LinuxQuestions.org - End-of-line Characters missing from last line of md5 file. Md5sum fails

- Linux - General (http://www.linuxquestions.org/questions/linux-general-1/)

- - End-of-line Characters missing from last line of md5 file. Md5sum fails (http://www.linuxquestions.org/questions/linux-general-1/end-of-line-characters-missing-from-last-line-of-md5-file-md5sum-fails-736546/)

End-of-line Characters missing from last line of md5 file. Md5sum fails

I have a number of md5 files that were either generated on windows machines or transfered in binary mode (bit for bit). As a result the end-of-line characters are formated with CRLF.

Code:

cat -e windows.md5

d4c638e28b1990c55b79e5249cf33038 *testfile_01.ext^M$

9e15845bdae8c1d1699dade90b73c744 *testfile_02.ext

When I translate them with

Code:

tr -d '\r' <windows.md5> linux.md5

the CR is removed

Code:

cat -e linux.md5

d4c638e28b1990c55b79e5249cf33038 *testfile_01.ext$ 

9e15845bdae8c1d1699dade90b73c744 *testfile_02.ext

and I can use

Code:

md5sum -c linux.md5

with success except for the last file/line to be checked.

My problem is that the last line in the file never had any end-of-line characters to begin with.

As a result, md5sum -c linux.md5 returns an error for the last line.

Code:

md5sum -c linux.md5

testfile_01.ext: OK

md5sum: testfile_02.ext : No such file or directory

testfile_02.ext : FAILED open or read

md5sum: WARNING: 1 of 2 listed files could not be read

If I open the linux.md5 file with EMACS, delete the last "blank" characters of the file then hit return, then save that I find the last line is fixed

Code:

9e15845bdae8c1d1699dade90b73c744 *testfile_02.ext$

and will be used by md5sum correctly.

I would like to be able to script the fix and not have to manually go through every md5 file by hand. It seems like an easy problem but it has got me vexed...

I have tried appending to the file:

Code:

cat >> linux.md5

Blah

^D

but that only returns

Code:

d4c638e28b1990c55b79e5249cf33038 *testfile_01.ext$ 

9e15845bdae8c1d1699dade90b73c744 *testfile_02.ext $

Blah$

Please notice the space between the "testfile_02.ext" and "$".
This returns the same error.

If I could only insert the correct end-of-line character for the last line I think I would be fixed.

Any suggestions?

To get you started, there is a tool/script already available, and included with many *nix systems, called 'dos2unix' and it has a partner tool called 'unix2dos' and it does the end-of-line conversions on text files. Check Google for it if it is not on your machine.

As to the missing end-of-line on the very last line, I'm not sure if dos2unix would insert one (I haven't used the tool myself) but it wouldn't be too much of a big leap to have a final newline appended to a file that doesn't have one.

Sasha

Quote:

Originally Posted by GrapefruiTgirl (Post 3590802)

Thanks,

I do have dos2unix installed and it gives the same result as

Code:

tr -d '\n' <oldfile> newfile

Both methods fail to leave any kind of EOL character on the final line.

Please make the leap and show me how to have a final newline appended to a file that doesn't have one!

I might also mention that I get the same results if I use vi and use 'set fileform=unix'. Same thing again when i use

Code:

sed -e 's/\r$//' old.md5 > new.md5

Seems like a job for sed or awk but I'm vexed..

sh# awk '{ printf "\n" >> $0 }' md5sum_file.txt

sh# echo >> md5sums_file.txt

There are a couple methods I just quickly grabbed off the 'net :) but haven't tested.
In the case of a file that already has a newline, a second one shouldn't hurt (I don't think it will :scratch:)

Sasha

Quote:

Originally Posted by GrapefruiTgirl (Post 3590823)

Allright!

Both methods work great!

Code:

tr -d '\r' <windows.md5> linux.md5; awk '{ printf "\n" >> $0 }' linux.md5

Code:

tr -d '\r' <windows.md5> linux.md5; echo >> linux.md5

Both methods remove CRLF from all lies and leave all lines with a LF (line feed).

The 2nd method using echo does insert an extra Character but that doesn't seem to effect the function of md5sum.

Thank-you!

No problem; I learned something myself too: a simple AWK function. I have spent much time learning what I can of GAWK, but AWK has stuff too that can be handy-- if only there weren't SO MANY permutations of AWK/GAWK/SED they would be easier to get a handle on.

Cheers,
Sasha