2 This file is part of GNUnet.
3 (C) 2005-2013 Christian Grothoff (and other contributing authors)
5 GNUnet is free software; you can redistribute it and/or modify
6 it under the terms of the GNU General Public License as published
7 by the Free Software Foundation; either version 3, or (at your
8 option) any later version.
10 GNUnet is distributed in the hope that it will be useful, but
11 WITHOUT ANY WARRANTY; without even the implied warranty of
12 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 General Public License for more details.
15 You should have received a copy of the GNU General Public License
16 along with GNUnet; see the file COPYING. If not, write to the
17 Free Software Foundation, Inc., 59 Temple Place - Suite 330,
18 Boston, MA 02111-1307, USA.
22 * @file util/strings.c
23 * @brief string functions
25 * @author Christian Grothoff
32 #include "gnunet_util_lib.h"
37 #define LOG(kind,...) GNUNET_log_from (kind, "util", __VA_ARGS__)
39 #define LOG_STRERROR(kind,syscall) GNUNET_log_from_strerror (kind, "util", syscall)
43 * Fill a buffer of the given size with
44 * count 0-terminated strings (given as varargs).
45 * If "buffer" is NULL, only compute the amount of
46 * space required (sum of "strlen(arg)+1").
48 * Unlike using "snprintf" with "%s", this function
49 * will add 0-terminators after each string. The
50 * "GNUNET_string_buffer_tokenize" function can be
51 * used to parse the buffer back into individual
54 * @param buffer the buffer to fill with strings, can
55 * be NULL in which case only the necessary
56 * amount of space will be calculated
57 * @param size number of bytes available in buffer
58 * @param count number of strings that follow
59 * @param ... count 0-terminated strings to copy to buffer
60 * @return number of bytes written to the buffer
61 * (or number of bytes that would have been written)
64 GNUNET_STRINGS_buffer_fill (char *buffer, size_t size, unsigned int count, ...)
75 s = va_arg (ap, const char *);
77 slen = strlen (s) + 1;
80 GNUNET_assert (needed + slen <= size);
81 memcpy (&buffer[needed], s, slen);
92 * Given a buffer of a given size, find "count"
93 * 0-terminated strings in the buffer and assign
94 * the count (varargs) of type "const char**" to the
95 * locations of the respective strings in the
98 * @param buffer the buffer to parse
99 * @param size size of the buffer
100 * @param count number of strings to locate
101 * @return offset of the character after the last 0-termination
102 * in the buffer, or 0 on error.
105 GNUNET_STRINGS_buffer_tokenize (const char *buffer, size_t size,
106 unsigned int count, ...)
114 va_start (ap, count);
117 r = va_arg (ap, const char **);
120 while ((needed < size) && (buffer[needed] != '\0'))
125 return 0; /* error */
128 needed++; /* skip 0-termination */
137 * Convert a given filesize into a fancy human-readable format.
139 * @param size number of bytes
140 * @return fancy representation of the size (possibly rounded) for humans
143 GNUNET_STRINGS_byte_size_fancy (unsigned long long size)
145 const char *unit = _( /* size unit */ "b");
168 ret = GNUNET_malloc (32);
169 GNUNET_snprintf (ret, 32, "%llu %s", size, unit);
175 * Unit conversion table entry for 'convert_with_table'.
177 struct ConversionTable
180 * Name of the unit (or NULL for end of table).
185 * Factor to apply for this unit.
187 unsigned long long value;
192 * Convert a string of the form "4 X 5 Y" into a numeric value
193 * by interpreting "X" and "Y" as units and then multiplying
194 * the numbers with the values associated with the respective
195 * unit from the conversion table.
197 * @param input input string to parse
198 * @param table table with the conversion of unit names to numbers
199 * @param output where to store the result
200 * @return #GNUNET_OK on success, #GNUNET_SYSERR on error
203 convert_with_table (const char *input,
204 const struct ConversionTable *table,
205 unsigned long long *output)
207 unsigned long long ret;
210 unsigned long long last;
215 in = GNUNET_strdup (input);
216 for (tok = strtok (in, " "); tok != NULL; tok = strtok (NULL, " "))
221 while ((table[i].name != NULL) && (0 != strcasecmp (table[i].name, tok)))
223 if (table[i].name != NULL)
225 last *= table[i].value;
226 break; /* next tok */
233 last = strtoull (tok, &endptr, 10);
234 if ((0 != errno) || (endptr == tok))
237 return GNUNET_SYSERR; /* expected number */
239 if ('\0' == endptr[0])
240 break; /* next tok */
242 tok = endptr; /* and re-check (handles times like "10s") */
244 } while (GNUNET_YES);
254 * Convert a given fancy human-readable size to bytes.
256 * @param fancy_size human readable string (i.e. 1 MB)
257 * @param size set to the size in bytes
258 * @return #GNUNET_OK on success, #GNUNET_SYSERR on error
261 GNUNET_STRINGS_fancy_size_to_bytes (const char *fancy_size,
262 unsigned long long *size)
264 static const struct ConversionTable table[] =
269 { "MiB", 1024 * 1024},
270 { "MB", 1000 * 1000},
271 { "GiB", 1024 * 1024 * 1024},
272 { "GB", 1000 * 1000 * 1000},
273 { "TiB", 1024LL * 1024LL * 1024LL * 1024LL},
274 { "TB", 1000LL * 1000LL * 1000LL * 1024LL},
275 { "PiB", 1024LL * 1024LL * 1024LL * 1024LL * 1024LL},
276 { "PB", 1000LL * 1000LL * 1000LL * 1024LL * 1000LL},
277 { "EiB", 1024LL * 1024LL * 1024LL * 1024LL * 1024LL * 1024LL},
278 { "EB", 1000LL * 1000LL * 1000LL * 1024LL * 1000LL * 1000LL},
282 return convert_with_table (fancy_size,
289 * Convert a given fancy human-readable time to our internal
292 * @param fancy_time human readable string (i.e. 1 minute)
293 * @param rtime set to the relative time
294 * @return #GNUNET_OK on success, #GNUNET_SYSERR on error
297 GNUNET_STRINGS_fancy_time_to_relative (const char *fancy_time,
298 struct GNUNET_TIME_Relative *rtime)
300 static const struct ConversionTable table[] =
304 { "s", 1000 * 1000LL },
305 { "\"", 1000 * 1000LL },
306 { "m", 60 * 1000 * 1000LL},
307 { "min", 60 * 1000 * 1000LL},
308 { "minutes", 60 * 1000 * 1000LL},
309 { "'", 60 * 1000 * 1000LL},
310 { "h", 60 * 60 * 1000 * 1000LL},
311 { "d", 24 * 60 * 60 * 1000LL * 1000LL},
312 { "day", 24 * 60 * 60 * 1000LL * 1000LL},
313 { "days", 24 * 60 * 60 * 1000LL * 1000LL},
314 { "a", 31536000000000LL /* year */ },
318 unsigned long long val;
320 if (0 == strcasecmp ("forever", fancy_time))
322 *rtime = GNUNET_TIME_UNIT_FOREVER_REL;
325 ret = convert_with_table (fancy_time,
328 rtime->rel_value_us = (uint64_t) val;
334 * Convert a given fancy human-readable time to our internal
335 * representation. The human-readable time is expected to be
336 * in local time, whereas the returned value will be in UTC.
338 * @param fancy_time human readable string (i.e. %Y-%m-%d %H:%M:%S)
339 * @param atime set to the absolute time
340 * @return #GNUNET_OK on success, #GNUNET_SYSERR on error
343 GNUNET_STRINGS_fancy_time_to_absolute (const char *fancy_time,
344 struct GNUNET_TIME_Absolute *atime)
352 if (0 == strcasecmp ("end of time", fancy_time))
354 *atime = GNUNET_TIME_UNIT_FOREVER_ABS;
357 memset (&tv, 0, sizeof (tv));
358 if ( (NULL == strptime (fancy_time, "%a %b %d %H:%M:%S %Y", &tv)) &&
359 (NULL == strptime (fancy_time, "%c", &tv)) &&
360 (NULL == strptime (fancy_time, "%Ec", &tv)) &&
361 (NULL == strptime (fancy_time, "%Y-%m-%d %H:%M:%S", &tv)) &&
362 (NULL == strptime (fancy_time, "%Y-%m-%d %H:%M", &tv)) &&
363 (NULL == strptime (fancy_time, "%x", &tv)) &&
364 (NULL == strptime (fancy_time, "%Ex", &tv)) &&
365 (NULL == strptime (fancy_time, "%Y-%m-%d", &tv)) &&
366 (NULL == strptime (fancy_time, "%Y-%m", &tv)) &&
367 (NULL == strptime (fancy_time, "%Y", &tv)) )
368 return GNUNET_SYSERR;
370 atime->abs_value_us = (uint64_t) ((uint64_t) t * 1000LL * 1000LL);
373 atime->abs_value_us += 1000LL * 1000LL * tp->tm_gmtoff;
375 atime->abs_value_us -= 1000LL * 1000LL * timezone;
376 #elif defined WINDOWS
379 TIME_ZONE_INFORMATION tzi;
380 tzv = GetTimeZoneInformation (&tzi);
381 if (TIME_ZONE_ID_INVALID != tzv)
383 atime->abs_value_us -= 1000LL * 1000LL * tzi.Bias * 60LL;
392 * Convert the len characters long character sequence
393 * given in input that is in the given input charset
394 * to a string in given output charset.
396 * @param input input string
397 * @param len number of bytes in @a input
398 * @param input_charset character set used for @a input
399 * @param output_charset desired character set for the return value
400 * @return the converted string (0-terminated),
401 * if conversion fails, a copy of the orignal
402 * string is returned.
405 GNUNET_STRINGS_conv (const char *input,
407 const char *input_charset,
408 const char *output_charset)
412 char *encoded_string;
413 size_t u8_string_length;
414 size_t encoded_string_length;
416 u8_string = u8_conv_from_encoding (input_charset,
421 if (NULL == u8_string)
423 LOG_STRERROR (GNUNET_ERROR_TYPE_WARNING, "u8_conv_from_encoding");
426 if (0 == strcmp (output_charset, "UTF-8"))
428 ret = GNUNET_malloc (u8_string_length + 1);
429 memcpy (ret, u8_string, u8_string_length);
430 ret[u8_string_length] = '\0';
434 encoded_string = u8_conv_to_encoding (output_charset, iconveh_error,
435 u8_string, u8_string_length,
437 &encoded_string_length);
439 if (NULL == encoded_string)
441 LOG_STRERROR (GNUNET_ERROR_TYPE_WARNING, "u8_conv_to_encoding");
444 ret = GNUNET_malloc (encoded_string_length + 1);
445 memcpy (ret, encoded_string, encoded_string_length);
446 ret[encoded_string_length] = '\0';
447 free (encoded_string);
450 LOG (GNUNET_ERROR_TYPE_WARNING, _("Character sets requested were `%s'->`%s'\n"),
451 "UTF-8", output_charset);
452 ret = GNUNET_malloc (len + 1);
453 memcpy (ret, input, len);
460 * Convert the len characters long character sequence
461 * given in input that is in the given charset
464 * @param input the input string (not necessarily 0-terminated)
465 * @param len the number of bytes in the @a input
466 * @param charset character set to convert from
467 * @return the converted string (0-terminated),
468 * if conversion fails, a copy of the orignal
469 * string is returned.
472 GNUNET_STRINGS_to_utf8 (const char *input,
476 return GNUNET_STRINGS_conv (input, len, charset, "UTF-8");
481 * Convert the len bytes-long UTF-8 string
482 * given in input to the given charset.
484 * @param input the input string (not necessarily 0-terminated)
485 * @param len the number of bytes in the @a input
486 * @param charset character set to convert to
487 * @return the converted string (0-terminated),
488 * if conversion fails, a copy of the orignal
489 * string is returned.
492 GNUNET_STRINGS_from_utf8 (const char *input,
496 return GNUNET_STRINGS_conv (input, len, "UTF-8", charset);
501 * Convert the utf-8 input string to lowercase.
502 * Output needs to be allocated appropriately.
504 * @param input input string
505 * @param output output buffer
508 GNUNET_STRINGS_utf8_tolower (const char *input,
514 tmp_in = u8_tolower ((uint8_t*)input, strlen ((char *) input),
515 NULL, UNINORM_NFD, NULL, &len);
516 memcpy(output, tmp_in, len);
523 * Convert the utf-8 input string to uppercase.
524 * Output needs to be allocated appropriately.
526 * @param input input string
527 * @param output output buffer
530 GNUNET_STRINGS_utf8_toupper(const char *input,
536 tmp_in = u8_toupper ((uint8_t*)input, strlen ((char *) input),
537 NULL, UNINORM_NFD, NULL, &len);
538 memcpy (output, tmp_in, len);
545 * Complete filename (a la shell) from abbrevition.
546 * @param fil the name of the file, may contain ~/ or
547 * be relative to the current directory
548 * @returns the full file name,
549 * NULL is returned on error
552 GNUNET_STRINGS_filename_expand (const char *fil)
569 if (fil[0] == DIR_SEPARATOR)
570 /* absolute path, just copy */
571 return GNUNET_strdup (fil);
574 fm = getenv ("HOME");
577 LOG (GNUNET_ERROR_TYPE_WARNING,
578 _("Failed to expand `$HOME': environment variable `HOME' not set"));
581 fm = GNUNET_strdup (fm);
582 /* do not copy '~' */
585 /* skip over dir seperator to be consistent */
586 if (fil_ptr[0] == DIR_SEPARATOR)
597 buffer = GNUNET_malloc (len);
598 if (getcwd (buffer, len) != NULL)
603 if ((errno == ERANGE) && (len < 1024 * 1024 * 4))
606 GNUNET_free (buffer);
609 GNUNET_free (buffer);
614 LOG_STRERROR (GNUNET_ERROR_TYPE_WARNING, "getcwd");
615 buffer = getenv ("PWD"); /* alternative */
617 fm = GNUNET_strdup (buffer);
620 fm = GNUNET_strdup ("./"); /* give up */
622 n = strlen (fm) + 1 + strlen (fil_ptr) + 1;
623 buffer = GNUNET_malloc (n);
624 GNUNET_snprintf (buffer, n, "%s%s%s", fm,
625 (fm[strlen (fm) - 1] ==
626 DIR_SEPARATOR) ? "" : DIR_SEPARATOR_STR, fil_ptr);
630 fn = GNUNET_malloc (MAX_PATH + 1);
632 if ((lRet = plibc_conv_to_win_path (fil, fn)) != ERROR_SUCCESS)
634 SetErrnoFromWinError (lRet);
635 LOG_STRERROR (GNUNET_ERROR_TYPE_WARNING, "plibc_conv_to_win_path");
638 /* is the path relative? */
639 if ((strncmp (fn + 1, ":\\", 2) != 0) && (strncmp (fn, "\\\\", 2) != 0))
641 char szCurDir[MAX_PATH + 1];
643 lRet = GetCurrentDirectory (MAX_PATH + 1, szCurDir);
644 if (lRet + strlen (fn) + 1 > (MAX_PATH + 1))
646 SetErrnoFromWinError (ERROR_BUFFER_OVERFLOW);
647 LOG_STRERROR (GNUNET_ERROR_TYPE_WARNING, "GetCurrentDirectory");
650 buffer = GNUNET_malloc (MAX_PATH + 1);
651 GNUNET_snprintf (buffer, MAX_PATH + 1, "%s\\%s", szCurDir, fn);
662 * Give relative time in human-readable fancy format.
663 * This is one of the very few calls in the entire API that is
666 * @param delta time in milli seconds
667 * @param do_round are we allowed to round a bit?
668 * @return time as human-readable string
671 GNUNET_STRINGS_relative_time_to_string (struct GNUNET_TIME_Relative delta,
674 static char buf[128];
675 const char *unit = _( /* time unit */ "µs");
676 uint64_t dval = delta.rel_value_us;
678 if (GNUNET_TIME_UNIT_FOREVER_REL.rel_value_us == delta.rel_value_us)
680 if (0 == delta.rel_value_us)
682 if ( ( (GNUNET_YES == do_round) &&
683 (dval > 5 * 1000) ) ||
684 (0 == (dval % 1000) ))
687 unit = _( /* time unit */ "ms");
688 if ( ( (GNUNET_YES == do_round) &&
689 (dval > 5 * 1000) ) ||
690 (0 == (dval % 1000) ))
693 unit = _( /* time unit */ "s");
694 if ( ( (GNUNET_YES == do_round) &&
696 (0 == (dval % 60) ) )
699 unit = _( /* time unit */ "m");
700 if ( ( (GNUNET_YES == do_round) &&
705 unit = _( /* time unit */ "h");
706 if ( ( (GNUNET_YES == do_round) &&
712 unit = _( /* time unit */ "day");
714 unit = _( /* time unit */ "days");
720 GNUNET_snprintf (buf, sizeof (buf),
721 "%llu %s", dval, unit);
727 * "asctime", except for GNUnet time. Converts a GNUnet internal
728 * absolute time (which is in UTC) to a string in local time.
729 * Note that the returned value will be overwritten if this function
732 * @param t the absolute time to convert
733 * @return timestamp in human-readable form in local time
736 GNUNET_STRINGS_absolute_time_to_string (struct GNUNET_TIME_Absolute t)
738 static char buf[255];
742 if (t.abs_value_us == GNUNET_TIME_UNIT_FOREVER_ABS.abs_value_us)
743 return _("end of time");
744 tt = t.abs_value_us / 1000LL / 1000LL;
746 strftime (buf, sizeof (buf), "%a %b %d %H:%M:%S %Y", tp);
753 * Returns a pointer to a part of filename (allocates nothing)!
755 * @param filename filename to extract basename from
756 * @return short (base) name of the file (that is, everything following the
757 * last directory separator in filename. If filename ends with a
758 * directory separator, the result will be a zero-length string.
759 * If filename has no directory separators, the result is filename
763 GNUNET_STRINGS_get_short_name (const char *filename)
765 const char *short_fn = filename;
767 while (NULL != (ss = strstr (short_fn, DIR_SEPARATOR_STR))
775 * Get the numeric value corresponding to a character.
777 * @param a a character
778 * @return corresponding numeric value
781 getValue__ (unsigned char a)
783 if ((a >= '0') && (a <= '9'))
785 if ((a >= 'A') && (a <= 'V'))
786 return (a - 'A' + 10);
787 if ((a >= 'a') && (a <= 'v'))
788 return (a - 'a' + 10);
794 * Convert binary data to ASCII encoding. The ASCII encoding is rather
795 * GNUnet specific. It was chosen such that it only uses characters
796 * in [0-9A-V], can be produced without complex arithmetics and uses a
797 * small number of characters.
798 * Does not append 0-terminator, but returns a pointer to the place where
799 * it should be placed, if needed.
801 * @param data data to encode
802 * @param size size of data (in bytes)
803 * @param out buffer to fill
804 * @param out_size size of the buffer. Must be large enough to hold
805 * ((size*8) + (((size*8) % 5) > 0 ? 5 - ((size*8) % 5) : 0)) / 5 bytes
806 * @return pointer to the next byte in 'out' or NULL on error.
809 GNUNET_STRINGS_data_to_string (const void *data, size_t size, char *out, size_t out_size)
812 * 32 characters for encoding
814 static char *encTable__ = "0123456789ABCDEFGHIJKLMNOPQRSTUV";
819 const unsigned char *udata;
821 GNUNET_assert (data != NULL);
822 GNUNET_assert (out != NULL);
824 if (out_size < (((size*8) + ((size*8) % 5)) % 5))
833 while ((rpos < size) || (vbit > 0))
835 if ((rpos < size) && (vbit < 5))
837 bits = (bits << 8) | udata[rpos++]; /* eat 8 more bits */
842 bits <<= (5 - vbit); /* zero-padding */
843 GNUNET_assert (vbit == ((size * 8) % 5));
846 if (wpos >= out_size)
851 out[wpos++] = encTable__[(bits >> (vbit - 5)) & 31];
854 GNUNET_assert (vbit == 0);
862 * Convert ASCII encoding back to data
863 * out_size must match exactly the size of the data before it was encoded.
865 * @param enc the encoding
866 * @param enclen number of characters in @a enc (without 0-terminator, which can be missing)
867 * @param out location where to store the decoded data
868 * @param out_size size of the output buffer @a out
869 * @return #GNUNET_OK on success, #GNUNET_SYSERR if result has the wrong encoding
872 GNUNET_STRINGS_string_to_data (const char *enc, size_t enclen,
873 void *out, size_t out_size)
882 unsigned int encoded_len = out_size * 8;
888 return GNUNET_SYSERR;
893 if ((encoded_len % 5) > 0)
895 vbit = encoded_len % 5; /* padding! */
897 bits = (ret = getValue__ (enc[--rpos])) >> (5 - (encoded_len % 5));
903 bits = (ret = getValue__ (enc[--rpos]));
905 if ((encoded_len + shift) / 5 != enclen)
906 return GNUNET_SYSERR;
908 return GNUNET_SYSERR;
914 return GNUNET_SYSERR;
916 bits = ((ret = getValue__ (enc[--rpos])) << vbit) | bits;
918 return GNUNET_SYSERR;
922 uout[--wpos] = (unsigned char) bits;
929 return GNUNET_SYSERR;
935 * Parse a path that might be an URI.
937 * @param path path to parse. Must be NULL-terminated.
938 * @param scheme_part a pointer to 'char *' where a pointer to a string that
939 * represents the URI scheme will be stored. Can be NULL. The string is
940 * allocated by the function, and should be freed by GNUNET_free() when
941 * it is no longer needed.
942 * @param path_part a pointer to 'const char *' where a pointer to the path
943 * part of the URI will be stored. Can be NULL. Points to the same block
944 * of memory as 'path', and thus must not be freed. Might point to '\0',
945 * if path part is zero-length.
946 * @return GNUNET_YES if it's an URI, GNUNET_NO otherwise. If 'path' is not
947 * an URI, '* scheme_part' and '*path_part' will remain unchanged
948 * (if they weren't NULL).
951 GNUNET_STRINGS_parse_uri (const char *path, char **scheme_part,
952 const char **path_part)
957 const char *post_scheme_part = NULL;
959 for (end = 0, i = 0; !end && i < len; i++)
964 if (path[i] == ':' && i > 0)
969 if (!((path[i] >= 'A' && path[i] <= 'Z') || (path[i] >= 'a' && path[i] <= 'z')
970 || (path[i] >= '0' && path[i] <= '9') || path[i] == '+' || path[i] == '-'
971 || (path[i] == '.')))
984 post_scheme_part = &path[i];
991 if (post_scheme_part == NULL)
995 *scheme_part = GNUNET_malloc (post_scheme_part - path + 1);
996 memcpy (*scheme_part, path, post_scheme_part - path);
997 (*scheme_part)[post_scheme_part - path] = '\0';
1000 *path_part = post_scheme_part;
1006 * Check whether @a filename is absolute or not, and if it's an URI
1008 * @param filename filename to check
1009 * @param can_be_uri #GNUNET_YES to check for being URI, #GNUNET_NO - to
1010 * assume it's not URI
1011 * @param r_is_uri a pointer to an int that is set to #GNUNET_YES if @a filename
1012 * is URI and to #GNUNET_NO otherwise. Can be NULL. If @a can_be_uri is
1013 * not #GNUNET_YES, `* r_is_uri` is set to #GNUNET_NO.
1014 * @param r_uri_scheme a pointer to a char * that is set to a pointer to URI scheme.
1015 * The string is allocated by the function, and should be freed with
1016 * GNUNET_free(). Can be NULL.
1017 * @return #GNUNET_YES if @a filename is absolute, #GNUNET_NO otherwise.
1020 GNUNET_STRINGS_path_is_absolute (const char *filename,
1023 char **r_uri_scheme)
1028 const char *post_scheme_path;
1031 /* consider POSIX paths to be absolute too, even on W32,
1032 * as plibc expansion will fix them for us.
1034 if (filename[0] == '/')
1038 is_uri = GNUNET_STRINGS_parse_uri (filename, &uri, &post_scheme_path);
1044 *r_uri_scheme = uri;
1046 GNUNET_free_non_null (uri);
1048 len = strlen(post_scheme_path);
1049 /* Special check for file:///c:/blah
1050 * We want to parse 'c:/', not '/c:/'
1052 if (post_scheme_path[0] == '/' && len >= 3 && post_scheme_path[2] == ':')
1053 post_scheme_path = &post_scheme_path[1];
1055 return GNUNET_STRINGS_path_is_absolute (post_scheme_path, GNUNET_NO, NULL, NULL);
1061 *r_is_uri = GNUNET_NO;
1064 len = strlen (filename);
1066 ((filename[0] >= 'A' && filename[0] <= 'Z')
1067 || (filename[0] >= 'a' && filename[0] <= 'z'))
1068 && filename[1] == ':' && (filename[2] == '/' || filename[2] == '\\'))
1075 #define _IFMT 0170000 /* type of file */
1076 #define _IFLNK 0120000 /* symbolic link */
1077 #define S_ISLNK(m) (((m)&_IFMT) == _IFLNK)
1082 * Perform @a checks on @a filename.
1084 * @param filename file to check
1085 * @param checks checks to perform
1086 * @return #GNUNET_YES if all checks pass, #GNUNET_NO if at least one of them
1087 * fails, #GNUNET_SYSERR when a check can't be performed
1090 GNUNET_STRINGS_check_filename (const char *filename,
1091 enum GNUNET_STRINGS_FilenameCheck checks)
1094 if ( (NULL == filename) || (filename[0] == '\0') )
1095 return GNUNET_SYSERR;
1096 if (0 != (checks & GNUNET_STRINGS_CHECK_IS_ABSOLUTE))
1097 if (!GNUNET_STRINGS_path_is_absolute (filename, GNUNET_NO, NULL, NULL))
1099 if (0 != (checks & (GNUNET_STRINGS_CHECK_EXISTS
1100 | GNUNET_STRINGS_CHECK_IS_DIRECTORY
1101 | GNUNET_STRINGS_CHECK_IS_LINK)))
1103 if (0 != STAT (filename, &st))
1105 if (0 != (checks & GNUNET_STRINGS_CHECK_EXISTS))
1108 return GNUNET_SYSERR;
1111 if (0 != (checks & GNUNET_STRINGS_CHECK_IS_DIRECTORY))
1112 if (!S_ISDIR (st.st_mode))
1114 if (0 != (checks & GNUNET_STRINGS_CHECK_IS_LINK))
1115 if (!S_ISLNK (st.st_mode))
1122 * Tries to convert 'zt_addr' string to an IPv6 address.
1123 * The string is expected to have the format "[ABCD::01]:80".
1125 * @param zt_addr 0-terminated string. May be mangled by the function.
1126 * @param addrlen length of @a zt_addr (not counting 0-terminator).
1127 * @param r_buf a buffer to fill. Initially gets filled with zeroes,
1128 * then its sin6_port, sin6_family and sin6_addr are set appropriately.
1129 * @return #GNUNET_OK if conversion succeded.
1130 * #GNUNET_SYSERR otherwise, in which
1131 * case the contents of @a r_buf are undefined.
1134 GNUNET_STRINGS_to_address_ipv6 (const char *zt_addr,
1136 struct sockaddr_in6 *r_buf)
1138 char zbuf[addrlen + 1];
1144 return GNUNET_SYSERR;
1145 memcpy (zbuf, zt_addr, addrlen);
1148 GNUNET_log (GNUNET_ERROR_TYPE_WARNING,
1149 _("IPv6 address did not start with `['\n"));
1150 return GNUNET_SYSERR;
1152 zbuf[addrlen] = '\0';
1153 port_colon = strrchr (zbuf, ':');
1154 if (NULL == port_colon)
1156 GNUNET_log (GNUNET_ERROR_TYPE_WARNING,
1157 _("IPv6 address did contain ':' to separate port number\n"));
1158 return GNUNET_SYSERR;
1160 if (']' != *(port_colon - 1))
1162 GNUNET_log (GNUNET_ERROR_TYPE_WARNING,
1163 _("IPv6 address did contain ']' before ':' to separate port number\n"));
1164 return GNUNET_SYSERR;
1166 ret = SSCANF (port_colon, ":%u", &port);
1167 if ( (1 != ret) || (port > 65535) )
1169 GNUNET_log (GNUNET_ERROR_TYPE_WARNING,
1170 _("IPv6 address did contain a valid port number after the last ':'\n"));
1171 return GNUNET_SYSERR;
1173 *(port_colon-1) = '\0';
1174 memset (r_buf, 0, sizeof (struct sockaddr_in6));
1175 ret = inet_pton (AF_INET6, &zbuf[1], &r_buf->sin6_addr);
1178 GNUNET_log (GNUNET_ERROR_TYPE_WARNING,
1179 _("Invalid IPv6 address `%s': %s\n"),
1182 return GNUNET_SYSERR;
1184 r_buf->sin6_port = htons (port);
1185 r_buf->sin6_family = AF_INET6;
1186 #if HAVE_SOCKADDR_IN_SIN_LEN
1187 r_buf->sin6_len = (u_char) sizeof (struct sockaddr_in6);
1194 * Tries to convert 'zt_addr' string to an IPv4 address.
1195 * The string is expected to have the format "1.2.3.4:80".
1197 * @param zt_addr 0-terminated string. May be mangled by the function.
1198 * @param addrlen length of @a zt_addr (not counting 0-terminator).
1199 * @param r_buf a buffer to fill.
1200 * @return #GNUNET_OK if conversion succeded.
1201 * #GNUNET_SYSERR otherwise, in which case
1202 * the contents of @a r_buf are undefined.
1205 GNUNET_STRINGS_to_address_ipv4 (const char *zt_addr, uint16_t addrlen,
1206 struct sockaddr_in *r_buf)
1208 unsigned int temps[4];
1213 return GNUNET_SYSERR;
1214 cnt = SSCANF (zt_addr, "%u.%u.%u.%u:%u", &temps[0], &temps[1], &temps[2], &temps[3], &port);
1216 return GNUNET_SYSERR;
1217 for (cnt = 0; cnt < 4; cnt++)
1218 if (temps[cnt] > 0xFF)
1219 return GNUNET_SYSERR;
1221 return GNUNET_SYSERR;
1222 r_buf->sin_family = AF_INET;
1223 r_buf->sin_port = htons (port);
1224 r_buf->sin_addr.s_addr = htonl ((temps[0] << 24) + (temps[1] << 16) +
1225 (temps[2] << 8) + temps[3]);
1226 #if HAVE_SOCKADDR_IN_SIN_LEN
1227 r_buf->sin_len = (u_char) sizeof (struct sockaddr_in);
1234 * Tries to convert @a addr string to an IP (v4 or v6) address.
1235 * Will automatically decide whether to treat 'addr' as v4 or v6 address.
1237 * @param addr a string, may not be 0-terminated.
1238 * @param addrlen number of bytes in @a addr (if addr is 0-terminated,
1239 * 0-terminator should not be counted towards addrlen).
1240 * @param r_buf a buffer to fill.
1241 * @return #GNUNET_OK if conversion succeded. GNUNET_SYSERR otherwise, in which
1242 * case the contents of r_buf are undefined.
1245 GNUNET_STRINGS_to_address_ip (const char *addr,
1247 struct sockaddr_storage *r_buf)
1250 return GNUNET_STRINGS_to_address_ipv6 (addr,
1252 (struct sockaddr_in6 *) r_buf);
1253 return GNUNET_STRINGS_to_address_ipv4 (addr,
1255 (struct sockaddr_in *) r_buf);
1260 * Makes a copy of argv that consists of a single memory chunk that can be
1261 * freed with a single call to GNUNET_free ();
1263 static char *const *
1264 _make_continuous_arg_copy (int argc,
1267 size_t argvsize = 0;
1271 for (i = 0; i < argc; i++)
1272 argvsize += strlen (argv[i]) + 1 + sizeof (char *);
1273 new_argv = GNUNET_malloc (argvsize + sizeof (char *));
1274 p = (char *) &new_argv[argc + 1];
1275 for (i = 0; i < argc; i++)
1278 strcpy (p, argv[i]);
1279 p += strlen (argv[i]) + 1;
1281 new_argv[argc] = NULL;
1282 return (char *const *) new_argv;
1287 * Returns utf-8 encoded arguments.
1288 * Does nothing (returns a copy of argc and argv) on any platform
1290 * Returned argv has u8argv[u8argc] == NULL.
1291 * Returned argv is a single memory block, and can be freed with a single
1292 * GNUNET_free () call.
1294 * @param argc argc (as given by main())
1295 * @param argv argv (as given by main())
1296 * @param u8argc a location to store new argc in (though it's th same as argc)
1297 * @param u8argv a location to store new argv in
1298 * @return #GNUNET_OK on success, #GNUNET_SYSERR on failure
1301 GNUNET_STRINGS_get_utf8_args (int argc, char *const *argv, int *u8argc, char *const **u8argv)
1308 char **split_u8argv;
1310 wcmd = GetCommandLineW ();
1312 return GNUNET_SYSERR;
1313 wargv = CommandLineToArgvW (wcmd, &wargc);
1315 return GNUNET_SYSERR;
1317 split_u8argv = GNUNET_malloc (argc * sizeof (char *));
1319 for (i = 0; i < wargc; i++)
1322 /* Hopefully it will allocate us NUL-terminated strings... */
1323 split_u8argv[i] = (char *) u16_to_u8 (wargv[i], wcslen (wargv[i]) + 1, NULL, &strl);
1324 if (NULL == split_u8argv[i])
1327 for (j = 0; j < i; j++)
1328 free (split_u8argv[j]);
1329 GNUNET_free (split_u8argv);
1331 return GNUNET_SYSERR;
1335 *u8argv = _make_continuous_arg_copy (wargc, split_u8argv);
1338 for (i = 0; i < wargc; i++)
1339 free (split_u8argv[i]);
1340 free (split_u8argv);
1343 char *const *new_argv = (char *const *) _make_continuous_arg_copy (argc, argv);
1352 * Parse the given port policy. The format is
1353 * "[!]SPORT[-DPORT]".
1355 * @param port_policy string to parse
1356 * @param pp policy to fill in
1357 * @return #GNUNET_OK on success, #GNUNET_SYSERR if the
1358 * @a port_policy is malformed
1361 parse_port_policy (const char *port_policy,
1362 struct GNUNET_STRINGS_PortPolicy *pp)
1372 pp->negate_portrange = GNUNET_YES;
1375 if (2 == sscanf (pos,
1384 GNUNET_log (GNUNET_ERROR_TYPE_WARNING,
1385 _("Port not in range\n"));
1386 return GNUNET_SYSERR;
1388 pp->start_port = (uint16_t) s;
1389 pp->end_port = (uint16_t) e;
1392 if (1 == sscanf (pos,
1400 GNUNET_log (GNUNET_ERROR_TYPE_WARNING,
1401 _("Port not in range\n"));
1402 return GNUNET_SYSERR;
1405 pp->start_port = (uint16_t) s;
1406 pp->end_port = (uint16_t) s;
1409 GNUNET_log (GNUNET_ERROR_TYPE_WARNING,
1410 _("Malformed port policy `%s'\n"),
1412 return GNUNET_SYSERR;
1417 * Parse an IPv4 network policy. The argument specifies a list of
1418 * subnets. The format is
1419 * <tt>(network[/netmask][:SPORT[-DPORT]];)*</tt> (no whitespace, must
1420 * be terminated with a semicolon). The network must be given in
1421 * dotted-decimal notation. The netmask can be given in CIDR notation
1422 * (/16) or in dotted-decimal (/255.255.0.0).
1424 * @param routeListX a string specifying the IPv4 subnets
1425 * @return the converted list, terminated with all zeros;
1426 * NULL if the synatx is flawed
1428 struct GNUNET_STRINGS_IPv4NetworkPolicy *
1429 GNUNET_STRINGS_parse_ipv4_policy (const char *routeListX)
1437 unsigned int temps[8];
1439 struct GNUNET_STRINGS_IPv4NetworkPolicy *result;
1444 if (NULL == routeListX)
1446 len = strlen (routeListX);
1449 routeList = GNUNET_strdup (routeListX);
1451 for (i = 0; i < len; i++)
1452 if (routeList[i] == ';')
1454 result = GNUNET_malloc (sizeof (struct GNUNET_STRINGS_IPv4NetworkPolicy) * (count + 1));
1459 for (colon = pos; ':' != routeList[colon]; colon++)
1460 if ( (';' == routeList[colon]) ||
1461 ('\0' == routeList[colon]) )
1463 for (end = colon; ';' != routeList[end]; end++)
1464 if ('\0' == routeList[end])
1466 if ('\0' == routeList[end])
1468 routeList[end] = '\0';
1469 if (':' == routeList[colon])
1471 routeList[colon] = '\0';
1472 if (GNUNET_OK != parse_port_policy (&routeList[colon + 1],
1477 SSCANF (&routeList[pos],
1478 "%u.%u.%u.%u/%u.%u.%u.%u",
1489 for (j = 0; j < 8; j++)
1490 if (temps[j] > 0xFF)
1492 LOG (GNUNET_ERROR_TYPE_WARNING,
1493 _("Invalid format for IP: `%s'\n"),
1495 GNUNET_free (result);
1496 GNUNET_free (routeList);
1499 result[i].network.s_addr =
1500 htonl ((temps[0] << 24) + (temps[1] << 16) + (temps[2] << 8) +
1502 result[i].netmask.s_addr =
1503 htonl ((temps[4] << 24) + (temps[5] << 16) + (temps[6] << 8) +
1509 /* try second notation */
1511 SSCANF (&routeList[pos],
1520 for (j = 0; j < 4; j++)
1521 if (temps[j] > 0xFF)
1523 LOG (GNUNET_ERROR_TYPE_WARNING,
1524 _("Invalid format for IP: `%s'\n"),
1526 GNUNET_free (result);
1527 GNUNET_free (routeList);
1530 result[i].network.s_addr =
1531 htonl ((temps[0] << 24) + (temps[1] << 16) + (temps[2] << 8) +
1533 if ((slash <= 32) && (slash >= 0))
1535 result[i].netmask.s_addr = 0;
1538 result[i].netmask.s_addr =
1539 (result[i].netmask.s_addr >> 1) + 0x80000000;
1542 result[i].netmask.s_addr = htonl (result[i].netmask.s_addr);
1549 LOG (GNUNET_ERROR_TYPE_WARNING,
1550 _("Invalid network notation ('/%d' is not legal in IPv4 CIDR)."),
1552 GNUNET_free (result);
1553 GNUNET_free (routeList);
1554 return NULL; /* error */
1557 /* try third notation */
1560 SSCANF (&routeList[pos],
1568 for (j = 0; j < 4; j++)
1569 if (temps[j] > 0xFF)
1571 LOG (GNUNET_ERROR_TYPE_WARNING,
1572 _("Invalid format for IP: `%s'\n"),
1574 GNUNET_free (result);
1575 GNUNET_free (routeList);
1578 result[i].network.s_addr =
1579 htonl ((temps[0] << 24) + (temps[1] << 16) + (temps[2] << 8) +
1581 result[i].netmask.s_addr = 0;
1584 result[i].netmask.s_addr = (result[i].netmask.s_addr >> 1) + 0x80000000;
1587 result[i].netmask.s_addr = htonl (result[i].netmask.s_addr);
1592 LOG (GNUNET_ERROR_TYPE_WARNING,
1593 _("Invalid format for IP: `%s'\n"),
1595 GNUNET_free (result);
1596 GNUNET_free (routeList);
1597 return NULL; /* error */
1599 if (pos < strlen (routeList))
1601 LOG (GNUNET_ERROR_TYPE_WARNING,
1602 _("Invalid format: `%s'\n"),
1604 GNUNET_free (result);
1605 GNUNET_free (routeList);
1606 return NULL; /* oops */
1608 GNUNET_free (routeList);
1609 return result; /* ok */
1614 * Parse an IPv6 network policy. The argument specifies a list of
1615 * subnets. The format is <tt>(network[/netmask[:SPORT[-DPORT]]];)*</tt>
1616 * (no whitespace, must be terminated with a semicolon). The network
1617 * must be given in colon-hex notation. The netmask must be given in
1618 * CIDR notation (/16) or can be omitted to specify a single host.
1619 * Note that the netmask is mandatory if ports are specified.
1621 * @param routeListX a string specifying the policy
1622 * @return the converted list, 0-terminated, NULL if the synatx is flawed
1624 struct GNUNET_STRINGS_IPv6NetworkPolicy *
1625 GNUNET_STRINGS_parse_ipv6_policy (const char *routeListX)
1635 struct GNUNET_STRINGS_IPv6NetworkPolicy *result;
1641 if (NULL == routeListX)
1643 len = strlen (routeListX);
1646 routeList = GNUNET_strdup (routeListX);
1648 for (i = 0; i < len; i++)
1649 if (';' == routeList[i])
1651 if (';' != routeList[len - 1])
1653 LOG (GNUNET_ERROR_TYPE_WARNING,
1654 _("Invalid network notation (does not end with ';': `%s')\n"),
1656 GNUNET_free (routeList);
1660 result = GNUNET_malloc (sizeof (struct GNUNET_STRINGS_IPv6NetworkPolicy) * (count + 1));
1666 while (';' != routeList[pos])
1669 while ((slash >= start) && (routeList[slash] != '/'))
1674 memset (&result[i].netmask,
1676 sizeof (struct in6_addr));
1681 routeList[pos] = '\0';
1682 for (colon = pos; ':' != routeList[colon]; colon--)
1683 if ('/' == routeList[colon])
1685 if (':' == routeList[colon])
1687 routeList[colon] = '\0';
1688 if (GNUNET_OK != parse_port_policy (&routeList[colon + 1],
1691 GNUNET_free (result);
1692 GNUNET_free (routeList);
1696 ret = inet_pton (AF_INET6, &routeList[slash + 1], &result[i].netmask);
1700 if ((1 != SSCANF (&routeList[slash + 1], "%u", &bits)) || (bits > 128))
1703 LOG (GNUNET_ERROR_TYPE_WARNING,
1704 _("Wrong format `%s' for netmask\n"),
1705 &routeList[slash + 1]);
1709 LOG_STRERROR (GNUNET_ERROR_TYPE_WARNING, "inet_pton");
1711 GNUNET_free (result);
1712 GNUNET_free (routeList);
1718 result[i].netmask.s6_addr[off++] = 0xFF;
1723 result[i].netmask.s6_addr[off] =
1724 (result[i].netmask.s6_addr[off] >> 1) + 0x80;
1729 routeList[slash] = '\0';
1730 ret = inet_pton (AF_INET6, &routeList[start], &result[i].network);
1734 LOG (GNUNET_ERROR_TYPE_WARNING,
1735 _("Wrong format `%s' for network\n"),
1736 &routeList[slash + 1]);
1738 LOG_STRERROR (GNUNET_ERROR_TYPE_ERROR,
1740 GNUNET_free (result);
1741 GNUNET_free (routeList);
1747 GNUNET_free (routeList);
1753 /** ******************** Base64 encoding ***********/
1755 #define FILLCHAR '='
1757 "ABCDEFGHIJKLMNOPQRSTUVWXYZ" "abcdefghijklmnopqrstuvwxyz" "0123456789+/";
1761 * Encode into Base64.
1763 * @param data the data to encode
1764 * @param len the length of the input
1765 * @param output where to write the output (*output should be NULL,
1767 * @return the size of the output
1770 GNUNET_STRINGS_base64_encode (const char *data,
1780 opt = GNUNET_malloc (2 + (len * 4 / 3) + 8);
1782 for (i = 0; i < len; ++i)
1784 c = (data[i] >> 2) & 0x3f;
1785 opt[ret++] = cvt[(int) c];
1786 c = (data[i] << 4) & 0x3f;
1788 c |= (data[i] >> 4) & 0x0f;
1789 opt[ret++] = cvt[(int) c];
1792 c = (data[i] << 2) & 0x3f;
1794 c |= (data[i] >> 6) & 0x03;
1795 opt[ret++] = cvt[(int) c];
1800 opt[ret++] = FILLCHAR;
1805 opt[ret++] = cvt[(int) c];
1809 opt[ret++] = FILLCHAR;
1812 opt[ret++] = FILLCHAR;
1816 #define cvtfind(a)( (((a) >= 'A')&&((a) <= 'Z'))? (a)-'A'\
1817 :(((a)>='a')&&((a)<='z')) ? (a)-'a'+26\
1818 :(((a)>='0')&&((a)<='9')) ? (a)-'0'+52\
1820 :((a) == '/') ? 63 : -1)
1824 * Decode from Base64.
1826 * @param data the data to encode
1827 * @param len the length of the input
1828 * @param output where to write the output (*output should be NULL,
1830 * @return the size of the output
1833 GNUNET_STRINGS_base64_decode (const char *data,
1834 size_t len, char **output)
1841 #define CHECK_CRLF while (data[i] == '\r' || data[i] == '\n') {\
1842 GNUNET_log(GNUNET_ERROR_TYPE_DEBUG | GNUNET_ERROR_TYPE_BULK, "ignoring CR/LF\n"); \
1844 if (i >= len) goto END; \
1847 *output = GNUNET_malloc ((len * 3 / 4) + 8);
1848 GNUNET_log (GNUNET_ERROR_TYPE_DEBUG,
1849 "base64_decode decoding len=%d\n",
1851 for (i = 0; i < len; ++i)
1854 if (FILLCHAR == data[i])
1856 c = (char) cvtfind (data[i]);
1859 c1 = (char) cvtfind (data[i]);
1860 c = (c << 2) | ((c1 >> 4) & 0x3);
1861 (*output)[ret++] = c;
1868 c = (char) cvtfind (c);
1869 c1 = ((c1 << 4) & 0xf0) | ((c >> 2) & 0xf);
1870 (*output)[ret++] = c1;
1879 c1 = (char) cvtfind (c1);
1880 c = ((c << 6) & 0xc0) | c1;
1881 (*output)[ret++] = c;
1892 /* end of strings.c */