just for the fun and to get an idea of how long these commands last ... I made the tests three times (just extract one representative), each followed by a sync and the computer has enough free memory to load the file into it. Here are numbers for a file of 179M
time tr -d '\015' <file.dos >file.unix real 0m1.323s user 0m0.390s sys 0m0.857s time sed -e 's/^M$//' file.dos >file.unix real 0m4.518s user 0m3.458s sys 0m0.856s time perl -pi -e 's/\012\015/\015/;' <file.dos >file.unix real 0m5.100s user 0m3.803s sys 0m0.905s Regexp is definitely not a good idea for dos2unix ... :) Regards, Sebastien Tandel Sake Blok wrote: > On Thu, Jan 18, 2007 at 04:20:40PM -0800, Guy Harris wrote: > >> On Jan 18, 2007, at 4:08 PM, Sebastien Tandel wrote: >> >> >>>> Is it safe to assume that dos2unix is available on a common UNIX >>>> developer machine?!? >>>> >>> Nope, it is not ... :-/ >>> >> No, but >> >> tr -d '\015' <file_with_CR_LF_line_endings >file_with_LF_line_endings >> >> will probably be available. (Unfortunately, "tr" isn't able to do >> unix2dos.) >> > > Since 'sed' is a requirement for a development-environment, it can also > be used for EOL transformations: > > dos2unix: sed -e 's/^M$//' <-- Use ctrl-v ctrl-m to create the ^M > unix2dos: sed -e 's/$/^M/' <-- Use ctrl-v ctrl-m to create the ^M > > Unfortunately "\r" is not recognized by sed on all platforms and > neither is \015. Therefore we have to use the raw CR character. > > Cheers, > > > Sake > > PS To prevent to over-dosizise the EOL's, you can also check whether > there is already a CR by using: sed -e 's/[^^M]$/^M/' > > _______________________________________________ > Wireshark-dev mailing list > Wireshark-dev@wireshark.org > http://www.wireshark.org/mailman/listinfo/wireshark-dev > _______________________________________________ Wireshark-dev mailing list Wireshark-dev@wireshark.org http://www.wireshark.org/mailman/listinfo/wireshark-dev
