I'm attempting to remove a line from several hundred files. The following does exactly what I need but, it doesn't save changes (as expected).
$ grep -v meow src/files
I've seen that appending > to the end of a given command will specify where the output buffer should save but, does this work for multiple files?
So I'd like to know if there's an elegant way to mass edit via the terminal. All of the examples I've come across using awk or sed only provide solutions for editing one file at a time.
One way to do this is using the following Perl one-liner:
perl -i.bak -n -e 'print unless /meow/' src/files
This should do in-place editing of multiple files. The originals are saved in .bak files.
Another way to do it is to do a similar operation with sed:
sed -i .bak '/meow/d' src/files/*
Perl got its -i option from sed, after all. Note that to use no backup file, you need an explicit empty extension with at least some versions of sed:
sed -i '' '/meow/d' src/files/*
Related
On my MacBook Pro, I'm trying to delete two ranges from multiple files in a directory. Searching thus far, it seems like sed is a way to do it, but if I try to delete one of the ranges with the command:
sed -i '5825,6144d' *.tab.txt
I get the error message:
sed: 1: "cl9_408230953ref_f00507 ...": command c expects \ followed by text
Alternatively, I've tried other variants where the command seems to work (i.e., no error message), but the files haven't changed. Suggestions?
Use
sed -i '' '5825,6144d' *.tab.txt
Or
sed -i.bak '5825,6144d' *.tab.txt
I have a file and I need to use sed to process it.
Here is my command: sed -i -e '/.*tour\.html\|.*Thumb[^\/]*\.jpg/!d'.
Now if I execute this command from the terminal, for example, sed -i -e '/.*tour\.html\|.*Thumb[^\/]*\.jpg/!d' myfile.txt, it works well. But if I write a bash script with the same command, it will delete all lines.
#!/bin/bash
sed -i -e '/.*tour\.html\|.*Thumb[^\/]*\.jpg/!d' "$1"
This script will delete all lines in file.
My PC is Mac OS.
As far as I understand getting both sed instances output of sed --help and sed --version showed that you have actually two different sed versions reacting to the two different ways of execuing your code.
Sed is a little inconsistent concerning the syntax, especially when it comes to commandline options.
For example, I know of an important difference for the -i switch, which in some Mac versions requires a file extension for backups being given explicitly. Others allow one optionally. This difference could explain why somethign involving a -i without backup extension works in one case and fails in another.
Anishsane suggested that different "PATH" variables could in turn be part of the mechanism to have two different sed versions executed.
I invite OP to edit the output of --help and --version (where possible, there should be a way to get the version out of both sed instances) here into this answer. I do not have those details actually. Which makes this answer seem a little "guessing".
I want to find specific text in a directory, and then delete the lines from the files that include the specific text.
Now I have two questions:
How can I achieve the task?
What is wrong with What I have tried? I have tried the methods below, but failed. the details are following:
grep -rnw "./" -e "webdesign"
This searches the current directory with pattern "webdesign", and I get the result:
.//pages/index.html:1:{% load webdesign %}
.//pages/pricing.html:1:{% load webdesign %}
.//prototypes.py:16: 'django.contrib.webdesign',
Then I use sed to remove the lines from those files, which doesn't work, only get blank file ( I mean it deletes all my file content):
sed -i "/webdesign/d" ./pages/index.html
or
sed "/webdesign/d" ./pages/index.html > ./pages/index.html
My software environment is: OS X Yosemite, Mac Terminal, Bash
A loop in bash will do the trick provided that there are no filenames with spaces (in which case other solutions are possible, but this is the simplest)
for i in `grep -lrnw "yourdirectory/" -e "webdesign"`
do
sed "/webdesign/d" $i > $i.tmp
# safety to avoid destroying the file if problem arises (disk full?)
if [ $? = 0 ] ; then
mv -f $i.tmp $i
fi
done
note that you should not locate this script in the current directory because it contains webdesign and it will be modified as well :)
Thanks to choroba, I know that -i option doesn't work like wished. But it has another meaning or it would be rejected by the opt parser. It has something to do with suffixes, well, it doesn't matter now, but it's difficult to see the problem at first.
Without -i you cannot work on a file in-place. And redirecting output to the input just destroys the input file (!). That's why your solution did not work.
You can install GNU sed that supports the -i option, then
sed -i '/webdesign/d' files
should work. Note that it's safer to use -i~ to create a backup.
You cannot write to the same file you're reading from, that's why
sed /webdesign/d file > file
doesn't work (it overwrites the file before you can read anything from it). Create a temporary file
sed /webdesign/d file > file.tmp
mv file.tmp file
On my Mac, when I try to run:
sed -i 2d file.csv
from the answer to Unix script to remove the first line of a CSV file, I get the following error:
sed: 1: "file.csv": invalid command code f
What can I do if I would like to delete the first two lines of file.csv?
On a Mac, the BSD sed requires the suffix for the backup (but the suffix can be an empty string, '') — it is not optional as it is with GNU sed. Hence, your command is being interpreted as "backup the file with the suffix 2d and … oops, the script you gave is file.csv, but f isn't a sed command".
sed -i .bak -e 2d file.csv
This deletes the first line of data from the CSV file (leaving the heading line in place).
If you want to write the code so it works with both BSD and GNU sed, then you have to attach the suffix to the -i option (GNU sed requires the suffix attached to the -i option; that's how it identifies whether there's an optional suffix or not):
sed -i.bak -e 2d file.csv
Note that you can't use an empty suffix and have the command work with both BSD sed and GNU sed.
The -e isn't necessary in either command line, but I quite often use it. I also often quote the command in single quotes, though it isn't necessary here.
If you want to delete the first two data lines, use 2,3d as the command. If you want to delete the first two lines, use 1,2d.
If you don't want the backup, then you can either remove it after the sed command completes (easiest) or use the two stage or three stage dance:
sed 2d file.csv > file.csv.bak &&
mv file.csv.bak file.csv # Oops; there went the links
sed 2d file.csv > file.csv.bak &&
cp file.csv.bak file.csv
rm -f file.csv.bak
With these, you might need to add trap commands to clean up the intermediate .bak file if an interrupt or other signal terminates the script.
To quote from the Apple documentation for sed — which was originally quoted by Diego in an answer which he chose to delete, the -i option takes an argument which indicates the extension to use for backup copies.
-i extension
Edit files in-place, saving backups with the specified extension. If a zero-length extension is given, no backup will be saved. It is not recommended to give a zero-length extension when in-place editing files, as you risk corruption or partial content in situations where disk space is exhausted, etc.
sed -i.bak '2,3d' filename.csv will delete lines 1 and 2 (assuming you have headers)
What this command does is edits the file being referenced while simultaneously creating a backup of the original file (hence the usage of .bak). The line deletion '2,3d' will delete the 2nd and 3rd line of the original file.
This question already has answers here:
How to delete from a text file, all lines that contain a specific string?
(21 answers)
Closed 5 years ago.
I want to delete a line containing a specific string from the file. How can I do this without using awk? I tried to use sed but I could not achieve it.
This should do it:
sed -e s/deletethis//g -i *
sed -e "s/deletethis//g" -i.backup *
sed -e "s/deletethis//g" -i .backup *
it will replace all occurrences of "deletethis" with "" (nothing) in all files (*), editing them in place.
In the second form the pattern can be edited a little safer, and it makes backups of any modified files, by suffixing them with ".backup".
The third form is the way some versions of sed like it. (e.g. Mac OS X)
man sed for more information.
sed -i '/pattern/d' file
Use 'd' to delete a line. This works at least with GNU-Sed.
If your Sed doesn't have the option, to change a file in place, maybe you can use an intermediate file, to store the modification:
sed '/pattern/d' file > tmpfile && mv tmpfile file
Writing directly to the source doesn't work: sed '/pattern/d' FILE > FILE so make a copy before trying out, if you doubt it. The redirection to a new file FILE will open a new file FILE before reading from it, so reading from it will result in an empty input.
Try the vim-way:
ex -s +"g/foo/d" -cwq file.txt