Hello I am using bash to create CSV file by extracting data from an html file using grep. The problem is after getting the data then using sed to take out , in it and put a word like My_com it gose a crazy on me. here is my code.
time=$(grep -oP 'data-context-item-time=.*.data-context-item-views' index.html \
| cut -d'"' -f2)
title=$(grep -oP 'data-context-item-title=.*.data-context-item-id' index.html |\
cut -d'"' -f2)
sed "s/,/\My_commoms/g" $title
echo "$user,$views,$time,$title" >> test
I keep getting this error
sed: can't read Flipping: No such file or directory
sed: can't read the: No such file or directory
and so on
any advice on what wrong with my code
You can't use sed on text directly on the command line like that; sed expects a file, so it is reading your text as a file name. Try this for your second to last line:
echo $title | sed 's/,/My_com/g'
that way sed sees the text on a file (stdin in this case). Also note that I've used single quotes in the argument to sed; in this case I don't think it will make any difference, but in general it is good practice to make sure bash doesn't mess with the command at all.
If you don't want to use the echo | sed chain, you might also be able to rewrite it like this:
sed 's/,/My_com/g' <<< "$title"
I think that only works in bash, not dash etc. This is called a 'here-string', and bash passes the stuff on the right of the <<< to the command on its stdin, so you get the same effect.
Related
I have a folder full of textfiles which i want to grep for a specific String, then add a String (let's say a "+ ") via sed at the beginning of that specific line (at all places where the search-String occurs).
So far i got this:
if grep -q &MyString /path/to/folder/* ; then
echo "String detected"
sed 's/^/+ /g' -i [what to put in here?]
fi
depending on what i put in the brackets it either adds the String in front of every line of every textfile or does nothing because i dont define a correct file. How would i reference the file i detected the String in before?
Thanks
grep -l &MyString /path/to/folder/* | uniq | while read line
do
echo "String detected"
sed '/&MyString/ s/^/+ /g' -i $line
done
Using grep with -l to print just the filename, use uniq to deduplicate and then read each line of the output. Taking each line from the output, use this to represent the filename in sed ($line).
Read line is used as opposed for var in $(grep ...) as this would cause issues for files with spaces in them as each space delimited variable in the output would be used to represent the filename.
I have about 700 text files that consist of config output which uses various special characters. I am using this script to remove the special characters so I can then run a different script referencing an SED file to remove the commands that should be there leaving what should not be in the config.
I got the below from Remove all special characters and case from string in bash but am hitting a wall.
When I run the script it continues to loop and writes the script into the output file. Ideally, it just takes out the special characters and creates a new file with the updated information. I have not gotten to the point to remove the previous text file since it probably wont be needed. Any insight is greatly appreciated.
for file in *.txt for file in *.txt
do
cat * | tr -cd '[:alnum:]\n\r' | tr '[:upper:]' '[:lower:]' >> "$file" >> "$file".new_file.txt
done
A less-broken version of this might look like:
#!/usr/bin/env bash
for file in *.txt; do
[[ $file = *.new_file.txt ]] && continue ## skip files created by this same script
tr -cd '[:alnum:]\n\r' <"$file" \
| tr '[:upper:]' '[:lower:]' \
>> "$file".new_file.txt
done
Note:
We're referring to the "$file" variable being set by for.
We aren't using cat. It slows your script down with no compensating benefits whatsoever. Instead, using <"$file" redirects from the specific input file being iterated over at present.
We're skipping files that already have .new_file.txt extensions.
We only have one output redirection (to the new_file.txt version of the file; you can't safely write to the file you're using as input in the same pipeline).
Using GNU sed:
sed -i 's/[^[:alnum:]\n\r]//g;s/./\l&/g' *.txt
I've been looking around for a while on this and can't seem to find a solution on how to use sed to do this. I have a file that is named:
FILE_772829345_D594_242_25kd_kljd.mov
that I want to be renamed
D594_242_25kd_kljd.mov
I currently have been trying to get sed to work for this but have only been able to remove the first second of the file:
echo 'FILE_772829345_D594_242_25kd_kljd.mov' | sed 's/[^_]*//'
_772829345_D594_242_25kd_kljd.mov
How would I get sed to do the same instruction again, up to the second underscore?
If the filename is in a shell variable, you don't even need to use sed, just use a shell expansion with # to trim through the second underscore:
filename="FILE_772829345_D594_242_25kd_kljd.mov"
echo "${filename#*_*_}" # prints "D594_242_25kd_kljd.mov"
BTW, if you're going to use mv to rename the file, use its -i option to avoid file getting overwritten if there are any name conflicts:
mv -i "$filename" "${filename#*_*_}"
If all your files are named similarly, you can use cut which would be a lot simpler than sed with a regex:
cut -f3- -d_ <<< "FILE_772829345_D594_242_25kd_kljd.mov"
Output:
D594_242_25kd_kljd.mov
I am writing a small script which is getting some configuration options from a settings file with a certain format (option=value or option=value1 value2 ...).
settings-file:
SomeOption=asdf
IFS=HDMI1 HDMI2 VGA1 DP1
SomeOtherOption=ghjk
Script:
for VALUE in $(cat settings | grep IFS | sed 's/.*=\(.*\)/\1/'); do
echo "$VALUE"x
done
Now I get the following output:
HDMI1x
HDMI2x
VGA1x
xP1
Expected output:
HDMI1x
HDMI2x
VGA1x
DP1x
I obviously can't use the data like this since the last read entry is mangled up somehow. What is going on and how do I stop this from happening?
Regards
Generally you can use awk like this:
awk -F'[= ]' '$1=="IFS"{for(i=2;i<=NF;i++)print $i"x"}' settings
-F'[= ] splits the line by = or space. The following awk program checks if the first field, the variable name equals IFS and then iterates trough column 2 to the end and prints them.
However, in comments you said that the file is using Windows line endings. In this case you need to pre-process the file before using awk. You can use tr to remove the carriage return symbols:
tr -d '\r' settings | awk -F'[= ]' '$1=="IFS"{for(i=2;i<=NF;i++)print $i"x"}'
The reason is likely that your settings file uses DOS line endings.
Once you've fixed that (with dos2unix for example), your loop can also be modified to the following, removing two utility invocations:
for value in $( sed -n -e 's/^IFS.*=\(.*\)/\1/p' settings ); do
echo "$value"x
done
Or you can do it all in one go, removing the need to modify the settings file at all:
tr -d '\r' <settings |
for value in $( sed -n -e 's/^IFS.*=\(.*\)/\1/p' ); do
echo "$value"x
done
i was trying to add the lines from the text file to the sed command
observered_list.txt
Uncaught SlingException
cannot render resource
IncludeTag Error
Recursive invocation
Reference component error
i need it to be coded like the following
sed '/Uncaught SlingException\|cannot render resource\|IncludeTag Error\|Recursive invocation\|Reference component error/ d'
help me to do this.
I would suggest you create a sed script and delete each pattern consecutively:
while read -r pattern; do
printf "/%s/ d;\n" "$pattern"
done < observered_list.txt >> remove_patterns.sed
# now invoke sed on the file you want to modify
sed -f remove_patterns.sed file_to_clean
Alternatively you could construct the sed command like this:
pattern=
while read -r line; do
pattern=$pattern'\|'$line
done < observered_list.txt
# strip of first and last \|
pattern=${pattern#\\\|}
pattern=${pattern%\\\|}
printf "sed '/%s/ d'\n" "$pattern"
# you still need to invoke the command, it's just printed
You can use grep for that:
grep -vFf /file/with/patterns.txt /file/to/process.txt
Explanation:
-v excludes lines of process.txt which match one of the patterns from output
-F treats patterns in patterns.txt as fixed strings instead of regexes (looks like this is desired here)
-f reads patterns from patterns.txt
Check man grep for further information.