awk output with spaces in first column - awk

Question

I tried using awk splitting the columns to print a sentence but the first column has spaces.
Sample of my beginner code:
$ awk '/Linux/ { print "The filename","\""$1"\"","is located in",$2 }' test.txt
The filename "The" is located in test
The filename "Some" is located in file
The filename "File" is located in name
The filename "Something_here" is located in /ABC
The filename "Another_test" is located in /DEFG
The filename "Label" is located in test
From file: test.txt
Filename Folder Type
-------------------------------------- -------------- ------
The test file /test/folder Linux
Some file / Linux
File name /Temp Linux
Something_here /ABC Linux
Another_test /DEFG Linux
Label test /HIJK Linux
what I want to achieve: (with quotes inclusive)
The filename "Default file" is located in /
The filename "The test file" is located in /test/folder
issue is when i use 'space' or '/' as delimiter i cannot get the whole line when printing

Answer 1

If you have GNU AWK, this should do the trick:
awk 'match($0, /([^\/]+)([^ ]+) *Linux/, arr) { sub(/ +$/, "", arr[1]); printf("The filename \"%s\" is located in %s\n", arr[1], arr[2]) }' test.txt
Explanation:
# match and store groups in 'arr'
# - arr[1]: everything up until the first slash (including a lot of whitespace)
# - arr[2]: first slash until space
# - rest: also ensure there's 'Linux' after that
match($0, /([^\/]+)([^ ]+) *Linux/, arr) {
# trim whitespace from the right hand side of the filename
sub(/ +$/, "", arr[1]);
# print
printf("The filename \"%s\" is located in %s\n", arr[1], arr[2])
}
Note that there is also a less powerful version of match in other flavors of AWK and the same thing could be achieve with those, but you'd have to write a bit more code.

Answer 2

GNU awk has regex field separators, so just require multiple spaces separating your columns.
awk '/Linux/ { print "The file \""$1"\" is in "$2"." }' FS=" *" test.txt
It also offers fixed-width fields, say info gawk fieldwidths, you could use the lengths of the dash lines to set those on the fly.

Answer 3

I would propose sed with the substitution based on regular expression and back references plus a grep command to eliminate the header lines of the source file:
$ cat test.txt | grep -E 'Linux[ ]*$' | sed -E 's%(.+)([^ ])([ ]+)(/.+)[ ]+Linux[ ]*$%The filename "\1\2" is located in \4%'
The filename "The test file" is located in /test/folder
The filename "Some file" is located in /
The filename "File name" is located in /Temp
The filename "Something_here" is located in /ABC
The filename "Another_test" is located in /DEFG
The filename "Label test" is located in /HIJK
A good reference for the regular expressions (regex) is in the Linux manuals
The detailed description as requested in the comment:
grep with -E option accepts the extended regex (reference document above). Here it is used to filter the lines containing "Linux" word followed by some spaces if any at the end of each line
The output of grep goes into the input of sed
sed is passed the -E option like grep to accept extended regex. The s command substitutes a characters matching the regex (first part between % chars = "(.+)([^ ])([ ]+)(/.+)[ ]+Linux[ ]*$") by others (second part between % chars = "The filename "\1\2" is located in \4").
The second part uses back references: '\' followed by a nonzero decimal digit n is substituted by the nth parenthesized sub-expression of the regex. Here, \1 is substituted by the string which matches the 1st "(.+)" which is the filename here, \2 is substituted by the following "([^ ])" which is the last char of the filename (trick to suppress the following blanks from the name)...
This is not a rigorous explanation but at least it provides some inputs to go farther.
Another solution is to pass multiple actions on the sed command line. Hence, you can add a query to delete the first 2 header lines to suppress the pipe with cat and grep. Here "1,2d', means "delete lines numbers 1 and 2":
$ sed -E '1,2d;s%(.+)([^ ])([ ]+)(/.+)[ ]+Linux[ ]*$%The filename "\1\2" is located in \4%' test.txt
The filename "The test file" is located in /test/folder
The filename "Some file" is located in /
The filename "File name" is located in /Temp
The filename "Something_here" is located in /ABC
The filename "Another_test" is located in /DEFG
The filename "Label test" is located in /HIJK
NOTES: According to the manual, the -E option switches to using extended regular expressions. It has been supported for years by GNU sed, and is now included in POSIX.
On older systems, -r may be used if -E is not supported:
$ sed -r '1,2d;s%(.+)([^ ])([ ]+)(/.+)[ ]+Linux[ ]*$%The filename "\1\2" is located in \4%' test.txt
The filename "The test file" is located in /test/folder
The filename "Some file" is located in /
The filename "File name" is located in /Temp
The filename "Something_here" is located in /ABC
The filename "Another_test" is located in /DEFG
The filename "Label test" is located in /HIJK