1 /* xbt_str.cpp - various helping functions to deal with strings */
3 /* Copyright (c) 2007-2018. The SimGrid Team. All rights reserved. */
5 /* This program is free software; you can redistribute it and/or modify it
6 * under the terms of the license (GNU LGPL) which comes with this package. */
8 #include "simgrid/exception.hpp"
10 #include "xbt/str.h" /* headers of these functions */
11 #include "xbt/sysdep.h"
13 /** @brief Splits a string into a dynar of strings
15 * @param s: the string to split
16 * @param sep: a string of all chars to consider as separator.
18 * By default (with sep=nullptr), these characters are used as separator:
20 * - " " (ASCII 32 (0x20)) space.
21 * - "\t" (ASCII 9 (0x09)) tab.
22 * - "\n" (ASCII 10 (0x0A)) line feed.
23 * - "\r" (ASCII 13 (0x0D)) carriage return.
24 * - "\0" (ASCII 0 (0x00)) nullptr.
25 * - "\x0B" (ASCII 11 (0x0B)) vertical tab.
27 xbt_dynar_t xbt_str_split(const char *s, const char *sep)
29 xbt_dynar_t res = xbt_dynar_new(sizeof(char *), &xbt_free_ref);
30 const char *sep_dflt = " \t\n\r\x0B";
31 char is_sep[256] = { 1, 0 };
33 /* check what are the separators */
34 memset(is_sep, 0, sizeof(is_sep));
37 is_sep[(unsigned char) *sep_dflt++] = 1;
40 is_sep[(unsigned char) *sep++] = 1;
42 is_sep[0] = 1; /* End of string is also separator */
54 while (not is_sep[(unsigned char)*q]) {
60 topush = (char*) xbt_malloc(q - p + 1);
61 memcpy(topush, p, q - p);
63 xbt_dynar_push(res, &topush);
70 /** @brief Just like @ref xbt_str_split_quoted (Splits a string into a dynar of strings), but without memory allocation
72 * The string passed as argument must be writable (not const)
73 * The elements of the dynar are just parts of the string passed as argument.
74 * So if you don't store that argument elsewhere, you should free it in addition to freeing the dynar. This can be done
75 * by simply freeing the first argument of the dynar:
76 * free(xbt_dynar_get_ptr(dynar,0));
78 * Actually this function puts a bunch of \0 in the memory area you passed as argument to separate the elements, and
79 * pushes the address of each chunk in the resulting dynar. Yes, that's uneven. Yes, that's gory. But that's efficient.
81 xbt_dynar_t xbt_str_split_quoted_in_place(char *s) {
82 xbt_dynar_t res = xbt_dynar_new(sizeof(char *), nullptr);
84 char* end; /* pointers around the parsed chunk */
85 int in_simple_quote = 0;
86 int in_double_quote = 0;
88 int ctn = 0; /* Got something in this block */
95 /* do not trim leading spaces: caller responsibility to clean his cruft */
102 /* Protected char; move it closer */
103 memmove(end, end + 1, strlen(end));
105 THROWF(arg_error, 0, "String ends with \\");
106 end++; /* Pass the protected char */
110 if (not in_double_quote) {
111 in_simple_quote = not in_simple_quote;
112 memmove(end, end + 1, strlen(end));
114 /* simple quote protected by double ones */
120 if (not in_simple_quote) {
121 in_double_quote = not in_double_quote;
122 memmove(end, end + 1, strlen(end));
124 /* double quote protected by simple ones */
132 if (*end == '\0' && (in_simple_quote || in_double_quote)) {
133 THROWF(arg_error, 0, "End of string found while searching for %c in %s", (in_simple_quote ? '\'' : '"'), s);
135 if (in_simple_quote || in_double_quote) {
143 /* Found a separator. Push the string if contains something */
144 xbt_dynar_push(res, &beg);
152 /* trim within the string, manually to speed things up */
166 /** @brief Splits a string into a dynar of strings, taking quotes into account
168 * It basically does the same argument separation than the shell, where white spaces can be escaped and where arguments
169 * are never split within a quote group.
170 * Several subsequent spaces are ignored (unless within quotes, of course).
171 * You may want to trim the input string, if you want to avoid empty entries
173 xbt_dynar_t xbt_str_split_quoted(const char *s)
175 xbt_dynar_t res = xbt_dynar_new(sizeof(char *), &xbt_free_ref);
177 char *str_to_free; /* we have to copy the string before, to handle backslashes */
183 str_to_free = xbt_strdup(s);
185 parsed = xbt_str_split_quoted_in_place(str_to_free);
186 xbt_dynar_foreach(parsed,cursor,p) {
187 char *q=xbt_strdup(p);
188 xbt_dynar_push(res,&q);
190 xbt_free(str_to_free);
191 xbt_dynar_shrink(res, 0);
192 xbt_dynar_free(&parsed);
196 /** @brief Join a set of strings as a single string
198 * The parameter must be a nullptr-terminated array of chars,
199 * just like xbt_dynar_to_array() produces
201 char *xbt_str_join_array(const char *const *strs, const char *sep)
203 int amount_strings=0;
206 if ((not strs) || (not strs[0]))
207 return xbt_strdup("");
209 /* compute the length before malloc */
210 for (int i = 0; strs[i]; i++) {
211 len += strlen(strs[i]);
214 len += strlen(sep) * amount_strings;
217 char* res = (char*)xbt_malloc(len);
219 for (int i = 0; strs[i]; i++) {
220 if (i != 0) { // not first loop
221 q += snprintf(q,len, "%s%s", sep, strs[i]);
223 q += snprintf(q,len, "%s",strs[i]);
229 /** @brief Parse an integer out of a string, or raise an error
231 * The @a str is passed as argument to your @a error_msg, as follows:
232 * @verbatim THROWF(arg_error, 0, error_msg, str); @endverbatim
234 long int xbt_str_parse_int(const char* str, const char* error_msg)
237 if (str == nullptr || str[0] == '\0')
238 THROWF(arg_error, 0, error_msg, str);
240 long int res = strtol(str, &endptr, 10);
241 if (endptr[0] != '\0')
242 THROWF(arg_error, 0, error_msg, str);
247 /** @brief Parse a double out of a string, or raise an error
249 * The @a str is passed as argument to your @a error_msg, as follows:
250 * @verbatim THROWF(arg_error, 0, error_msg, str); @endverbatim
252 double xbt_str_parse_double(const char* str, const char* error_msg)
255 if (str == nullptr || str[0] == '\0')
256 THROWF(arg_error, 0, error_msg, str);
258 double res = strtod(str, &endptr);
259 if (endptr[0] != '\0')
260 THROWF(arg_error, 0, error_msg, str);
266 #include "simgrid/exception.hpp"
269 XBT_TEST_SUITE("xbt_str", "String Handling");
271 #define mytest(name, input, expected) \
272 xbt_test_add(name); \
273 a = static_cast<char**>(xbt_dynar_to_array(xbt_str_split_quoted(input))); \
274 s = xbt_str_join_array(a, "XXX"); \
275 xbt_test_assert(not strcmp(s, expected), "Input (%s) leads to (%s) instead of (%s)", input, s, expected); \
277 for (int i = 0; a[i] != nullptr; i++) \
280 XBT_TEST_UNIT("xbt_str_split_quoted", test_split_quoted, "Test the function xbt_str_split_quoted")
285 mytest("Empty", "", "");
286 mytest("Basic test", "toto tutu", "totoXXXtutu");
287 mytest("Useless backslashes", "\\t\\o\\t\\o \\t\\u\\t\\u", "totoXXXtutu");
288 mytest("Protected space", "toto\\ tutu", "toto tutu");
289 mytest("Several spaces", "toto tutu", "totoXXXtutu");
290 mytest("LTriming", " toto tatu", "totoXXXtatu");
291 mytest("Triming", " toto tutu ", "totoXXXtutu");
292 mytest("Single quotes", "'toto tutu' tata", "toto tutuXXXtata");
293 mytest("Double quotes", "\"toto tutu\" tata", "toto tutuXXXtata");
294 mytest("Mixed quotes", "\"toto' 'tutu\" tata", "toto' 'tutuXXXtata");
295 mytest("Backslashed quotes", "\\'toto tutu\\' tata", "'totoXXXtutu'XXXtata");
296 mytest("Backslashed quotes + quotes", "'toto \\'tutu' tata", "toto 'tutuXXXtata");
299 #define test_parse_error(function, name, variable, str) \
301 xbt_test_add(name); \
303 variable = function(str, "Parse error"); \
304 xbt_test_fail("The test '%s' did not detect the problem",name ); \
305 } catch(xbt_ex& e) { \
306 if (e.category != arg_error) { \
307 xbt_test_exception(e); \
311 #define test_parse_ok(function, name, variable, str, value) \
313 xbt_test_add(name); \
315 variable = function(str, "Parse error"); \
316 } catch(xbt_ex& e) { \
317 xbt_test_exception(e); \
319 xbt_test_assert(variable == value, "Fail to parse '%s'", str); \
322 XBT_TEST_UNIT("xbt_str_parse", test_parse, "Test the parsing functions")
325 test_parse_ok(xbt_str_parse_int, "Parse int", rint, "42", 42);
326 test_parse_ok(xbt_str_parse_int, "Parse 0 as an int", rint, "0", 0);
327 test_parse_ok(xbt_str_parse_int, "Parse -1 as an int", rint, "-1", -1);
329 test_parse_error(xbt_str_parse_int, "Parse int + noise", rint, "342 cruft");
330 test_parse_error(xbt_str_parse_int, "Parse nullptr as an int", rint, nullptr);
331 test_parse_error(xbt_str_parse_int, "Parse '' as an int", rint, "");
332 test_parse_error(xbt_str_parse_int, "Parse cruft as an int", rint, "cruft");
334 double rdouble = -9999;
335 test_parse_ok(xbt_str_parse_double, "Parse 42 as a double", rdouble, "42", 42);
336 test_parse_ok(xbt_str_parse_double, "Parse 42.5 as a double", rdouble, "42.5", 42.5);
337 test_parse_ok(xbt_str_parse_double, "Parse 0 as a double", rdouble, "0", 0);
338 test_parse_ok(xbt_str_parse_double, "Parse -1 as a double", rdouble, "-1", -1);
340 test_parse_error(xbt_str_parse_double, "Parse double + noise", rdouble, "342 cruft");
341 test_parse_error(xbt_str_parse_double, "Parse nullptr as a double", rdouble, nullptr);
342 test_parse_error(xbt_str_parse_double, "Parse '' as a double", rdouble, "");
343 test_parse_error(xbt_str_parse_double, "Parse cruft as a double", rdouble, "cruft");
345 #endif /* SIMGRID_TEST */