I am trying to get a specific line from a text file.
So far, online I have only seen stuff like sed, (I can only use the sh -not bash or sed or anything like that). I need to do this only using a basic shell script.
cat file | while read line do #do something done I know how to iterate through lines, as shown above, but what if I just need to get the contents of a particular line
sed:
sed '5!d' file awk:
awk 'NR==5' file 
5!d means delete all lines except 5. shell var is possible, you need double quotes.
sed -n 5p This seems more logical to remember for newbies, because -n means "no output by default" and p stands for "print", and there's no potentially confusing mention of deleting (when people talk of files, deleting lines tends to mean something different).-n '5p' works for this problem too. The difference here is, with 5!d you can add -i to write the change back to the file. however, with -n 5p you have to sed -n '5p' f > f2&& mv f2 f again, for this question, I am agree with your opinion.You could use sed -n 5p file.
You can also get a range, e.g., sed -n 5,10p file.
grep -C? Would be useful to easily seek for errors on source code.Assuming line is a variable which holds your required line number, if you can use head and tail, then it is quite simple:
head -n $line file | tail -1 If not, this should work:
x=0 want=5 cat lines | while read line; do x=$(( x+1 )) if [ $x -eq "$want" ]; then echo $line break fi done 
-eq comparison is for integers, so it wants a line number, not line content ($line). This has to be fixed by defining e.g. want=5 prior to the loop, and then using the -eq comparison on $want. [moved from a rejected edit]
sed is obviously faster, it does not provide an answer to the question. If you missed it, the question explicitly states that sed is not usable, and sh is the only tool available.Best performance method
sed '5q;d' file Because sed stops reading any lines after the 5th one
Update experiment from Mr. Roger Dueck
I installed wcanadian-insane (6.6MB) and compared sed -n 1p /usr/share/dict/words and sed '1q;d' /usr/share/dict/words using the time command; the first took 0.043s, the second only 0.002s, so using 'q' is definitely a performance improvement!
sed -n 5q
sed stops reading any lines after the 5th one.
sed -n 1p /usr/share/dict/words and sed '1q;d' /usr/share/dict/words using the time command; the first took 0.043s, the second only 0.002s, so using 'q' is definitely a performance improvement!
q in a sed command from a pipe will result in broken pipe. In that case, have to resort to sed -n 5p
sed -n 5q prints nothing for me. I think the answer is the best variant, but sed -n '5{p;q}' also works for me.If for example you want to get the lines 10 to 20 of a file you can use each of these two methods:
head -n 20 york.txt | tail -11 or
sed -n '10,20p' york.txt p in above command stands for printing.
Here's what you'll see: 
You could use sed command.
If the preferred line number is 5:
sed -n '5p' filename #get the 5th line and prints the value (p stands for print) If the preferred line number is a range, e.g. 1-5 lines:
sed -n '1,5p' filename #get the 1 to 5th line and prints the values If need to get 1st and 5th line only, e.g. 1st Line, 5th Line:
sed -n '1p;5p;' filename #get the 1st and 5th line values only The standard way to do this sort of thing is to use external tools. Disallowing the use of external tools while writing a shell script is absurd. However, if you really don't want to use external tools, you can print line 5 with:
i=0; while read line; do test $((++i)) = 5 && echo "$line"; done < input-file Note that this will print logical line 5. That is, if input-file contains line continuations, they will be counted as a single line. You can change this behavior by adding -r to the read command. (Which is probably the desired behavior.)
$((++i)) appears to be a bashism; if the OP is restricted in using external tools, I wouldn't assume they'll have access to more than a plain /bin/sh
$((i+=1)) works in Dash, too.$(($i+1)) is the simple workaround I was thinking of.Assuming the question was asked for bash, here is the fastest simplest way to do this.
readarray -t a <file ; echo ${a[5-1]} You may may discard array a when not needed anymore.
In parallel with William Pursell's answer, here is a simple construct which should work even in the original v7 Bourne shell (and thus also places where Bash is not available).
i=0 while read line; do i=`expr "$i" + 1` case $i in 5) echo "$line"; break;; esac done <file Notice also the optimization to break out of the loop when we have obtained the line we were looking for.
#!/bin/bash for i in {1..50} do line=$(sed "${i}q;d" file.txt) echo $line done Easy with perl! If you want to get line 1, 3 and 5 from a file, say /etc/passwd:
perl -e 'while(<>){if(++$l~~[1,3,5]){print}}' < /etc/passwd 
seq 5 | perl -ne 'print if $. ~~ [1, 4, 5]' but smartmatch is experimental and it's use discouraged
cat /etc/password | perl -ne 'next unless ++$l == 4; print' if you are interested in one single line. To keep your flexibility without experimental -> cat /etc/password | perl -ne 'next unless $ref{++$l}; print; BEGIN{ %ref = map{$_=>1} (1,4,5)}line=5; prep=`grep -ne ^ file.txt | grep -e ^$line:`; echo "${prep#$line:}" 
sed -n 5p, which of course can still be optimized to something like sed -n '5!d;p;q'
catokay butsedis not? That makes no sense.cat. Aw... cutecat!