It seems like the posix sort
command line utility will do some fancy locale based shenanegans to compare the given strings.
I scanned the man page but could not seem to find a way to get it to use the raw byte values instead.
Is there a way to get sort
(I have the GNU coreutils version) to behave like
qsort(array_of_my_strings, N, strcmp)
would in C
? Solutions using another tool then sort
would be fine too.
For demonstration, I currently get:
printf "\xC3\xBC\n\x76\n" | sort
ü
v
because the german umlaut ü
seems to be compared as u which comes before v, despite \xC3
being larger than \x76
.
What i want is
printf "\xC3\xBC\n\x76\n" | sort --raw-bytes-please
v
ü
Collation order and (multi-byte) character type are influenced by your locale. The locale name for disabling multibyte and locale-aware behaviors is C
.
Thus:
LC_COLLATE=C LC_CTYPE=C sort
...will set only the character type and the collation order (assuming LC_ALL
isn't set, in which case they would be ignored).
As a big hammer, you can also use:
LC_ALL=C sort
albeit with side effects such as changing the language used for printing error messages &c to the strings originally written by sort
's developers with no translation tables in effect.