mjhjmtu

I got the following error:

./assemblyDB.116.las
test.sh: line 9: $ sed 's/.las//': bad substitution

and this is the script:

for filename in $(find . -type f -name "assemblyDB.*.las"); do
 echo $filename
 no=$ sed 's/.las//'
 echo $no
done

Thank you in advance.

$ ... (curly braces) marks several sorts of parameter expansion, the simplest of which is just expanding the value of a variable. The stuff inside braces isn't a valid parameter name, or any other expansion, so the shell complains.

You seem to want command substitution instead, for that, the syntax is $( ... ) (regular parenthesis).

Also, the ls in ls $filename | sed... seems a bit unnecessary, the variable expands to your filename, and ls just passes it through. You could just use echo "$filename" | sed ... instead.

That said, you could do those modifications directly in the shell:

no=$filename/assemblyDB. # remove first match
no=$no/.las

or, using the standard operators:

no=$filename#assemblyDB. # remove from start of string
no=$no%.las # remove from end of string

If you do run sed, you may want to note that . matches any character in regular expressions, so it would be more correct to quote it with a backslash. Also you can give one sed instance both commands: sed -e 's/assemblyDB.//' -e 's/.las//'.

And then for filename in $(find . -type f -name "assemblyDB.*.las"); do has the issues of parsing ls, mostly the fact that whitespace and wildcards in the file names will break it. In ksh/Bash/zsh, you could do that whole loop in the shell:

shopt -s globstar # in Bash
for filename in **/assemblyDB.*.las; do
 ...

Command substitution (line 3)

for filename in $(find . -type f -name "assemblyDB.*.las"); do
 echo $filename
 no=$(ls $filename | sed 's/assemblyDB.//' | sed 's/.las//')
 echo $no
done

is equivalent, as already pointed out, to (line 3, no ls)

for filename in $(find . -type f -name "assemblyDB.*.las"); do
 echo $filename
 no=$(echo $filename | sed 's/assemblyDB.//' | sed 's/.las//')
 echo $no
done

or shorter (line 4 gone, now echo directly)

for filename in $(find . -type f -name "assemblyDB.*.las"); do
 echo $filename
 echo $filename | sed 's/assemblyDB.//' | sed 's/.las//'
done

and the sed command can be reduced to (line 3)

for filename in $(find . -type f -name "assemblyDB.*.las"); do
 echo $filename
 echo $filename | sed 's/assemblyDB.//;s/.las//'
done

or maybe extract the middle part with sed: (still line 3)

for filename in $(find . -type f -name "assemblyDB.*.las"); do
 echo $filename
 echo $filename | sed -r 's/.*assemblyDB.(.*).las/1/'" 
done

now find-iterator instead of for-iterator: (line 1 to 3, 4 gone)

find . -type f -name "assemblyDB.*.las" -printf "%fn" -exec sh -c "
 echo | sed -r 's/.*assemblyDB.(.*).las/1/'" ";"

assemblyDB.11.las
11
assemblyDB.9.las
9
assemblyDB.10.las
10

If your filenames are in such or a similar order, the following might work too:

for i in 8..12 ; do ls assemblyDB.$i.las && echo $i ; done 2>/dev/null

assemblyDB.9.las
9
assemblyDB.10.las
10
assemblyDB.11.las
11

(8 and 12 included, to demonstrate missing files in the sequence).

Instead of $ use backticks ` [The button below Escape on your keyboard.]