Grouping lines into heterogeneous subsets

Question

I have file with n lines.  (Each line refers to a “question”, and therefore they are labeled Q.1, Q.2, Q.3, ..., Q.n.)  Each line (question) has a “Marks” attribute, which has the value 2, 3, 4, 5, or 6.  There are ⁿ⁄₅ lines with each value.

For example: A 10-line file (i.e., n=10) might look like

amol@mypc:~$ cat questions.txt
Q.1 2 Marks
Q.2 5 Marks
Q.3 4 Marks
Q.4 3 Marks
Q.5 6 Marks
Q.6 4 Marks
Q.7 3 Marks
Q.8 2 Marks
Q.9 6 Marks
Q.10 5 Marks

I know I can split this into five homogeneous (i.e., all the same) files with something like

amol@mypc:~$ grep " 2 Marks" questions.txt > questions2Marks.txt
amol@mypc:~$ grep " 3 Marks" questions.txt > questions3Marks.txt
amol@mypc:~$ grep " 4 Marks" questions.txt > questions4Marks.txt
amol@mypc:~$ grep " 5 Marks" questions.txt > questions5Marks.txt
amol@mypc:~$ grep " 6 Marks" questions.txt > questions6Marks.txt

Each of the resulting files will have ⁿ⁄₅ lines.

I want to do the inverse operation – i.e., produce a transpose of the above result.  I want to split my questions.txt file into ⁿ⁄₅ files: questions1.txt, questions2.txt, questions3.txt, ..., questionsM.txt (using M to represent ⁿ⁄₅) where each file is five lines long and is heterogeneous (i.e., all different).

questions1.txt should contain

• the first line in questions.txt with 2 Marks,
• the first line in questions.txt with 3 Marks,
• the first line in questions.txt with 4 Marks,
• the first line in questions.txt with 5 Marks, and
• the first line in questions.txt with 6 Marks,

in that order.  questions2.txt should contain the second line of each, etc.

So, for n=10, M obviously is 2.  I would want my example questions.txt from above broken down into these two files:

amol@mypc:~$ cat questions1.txt            
Q.1 2 Marks
Q.4 3 Marks
Q.3 4 Marks
Q.2 5 Marks
Q.5 6 Marks

amol@mypc:~$ cat questions2.txt            
Q.8 2 Marks
Q.7 3 Marks
Q.6 4 Marks
Q.10 5 Marks
Q.9 6 Marks

How can I achieve that using *nix tools (sed, awk, perl, shell script, etc...)?

Answer 1

sort -n -k2 -k1.3 file | awk '{$2!=a?x=1:x++} {print > "file"x; a=$2}'

First , we need to sort the file correctly. -n sorts the file numerically, -k2 sorts according to the second field (the marks 2-6), -k1.3 then sorts within this order the first field starting from the 3rd character numerically (irgnoring the leading Q.). Now awk splits the output between ascending files (file1, file2, file3, filen....).

The output looks like this, file1:

$ cat file1
Q.1 2 Marks
Q.4 3 Marks
Q.3 4 Marks
Q.2 5 Marks
Q.5 6 Marks

And file2:

$ cat file2
Q.8 2 Marks
Q.7 3 Marks
Q.6 4 Marks
Q.10 5 Marks
Q.9 6 Marks

Answer 2

an awk answer: this will keep the order the questions the same as in the source file.

$ awk '{filename = "questions" ++n[$2] ".txt"; print > filename}' questions.txt 
$ cat questions1.txt 
Q.1 2 Marks
Q.2 5 Marks
Q.3 4 Marks
Q.4 3 Marks
Q.5 6 Marks
$ cat questions2.txt 
Q.6 4 Marks
Q.7 3 Marks
Q.8 2 Marks
Q.9 6 Marks
Q.10 5 Marks