1 /* xbt_str.cpp - various helping functions to deal with strings */
3 /* Copyright (c) 2007-2018. The SimGrid Team.
4 * All rights reserved. */
6 /* This program is free software; you can redistribute it and/or modify it
7 * under the terms of the license (GNU LGPL) which comes with this package. */
11 #include "xbt/sysdep.h"
12 #include "xbt/str.h" /* headers of these functions */
14 /** @brief Splits a string into a dynar of strings
16 * @param s: the string to split
17 * @param sep: a string of all chars to consider as separator.
19 * By default (with sep=nullptr), these characters are used as separator:
21 * - " " (ASCII 32 (0x20)) space.
22 * - "\t" (ASCII 9 (0x09)) tab.
23 * - "\n" (ASCII 10 (0x0A)) line feed.
24 * - "\r" (ASCII 13 (0x0D)) carriage return.
25 * - "\0" (ASCII 0 (0x00)) nullptr.
26 * - "\x0B" (ASCII 11 (0x0B)) vertical tab.
28 xbt_dynar_t xbt_str_split(const char *s, const char *sep)
30 xbt_dynar_t res = xbt_dynar_new(sizeof(char *), &xbt_free_ref);
31 const char *sep_dflt = " \t\n\r\x0B";
32 char is_sep[256] = { 1, 0 };
34 /* check what are the separators */
35 memset(is_sep, 0, sizeof(is_sep));
38 is_sep[(unsigned char) *sep_dflt++] = 1;
41 is_sep[(unsigned char) *sep++] = 1;
43 is_sep[0] = 1; /* End of string is also separator */
55 while (not is_sep[(unsigned char)*q]) {
61 topush = (char*) xbt_malloc(q - p + 1);
62 memcpy(topush, p, q - p);
64 xbt_dynar_push(res, &topush);
71 /** @brief Just like @ref xbt_str_split_quoted (Splits a string into a dynar of strings), but without memory allocation
73 * The string passed as argument must be writable (not const)
74 * The elements of the dynar are just parts of the string passed as argument.
75 * So if you don't store that argument elsewhere, you should free it in addition to freeing the dynar. This can be done
76 * by simply freeing the first argument of the dynar:
77 * free(xbt_dynar_get_ptr(dynar,0));
79 * Actually this function puts a bunch of \0 in the memory area you passed as argument to separate the elements, and
80 * pushes the address of each chunk in the resulting dynar. Yes, that's uneven. Yes, that's gory. But that's efficient.
82 xbt_dynar_t xbt_str_split_quoted_in_place(char *s) {
83 xbt_dynar_t res = xbt_dynar_new(sizeof(char *), nullptr);
85 char* end; /* pointers around the parsed chunk */
86 int in_simple_quote = 0;
87 int in_double_quote = 0;
89 int ctn = 0; /* Got something in this block */
96 /* do not trim leading spaces: caller responsibility to clean his cruft */
103 /* Protected char; move it closer */
104 memmove(end, end + 1, strlen(end));
106 THROWF(arg_error, 0, "String ends with \\");
107 end++; /* Pass the protected char */
111 if (not in_double_quote) {
112 in_simple_quote = not in_simple_quote;
113 memmove(end, end + 1, strlen(end));
115 /* simple quote protected by double ones */
121 if (not in_simple_quote) {
122 in_double_quote = not in_double_quote;
123 memmove(end, end + 1, strlen(end));
125 /* double quote protected by simple ones */
133 if (*end == '\0' && (in_simple_quote || in_double_quote)) {
134 THROWF(arg_error, 0, "End of string found while searching for %c in %s", (in_simple_quote ? '\'' : '"'), s);
136 if (in_simple_quote || in_double_quote) {
144 /* Found a separator. Push the string if contains something */
145 xbt_dynar_push(res, &beg);
153 /* trim within the string, manually to speed things up */
167 /** @brief Splits a string into a dynar of strings, taking quotes into account
169 * It basically does the same argument separation than the shell, where white spaces can be escaped and where arguments
170 * are never split within a quote group.
171 * Several subsequent spaces are ignored (unless within quotes, of course).
172 * You may want to trim the input string, if you want to avoid empty entries
174 xbt_dynar_t xbt_str_split_quoted(const char *s)
176 xbt_dynar_t res = xbt_dynar_new(sizeof(char *), &xbt_free_ref);
178 char *str_to_free; /* we have to copy the string before, to handle backslashes */
184 str_to_free = xbt_strdup(s);
186 parsed = xbt_str_split_quoted_in_place(str_to_free);
187 xbt_dynar_foreach(parsed,cursor,p) {
188 char *q=xbt_strdup(p);
189 xbt_dynar_push(res,&q);
191 xbt_free(str_to_free);
192 xbt_dynar_shrink(res, 0);
193 xbt_dynar_free(&parsed);
197 /** @brief Join a set of strings as a single string
199 * The parameter must be a nullptr-terminated array of chars,
200 * just like xbt_dynar_to_array() produces
202 char *xbt_str_join_array(const char *const *strs, const char *sep)
204 int amount_strings=0;
207 if ((not strs) || (not strs[0]))
208 return xbt_strdup("");
210 /* compute the length before malloc */
211 for (int i = 0; strs[i]; i++) {
212 len += strlen(strs[i]);
215 len += strlen(sep) * amount_strings;
218 char* res = (char*)xbt_malloc(len);
220 for (int i = 0; strs[i]; i++) {
221 if (i != 0) { // not first loop
222 q += snprintf(q,len, "%s%s", sep, strs[i]);
224 q += snprintf(q,len, "%s",strs[i]);
230 /** @brief Parse an integer out of a string, or raise an error
232 * The @a str is passed as argument to your @a error_msg, as follows:
233 * @verbatim THROWF(arg_error, 0, error_msg, str); @endverbatim
235 long int xbt_str_parse_int(const char* str, const char* error_msg)
238 if (str == nullptr || str[0] == '\0')
239 THROWF(arg_error, 0, error_msg, str);
241 long int res = strtol(str, &endptr, 10);
242 if (endptr[0] != '\0')
243 THROWF(arg_error, 0, error_msg, str);
248 /** @brief Parse a double out of a string, or raise an error
250 * The @a str is passed as argument to your @a error_msg, as follows:
251 * @verbatim THROWF(arg_error, 0, error_msg, str); @endverbatim
253 double xbt_str_parse_double(const char* str, const char* error_msg)
256 if (str == nullptr || str[0] == '\0')
257 THROWF(arg_error, 0, error_msg, str);
259 double res = strtod(str, &endptr);
260 if (endptr[0] != '\0')
261 THROWF(arg_error, 0, error_msg, str);
267 #include <xbt/ex.hpp>
270 XBT_TEST_SUITE("xbt_str", "String Handling");
272 #define mytest(name, input, expected) \
273 xbt_test_add(name); \
274 a = static_cast<char**>(xbt_dynar_to_array(xbt_str_split_quoted(input))); \
275 s = xbt_str_join_array(a, "XXX"); \
276 xbt_test_assert(not strcmp(s, expected), "Input (%s) leads to (%s) instead of (%s)", input, s, expected); \
278 for (int i = 0; a[i] != nullptr; i++) \
281 XBT_TEST_UNIT("xbt_str_split_quoted", test_split_quoted, "Test the function xbt_str_split_quoted")
286 mytest("Empty", "", "");
287 mytest("Basic test", "toto tutu", "totoXXXtutu");
288 mytest("Useless backslashes", "\\t\\o\\t\\o \\t\\u\\t\\u", "totoXXXtutu");
289 mytest("Protected space", "toto\\ tutu", "toto tutu");
290 mytest("Several spaces", "toto tutu", "totoXXXtutu");
291 mytest("LTriming", " toto tatu", "totoXXXtatu");
292 mytest("Triming", " toto tutu ", "totoXXXtutu");
293 mytest("Single quotes", "'toto tutu' tata", "toto tutuXXXtata");
294 mytest("Double quotes", "\"toto tutu\" tata", "toto tutuXXXtata");
295 mytest("Mixed quotes", "\"toto' 'tutu\" tata", "toto' 'tutuXXXtata");
296 mytest("Backslashed quotes", "\\'toto tutu\\' tata", "'totoXXXtutu'XXXtata");
297 mytest("Backslashed quotes + quotes", "'toto \\'tutu' tata", "toto 'tutuXXXtata");
300 #define test_parse_error(function, name, variable, str) \
302 xbt_test_add(name); \
304 variable = function(str, "Parse error"); \
305 xbt_test_fail("The test '%s' did not detect the problem",name ); \
306 } catch(xbt_ex& e) { \
307 if (e.category != arg_error) { \
308 xbt_test_exception(e); \
312 #define test_parse_ok(function, name, variable, str, value) \
314 xbt_test_add(name); \
316 variable = function(str, "Parse error"); \
317 } catch(xbt_ex& e) { \
318 xbt_test_exception(e); \
320 xbt_test_assert(variable == value, "Fail to parse '%s'", str); \
323 XBT_TEST_UNIT("xbt_str_parse", test_parse, "Test the parsing functions")
326 test_parse_ok(xbt_str_parse_int, "Parse int", rint, "42", 42);
327 test_parse_ok(xbt_str_parse_int, "Parse 0 as an int", rint, "0", 0);
328 test_parse_ok(xbt_str_parse_int, "Parse -1 as an int", rint, "-1", -1);
330 test_parse_error(xbt_str_parse_int, "Parse int + noise", rint, "342 cruft");
331 test_parse_error(xbt_str_parse_int, "Parse nullptr as an int", rint, nullptr);
332 test_parse_error(xbt_str_parse_int, "Parse '' as an int", rint, "");
333 test_parse_error(xbt_str_parse_int, "Parse cruft as an int", rint, "cruft");
335 double rdouble = -9999;
336 test_parse_ok(xbt_str_parse_double, "Parse 42 as a double", rdouble, "42", 42);
337 test_parse_ok(xbt_str_parse_double, "Parse 42.5 as a double", rdouble, "42.5", 42.5);
338 test_parse_ok(xbt_str_parse_double, "Parse 0 as a double", rdouble, "0", 0);
339 test_parse_ok(xbt_str_parse_double, "Parse -1 as a double", rdouble, "-1", -1);
341 test_parse_error(xbt_str_parse_double, "Parse double + noise", rdouble, "342 cruft");
342 test_parse_error(xbt_str_parse_double, "Parse nullptr as a double", rdouble, nullptr);
343 test_parse_error(xbt_str_parse_double, "Parse '' as a double", rdouble, "");
344 test_parse_error(xbt_str_parse_double, "Parse cruft as a double", rdouble, "cruft");
346 #endif /* SIMGRID_TEST */