I have a some file names in bash that I have acquired with
$ ones=$(find SRR*pass*1*.fq)
$ echo $ones
SRR6301033_pass_1_trimmed.fq
SRR6301034_pass_1_trimmed.fq
SRR6301037_pass_1_trimmed.fq
...
I then converted into an array so I can iterate over this list and perform some operations with filenames:
# convert to array
$ ones=(${ones// / })
and the iteration:
for i in $ones;
do
fle=$(basename $i)
out=$(echo $fle | grep -Po '(SRR\d*)')
echo "quants/$out.quant"
done
which produces:
quants/SRR6301033
SRR6301034
...
...
SRR6301220
SRR6301221.quant
However I want this:
quants/SRR6301033.quant
quants/SRR6301034.quant
...
...
quants/SRR6301220.quant
quants/SRR6301221.quant
Could somebody explain why what I'm doing doesn't work and how to correct it?
Why do you want this be done this complicated? You can get rid of all the unnecessary roundabouts and just use a for loop and built-in parameter expansion techniques to get this done.
# Initialize an empty indexed array
array=()
# Start a loop over files ending with '.fq' and if there are no such files
# the *.fq would be un-expanded and checking it against '-f' would fail and
# in-turn would cause the loop to break out
for file in *.fq; do
[ -f "$file" ] || continue
# Get the part of filename after the last '/' ( same as basename )
bName="${file##*/}"
# Remove the part after '.' (removing extension)
woExt="${bName%%.*}"
# In the resulting string, remove the part after first '_'
onlyFir="${woExt%%_*}"
# Append the result to the array, prefixing/suffixing strings 'quant'
array+=( quants/"$onlyFir".quant )
done
Now print the array to see the result
for entry in "${array[#]}"; do
printf '%s\n' "$entry"
done
Ways your attempt could fail
With ones=$(find SRR*pass*1*.fq) you are storing the results in a variable and not in an array. A variable has no way to distinguish if the contents are a list or a single string separated by spaces
With echo $ones i.e. an unquoted expansion, the string content is subject to word splitting. You might not see a difference as long as you have filenames with spaces, having one might let you interpret parts of the filename as different files
The part ${ones// / } makes no-sense in converting the string to an array as the attempt to use an unquoted variable $ones itself would be erroneous
for i in $ones; would be error prone for the said reasons above, the filenames with spaces could be interpreted as separated files instead of one.
Related
I am new to Bash coding. I would like to concatenate a string to each element of a comma-separated strings "array".
This is an example of what I have in mind:
s=a,b,c
# Here a function to concatenate the string "_string" to each of them.
# Expected result:
a_string,b_string,c_string
One way:
$ s=a,b,c
$ echo ${s//,/_string,}_string
a_string,b_string,c_string
Using a proper array is generally a much more robust solution. It allows the values to contain literal commas, whitespace, etc.
s=(a b c)
printf '%s\n' "${s[#]/%/_string}"
As suggested by chepner, you can use IFS="," to merge the result with commas.
(IFS=","; echo "${s[#]/%/_string}")
(The subshell is useful to keep the scope of the IFS reassignment from leaking to the current shell.)
Simply, you could use a for loop
main() {
local input='a,b,c'
local append='_string'
# create an 'output' variable that is empty
local output=
# convert the input into an array called 'items' (without the commas)
IFS=',' read -ra items <<< "$input"
# loop over each item in the array, and append whatever string we want, in this case, '_string'
for item in "${items[#]}"; do
output+="${item}${append},"
done
# in the loop, the comma was re-added back. now, we must remove the so there are only commas _in between_ elements
output=${output%,}
echo "$output"
}
main
I've split it up in three steps:
Make it into an actual array.
Append _string to each element in the array using Parameter expansion.
Turn it back into a scalar (for which I've made a function called turn_array_into_scalar).
#!/bin/bash
function turn_array_into_scalar() {
local -n arr=$1 # -n makes `arr` a reference the array `s`
local IFS=$2 # set the field separator to ,
arr="${arr[*]}" # "join" over IFS and assign it back to `arr`
}
s=a,b,c
# make it into an array by turning , into newline and reading into `s`
readarray -t s < <(tr , '\n' <<< "$s")
# append _string to each string in the array by using parameter expansion
s=( "${s[#]/%/_string}" )
# use the function to make it into a scalar again and join over ,
turn_array_into_scalar s ,
echo "$s"
I would like to add a name in the middle of dirPath
#!/bin/bash
name='agent_name-2'
dirPath='/var/azp/1/s'
I want to insert agent_name-2 after /var/azp in dirPath, and store it in a separate variable result like this
result=/var/azp/agent_name-2/1/s
If /var/azp is a hard coded string (i.e. constant), try:
name='agent_name-2'
dirPath='/var/azp/1/s'
result="/var/azp/$name${dirPath#/var/azp}"
Explanation: ${dirPath#/var/azp} removes the string /var/azp from the beginning of the string $dirPath.
Try this:
#!/bin/bash
name='agent_name-2'
dirPath='/var/azp/1/s'
Split dirPath by / and store it in the array dirs.
IFS=/ read -r -a dirs <<< "$dirPath"
Calculate the middle of the array.
middle=$(((${#dirs[#]}+1)/2))
Create two new arrays left and right with the left and right half of the dirs array.
left=("${dirs[#]:0:$middle}")
right=("${dirs[#]:$middle}")
Join the left and right half and put the name in between.
result="$(printf "%s/" "${left[#]}" "$name" "${right[#]}")"
Remove the trailing slash.
result=${result%/}
Bash search-replace
You can use Bash's search and replace syntax ${variable//search/replace}.
prefix='/var/azp'
result=${dirPath//$prefix/$prefix\/$name}
# > /var/azp/agent_name-2/1/s
sed s
If $name doesn't contain any special characters, you could inject it into a sed search-replace:
$ sed "s|/var/azp|\0/$name|" <<< "$dirPath"
/var/azp/agent_name-2/1/s
Then for saving the result to a variable, see How do I set a variable to the output of a command in Bash?
I'm writing a for loop in bash to run a command and I need to add a comma after one of my variables. I can't seem to do this without an extra space added. When I move "," right next to $bams then it outputs *.sorted,
#!/bin/bash
bams=*.sorted
for i in $bams
do echo $bams ","
done;
Output should be this:
'file1.sorted','file2.sorted','file3.sorted'
The eventual end goal is to be able to insert a list of files into a --flag in the format above. Not sure how to do that either.
First, a literal answer (if your goal were to generate a string of the form 'foo','bar','baz', rather than to run a program with a command line equivalent to somecommand --flag='foo','bar','baz', which is quite different):
shopt -s nullglob # generate a null result if no matches exist
printf -v var "'%s'," *.sorted # put list of files, each w/ a comma, in var
echo "${var%,}" # echo contents of var, with last comma removed
Or, if you don't need the literal single quotes (and if you're passing your result to another program on its command line with the single quotes being syntactic rather than literal, you absolutely don't want them):
files=( *.sorted ) # put *.sorted in an array
IFS=, # set the comma character as the field separator
somecommand --flag "${files[*]}" # run your program with the comma-separated list
try this -
lst=$( echo *.sorted | sed 's/ /,/g' ) # stack filenames with commas
echo $lst
if you really need the single-ticks around each filename, then
lst="'$( echo *.sorted | sed "s/ /','/g" )'" # commas AND quotes
#!/bin/bash
bams=*.sorted
for i in $bams
do flag+="${flag:+,}'$i'"
done
echo $flag
I am trying to split many folder names in a for loop and extract the element between first and last underscore of filename. Filenames can look like ENCSR000AMA_HepG2_CTCF or ENCSR000ALA_endothelial_cell_of_umbilical_vein_CTCF.
My problem is that folder names differ form each other in the total number of underscores, so I cannot use something like:
IN=$d
folderIN=(${IN//_/ })
tf_name=${folderIN[-1]%/*} #get last element which is the TF name
cell_line=${folderIN[-2]%/*}; #get second last element which is the cell line
dataset_name=${folderIN[0]%/*}; #get first element which is the dataset name
cell_line can be one or more words separated by underscore but it's allways between 1st and last underscore.
Any help?
Just do this in a two step bash parameter expansion ONLY because bash does not support nested parameter expansion unlike zsh or other shells.
"${string%_*}" to strip the everything after the last occurrence of '_' and "${tempString#*_}" to strip everything from beginning to first occurrence of '_'
string="ENCSR000ALA_endothelial_cell_of_umbilical_vein_CTCF"
tempString="${string%_*}"
printf "%s\n" "${tempString#*_}"
endothelial_cell_of_umbilical_vein
Another example,
string="ENCSR000AMA_HepG2_CTCF"
tempString="${string%_*}"
printf "%s\n" "${tempString#*_}"
HepG2
You can modify this logic to apply on each of the file-names in your folder.
Could use regex.
extract_words() {
[[ "$1" =~ ^([^_]+)_(.*)_([^_]+)$ ]] && echo "${BASH_REMATCH[2]}"
}
while read -r from_line
do
extracted=$(extract_words "$from_line")
echo "$from_line" "[$extracted]"
done < list_of_filenames.txt
EDIT: I moved the "extraction" into an alone bash function for reuse and easy modification for more complex cases, like:
extract_words() {
perl -lnE 'say $2 if /^([^_]+)_(.*)_([^_]+)$/' <<< "$1"
}
I'm writing the second version of my post-receive git hook.
I have a GL_REPO variable which conforms to:
/project.name/vhost-type/versioncodename
It may or may not have a trailing and/or preceding slash.
My current code misunderstood the function of the following code, and as a result it clearly duplicates $versioncodename into each variable:
# regex out project codename
PROJECT_NAME=${GL_REPO##*/}
echo "project codename is: $PROJECT_NAME"
# extract server target vhost-type -fix required
VHOST_TYPE=${GL_REPO##*/}
echo "server target is: $VHOST_TYPE"
# get server project - fix required
PROJECT_CODENAME=${GL_REPO##*/}
echo "server project is: $PROJECT_CODENAME"
What is the correct method for taking these elements one at a time from the back of the string, or guaranteeing that a three part string allocates these variables?
I guess it might be better to split into an array?
#!/bin/bash
GL_REPO=/project.name/vhost-type/versioncodename
GL_REPO=${GL_REPO#/} # remove preceding slash, if any
IFS=/ read -a arr <<< "$GL_REPO"
PROJECT_NAME="${arr[0]}"
VHOST_TYPE="${arr[1]}"
PROJECT_CODENAME="${arr[2]}"
UPDATE: an alternative solution by anishsane:
IFS=/ read PROJECT_NAME VHOST_TYPE PROJECT_CODENAME <<< "$GL_REPO"
You can use cut with a field separator to pull out items by order:
NAME=$(echo $GL_REPO | cut -d / -f 1)
You can repeat the same for other fields. The trailing/leading slash you can ignore (you'll get a NAME field being empty, for example) or you can strip off a leading slash with ${GL_REPO##/} (similarly, you can strip off a trailing slash with ${GL_REPO%%/}).
This is another way:
GL_REPO="/project.name/vhost-type/versioncodename"
GL_REPO="${GL_REPO/#\//}"
#^replace preceding slash (if any) with empty string.
IFS="/" arr=($GL_REPO)
echo "PN: ${arr[0]} VHT: ${arr[1]} VC: ${arr[2]}"
Using Bash Pattern Matching:
GL_REPO="/project.name/vhost-type/versioncodename"
patt="([^/]+)/([^/]+)/([^/]+)"
[[ $GL_REPO =~ $patt ]]
echo "PN: ${BASH_REMATCH[1]} VHT: ${BASH_REMATCH[2]} VC: ${BASH_REMATCH[3]}"