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)
354 if (0 == strcasecmp ("end of time", fancy_time))
356 *atime = GNUNET_TIME_UNIT_FOREVER_ABS;
359 memset (&tv, 0, sizeof (tv));
360 if ( (NULL == strptime (fancy_time, "%a %b %d %H:%M:%S %Y", &tv)) &&
361 (NULL == strptime (fancy_time, "%c", &tv)) &&
362 (NULL == strptime (fancy_time, "%Ec", &tv)) &&
363 (NULL == strptime (fancy_time, "%Y-%m-%d %H:%M:%S", &tv)) &&
364 (NULL == strptime (fancy_time, "%Y-%m-%d %H:%M", &tv)) &&
365 (NULL == strptime (fancy_time, "%x", &tv)) &&
366 (NULL == strptime (fancy_time, "%Ex", &tv)) &&
367 (NULL == strptime (fancy_time, "%Y-%m-%d", &tv)) &&
368 (NULL == strptime (fancy_time, "%Y-%m", &tv)) &&
369 (NULL == strptime (fancy_time, "%Y", &tv)) )
370 return GNUNET_SYSERR;
372 atime->abs_value_us = (uint64_t) ((uint64_t) t * 1000LL * 1000LL);
375 atime->abs_value_us += 1000LL * 1000LL * tp->tm_gmtoff;
377 atime->abs_value_us -= 1000LL * 1000LL * timezone;
378 #elif defined WINDOWS
381 TIME_ZONE_INFORMATION tzi;
382 tzv = GetTimeZoneInformation (&tzi);
383 if (TIME_ZONE_ID_INVALID != tzv)
385 atime->abs_value_us -= 1000LL * 1000LL * tzi.Bias * 60LL;
394 * Convert the len characters long character sequence
395 * given in input that is in the given input charset
396 * to a string in given output charset.
398 * @param input input string
399 * @param len number of bytes in @a input
400 * @param input_charset character set used for @a input
401 * @param output_charset desired character set for the return value
402 * @return the converted string (0-terminated),
403 * if conversion fails, a copy of the orignal
404 * string is returned.
407 GNUNET_STRINGS_conv (const char *input,
409 const char *input_charset,
410 const char *output_charset)
414 char *encoded_string;
415 size_t u8_string_length;
416 size_t encoded_string_length;
418 u8_string = u8_conv_from_encoding (input_charset,
423 if (NULL == u8_string)
425 LOG_STRERROR (GNUNET_ERROR_TYPE_WARNING, "u8_conv_from_encoding");
428 if (0 == strcmp (output_charset, "UTF-8"))
430 ret = GNUNET_malloc (u8_string_length + 1);
431 memcpy (ret, u8_string, u8_string_length);
432 ret[u8_string_length] = '\0';
436 encoded_string = u8_conv_to_encoding (output_charset, iconveh_error,
437 u8_string, u8_string_length,
439 &encoded_string_length);
441 if (NULL == encoded_string)
443 LOG_STRERROR (GNUNET_ERROR_TYPE_WARNING, "u8_conv_to_encoding");
446 ret = GNUNET_malloc (encoded_string_length + 1);
447 memcpy (ret, encoded_string, encoded_string_length);
448 ret[encoded_string_length] = '\0';
449 free (encoded_string);
452 LOG (GNUNET_ERROR_TYPE_WARNING, _("Character sets requested were `%s'->`%s'\n"),
453 "UTF-8", output_charset);
454 ret = GNUNET_malloc (len + 1);
455 memcpy (ret, input, len);
462 * Convert the len characters long character sequence
463 * given in input that is in the given charset
466 * @param input the input string (not necessarily 0-terminated)
467 * @param len the number of bytes in the @a input
468 * @param charset character set to convert from
469 * @return the converted string (0-terminated),
470 * if conversion fails, a copy of the orignal
471 * string is returned.
474 GNUNET_STRINGS_to_utf8 (const char *input,
478 return GNUNET_STRINGS_conv (input, len, charset, "UTF-8");
483 * Convert the len bytes-long UTF-8 string
484 * given in input to the given charset.
486 * @param input the input string (not necessarily 0-terminated)
487 * @param len the number of bytes in the @a input
488 * @param charset character set to convert to
489 * @return the converted string (0-terminated),
490 * if conversion fails, a copy of the orignal
491 * string is returned.
494 GNUNET_STRINGS_from_utf8 (const char *input,
498 return GNUNET_STRINGS_conv (input, len, "UTF-8", charset);
503 * Convert the utf-8 input string to lowercase.
504 * Output needs to be allocated appropriately.
506 * @param input input string
507 * @param output output buffer
510 GNUNET_STRINGS_utf8_tolower (const char *input,
516 tmp_in = u8_tolower ((uint8_t*)input, strlen ((char *) input),
517 NULL, UNINORM_NFD, NULL, &len);
518 memcpy(output, tmp_in, len);
525 * Convert the utf-8 input string to uppercase.
526 * Output needs to be allocated appropriately.
528 * @param input input string
529 * @param output output buffer
532 GNUNET_STRINGS_utf8_toupper(const char *input,
538 tmp_in = u8_toupper ((uint8_t*)input, strlen ((char *) input),
539 NULL, UNINORM_NFD, NULL, &len);
540 memcpy (output, tmp_in, len);
547 * Complete filename (a la shell) from abbrevition.
548 * @param fil the name of the file, may contain ~/ or
549 * be relative to the current directory
550 * @returns the full file name,
551 * NULL is returned on error
554 GNUNET_STRINGS_filename_expand (const char *fil)
571 if (fil[0] == DIR_SEPARATOR)
572 /* absolute path, just copy */
573 return GNUNET_strdup (fil);
576 fm = getenv ("HOME");
579 LOG (GNUNET_ERROR_TYPE_WARNING,
580 _("Failed to expand `$HOME': environment variable `HOME' not set"));
583 fm = GNUNET_strdup (fm);
584 /* do not copy '~' */
587 /* skip over dir seperator to be consistent */
588 if (fil_ptr[0] == DIR_SEPARATOR)
599 buffer = GNUNET_malloc (len);
600 if (getcwd (buffer, len) != NULL)
605 if ((errno == ERANGE) && (len < 1024 * 1024 * 4))
608 GNUNET_free (buffer);
611 GNUNET_free (buffer);
616 LOG_STRERROR (GNUNET_ERROR_TYPE_WARNING, "getcwd");
617 buffer = getenv ("PWD"); /* alternative */
619 fm = GNUNET_strdup (buffer);
622 fm = GNUNET_strdup ("./"); /* give up */
624 n = strlen (fm) + 1 + strlen (fil_ptr) + 1;
625 buffer = GNUNET_malloc (n);
626 GNUNET_snprintf (buffer, n, "%s%s%s", fm,
627 (fm[strlen (fm) - 1] ==
628 DIR_SEPARATOR) ? "" : DIR_SEPARATOR_STR, fil_ptr);
632 fn = GNUNET_malloc (MAX_PATH + 1);
634 if ((lRet = plibc_conv_to_win_path (fil, fn)) != ERROR_SUCCESS)
636 SetErrnoFromWinError (lRet);
637 LOG_STRERROR (GNUNET_ERROR_TYPE_WARNING, "plibc_conv_to_win_path");
640 /* is the path relative? */
641 if ((strncmp (fn + 1, ":\\", 2) != 0) && (strncmp (fn, "\\\\", 2) != 0))
643 char szCurDir[MAX_PATH + 1];
645 lRet = GetCurrentDirectory (MAX_PATH + 1, szCurDir);
646 if (lRet + strlen (fn) + 1 > (MAX_PATH + 1))
648 SetErrnoFromWinError (ERROR_BUFFER_OVERFLOW);
649 LOG_STRERROR (GNUNET_ERROR_TYPE_WARNING, "GetCurrentDirectory");
652 buffer = GNUNET_malloc (MAX_PATH + 1);
653 GNUNET_snprintf (buffer, MAX_PATH + 1, "%s\\%s", szCurDir, fn);
664 * Give relative time in human-readable fancy format.
665 * This is one of the very few calls in the entire API that is
668 * @param delta time in milli seconds
669 * @param do_round are we allowed to round a bit?
670 * @return time as human-readable string
673 GNUNET_STRINGS_relative_time_to_string (struct GNUNET_TIME_Relative delta,
676 static char buf[128];
677 const char *unit = _( /* time unit */ "µs");
678 uint64_t dval = delta.rel_value_us;
680 if (GNUNET_TIME_UNIT_FOREVER_REL.rel_value_us == delta.rel_value_us)
682 if (0 == delta.rel_value_us)
684 if ( ( (GNUNET_YES == do_round) &&
685 (dval > 5 * 1000) ) ||
686 (0 == (dval % 1000) ))
689 unit = _( /* time unit */ "ms");
690 if ( ( (GNUNET_YES == do_round) &&
691 (dval > 5 * 1000) ) ||
692 (0 == (dval % 1000) ))
695 unit = _( /* time unit */ "s");
696 if ( ( (GNUNET_YES == do_round) &&
698 (0 == (dval % 60) ) )
701 unit = _( /* time unit */ "m");
702 if ( ( (GNUNET_YES == do_round) &&
707 unit = _( /* time unit */ "h");
708 if ( ( (GNUNET_YES == do_round) &&
714 unit = _( /* time unit */ "day");
716 unit = _( /* time unit */ "days");
722 GNUNET_snprintf (buf, sizeof (buf),
723 "%llu %s", dval, unit);
729 * "asctime", except for GNUnet time. Converts a GNUnet internal
730 * absolute time (which is in UTC) to a string in local time.
731 * Note that the returned value will be overwritten if this function
734 * @param t the absolute time to convert
735 * @return timestamp in human-readable form in local time
738 GNUNET_STRINGS_absolute_time_to_string (struct GNUNET_TIME_Absolute t)
740 static char buf[255];
744 if (t.abs_value_us == GNUNET_TIME_UNIT_FOREVER_ABS.abs_value_us)
745 return _("end of time");
746 tt = t.abs_value_us / 1000LL / 1000LL;
748 /* This is hacky, but i don't know a way to detect libc character encoding.
749 * Just expect utf8 from glibc these days.
750 * As for msvcrt, use the wide variant, which always returns utf16
751 * (otherwise we'd have to detect current codepage or use W32API character
752 * set conversion routines to convert to UTF8).
755 strftime (buf, sizeof (buf), "%a %b %d %H:%M:%S %Y", tp);
758 static wchar_t wbuf[255];
762 wcsftime (wbuf, sizeof (wbuf) / sizeof (wchar_t),
763 L"%a %b %d %H:%M:%S %Y", tp);
765 ssize = sizeof (buf);
766 conved = u16_to_u8 (wbuf, sizeof (wbuf) / sizeof (wchar_t),
767 (uint8_t *) buf, &ssize);
768 if (conved != (uint8_t *) buf)
770 strncpy (buf, (char *) conved, sizeof (buf));
782 * Returns a pointer to a part of filename (allocates nothing)!
784 * @param filename filename to extract basename from
785 * @return short (base) name of the file (that is, everything following the
786 * last directory separator in filename. If filename ends with a
787 * directory separator, the result will be a zero-length string.
788 * If filename has no directory separators, the result is filename
792 GNUNET_STRINGS_get_short_name (const char *filename)
794 const char *short_fn = filename;
796 while (NULL != (ss = strstr (short_fn, DIR_SEPARATOR_STR))
804 * Get the numeric value corresponding to a character.
806 * @param a a character
807 * @return corresponding numeric value
810 getValue__ (unsigned char a)
812 if ((a >= '0') && (a <= '9'))
814 if ((a >= 'A') && (a <= 'V'))
815 return (a - 'A' + 10);
816 if ((a >= 'a') && (a <= 'v'))
817 return (a - 'a' + 10);
823 * Convert binary data to ASCII encoding. The ASCII encoding is rather
824 * GNUnet specific. It was chosen such that it only uses characters
825 * in [0-9A-V], can be produced without complex arithmetics and uses a
826 * small number of characters.
827 * Does not append 0-terminator, but returns a pointer to the place where
828 * it should be placed, if needed.
830 * @param data data to encode
831 * @param size size of data (in bytes)
832 * @param out buffer to fill
833 * @param out_size size of the buffer. Must be large enough to hold
834 * ((size*8) + (((size*8) % 5) > 0 ? 5 - ((size*8) % 5) : 0)) / 5 bytes
835 * @return pointer to the next byte in 'out' or NULL on error.
838 GNUNET_STRINGS_data_to_string (const void *data, size_t size, char *out, size_t out_size)
841 * 32 characters for encoding
843 static char *encTable__ = "0123456789ABCDEFGHIJKLMNOPQRSTUV";
848 const unsigned char *udata;
850 GNUNET_assert (data != NULL);
851 GNUNET_assert (out != NULL);
853 if (out_size < (((size*8) + ((size*8) % 5)) % 5))
862 while ((rpos < size) || (vbit > 0))
864 if ((rpos < size) && (vbit < 5))
866 bits = (bits << 8) | udata[rpos++]; /* eat 8 more bits */
871 bits <<= (5 - vbit); /* zero-padding */
872 GNUNET_assert (vbit == ((size * 8) % 5));
875 if (wpos >= out_size)
880 out[wpos++] = encTable__[(bits >> (vbit - 5)) & 31];
883 GNUNET_assert (vbit == 0);
891 * Convert ASCII encoding back to data
892 * out_size must match exactly the size of the data before it was encoded.
894 * @param enc the encoding
895 * @param enclen number of characters in @a enc (without 0-terminator, which can be missing)
896 * @param out location where to store the decoded data
897 * @param out_size size of the output buffer @a out
898 * @return #GNUNET_OK on success, #GNUNET_SYSERR if result has the wrong encoding
901 GNUNET_STRINGS_string_to_data (const char *enc, size_t enclen,
902 void *out, size_t out_size)
911 unsigned int encoded_len = out_size * 8;
917 return GNUNET_SYSERR;
922 if ((encoded_len % 5) > 0)
924 vbit = encoded_len % 5; /* padding! */
926 bits = (ret = getValue__ (enc[--rpos])) >> (5 - (encoded_len % 5));
932 bits = (ret = getValue__ (enc[--rpos]));
934 if ((encoded_len + shift) / 5 != enclen)
935 return GNUNET_SYSERR;
937 return GNUNET_SYSERR;
943 return GNUNET_SYSERR;
945 bits = ((ret = getValue__ (enc[--rpos])) << vbit) | bits;
947 return GNUNET_SYSERR;
951 uout[--wpos] = (unsigned char) bits;
958 return GNUNET_SYSERR;
964 * Parse a path that might be an URI.
966 * @param path path to parse. Must be NULL-terminated.
967 * @param scheme_part a pointer to 'char *' where a pointer to a string that
968 * represents the URI scheme will be stored. Can be NULL. The string is
969 * allocated by the function, and should be freed by GNUNET_free() when
970 * it is no longer needed.
971 * @param path_part a pointer to 'const char *' where a pointer to the path
972 * part of the URI will be stored. Can be NULL. Points to the same block
973 * of memory as 'path', and thus must not be freed. Might point to '\0',
974 * if path part is zero-length.
975 * @return GNUNET_YES if it's an URI, GNUNET_NO otherwise. If 'path' is not
976 * an URI, '* scheme_part' and '*path_part' will remain unchanged
977 * (if they weren't NULL).
980 GNUNET_STRINGS_parse_uri (const char *path, char **scheme_part,
981 const char **path_part)
986 const char *post_scheme_part = NULL;
988 for (end = 0, i = 0; !end && i < len; i++)
993 if (path[i] == ':' && i > 0)
998 if (!((path[i] >= 'A' && path[i] <= 'Z') || (path[i] >= 'a' && path[i] <= 'z')
999 || (path[i] >= '0' && path[i] <= '9') || path[i] == '+' || path[i] == '-'
1000 || (path[i] == '.')))
1013 post_scheme_part = &path[i];
1020 if (post_scheme_part == NULL)
1024 *scheme_part = GNUNET_malloc (post_scheme_part - path + 1);
1025 memcpy (*scheme_part, path, post_scheme_part - path);
1026 (*scheme_part)[post_scheme_part - path] = '\0';
1029 *path_part = post_scheme_part;
1035 * Check whether @a filename is absolute or not, and if it's an URI
1037 * @param filename filename to check
1038 * @param can_be_uri #GNUNET_YES to check for being URI, #GNUNET_NO - to
1039 * assume it's not URI
1040 * @param r_is_uri a pointer to an int that is set to #GNUNET_YES if @a filename
1041 * is URI and to #GNUNET_NO otherwise. Can be NULL. If @a can_be_uri is
1042 * not #GNUNET_YES, `* r_is_uri` is set to #GNUNET_NO.
1043 * @param r_uri_scheme a pointer to a char * that is set to a pointer to URI scheme.
1044 * The string is allocated by the function, and should be freed with
1045 * GNUNET_free(). Can be NULL.
1046 * @return #GNUNET_YES if @a filename is absolute, #GNUNET_NO otherwise.
1049 GNUNET_STRINGS_path_is_absolute (const char *filename,
1052 char **r_uri_scheme)
1057 const char *post_scheme_path;
1060 /* consider POSIX paths to be absolute too, even on W32,
1061 * as plibc expansion will fix them for us.
1063 if (filename[0] == '/')
1067 is_uri = GNUNET_STRINGS_parse_uri (filename, &uri, &post_scheme_path);
1073 *r_uri_scheme = uri;
1075 GNUNET_free_non_null (uri);
1077 len = strlen(post_scheme_path);
1078 /* Special check for file:///c:/blah
1079 * We want to parse 'c:/', not '/c:/'
1081 if (post_scheme_path[0] == '/' && len >= 3 && post_scheme_path[2] == ':')
1082 post_scheme_path = &post_scheme_path[1];
1084 return GNUNET_STRINGS_path_is_absolute (post_scheme_path, GNUNET_NO, NULL, NULL);
1090 *r_is_uri = GNUNET_NO;
1093 len = strlen (filename);
1095 ((filename[0] >= 'A' && filename[0] <= 'Z')
1096 || (filename[0] >= 'a' && filename[0] <= 'z'))
1097 && filename[1] == ':' && (filename[2] == '/' || filename[2] == '\\'))
1104 #define _IFMT 0170000 /* type of file */
1105 #define _IFLNK 0120000 /* symbolic link */
1106 #define S_ISLNK(m) (((m)&_IFMT) == _IFLNK)
1111 * Perform @a checks on @a filename.
1113 * @param filename file to check
1114 * @param checks checks to perform
1115 * @return #GNUNET_YES if all checks pass, #GNUNET_NO if at least one of them
1116 * fails, #GNUNET_SYSERR when a check can't be performed
1119 GNUNET_STRINGS_check_filename (const char *filename,
1120 enum GNUNET_STRINGS_FilenameCheck checks)
1123 if ( (NULL == filename) || (filename[0] == '\0') )
1124 return GNUNET_SYSERR;
1125 if (0 != (checks & GNUNET_STRINGS_CHECK_IS_ABSOLUTE))
1126 if (!GNUNET_STRINGS_path_is_absolute (filename, GNUNET_NO, NULL, NULL))
1128 if (0 != (checks & (GNUNET_STRINGS_CHECK_EXISTS
1129 | GNUNET_STRINGS_CHECK_IS_DIRECTORY
1130 | GNUNET_STRINGS_CHECK_IS_LINK)))
1132 if (0 != STAT (filename, &st))
1134 if (0 != (checks & GNUNET_STRINGS_CHECK_EXISTS))
1137 return GNUNET_SYSERR;
1140 if (0 != (checks & GNUNET_STRINGS_CHECK_IS_DIRECTORY))
1141 if (!S_ISDIR (st.st_mode))
1143 if (0 != (checks & GNUNET_STRINGS_CHECK_IS_LINK))
1144 if (!S_ISLNK (st.st_mode))
1151 * Tries to convert 'zt_addr' string to an IPv6 address.
1152 * The string is expected to have the format "[ABCD::01]:80".
1154 * @param zt_addr 0-terminated string. May be mangled by the function.
1155 * @param addrlen length of @a zt_addr (not counting 0-terminator).
1156 * @param r_buf a buffer to fill. Initially gets filled with zeroes,
1157 * then its sin6_port, sin6_family and sin6_addr are set appropriately.
1158 * @return #GNUNET_OK if conversion succeded.
1159 * #GNUNET_SYSERR otherwise, in which
1160 * case the contents of @a r_buf are undefined.
1163 GNUNET_STRINGS_to_address_ipv6 (const char *zt_addr,
1165 struct sockaddr_in6 *r_buf)
1167 char zbuf[addrlen + 1];
1173 return GNUNET_SYSERR;
1174 memcpy (zbuf, zt_addr, addrlen);
1177 GNUNET_log (GNUNET_ERROR_TYPE_WARNING,
1178 _("IPv6 address did not start with `['\n"));
1179 return GNUNET_SYSERR;
1181 zbuf[addrlen] = '\0';
1182 port_colon = strrchr (zbuf, ':');
1183 if (NULL == port_colon)
1185 GNUNET_log (GNUNET_ERROR_TYPE_WARNING,
1186 _("IPv6 address did contain ':' to separate port number\n"));
1187 return GNUNET_SYSERR;
1189 if (']' != *(port_colon - 1))
1191 GNUNET_log (GNUNET_ERROR_TYPE_WARNING,
1192 _("IPv6 address did contain ']' before ':' to separate port number\n"));
1193 return GNUNET_SYSERR;
1195 ret = SSCANF (port_colon, ":%u", &port);
1196 if ( (1 != ret) || (port > 65535) )
1198 GNUNET_log (GNUNET_ERROR_TYPE_WARNING,
1199 _("IPv6 address did contain a valid port number after the last ':'\n"));
1200 return GNUNET_SYSERR;
1202 *(port_colon-1) = '\0';
1203 memset (r_buf, 0, sizeof (struct sockaddr_in6));
1204 ret = inet_pton (AF_INET6, &zbuf[1], &r_buf->sin6_addr);
1207 GNUNET_log (GNUNET_ERROR_TYPE_WARNING,
1208 _("Invalid IPv6 address `%s': %s\n"),
1211 return GNUNET_SYSERR;
1213 r_buf->sin6_port = htons (port);
1214 r_buf->sin6_family = AF_INET6;
1215 #if HAVE_SOCKADDR_IN_SIN_LEN
1216 r_buf->sin6_len = (u_char) sizeof (struct sockaddr_in6);
1223 * Tries to convert 'zt_addr' string to an IPv4 address.
1224 * The string is expected to have the format "1.2.3.4:80".
1226 * @param zt_addr 0-terminated string. May be mangled by the function.
1227 * @param addrlen length of @a zt_addr (not counting 0-terminator).
1228 * @param r_buf a buffer to fill.
1229 * @return #GNUNET_OK if conversion succeded.
1230 * #GNUNET_SYSERR otherwise, in which case
1231 * the contents of @a r_buf are undefined.
1234 GNUNET_STRINGS_to_address_ipv4 (const char *zt_addr, uint16_t addrlen,
1235 struct sockaddr_in *r_buf)
1237 unsigned int temps[4];
1242 return GNUNET_SYSERR;
1243 cnt = SSCANF (zt_addr, "%u.%u.%u.%u:%u", &temps[0], &temps[1], &temps[2], &temps[3], &port);
1245 return GNUNET_SYSERR;
1246 for (cnt = 0; cnt < 4; cnt++)
1247 if (temps[cnt] > 0xFF)
1248 return GNUNET_SYSERR;
1250 return GNUNET_SYSERR;
1251 r_buf->sin_family = AF_INET;
1252 r_buf->sin_port = htons (port);
1253 r_buf->sin_addr.s_addr = htonl ((temps[0] << 24) + (temps[1] << 16) +
1254 (temps[2] << 8) + temps[3]);
1255 #if HAVE_SOCKADDR_IN_SIN_LEN
1256 r_buf->sin_len = (u_char) sizeof (struct sockaddr_in);
1263 * Tries to convert @a addr string to an IP (v4 or v6) address.
1264 * Will automatically decide whether to treat 'addr' as v4 or v6 address.
1266 * @param addr a string, may not be 0-terminated.
1267 * @param addrlen number of bytes in @a addr (if addr is 0-terminated,
1268 * 0-terminator should not be counted towards addrlen).
1269 * @param r_buf a buffer to fill.
1270 * @return #GNUNET_OK if conversion succeded. GNUNET_SYSERR otherwise, in which
1271 * case the contents of r_buf are undefined.
1274 GNUNET_STRINGS_to_address_ip (const char *addr,
1276 struct sockaddr_storage *r_buf)
1279 return GNUNET_STRINGS_to_address_ipv6 (addr,
1281 (struct sockaddr_in6 *) r_buf);
1282 return GNUNET_STRINGS_to_address_ipv4 (addr,
1284 (struct sockaddr_in *) r_buf);
1289 * Makes a copy of argv that consists of a single memory chunk that can be
1290 * freed with a single call to GNUNET_free();
1292 static char *const *
1293 _make_continuous_arg_copy (int argc,
1296 size_t argvsize = 0;
1300 for (i = 0; i < argc; i++)
1301 argvsize += strlen (argv[i]) + 1 + sizeof (char *);
1302 new_argv = GNUNET_malloc (argvsize + sizeof (char *));
1303 p = (char *) &new_argv[argc + 1];
1304 for (i = 0; i < argc; i++)
1307 strcpy (p, argv[i]);
1308 p += strlen (argv[i]) + 1;
1310 new_argv[argc] = NULL;
1311 return (char *const *) new_argv;
1316 * Returns utf-8 encoded arguments.
1317 * Does nothing (returns a copy of argc and argv) on any platform
1319 * Returned argv has u8argv[u8argc] == NULL.
1320 * Returned argv is a single memory block, and can be freed with a single
1321 * GNUNET_free() call.
1323 * @param argc argc (as given by main())
1324 * @param argv argv (as given by main())
1325 * @param u8argc a location to store new argc in (though it's th same as argc)
1326 * @param u8argv a location to store new argv in
1327 * @return #GNUNET_OK on success, #GNUNET_SYSERR on failure
1330 GNUNET_STRINGS_get_utf8_args (int argc, char *const *argv, int *u8argc, char *const **u8argv)
1337 char **split_u8argv;
1339 wcmd = GetCommandLineW ();
1341 return GNUNET_SYSERR;
1342 wargv = CommandLineToArgvW (wcmd, &wargc);
1344 return GNUNET_SYSERR;
1346 split_u8argv = GNUNET_malloc (argc * sizeof (char *));
1348 for (i = 0; i < wargc; i++)
1351 /* Hopefully it will allocate us NUL-terminated strings... */
1352 split_u8argv[i] = (char *) u16_to_u8 (wargv[i], wcslen (wargv[i]) + 1, NULL, &strl);
1353 if (NULL == split_u8argv[i])
1356 for (j = 0; j < i; j++)
1357 free (split_u8argv[j]);
1358 GNUNET_free (split_u8argv);
1360 return GNUNET_SYSERR;
1364 *u8argv = _make_continuous_arg_copy (wargc, split_u8argv);
1367 for (i = 0; i < wargc; i++)
1368 free (split_u8argv[i]);
1369 free (split_u8argv);
1372 char *const *new_argv = (char *const *) _make_continuous_arg_copy (argc, argv);
1381 * Parse the given port policy. The format is
1382 * "[!]SPORT[-DPORT]".
1384 * @param port_policy string to parse
1385 * @param pp policy to fill in
1386 * @return #GNUNET_OK on success, #GNUNET_SYSERR if the
1387 * @a port_policy is malformed
1390 parse_port_policy (const char *port_policy,
1391 struct GNUNET_STRINGS_PortPolicy *pp)
1401 pp->negate_portrange = GNUNET_YES;
1404 if (2 == sscanf (pos,
1413 GNUNET_log (GNUNET_ERROR_TYPE_WARNING,
1414 _("Port not in range\n"));
1415 return GNUNET_SYSERR;
1417 pp->start_port = (uint16_t) s;
1418 pp->end_port = (uint16_t) e;
1421 if (1 == sscanf (pos,
1429 GNUNET_log (GNUNET_ERROR_TYPE_WARNING,
1430 _("Port not in range\n"));
1431 return GNUNET_SYSERR;
1434 pp->start_port = (uint16_t) s;
1435 pp->end_port = (uint16_t) s;
1438 GNUNET_log (GNUNET_ERROR_TYPE_WARNING,
1439 _("Malformed port policy `%s'\n"),
1441 return GNUNET_SYSERR;
1446 * Parse an IPv4 network policy. The argument specifies a list of
1447 * subnets. The format is
1448 * <tt>(network[/netmask][:SPORT[-DPORT]];)*</tt> (no whitespace, must
1449 * be terminated with a semicolon). The network must be given in
1450 * dotted-decimal notation. The netmask can be given in CIDR notation
1451 * (/16) or in dotted-decimal (/255.255.0.0).
1453 * @param routeListX a string specifying the IPv4 subnets
1454 * @return the converted list, terminated with all zeros;
1455 * NULL if the synatx is flawed
1457 struct GNUNET_STRINGS_IPv4NetworkPolicy *
1458 GNUNET_STRINGS_parse_ipv4_policy (const char *routeListX)
1466 unsigned int temps[8];
1468 struct GNUNET_STRINGS_IPv4NetworkPolicy *result;
1473 if (NULL == routeListX)
1475 len = strlen (routeListX);
1478 routeList = GNUNET_strdup (routeListX);
1480 for (i = 0; i < len; i++)
1481 if (routeList[i] == ';')
1483 result = GNUNET_malloc (sizeof (struct GNUNET_STRINGS_IPv4NetworkPolicy) * (count + 1));
1488 for (colon = pos; ':' != routeList[colon]; colon++)
1489 if ( (';' == routeList[colon]) ||
1490 ('\0' == routeList[colon]) )
1492 for (end = colon; ';' != routeList[end]; end++)
1493 if ('\0' == routeList[end])
1495 if ('\0' == routeList[end])
1497 routeList[end] = '\0';
1498 if (':' == routeList[colon])
1500 routeList[colon] = '\0';
1501 if (GNUNET_OK != parse_port_policy (&routeList[colon + 1],
1506 SSCANF (&routeList[pos],
1507 "%u.%u.%u.%u/%u.%u.%u.%u",
1518 for (j = 0; j < 8; j++)
1519 if (temps[j] > 0xFF)
1521 LOG (GNUNET_ERROR_TYPE_WARNING,
1522 _("Invalid format for IP: `%s'\n"),
1524 GNUNET_free (result);
1525 GNUNET_free (routeList);
1528 result[i].network.s_addr =
1529 htonl ((temps[0] << 24) + (temps[1] << 16) + (temps[2] << 8) +
1531 result[i].netmask.s_addr =
1532 htonl ((temps[4] << 24) + (temps[5] << 16) + (temps[6] << 8) +
1538 /* try second notation */
1540 SSCANF (&routeList[pos],
1549 for (j = 0; j < 4; j++)
1550 if (temps[j] > 0xFF)
1552 LOG (GNUNET_ERROR_TYPE_WARNING,
1553 _("Invalid format for IP: `%s'\n"),
1555 GNUNET_free (result);
1556 GNUNET_free (routeList);
1559 result[i].network.s_addr =
1560 htonl ((temps[0] << 24) + (temps[1] << 16) + (temps[2] << 8) +
1562 if ((slash <= 32) && (slash >= 0))
1564 result[i].netmask.s_addr = 0;
1567 result[i].netmask.s_addr =
1568 (result[i].netmask.s_addr >> 1) + 0x80000000;
1571 result[i].netmask.s_addr = htonl (result[i].netmask.s_addr);
1578 LOG (GNUNET_ERROR_TYPE_WARNING,
1579 _("Invalid network notation ('/%d' is not legal in IPv4 CIDR)."),
1581 GNUNET_free (result);
1582 GNUNET_free (routeList);
1583 return NULL; /* error */
1586 /* try third notation */
1589 SSCANF (&routeList[pos],
1597 for (j = 0; j < 4; j++)
1598 if (temps[j] > 0xFF)
1600 LOG (GNUNET_ERROR_TYPE_WARNING,
1601 _("Invalid format for IP: `%s'\n"),
1603 GNUNET_free (result);
1604 GNUNET_free (routeList);
1607 result[i].network.s_addr =
1608 htonl ((temps[0] << 24) + (temps[1] << 16) + (temps[2] << 8) +
1610 result[i].netmask.s_addr = 0;
1613 result[i].netmask.s_addr = (result[i].netmask.s_addr >> 1) + 0x80000000;
1616 result[i].netmask.s_addr = htonl (result[i].netmask.s_addr);
1621 LOG (GNUNET_ERROR_TYPE_WARNING,
1622 _("Invalid format for IP: `%s'\n"),
1624 GNUNET_free (result);
1625 GNUNET_free (routeList);
1626 return NULL; /* error */
1628 if (pos < strlen (routeList))
1630 LOG (GNUNET_ERROR_TYPE_WARNING,
1631 _("Invalid format: `%s'\n"),
1633 GNUNET_free (result);
1634 GNUNET_free (routeList);
1635 return NULL; /* oops */
1637 GNUNET_free (routeList);
1638 return result; /* ok */
1643 * Parse an IPv6 network policy. The argument specifies a list of
1644 * subnets. The format is <tt>(network[/netmask[:SPORT[-DPORT]]];)*</tt>
1645 * (no whitespace, must be terminated with a semicolon). The network
1646 * must be given in colon-hex notation. The netmask must be given in
1647 * CIDR notation (/16) or can be omitted to specify a single host.
1648 * Note that the netmask is mandatory if ports are specified.
1650 * @param routeListX a string specifying the policy
1651 * @return the converted list, 0-terminated, NULL if the synatx is flawed
1653 struct GNUNET_STRINGS_IPv6NetworkPolicy *
1654 GNUNET_STRINGS_parse_ipv6_policy (const char *routeListX)
1664 struct GNUNET_STRINGS_IPv6NetworkPolicy *result;
1670 if (NULL == routeListX)
1672 len = strlen (routeListX);
1675 routeList = GNUNET_strdup (routeListX);
1677 for (i = 0; i < len; i++)
1678 if (';' == routeList[i])
1680 if (';' != routeList[len - 1])
1682 LOG (GNUNET_ERROR_TYPE_WARNING,
1683 _("Invalid network notation (does not end with ';': `%s')\n"),
1685 GNUNET_free (routeList);
1689 result = GNUNET_malloc (sizeof (struct GNUNET_STRINGS_IPv6NetworkPolicy) * (count + 1));
1695 while (';' != routeList[pos])
1698 while ((slash >= start) && (routeList[slash] != '/'))
1703 memset (&result[i].netmask,
1705 sizeof (struct in6_addr));
1710 routeList[pos] = '\0';
1711 for (colon = pos; ':' != routeList[colon]; colon--)
1712 if ('/' == routeList[colon])
1714 if (':' == routeList[colon])
1716 routeList[colon] = '\0';
1717 if (GNUNET_OK != parse_port_policy (&routeList[colon + 1],
1720 GNUNET_free (result);
1721 GNUNET_free (routeList);
1725 ret = inet_pton (AF_INET6, &routeList[slash + 1], &result[i].netmask);
1729 if ((1 != SSCANF (&routeList[slash + 1], "%u", &bits)) || (bits > 128))
1732 LOG (GNUNET_ERROR_TYPE_WARNING,
1733 _("Wrong format `%s' for netmask\n"),
1734 &routeList[slash + 1]);
1738 LOG_STRERROR (GNUNET_ERROR_TYPE_WARNING, "inet_pton");
1740 GNUNET_free (result);
1741 GNUNET_free (routeList);
1747 result[i].netmask.s6_addr[off++] = 0xFF;
1752 result[i].netmask.s6_addr[off] =
1753 (result[i].netmask.s6_addr[off] >> 1) + 0x80;
1758 routeList[slash] = '\0';
1759 ret = inet_pton (AF_INET6, &routeList[start], &result[i].network);
1763 LOG (GNUNET_ERROR_TYPE_WARNING,
1764 _("Wrong format `%s' for network\n"),
1765 &routeList[slash + 1]);
1767 LOG_STRERROR (GNUNET_ERROR_TYPE_ERROR,
1769 GNUNET_free (result);
1770 GNUNET_free (routeList);
1776 GNUNET_free (routeList);
1782 /** ******************** Base64 encoding ***********/
1784 #define FILLCHAR '='
1786 "ABCDEFGHIJKLMNOPQRSTUVWXYZ" "abcdefghijklmnopqrstuvwxyz" "0123456789+/";
1790 * Encode into Base64.
1792 * @param data the data to encode
1793 * @param len the length of the input
1794 * @param output where to write the output (*output should be NULL,
1796 * @return the size of the output
1799 GNUNET_STRINGS_base64_encode (const char *data,
1809 opt = GNUNET_malloc (2 + (len * 4 / 3) + 8);
1811 for (i = 0; i < len; ++i)
1813 c = (data[i] >> 2) & 0x3f;
1814 opt[ret++] = cvt[(int) c];
1815 c = (data[i] << 4) & 0x3f;
1817 c |= (data[i] >> 4) & 0x0f;
1818 opt[ret++] = cvt[(int) c];
1821 c = (data[i] << 2) & 0x3f;
1823 c |= (data[i] >> 6) & 0x03;
1824 opt[ret++] = cvt[(int) c];
1829 opt[ret++] = FILLCHAR;
1834 opt[ret++] = cvt[(int) c];
1838 opt[ret++] = FILLCHAR;
1841 opt[ret++] = FILLCHAR;
1845 #define cvtfind(a)( (((a) >= 'A')&&((a) <= 'Z'))? (a)-'A'\
1846 :(((a)>='a')&&((a)<='z')) ? (a)-'a'+26\
1847 :(((a)>='0')&&((a)<='9')) ? (a)-'0'+52\
1849 :((a) == '/') ? 63 : -1)
1853 * Decode from Base64.
1855 * @param data the data to encode
1856 * @param len the length of the input
1857 * @param output where to write the output (*output should be NULL,
1859 * @return the size of the output
1862 GNUNET_STRINGS_base64_decode (const char *data,
1863 size_t len, char **output)
1870 #define CHECK_CRLF while (data[i] == '\r' || data[i] == '\n') {\
1871 GNUNET_log(GNUNET_ERROR_TYPE_DEBUG | GNUNET_ERROR_TYPE_BULK, "ignoring CR/LF\n"); \
1873 if (i >= len) goto END; \
1876 *output = GNUNET_malloc ((len * 3 / 4) + 8);
1877 GNUNET_log (GNUNET_ERROR_TYPE_DEBUG,
1878 "base64_decode decoding len=%d\n",
1880 for (i = 0; i < len; ++i)
1883 if (FILLCHAR == data[i])
1885 c = (char) cvtfind (data[i]);
1888 c1 = (char) cvtfind (data[i]);
1889 c = (c << 2) | ((c1 >> 4) & 0x3);
1890 (*output)[ret++] = c;
1897 c = (char) cvtfind (c);
1898 c1 = ((c1 << 4) & 0xf0) | ((c >> 2) & 0xf);
1899 (*output)[ret++] = c1;
1908 c1 = (char) cvtfind (c1);
1909 c = ((c << 6) & 0xc0) | c1;
1910 (*output)[ret++] = c;
1921 /* end of strings.c */