | |
86 | ||
87 | goto end; | |
88 | case '\\': | |
89 | p++; | |
90 | ||
91 | if (*p == '\0') { | |
92 | goto end; | |
93 | } | |
94 | break; | |
95 | default: | |
96 | break; | |
97 | } | |
98 | } | |
99 | ||
100 | end: | |
101 | return ret; | |
102 | } | |
103 | ||
104 | /* | |
105 | * Returns true if `pattern` is a star-only globbing pattern, that is, | |
106 | * it contains at least one non-escaped `*`. | |
107 | */ | |
108 | bool strutils_is_star_glob_pattern(const char *pattern) | |
109 | { | |
110 | return strutils_test_glob_pattern(pattern) & | |
111 | STAR_GLOB_PATTERN_TYPE_FLAG_PATTERN; | |
112 | } | |
113 | ||
114 | /* | |
115 | * Returns true if `pattern` is a globbing pattern with a globbing, | |
116 | * non-escaped star only at its very end. | |
117 | */ | |
118 | bool strutils_is_star_at_the_end_only_glob_pattern(const char *pattern) | |
119 | { | |
120 | return strutils_test_glob_pattern(pattern) & | |
121 | STAR_GLOB_PATTERN_TYPE_FLAG_END_ONLY; | |
122 | } | |
123 | ||
124 | /* | |
125 | * Unescapes the input string `input`, that is, in a `\x` sequence, | |
126 | * removes `\`. If `only_char` is not 0, only this character is | |
127 | * escaped. | |
128 | */ | |
9c55c241 PP |
129 | char *strutils_unescape_string(const char *input, char only_char) |
130 | { | |
131 | char *output; | |
132 | char *o; | |
133 | const char *i; | |
134 | ||
a0377dfe | 135 | LTTNG_ASSERT(input); |
9c55c241 PP |
136 | output = zmalloc(strlen(input) + 1); |
137 | if (!output) { | |
138 | goto end; | |
139 | } | |
140 | ||
141 | for (i = input, o = output; *i != '\0'; i++) { | |
142 | switch (*i) { | |
143 | case '\\': | |
144 | if (only_char && i[1] != only_char) { | |
145 | break; | |
146 | } | |
147 | ||
148 | i++; | |
149 | ||
150 | if (*i == '\0') { | |
151 | /* Copy last `\`. */ | |
152 | *o = '\\'; | |
153 | o++; | |
154 | goto end; | |
155 | } | |
156 | default: | |
157 | break; | |
158 | } | |
159 | ||
160 | /* Copy single character. */ | |
161 | *o = *i; | |
162 | o++; | |
163 | } | |
164 | ||
165 | end: | |
166 | return output; | |
167 | } | |
168 | ||
169 | /* | |
170 | * Frees a null-terminated array of strings, including each contained | |
171 | * string. | |
172 | */ | |
9c55c241 PP |
173 | void strutils_free_null_terminated_array_of_strings(char **array) |
174 | { | |
175 | char **item; | |
176 | ||
177 | if (!array) { | |
178 | return; | |
179 | } | |
180 | ||
181 | for (item = array; *item; item++) { | |
182 | free(*item); | |
183 | } | |
184 | ||
185 | free(array); | |
186 | } | |
187 | ||
188 | /* | |
189 | * Splits the input string `input` using the given delimiter `delim`. | |
190 | * | |
e358ddd5 JG |
191 | * The return value is a dynamic pointer array that is assumed to be empty. The |
192 | * array must be discarded by the caller by invoking | |
193 | * lttng_dynamic_pointer_array_reset(). | |
9c55c241 PP |
194 | * |
195 | * Empty substrings are part of the result. For example: | |
196 | * | |
197 | * Input: ,hello,,there, | |
198 | * Result: | |
199 | * `` | |
200 | * `hello` | |
201 | * `` | |
202 | * `there` | |
203 | * `` | |
204 | * | |
205 | * If `escape_delim` is true, then `\,`, where `,` is the delimiter, | |
206 | * escapes the delimiter and is copied as `,` only in the resulting | |
207 | * substring. For example: | |
208 | * | |
209 | * Input: hello\,world,zoom,\,hi | |
210 | * Result: | |
211 | * `hello,world` | |
212 | * `zoom` | |
213 | * `,hi` | |
214 | * | |
215 | * Other characters are not escaped (this is the caller's job if | |
216 | * needed). However they are considering during the parsing, that is, | |
217 | * `\x`, where `x` is any character, is copied as is to the resulting | |
218 | * substring, e.g.: | |
219 | * | |
220 | * Input: hello\,wo\rld\\,zoom\, | |
221 | * Result: | |
222 | * `hello,wo\rld\\` | |
223 | * `zoom,` | |
224 | * | |
225 | * If `escape_delim` is false, nothing at all is escaped, and `delim`, | |
226 | * when found in `input`, is always a delimiter, e.g.: | |
227 | * | |
228 | * Input: hello\,world,zoom,\,hi | |
229 | * Result: | |
230 | * `hello\` | |
231 | * `world` | |
232 | * `zoom` | |
233 | * `\` | |
234 | * `hi` | |
235 | * | |
e358ddd5 | 236 | * Returns -1 if there's an error. |
9c55c241 | 237 | */ |
e358ddd5 JG |
238 | int strutils_split(const char *input, |
239 | char delim, | |
240 | bool escape_delim, | |
241 | struct lttng_dynamic_pointer_array *out_strings) | |
9c55c241 | 242 | { |
e358ddd5 | 243 | int ret; |
9c55c241 PP |
244 | size_t at; |
245 | size_t number_of_substrings = 1; | |
246 | size_t longest_substring_len = 0; | |
247 | const char *s; | |
248 | const char *last; | |
9c55c241 | 249 | |
a0377dfe FD |
250 | LTTNG_ASSERT(input); |
251 | LTTNG_ASSERT(!(escape_delim && delim == '\\')); | |
252 | LTTNG_ASSERT(delim != '\0'); | |
e358ddd5 | 253 | lttng_dynamic_pointer_array_init(out_strings, free); |
9c55c241 PP |
254 | |
255 | /* First pass: count the number of substrings. */ | |
256 | for (s = input, last = input - 1; *s != '\0'; s++) { | |
257 | if (escape_delim && *s == '\\') { | |
258 | /* Ignore following (escaped) character. */ | |
259 | s++; | |
260 | ||
261 | if (*s == '\0') { | |
262 | break; | |
263 | } | |
264 | ||
265 | continue; | |
266 | } | |
267 | ||
268 | if (*s == delim) { | |
269 | size_t last_len = s - last - 1; | |
270 | last = s; | |
271 | number_of_substrings++; | |
272 | ||
273 | if (last_len > longest_substring_len) { | |
274 | longest_substring_len = last_len; | |
275 | } | |
276 | } | |
277 | } | |
278 | ||
279 | if ((s - last - 1) > longest_substring_len) { | |
280 | longest_substring_len = s - last - 1; | |
281 | } | |
282 | ||
9c55c241 PP |
283 | /* Second pass: actually split and copy substrings. */ |
284 | for (at = 0, s = input; at < number_of_substrings; at++) { | |
285 | const char *ss; | |
286 | char *d; | |
e358ddd5 JG |
287 | char *substring = zmalloc(longest_substring_len + 1); |
288 | ||
289 | if (!substring) { | |
290 | goto error; | |
291 | } | |
9c55c241 | 292 | |
e358ddd5 JG |
293 | ret = lttng_dynamic_pointer_array_add_pointer( |
294 | out_strings, substring); | |
295 | if (ret) { | |
296 | free(substring); | |
9c55c241 PP |
297 | goto error; |
298 | } | |
299 | ||
300 | /* | |
301 | * Copy characters to substring until we find the next | |
302 | * delimiter or the end of the input string. | |
303 | */ | |
e358ddd5 | 304 | for (ss = s, d = substring; *ss != '\0'; ss++) { |
9c55c241 PP |
305 | if (escape_delim && *ss == '\\') { |
306 | if (ss[1] == delim) { | |
307 | /* | |
308 | * '\' followed by delimiter and | |
309 | * we need to escape this ('\' | |
310 | * won't be part of the | |
311 | * resulting substring). | |
312 | */ | |
313 | ss++; | |
314 | *d = *ss; | |
315 | d++; | |
316 | continue; | |
317 | } else { | |
318 | /* | |
319 | * Copy '\' and the following | |
320 | * character. | |
321 | */ | |
322 | *d = *ss; | |
323 | d++; | |
324 | ss++; | |
325 | ||
326 | if (*ss == '\0') { | |
327 | break; | |
328 | } | |
329 | } | |
330 | } else if (*ss == delim) { | |
331 | /* We're done with this substring. */ | |
332 | break; | |
333 | } | |
334 | ||
335 | *d = *ss; | |
336 | d++; | |
337 | } | |
338 | ||
339 | /* Next substring starts after the last delimiter. */ | |
340 | s = ss + 1; | |
341 | } | |
342 | ||
e358ddd5 | 343 | ret = 0; |
9c55c241 PP |
344 | goto end; |
345 | ||
346 | error: | |
e358ddd5 | 347 | ret = -1; |
9c55c241 | 348 | end: |
e358ddd5 | 349 | return ret; |
9c55c241 PP |
350 | } |
351 | ||
9c55c241 PP |
352 | size_t strutils_array_of_strings_len(char * const *array) |
353 | { | |
354 | char * const *item; | |
355 | size_t count = 0; | |
356 | ||
a0377dfe | 357 | LTTNG_ASSERT(array); |
9c55c241 PP |
358 | |
359 | for (item = array; *item; item++) { | |
360 | count++; | |
361 | } | |
362 | ||
363 | return count; | |
364 | } |