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 { "week", 7 * 24 * 60 * 60 * 1000LL * 1000LL},
315 { "weeks", 7 * 24 * 60 * 60 * 1000LL * 1000LL},
316 { "a", 31536000000000LL /* year */ },
320 unsigned long long val;
322 if (0 == strcasecmp ("forever", fancy_time))
324 *rtime = GNUNET_TIME_UNIT_FOREVER_REL;
327 ret = convert_with_table (fancy_time,
330 rtime->rel_value_us = (uint64_t) val;
336 * Convert a given fancy human-readable time to our internal
337 * representation. The human-readable time is expected to be
338 * in local time, whereas the returned value will be in UTC.
340 * @param fancy_time human readable string (i.e. %Y-%m-%d %H:%M:%S)
341 * @param atime set to the absolute time
342 * @return #GNUNET_OK on success, #GNUNET_SYSERR on error
345 GNUNET_STRINGS_fancy_time_to_absolute (const char *fancy_time,
346 struct GNUNET_TIME_Absolute *atime)
351 if (0 == strcasecmp ("end of time", fancy_time))
353 *atime = GNUNET_TIME_UNIT_FOREVER_ABS;
356 memset (&tv, 0, sizeof (tv));
357 if ( (NULL == strptime (fancy_time, "%a %b %d %H:%M:%S %Y", &tv)) &&
358 (NULL == strptime (fancy_time, "%c", &tv)) &&
359 (NULL == strptime (fancy_time, "%Ec", &tv)) &&
360 (NULL == strptime (fancy_time, "%Y-%m-%d %H:%M:%S", &tv)) &&
361 (NULL == strptime (fancy_time, "%Y-%m-%d %H:%M", &tv)) &&
362 (NULL == strptime (fancy_time, "%x", &tv)) &&
363 (NULL == strptime (fancy_time, "%Ex", &tv)) &&
364 (NULL == strptime (fancy_time, "%Y-%m-%d", &tv)) &&
365 (NULL == strptime (fancy_time, "%Y-%m", &tv)) &&
366 (NULL == strptime (fancy_time, "%Y", &tv)) )
367 return GNUNET_SYSERR;
369 atime->abs_value_us = (uint64_t) ((uint64_t) t * 1000LL * 1000LL);
373 TIME_ZONE_INFORMATION tzi;
374 tzv = GetTimeZoneInformation (&tzi);
375 if (TIME_ZONE_ID_INVALID != tzv)
377 atime->abs_value_us -= 1000LL * 1000LL * tzi.Bias * 60LL;
386 * Convert the len characters long character sequence
387 * given in input that is in the given input charset
388 * to a string in given output charset.
390 * @param input input string
391 * @param len number of bytes in @a input
392 * @param input_charset character set used for @a input
393 * @param output_charset desired character set for the return value
394 * @return the converted string (0-terminated),
395 * if conversion fails, a copy of the orignal
396 * string is returned.
399 GNUNET_STRINGS_conv (const char *input,
401 const char *input_charset,
402 const char *output_charset)
406 char *encoded_string;
407 size_t u8_string_length;
408 size_t encoded_string_length;
410 u8_string = u8_conv_from_encoding (input_charset,
415 if (NULL == u8_string)
417 LOG_STRERROR (GNUNET_ERROR_TYPE_WARNING, "u8_conv_from_encoding");
420 if (0 == strcmp (output_charset, "UTF-8"))
422 ret = GNUNET_malloc (u8_string_length + 1);
423 memcpy (ret, u8_string, u8_string_length);
424 ret[u8_string_length] = '\0';
428 encoded_string = u8_conv_to_encoding (output_charset, iconveh_error,
429 u8_string, u8_string_length,
431 &encoded_string_length);
433 if (NULL == encoded_string)
435 LOG_STRERROR (GNUNET_ERROR_TYPE_WARNING, "u8_conv_to_encoding");
438 ret = GNUNET_malloc (encoded_string_length + 1);
439 memcpy (ret, encoded_string, encoded_string_length);
440 ret[encoded_string_length] = '\0';
441 free (encoded_string);
444 LOG (GNUNET_ERROR_TYPE_WARNING, _("Character sets requested were `%s'->`%s'\n"),
445 "UTF-8", output_charset);
446 ret = GNUNET_malloc (len + 1);
447 memcpy (ret, input, len);
454 * Convert the len characters long character sequence
455 * given in input that is in the given charset
458 * @param input the input string (not necessarily 0-terminated)
459 * @param len the number of bytes in the @a input
460 * @param charset character set to convert from
461 * @return the converted string (0-terminated),
462 * if conversion fails, a copy of the orignal
463 * string is returned.
466 GNUNET_STRINGS_to_utf8 (const char *input,
470 return GNUNET_STRINGS_conv (input, len, charset, "UTF-8");
475 * Convert the len bytes-long UTF-8 string
476 * given in input to the given charset.
478 * @param input the input string (not necessarily 0-terminated)
479 * @param len the number of bytes in the @a input
480 * @param charset character set to convert to
481 * @return the converted string (0-terminated),
482 * if conversion fails, a copy of the orignal
483 * string is returned.
486 GNUNET_STRINGS_from_utf8 (const char *input,
490 return GNUNET_STRINGS_conv (input, len, "UTF-8", charset);
495 * Convert the utf-8 input string to lowercase.
496 * Output needs to be allocated appropriately.
498 * @param input input string
499 * @param output output buffer
502 GNUNET_STRINGS_utf8_tolower (const char *input,
508 tmp_in = u8_tolower ((uint8_t*)input, strlen ((char *) input),
509 NULL, UNINORM_NFD, NULL, &len);
510 memcpy(output, tmp_in, len);
517 * Convert the utf-8 input string to uppercase.
518 * Output needs to be allocated appropriately.
520 * @param input input string
521 * @param output output buffer
524 GNUNET_STRINGS_utf8_toupper(const char *input,
530 tmp_in = u8_toupper ((uint8_t*)input, strlen ((char *) input),
531 NULL, UNINORM_NFD, NULL, &len);
532 memcpy (output, tmp_in, len);
539 * Complete filename (a la shell) from abbrevition.
540 * @param fil the name of the file, may contain ~/ or
541 * be relative to the current directory
542 * @returns the full file name,
543 * NULL is returned on error
546 GNUNET_STRINGS_filename_expand (const char *fil)
563 if (fil[0] == DIR_SEPARATOR)
564 /* absolute path, just copy */
565 return GNUNET_strdup (fil);
568 fm = getenv ("HOME");
571 LOG (GNUNET_ERROR_TYPE_WARNING,
572 _("Failed to expand `$HOME': environment variable `HOME' not set"));
575 fm = GNUNET_strdup (fm);
576 /* do not copy '~' */
579 /* skip over dir seperator to be consistent */
580 if (fil_ptr[0] == DIR_SEPARATOR)
591 buffer = GNUNET_malloc (len);
592 if (getcwd (buffer, len) != NULL)
597 if ((errno == ERANGE) && (len < 1024 * 1024 * 4))
600 GNUNET_free (buffer);
603 GNUNET_free (buffer);
608 LOG_STRERROR (GNUNET_ERROR_TYPE_WARNING, "getcwd");
609 buffer = getenv ("PWD"); /* alternative */
611 fm = GNUNET_strdup (buffer);
614 fm = GNUNET_strdup ("./"); /* give up */
616 n = strlen (fm) + 1 + strlen (fil_ptr) + 1;
617 buffer = GNUNET_malloc (n);
618 GNUNET_snprintf (buffer, n, "%s%s%s", fm,
619 (fm[strlen (fm) - 1] ==
620 DIR_SEPARATOR) ? "" : DIR_SEPARATOR_STR, fil_ptr);
624 fn = GNUNET_malloc (MAX_PATH + 1);
626 if ((lRet = plibc_conv_to_win_path (fil, fn)) != ERROR_SUCCESS)
628 SetErrnoFromWinError (lRet);
629 LOG_STRERROR (GNUNET_ERROR_TYPE_WARNING, "plibc_conv_to_win_path");
632 /* is the path relative? */
633 if ((strncmp (fn + 1, ":\\", 2) != 0) && (strncmp (fn, "\\\\", 2) != 0))
635 char szCurDir[MAX_PATH + 1];
637 lRet = GetCurrentDirectory (MAX_PATH + 1, szCurDir);
638 if (lRet + strlen (fn) + 1 > (MAX_PATH + 1))
640 SetErrnoFromWinError (ERROR_BUFFER_OVERFLOW);
641 LOG_STRERROR (GNUNET_ERROR_TYPE_WARNING, "GetCurrentDirectory");
644 buffer = GNUNET_malloc (MAX_PATH + 1);
645 GNUNET_snprintf (buffer, MAX_PATH + 1, "%s\\%s", szCurDir, fn);
656 * Give relative time in human-readable fancy format.
657 * This is one of the very few calls in the entire API that is
660 * @param delta time in milli seconds
661 * @param do_round are we allowed to round a bit?
662 * @return time as human-readable string
665 GNUNET_STRINGS_relative_time_to_string (struct GNUNET_TIME_Relative delta,
668 static char buf[128];
669 const char *unit = _( /* time unit */ "µs");
670 uint64_t dval = delta.rel_value_us;
672 if (GNUNET_TIME_UNIT_FOREVER_REL.rel_value_us == delta.rel_value_us)
674 if (0 == delta.rel_value_us)
676 if ( ( (GNUNET_YES == do_round) &&
677 (dval > 5 * 1000) ) ||
678 (0 == (dval % 1000) ))
681 unit = _( /* time unit */ "ms");
682 if ( ( (GNUNET_YES == do_round) &&
683 (dval > 5 * 1000) ) ||
684 (0 == (dval % 1000) ))
687 unit = _( /* time unit */ "s");
688 if ( ( (GNUNET_YES == do_round) &&
690 (0 == (dval % 60) ) )
693 unit = _( /* time unit */ "m");
694 if ( ( (GNUNET_YES == do_round) &&
699 unit = _( /* time unit */ "h");
700 if ( ( (GNUNET_YES == do_round) &&
706 unit = _( /* time unit */ "day");
708 unit = _( /* time unit */ "days");
714 GNUNET_snprintf (buf, sizeof (buf),
715 "%llu %s", dval, unit);
721 * "asctime", except for GNUnet time. Converts a GNUnet internal
722 * absolute time (which is in UTC) to a string in local time.
723 * Note that the returned value will be overwritten if this function
726 * @param t the absolute time to convert
727 * @return timestamp in human-readable form in local time
730 GNUNET_STRINGS_absolute_time_to_string (struct GNUNET_TIME_Absolute t)
732 static char buf[255];
736 if (t.abs_value_us == GNUNET_TIME_UNIT_FOREVER_ABS.abs_value_us)
737 return _("end of time");
738 tt = t.abs_value_us / 1000LL / 1000LL;
739 tp = localtime (&tt);
740 /* This is hacky, but i don't know a way to detect libc character encoding.
741 * Just expect utf8 from glibc these days.
742 * As for msvcrt, use the wide variant, which always returns utf16
743 * (otherwise we'd have to detect current codepage or use W32API character
744 * set conversion routines to convert to UTF8).
747 strftime (buf, sizeof (buf), "%a %b %d %H:%M:%S %Y", tp);
750 static wchar_t wbuf[255];
754 wcsftime (wbuf, sizeof (wbuf) / sizeof (wchar_t),
755 L"%a %b %d %H:%M:%S %Y", tp);
757 ssize = sizeof (buf);
758 conved = u16_to_u8 (wbuf, sizeof (wbuf) / sizeof (wchar_t),
759 (uint8_t *) buf, &ssize);
760 if (conved != (uint8_t *) buf)
762 strncpy (buf, (char *) conved, sizeof (buf));
774 * Returns a pointer to a part of filename (allocates nothing)!
776 * @param filename filename to extract basename from
777 * @return short (base) name of the file (that is, everything following the
778 * last directory separator in filename. If filename ends with a
779 * directory separator, the result will be a zero-length string.
780 * If filename has no directory separators, the result is filename
784 GNUNET_STRINGS_get_short_name (const char *filename)
786 const char *short_fn = filename;
788 while (NULL != (ss = strstr (short_fn, DIR_SEPARATOR_STR))
796 * Get the numeric value corresponding to a character.
798 * @param a a character
799 * @return corresponding numeric value
802 getValue__ (unsigned char a)
804 if ((a >= '0') && (a <= '9'))
806 if ((a >= 'A') && (a <= 'V'))
807 return (a - 'A' + 10);
808 if ((a >= 'a') && (a <= 'v'))
809 return (a - 'a' + 10);
815 * Convert binary data to ASCII encoding. The ASCII encoding is rather
816 * GNUnet specific. It was chosen such that it only uses characters
817 * in [0-9A-V], can be produced without complex arithmetics and uses a
818 * small number of characters.
819 * Does not append 0-terminator, but returns a pointer to the place where
820 * it should be placed, if needed.
822 * @param data data to encode
823 * @param size size of data (in bytes)
824 * @param out buffer to fill
825 * @param out_size size of the buffer. Must be large enough to hold
826 * ((size*8) + (((size*8) % 5) > 0 ? 5 - ((size*8) % 5) : 0)) / 5 bytes
827 * @return pointer to the next byte in 'out' or NULL on error.
830 GNUNET_STRINGS_data_to_string (const void *data, size_t size, char *out, size_t out_size)
833 * 32 characters for encoding
835 static char *encTable__ = "0123456789ABCDEFGHIJKLMNOPQRSTUV";
840 const unsigned char *udata;
842 GNUNET_assert (data != NULL);
843 GNUNET_assert (out != NULL);
845 if (out_size < (((size*8) + ((size*8) % 5)) % 5))
854 while ((rpos < size) || (vbit > 0))
856 if ((rpos < size) && (vbit < 5))
858 bits = (bits << 8) | udata[rpos++]; /* eat 8 more bits */
863 bits <<= (5 - vbit); /* zero-padding */
864 GNUNET_assert (vbit == ((size * 8) % 5));
867 if (wpos >= out_size)
872 out[wpos++] = encTable__[(bits >> (vbit - 5)) & 31];
875 GNUNET_assert (vbit == 0);
883 * Convert ASCII encoding back to data
884 * out_size must match exactly the size of the data before it was encoded.
886 * @param enc the encoding
887 * @param enclen number of characters in @a enc (without 0-terminator, which can be missing)
888 * @param out location where to store the decoded data
889 * @param out_size size of the output buffer @a out
890 * @return #GNUNET_OK on success, #GNUNET_SYSERR if result has the wrong encoding
893 GNUNET_STRINGS_string_to_data (const char *enc, size_t enclen,
894 void *out, size_t out_size)
903 unsigned int encoded_len = out_size * 8;
909 return GNUNET_SYSERR;
914 if ((encoded_len % 5) > 0)
916 vbit = encoded_len % 5; /* padding! */
918 bits = (ret = getValue__ (enc[--rpos])) >> (5 - (encoded_len % 5));
924 bits = (ret = getValue__ (enc[--rpos]));
926 if ((encoded_len + shift) / 5 != enclen)
927 return GNUNET_SYSERR;
929 return GNUNET_SYSERR;
935 return GNUNET_SYSERR;
937 bits = ((ret = getValue__ (enc[--rpos])) << vbit) | bits;
939 return GNUNET_SYSERR;
943 uout[--wpos] = (unsigned char) bits;
950 return GNUNET_SYSERR;
956 * Parse a path that might be an URI.
958 * @param path path to parse. Must be NULL-terminated.
959 * @param scheme_part a pointer to 'char *' where a pointer to a string that
960 * represents the URI scheme will be stored. Can be NULL. The string is
961 * allocated by the function, and should be freed by GNUNET_free() when
962 * it is no longer needed.
963 * @param path_part a pointer to 'const char *' where a pointer to the path
964 * part of the URI will be stored. Can be NULL. Points to the same block
965 * of memory as 'path', and thus must not be freed. Might point to '\0',
966 * if path part is zero-length.
967 * @return GNUNET_YES if it's an URI, GNUNET_NO otherwise. If 'path' is not
968 * an URI, '* scheme_part' and '*path_part' will remain unchanged
969 * (if they weren't NULL).
972 GNUNET_STRINGS_parse_uri (const char *path, char **scheme_part,
973 const char **path_part)
978 const char *post_scheme_part = NULL;
980 for (end = 0, i = 0; !end && i < len; i++)
985 if (path[i] == ':' && i > 0)
990 if (!((path[i] >= 'A' && path[i] <= 'Z') || (path[i] >= 'a' && path[i] <= 'z')
991 || (path[i] >= '0' && path[i] <= '9') || path[i] == '+' || path[i] == '-'
992 || (path[i] == '.')))
1005 post_scheme_part = &path[i];
1012 if (post_scheme_part == NULL)
1016 *scheme_part = GNUNET_malloc (post_scheme_part - path + 1);
1017 memcpy (*scheme_part, path, post_scheme_part - path);
1018 (*scheme_part)[post_scheme_part - path] = '\0';
1021 *path_part = post_scheme_part;
1027 * Check whether @a filename is absolute or not, and if it's an URI
1029 * @param filename filename to check
1030 * @param can_be_uri #GNUNET_YES to check for being URI, #GNUNET_NO - to
1031 * assume it's not URI
1032 * @param r_is_uri a pointer to an int that is set to #GNUNET_YES if @a filename
1033 * is URI and to #GNUNET_NO otherwise. Can be NULL. If @a can_be_uri is
1034 * not #GNUNET_YES, `* r_is_uri` is set to #GNUNET_NO.
1035 * @param r_uri_scheme a pointer to a char * that is set to a pointer to URI scheme.
1036 * The string is allocated by the function, and should be freed with
1037 * GNUNET_free(). Can be NULL.
1038 * @return #GNUNET_YES if @a filename is absolute, #GNUNET_NO otherwise.
1041 GNUNET_STRINGS_path_is_absolute (const char *filename,
1044 char **r_uri_scheme)
1049 const char *post_scheme_path;
1052 /* consider POSIX paths to be absolute too, even on W32,
1053 * as plibc expansion will fix them for us.
1055 if (filename[0] == '/')
1059 is_uri = GNUNET_STRINGS_parse_uri (filename, &uri, &post_scheme_path);
1065 *r_uri_scheme = uri;
1067 GNUNET_free_non_null (uri);
1069 len = strlen(post_scheme_path);
1070 /* Special check for file:///c:/blah
1071 * We want to parse 'c:/', not '/c:/'
1073 if (post_scheme_path[0] == '/' && len >= 3 && post_scheme_path[2] == ':')
1074 post_scheme_path = &post_scheme_path[1];
1076 return GNUNET_STRINGS_path_is_absolute (post_scheme_path, GNUNET_NO, NULL, NULL);
1082 *r_is_uri = GNUNET_NO;
1085 len = strlen (filename);
1087 ((filename[0] >= 'A' && filename[0] <= 'Z')
1088 || (filename[0] >= 'a' && filename[0] <= 'z'))
1089 && filename[1] == ':' && (filename[2] == '/' || filename[2] == '\\'))
1096 #define _IFMT 0170000 /* type of file */
1097 #define _IFLNK 0120000 /* symbolic link */
1098 #define S_ISLNK(m) (((m)&_IFMT) == _IFLNK)
1103 * Perform @a checks on @a filename.
1105 * @param filename file to check
1106 * @param checks checks to perform
1107 * @return #GNUNET_YES if all checks pass, #GNUNET_NO if at least one of them
1108 * fails, #GNUNET_SYSERR when a check can't be performed
1111 GNUNET_STRINGS_check_filename (const char *filename,
1112 enum GNUNET_STRINGS_FilenameCheck checks)
1115 if ( (NULL == filename) || (filename[0] == '\0') )
1116 return GNUNET_SYSERR;
1117 if (0 != (checks & GNUNET_STRINGS_CHECK_IS_ABSOLUTE))
1118 if (!GNUNET_STRINGS_path_is_absolute (filename, GNUNET_NO, NULL, NULL))
1120 if (0 != (checks & (GNUNET_STRINGS_CHECK_EXISTS
1121 | GNUNET_STRINGS_CHECK_IS_DIRECTORY
1122 | GNUNET_STRINGS_CHECK_IS_LINK)))
1124 if (0 != STAT (filename, &st))
1126 if (0 != (checks & GNUNET_STRINGS_CHECK_EXISTS))
1129 return GNUNET_SYSERR;
1132 if (0 != (checks & GNUNET_STRINGS_CHECK_IS_DIRECTORY))
1133 if (!S_ISDIR (st.st_mode))
1135 if (0 != (checks & GNUNET_STRINGS_CHECK_IS_LINK))
1136 if (!S_ISLNK (st.st_mode))
1143 * Tries to convert 'zt_addr' string to an IPv6 address.
1144 * The string is expected to have the format "[ABCD::01]:80".
1146 * @param zt_addr 0-terminated string. May be mangled by the function.
1147 * @param addrlen length of @a zt_addr (not counting 0-terminator).
1148 * @param r_buf a buffer to fill. Initially gets filled with zeroes,
1149 * then its sin6_port, sin6_family and sin6_addr are set appropriately.
1150 * @return #GNUNET_OK if conversion succeded.
1151 * #GNUNET_SYSERR otherwise, in which
1152 * case the contents of @a r_buf are undefined.
1155 GNUNET_STRINGS_to_address_ipv6 (const char *zt_addr,
1157 struct sockaddr_in6 *r_buf)
1159 char zbuf[addrlen + 1];
1165 return GNUNET_SYSERR;
1166 memcpy (zbuf, zt_addr, addrlen);
1169 GNUNET_log (GNUNET_ERROR_TYPE_WARNING,
1170 _("IPv6 address did not start with `['\n"));
1171 return GNUNET_SYSERR;
1173 zbuf[addrlen] = '\0';
1174 port_colon = strrchr (zbuf, ':');
1175 if (NULL == port_colon)
1177 GNUNET_log (GNUNET_ERROR_TYPE_WARNING,
1178 _("IPv6 address did contain ':' to separate port number\n"));
1179 return GNUNET_SYSERR;
1181 if (']' != *(port_colon - 1))
1183 GNUNET_log (GNUNET_ERROR_TYPE_WARNING,
1184 _("IPv6 address did contain ']' before ':' to separate port number\n"));
1185 return GNUNET_SYSERR;
1187 ret = SSCANF (port_colon, ":%u", &port);
1188 if ( (1 != ret) || (port > 65535) )
1190 GNUNET_log (GNUNET_ERROR_TYPE_WARNING,
1191 _("IPv6 address did contain a valid port number after the last ':'\n"));
1192 return GNUNET_SYSERR;
1194 *(port_colon-1) = '\0';
1195 memset (r_buf, 0, sizeof (struct sockaddr_in6));
1196 ret = inet_pton (AF_INET6, &zbuf[1], &r_buf->sin6_addr);
1199 GNUNET_log (GNUNET_ERROR_TYPE_WARNING,
1200 _("Invalid IPv6 address `%s': %s\n"),
1203 return GNUNET_SYSERR;
1205 r_buf->sin6_port = htons (port);
1206 r_buf->sin6_family = AF_INET6;
1207 #if HAVE_SOCKADDR_IN_SIN_LEN
1208 r_buf->sin6_len = (u_char) sizeof (struct sockaddr_in6);
1215 * Tries to convert 'zt_addr' string to an IPv4 address.
1216 * The string is expected to have the format "1.2.3.4:80".
1218 * @param zt_addr 0-terminated string. May be mangled by the function.
1219 * @param addrlen length of @a zt_addr (not counting 0-terminator).
1220 * @param r_buf a buffer to fill.
1221 * @return #GNUNET_OK if conversion succeded.
1222 * #GNUNET_SYSERR otherwise, in which case
1223 * the contents of @a r_buf are undefined.
1226 GNUNET_STRINGS_to_address_ipv4 (const char *zt_addr, uint16_t addrlen,
1227 struct sockaddr_in *r_buf)
1229 unsigned int temps[4];
1234 return GNUNET_SYSERR;
1235 cnt = SSCANF (zt_addr, "%u.%u.%u.%u:%u", &temps[0], &temps[1], &temps[2], &temps[3], &port);
1237 return GNUNET_SYSERR;
1238 for (cnt = 0; cnt < 4; cnt++)
1239 if (temps[cnt] > 0xFF)
1240 return GNUNET_SYSERR;
1242 return GNUNET_SYSERR;
1243 r_buf->sin_family = AF_INET;
1244 r_buf->sin_port = htons (port);
1245 r_buf->sin_addr.s_addr = htonl ((temps[0] << 24) + (temps[1] << 16) +
1246 (temps[2] << 8) + temps[3]);
1247 #if HAVE_SOCKADDR_IN_SIN_LEN
1248 r_buf->sin_len = (u_char) sizeof (struct sockaddr_in);
1255 * Tries to convert @a addr string to an IP (v4 or v6) address.
1256 * Will automatically decide whether to treat 'addr' as v4 or v6 address.
1258 * @param addr a string, may not be 0-terminated.
1259 * @param addrlen number of bytes in @a addr (if addr is 0-terminated,
1260 * 0-terminator should not be counted towards addrlen).
1261 * @param r_buf a buffer to fill.
1262 * @return #GNUNET_OK if conversion succeded. GNUNET_SYSERR otherwise, in which
1263 * case the contents of r_buf are undefined.
1266 GNUNET_STRINGS_to_address_ip (const char *addr,
1268 struct sockaddr_storage *r_buf)
1271 return GNUNET_STRINGS_to_address_ipv6 (addr,
1273 (struct sockaddr_in6 *) r_buf);
1274 return GNUNET_STRINGS_to_address_ipv4 (addr,
1276 (struct sockaddr_in *) r_buf);
1281 * Makes a copy of argv that consists of a single memory chunk that can be
1282 * freed with a single call to GNUNET_free();
1284 static char *const *
1285 _make_continuous_arg_copy (int argc,
1288 size_t argvsize = 0;
1292 for (i = 0; i < argc; i++)
1293 argvsize += strlen (argv[i]) + 1 + sizeof (char *);
1294 new_argv = GNUNET_malloc (argvsize + sizeof (char *));
1295 p = (char *) &new_argv[argc + 1];
1296 for (i = 0; i < argc; i++)
1299 strcpy (p, argv[i]);
1300 p += strlen (argv[i]) + 1;
1302 new_argv[argc] = NULL;
1303 return (char *const *) new_argv;
1308 * Returns utf-8 encoded arguments.
1309 * Does nothing (returns a copy of argc and argv) on any platform
1311 * Returned argv has u8argv[u8argc] == NULL.
1312 * Returned argv is a single memory block, and can be freed with a single
1313 * GNUNET_free() call.
1315 * @param argc argc (as given by main())
1316 * @param argv argv (as given by main())
1317 * @param u8argc a location to store new argc in (though it's th same as argc)
1318 * @param u8argv a location to store new argv in
1319 * @return #GNUNET_OK on success, #GNUNET_SYSERR on failure
1322 GNUNET_STRINGS_get_utf8_args (int argc, char *const *argv, int *u8argc, char *const **u8argv)
1329 char **split_u8argv;
1331 wcmd = GetCommandLineW ();
1333 return GNUNET_SYSERR;
1334 wargv = CommandLineToArgvW (wcmd, &wargc);
1336 return GNUNET_SYSERR;
1338 split_u8argv = GNUNET_malloc (argc * sizeof (char *));
1340 for (i = 0; i < wargc; i++)
1343 /* Hopefully it will allocate us NUL-terminated strings... */
1344 split_u8argv[i] = (char *) u16_to_u8 (wargv[i], wcslen (wargv[i]) + 1, NULL, &strl);
1345 if (NULL == split_u8argv[i])
1348 for (j = 0; j < i; j++)
1349 free (split_u8argv[j]);
1350 GNUNET_free (split_u8argv);
1352 return GNUNET_SYSERR;
1356 *u8argv = _make_continuous_arg_copy (wargc, split_u8argv);
1359 for (i = 0; i < wargc; i++)
1360 free (split_u8argv[i]);
1361 free (split_u8argv);
1364 char *const *new_argv = (char *const *) _make_continuous_arg_copy (argc, argv);
1373 * Parse the given port policy. The format is
1374 * "[!]SPORT[-DPORT]".
1376 * @param port_policy string to parse
1377 * @param pp policy to fill in
1378 * @return #GNUNET_OK on success, #GNUNET_SYSERR if the
1379 * @a port_policy is malformed
1382 parse_port_policy (const char *port_policy,
1383 struct GNUNET_STRINGS_PortPolicy *pp)
1393 pp->negate_portrange = GNUNET_YES;
1396 if (2 == sscanf (pos,
1405 GNUNET_log (GNUNET_ERROR_TYPE_WARNING,
1406 _("Port not in range\n"));
1407 return GNUNET_SYSERR;
1409 pp->start_port = (uint16_t) s;
1410 pp->end_port = (uint16_t) e;
1413 if (1 == sscanf (pos,
1421 GNUNET_log (GNUNET_ERROR_TYPE_WARNING,
1422 _("Port not in range\n"));
1423 return GNUNET_SYSERR;
1426 pp->start_port = (uint16_t) s;
1427 pp->end_port = (uint16_t) s;
1430 GNUNET_log (GNUNET_ERROR_TYPE_WARNING,
1431 _("Malformed port policy `%s'\n"),
1433 return GNUNET_SYSERR;
1438 * Parse an IPv4 network policy. The argument specifies a list of
1439 * subnets. The format is
1440 * <tt>(network[/netmask][:SPORT[-DPORT]];)*</tt> (no whitespace, must
1441 * be terminated with a semicolon). The network must be given in
1442 * dotted-decimal notation. The netmask can be given in CIDR notation
1443 * (/16) or in dotted-decimal (/255.255.0.0).
1445 * @param routeListX a string specifying the IPv4 subnets
1446 * @return the converted list, terminated with all zeros;
1447 * NULL if the synatx is flawed
1449 struct GNUNET_STRINGS_IPv4NetworkPolicy *
1450 GNUNET_STRINGS_parse_ipv4_policy (const char *routeListX)
1458 unsigned int temps[8];
1460 struct GNUNET_STRINGS_IPv4NetworkPolicy *result;
1465 if (NULL == routeListX)
1467 len = strlen (routeListX);
1470 routeList = GNUNET_strdup (routeListX);
1472 for (i = 0; i < len; i++)
1473 if (routeList[i] == ';')
1475 result = GNUNET_malloc (sizeof (struct GNUNET_STRINGS_IPv4NetworkPolicy) * (count + 1));
1480 for (colon = pos; ':' != routeList[colon]; colon++)
1481 if ( (';' == routeList[colon]) ||
1482 ('\0' == routeList[colon]) )
1484 for (end = colon; ';' != routeList[end]; end++)
1485 if ('\0' == routeList[end])
1487 if ('\0' == routeList[end])
1489 routeList[end] = '\0';
1490 if (':' == routeList[colon])
1492 routeList[colon] = '\0';
1493 if (GNUNET_OK != parse_port_policy (&routeList[colon + 1],
1498 SSCANF (&routeList[pos],
1499 "%u.%u.%u.%u/%u.%u.%u.%u",
1510 for (j = 0; j < 8; j++)
1511 if (temps[j] > 0xFF)
1513 LOG (GNUNET_ERROR_TYPE_WARNING,
1514 _("Invalid format for IP: `%s'\n"),
1516 GNUNET_free (result);
1517 GNUNET_free (routeList);
1520 result[i].network.s_addr =
1521 htonl ((temps[0] << 24) + (temps[1] << 16) + (temps[2] << 8) +
1523 result[i].netmask.s_addr =
1524 htonl ((temps[4] << 24) + (temps[5] << 16) + (temps[6] << 8) +
1530 /* try second notation */
1532 SSCANF (&routeList[pos],
1541 for (j = 0; j < 4; j++)
1542 if (temps[j] > 0xFF)
1544 LOG (GNUNET_ERROR_TYPE_WARNING,
1545 _("Invalid format for IP: `%s'\n"),
1547 GNUNET_free (result);
1548 GNUNET_free (routeList);
1551 result[i].network.s_addr =
1552 htonl ((temps[0] << 24) + (temps[1] << 16) + (temps[2] << 8) +
1554 if ((slash <= 32) && (slash >= 0))
1556 result[i].netmask.s_addr = 0;
1559 result[i].netmask.s_addr =
1560 (result[i].netmask.s_addr >> 1) + 0x80000000;
1563 result[i].netmask.s_addr = htonl (result[i].netmask.s_addr);
1570 LOG (GNUNET_ERROR_TYPE_WARNING,
1571 _("Invalid network notation ('/%d' is not legal in IPv4 CIDR)."),
1573 GNUNET_free (result);
1574 GNUNET_free (routeList);
1575 return NULL; /* error */
1578 /* try third notation */
1581 SSCANF (&routeList[pos],
1589 for (j = 0; j < 4; j++)
1590 if (temps[j] > 0xFF)
1592 LOG (GNUNET_ERROR_TYPE_WARNING,
1593 _("Invalid format for IP: `%s'\n"),
1595 GNUNET_free (result);
1596 GNUNET_free (routeList);
1599 result[i].network.s_addr =
1600 htonl ((temps[0] << 24) + (temps[1] << 16) + (temps[2] << 8) +
1602 result[i].netmask.s_addr = 0;
1605 result[i].netmask.s_addr = (result[i].netmask.s_addr >> 1) + 0x80000000;
1608 result[i].netmask.s_addr = htonl (result[i].netmask.s_addr);
1613 LOG (GNUNET_ERROR_TYPE_WARNING,
1614 _("Invalid format for IP: `%s'\n"),
1616 GNUNET_free (result);
1617 GNUNET_free (routeList);
1618 return NULL; /* error */
1620 if (pos < strlen (routeList))
1622 LOG (GNUNET_ERROR_TYPE_WARNING,
1623 _("Invalid format: `%s'\n"),
1625 GNUNET_free (result);
1626 GNUNET_free (routeList);
1627 return NULL; /* oops */
1629 GNUNET_free (routeList);
1630 return result; /* ok */
1635 * Parse an IPv6 network policy. The argument specifies a list of
1636 * subnets. The format is <tt>(network[/netmask[:SPORT[-DPORT]]];)*</tt>
1637 * (no whitespace, must be terminated with a semicolon). The network
1638 * must be given in colon-hex notation. The netmask must be given in
1639 * CIDR notation (/16) or can be omitted to specify a single host.
1640 * Note that the netmask is mandatory if ports are specified.
1642 * @param routeListX a string specifying the policy
1643 * @return the converted list, 0-terminated, NULL if the synatx is flawed
1645 struct GNUNET_STRINGS_IPv6NetworkPolicy *
1646 GNUNET_STRINGS_parse_ipv6_policy (const char *routeListX)
1656 struct GNUNET_STRINGS_IPv6NetworkPolicy *result;
1662 if (NULL == routeListX)
1664 len = strlen (routeListX);
1667 routeList = GNUNET_strdup (routeListX);
1669 for (i = 0; i < len; i++)
1670 if (';' == routeList[i])
1672 if (';' != routeList[len - 1])
1674 LOG (GNUNET_ERROR_TYPE_WARNING,
1675 _("Invalid network notation (does not end with ';': `%s')\n"),
1677 GNUNET_free (routeList);
1681 result = GNUNET_malloc (sizeof (struct GNUNET_STRINGS_IPv6NetworkPolicy) * (count + 1));
1687 while (';' != routeList[pos])
1690 while ((slash >= start) && (routeList[slash] != '/'))
1695 memset (&result[i].netmask,
1697 sizeof (struct in6_addr));
1702 routeList[pos] = '\0';
1703 for (colon = pos; ':' != routeList[colon]; colon--)
1704 if ('/' == routeList[colon])
1706 if (':' == routeList[colon])
1708 routeList[colon] = '\0';
1709 if (GNUNET_OK != parse_port_policy (&routeList[colon + 1],
1712 GNUNET_free (result);
1713 GNUNET_free (routeList);
1717 ret = inet_pton (AF_INET6, &routeList[slash + 1], &result[i].netmask);
1721 if ((1 != SSCANF (&routeList[slash + 1], "%u", &bits)) || (bits > 128))
1724 LOG (GNUNET_ERROR_TYPE_WARNING,
1725 _("Wrong format `%s' for netmask\n"),
1726 &routeList[slash + 1]);
1730 LOG_STRERROR (GNUNET_ERROR_TYPE_WARNING, "inet_pton");
1732 GNUNET_free (result);
1733 GNUNET_free (routeList);
1739 result[i].netmask.s6_addr[off++] = 0xFF;
1744 result[i].netmask.s6_addr[off] =
1745 (result[i].netmask.s6_addr[off] >> 1) + 0x80;
1750 routeList[slash] = '\0';
1751 ret = inet_pton (AF_INET6, &routeList[start], &result[i].network);
1755 LOG (GNUNET_ERROR_TYPE_WARNING,
1756 _("Wrong format `%s' for network\n"),
1757 &routeList[slash + 1]);
1759 LOG_STRERROR (GNUNET_ERROR_TYPE_ERROR,
1761 GNUNET_free (result);
1762 GNUNET_free (routeList);
1768 GNUNET_free (routeList);
1774 /** ******************** Base64 encoding ***********/
1776 #define FILLCHAR '='
1778 "ABCDEFGHIJKLMNOPQRSTUVWXYZ" "abcdefghijklmnopqrstuvwxyz" "0123456789+/";
1782 * Encode into Base64.
1784 * @param data the data to encode
1785 * @param len the length of the input
1786 * @param output where to write the output (*output should be NULL,
1788 * @return the size of the output
1791 GNUNET_STRINGS_base64_encode (const char *data,
1801 opt = GNUNET_malloc (2 + (len * 4 / 3) + 8);
1803 for (i = 0; i < len; ++i)
1805 c = (data[i] >> 2) & 0x3f;
1806 opt[ret++] = cvt[(int) c];
1807 c = (data[i] << 4) & 0x3f;
1809 c |= (data[i] >> 4) & 0x0f;
1810 opt[ret++] = cvt[(int) c];
1813 c = (data[i] << 2) & 0x3f;
1815 c |= (data[i] >> 6) & 0x03;
1816 opt[ret++] = cvt[(int) c];
1821 opt[ret++] = FILLCHAR;
1826 opt[ret++] = cvt[(int) c];
1830 opt[ret++] = FILLCHAR;
1833 opt[ret++] = FILLCHAR;
1837 #define cvtfind(a)( (((a) >= 'A')&&((a) <= 'Z'))? (a)-'A'\
1838 :(((a)>='a')&&((a)<='z')) ? (a)-'a'+26\
1839 :(((a)>='0')&&((a)<='9')) ? (a)-'0'+52\
1841 :((a) == '/') ? 63 : -1)
1845 * Decode from Base64.
1847 * @param data the data to encode
1848 * @param len the length of the input
1849 * @param output where to write the output (*output should be NULL,
1851 * @return the size of the output
1854 GNUNET_STRINGS_base64_decode (const char *data,
1855 size_t len, char **output)
1862 #define CHECK_CRLF while (data[i] == '\r' || data[i] == '\n') {\
1863 GNUNET_log(GNUNET_ERROR_TYPE_DEBUG | GNUNET_ERROR_TYPE_BULK, "ignoring CR/LF\n"); \
1865 if (i >= len) goto END; \
1868 *output = GNUNET_malloc ((len * 3 / 4) + 8);
1869 GNUNET_log (GNUNET_ERROR_TYPE_DEBUG,
1870 "base64_decode decoding len=%d\n",
1872 for (i = 0; i < len; ++i)
1875 if (FILLCHAR == data[i])
1877 c = (char) cvtfind (data[i]);
1880 c1 = (char) cvtfind (data[i]);
1881 c = (c << 2) | ((c1 >> 4) & 0x3);
1882 (*output)[ret++] = c;
1889 c = (char) cvtfind (c);
1890 c1 = ((c1 << 4) & 0xf0) | ((c >> 2) & 0xf);
1891 (*output)[ret++] = c1;
1900 c1 = (char) cvtfind (c1);
1901 c = ((c << 6) & 0xc0) | c1;
1902 (*output)[ret++] = c;
1913 /* end of strings.c */