Distinguishing files on the basis of first column’s content

I have multiple tab-separated files with .cluster extension. I want to classify these files on the basis of their first column content using the following criteria: (2 and 3 are actual digits/content inside the files)

• class_1: Only 2 AND 3 present on successive lines
• class_2: Only 2 present
• class_3: Only 3 present

I want to write their file names in .txt files with the respective class name. How do I do it with shell scripting?

Answer

for filename in *.cluster
do
    class=$(cut -d$'t' -f1)                      # Part 1
    if [ $(wc -l "$filename") -eq 2 ]             # Part 2, start
    then
        class=1
    fi                                            # Part 2, end
    printf '%sn' "$filename" >> class_"$class".txt # Part 3
done

This has three parts:

1. By default, it classifies the file based on the first field of the only line: the class variable is set to whatever is in the file, up to the first tab character on each line. This will be either 2 or 3 for class 2 & 3, since those files have only one line.

   cut chops files up by delimiters, $'t' is a Bash way of writing a tab character, and -f1 asks cut to output only the first delimited field.

2. If the file has two lines ($(wc -l "$filename") -eq 2), it must be class 1, so the class variable is forcibly set to 1, replacing its value from step 1. The if … fi deals with this.
3. Finally, the filename is appended to the appropriate class file: printf '%sn' "$filename" >> class_"$class".txt

At the end you will have three files class_N.txt for each N in 1, 2, 3, with one filename per line. If any file has some other contents than what you outlined in the question, like a different first field or length, you will get extra class files created.

In the perverse case where a filename itself contains a newline character, this will fall apart (and give you an opportunity to reconsider your filename choices), but otherwise it should be fine.