I have the following script:
while read lineA && read lineB <&3 do echo "$lineA" echo "$lineB" echo done < file11 3< file22
This code is working fine. But I don't understand these:
done < file11 3< file22
The script reads pairs of strings from two files. In each iteration, it reads a string from
lineA, and from
lineB. It does this until it encounters the end of either file. The body of the loop outputs the two strings and a blank line.
<file11 means "connect/redirect standard input from
file11 into this command", where "this command" is the
while loop. This could also have been written
Standard input is file descriptor number 0. File descriptors 1 and 2 are reserved for the standard output and standard error streams.
3<file22 means "connect file descriptor 3 to
file22 for this command". The number 3 happens to be the next "free" file descriptor number, but the code could have used 4 or 5 or any small integer other than 0, 1 and 2 instead (the limit of the usable file descriptor numbers depends on what shell is being used, but numbers up to 9 are guaranteed to work).
In the loop,
read lineA will read a string from standard input by default, i.e. from the file
file11 (standard input is inherited from the loop). This could also have been written
read lineA <&0.
read lineB <&3, the
read command reads from file descriptor 3, i.e. from
file22, rather than from standard input (file descriptor 3 is available in the loop, just like standard input is).
In short, it uses the standard input stream and an "extra" file descriptor (number 3) to read from two files "simultaneously" (a string at a time from each file in turn). Usually, one only have access to one input stream, but this bypasses that restriction by explicitly creating a second input stream and connecting it to the
I use the word "string" above rather than "line". The things read from the two files will have flanking whitespaces stripped off, and certain escape sequences will be interpreted specially.
To read a line using
IFS= read -r varname