I am looking over a bash file and I don't know what the following lines do:
sed s/
//g < setenv.sh > setenv.sh2
When running the script, I always get an error on line 2:
//g: No such file or directory
So what it his trying to do here ?
Asked
Active
Viewed 101 times
0
PersianGulf
- 2,845
- 6
- 47
- 67
Adrian
- 19,440
- 34
- 112
- 219
-
2probably they were intended to be on same line. Try after joining the lines. – Amit Mar 05 '14 at 19:51
-
like this: sed s/ //g < setenv.sh > setenv.sh2 ? – Adrian Mar 05 '14 at 19:54
-
1It looks more like an attempt to embed a newline into a `sed` command. – chepner Mar 05 '14 at 19:54
-
Check out : http://stackoverflow.com/questions/1251999/sed-how-can-i-replace-a-newline-n – Gurubaran Mar 06 '14 at 00:54
3 Answers
4
It looks like that's a broken attempt to remove newlines from a file.
sed s/<enter|return>//g
Of course, it won't work – sed is newline aware. Just joining the lines probably won't do what you want either, unless there's an invisible character between the first slash and the line break. Otherwise you'd end up with
sed s///g
What you should do instead is identify which character (newline, carriage return, both?) you want, and use tr -d
tr -d $'\n' < file > file.nonewlines
tr -d $'\r' < file > file.nocarriagereturns
tr -d $'\r\n' < file > file.nocrlfs
kojiro
- 74,557
- 19
- 143
- 201
0
missing quote or double quote around the action and an escape of the new line also this is certainly part of a bigger script where working buffer could have more than 1 line in it. It is maybe coming from a print copy of a script
sed '#other action before this like H or N
s/\
//g' < setenv.sh > setenv.sh2
for the g it mean every occurence of the substitution (s///) but error is coming from the wrong first pattern (escaped enter)
NeronLeVelu
- 9,908
- 1
- 23
- 43
0
The /g flag cause not only the first occurence in each line to be replaced but all.
echo "ggg" | sed -e 's/g/j/'
# jgg
echo "ggg" | sed -e 's/g/j/g'
# jjj
MUY Belgium
- 2,330
- 4
- 30
- 46