Hi,
I tried the following options but was unable to delete blank lines from file
Input file = temp.hash.txt
temp.hash.txt content
90
0
89.56
0
0
57575.4544
56.89
0
0
0
0
----------------------------
Options tried
Nothing worked.
Any help is appreciated!
Thanks...
Any of your options tried does what you want them to do, so I guess your file has some "hidden" features like e.g. non-printing control characters. Try
od -tx1c temp.hash.txt
and post the result.
This is the result
[oratest@uswclora15 ~]$ od -tx1c temp.hash.txt
0000000 39 30 0d 0a 0d 0a 30 0d 0a 0d 0a 38 39 2e 35 36
9 0 \r \n \r \n 0 \r \n \r \n 8 9 . 5 6
0000020 0d 0a 30 0d 0a 30 0d 0a 0d 0a 0d 0a 35 37 35 37
\r \n 0 \r \n 0 \r \n \r \n \r \n 5 7 5 7
0000040 35 2e 34 35 34 34 0d 0a 35 36 2e 38 39 0d 0a 0d
5 . 4 5 4 4 \r \n 5 6 . 8 9 \r \n \r
0000060 0a 0d 0a 0d 0a 0d 0a 30 0d 0a 0d 0a 0d 0a 30 0d
\n \r \n \r \n \r \n 0 \r \n \r \n \r \n 0 \r
0000100 0a 30 0d 0a 30 0d 0a 0d 0a 0d 0a 35 37 35 36 35
\n 0 \r \n 0 \r \n \r \n \r \n 5 7 5 6 5
0000120
Hello uuuunnnn,
So it is confirmed now that \r carriage characters are present in your Input_file, so could you please first remove them and then try your any of the code mentioned in POST#1 and let us know how it goes then.
tr -d '\r' < Input_file > temp_Input_file && mv temp_Input_file Input_file
Thanks,
R. Singh
You can also use grep but with [[:space:]] instead of [[:blank:]] :
grep -v '^[[:space:]]*$' file
But this will leave the Windows carriage return character on the other lines.
--
You could do it all at once like so:
awk '{sub(/\r/,x)}NF' file
I did that, but it completely deleted all lines
[oratest@uswclora15 ~]$ tr -d '\r' < temp.hash.txt > temp_Input_file && mv temp_Input_file temp.hash.txt
[oratest@uswclora15 ~]$ vi temp.hash.txt
[oratest@uswclora15 ~]$ sed '/^$/d' temp.hash.txt > temp.hash.txt
[oratest@uswclora15 ~]$ vi temp.hash.txt
[oratest@uswclora15 ~]$
Thanks,
Niladri
Whoa!! I hope you have a backup of your file. 
You are essentially trying to read from a file, make some changes and write back to it. It doesn't work the way it looks. (*)
A few options are:
1) If your sed version allows it, try the "in-place edit" modifier; something like "sed -i". Check your sed manual "man sed".
2) Use a temporary file as you did for the "tr" command.
======
(*)
As for why it doesn't work the way it looks - you have two different processes (a) a read process and (b) a write process, both sharing the same resource - your file. When they are executed by the OS, a race condition occurs between them which renders the final output unpredictable.
If the writing is done before the reading, you get a zero-byte file. If the writing is done mid-way during the reading process, you may get a "half-baked" output file. The safest way, therefore, is to either use a temporary file explicitly or let sed do it via the "-i" modifier - which is not available in all sed versions.
You could also use other tools - Bash, awk, Perl etc. but the temporary file will be involved, either explicitly or implicitly.
Hello durden_tyler/uuuunnnn,
That tr command worked for me. Also && condition will make sure like first reading of Input_file and it's output's redirection is done then only it will go to next command which is mv temp_Input_file Input_file , ideally it shouldn't fail. Kindly do let me know your views on same.
Thanks,
R. Singh
The explanation was for the part in red in my post; I thought the post was clear about the erroneous part.
Thanks everyone.
Yes the issue was that I was copying the values from windows to unix and carriage return was entered. As suggested I entered the values in unix and used
sed '/^$/d' temp.hash.txt > temp.hash.out
It worked fine.
Thanks