script to read multi labes

[desp_to_learn]

hi guys, i am trying to write a shell script using bash shell. i want to write a script that can read multi labels which are placed on the top of a file (i used head command )and move the file in an appropriate category. so far i could only write a script which reads single labels only.this is the code i have
>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>>>>>>>>
#!/bin/bash
#This script reads the labels of documents and put them in their approriate label directory
for i in ./features/*;
do
head -1 $i > HeadFile.txt
FILE=HeadFile.txt
thfivecatDIR=category35
zerocatDIR=category0
if grep "35" $FILE;
then
mv $i $thfivecatDIR
#echo "there are $(ls -l $thfivecatDIR | wc -l) files of category35"
elif grep "0" $FILE;
then
echo "category 0 found"
mv $i $zerocatDIR
else
echo"doesnt belong to any category"
fi
done
>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>>>>>>>>

so how do i get it to read multi labels and move the files with more than one label to the directory multilabelDIR ?
for example if i have
0 35 place it in a multi label directory
0 place it in a 0label directory
i doubt whether i ve clarified my point enough

OH another problem with that script above is if i have a label 108 it places it in a olabel directory which is not my intention. I know its because i am using grep command , so is there any advice on which command to use.

help pleaseee
thank u in advance

[fiomba]

Hi desp_to_learn.
you are right... you didn't clarify enough!
Do you mean that in the 1st line of FILE you can have both "0" and "35"?
In this case your main problem (to capture only one label) is easily explained:

if "grep" finds "35", it doesn't test also for "0"

The solution is... (in the correct way to display a code)

# grep_labels.sh
# This script reads the labels of documents and
#     put them in their approriate label directory
for i in ./features/*;
do
    head -1 $i > HeadFile.txt   # ERROR! head -n 1 $i > HeadFile.txt
                                # capture the 1st line
    FILE=HeadFile.txt
    thfivecatDIR=category35
    zerocatDIR=category0        
    if grep "35" $FILE;          # ERROR! if grep -w "35" $FILE
    then                         # ... match only whole words
        mv $i $thfivecatDIR
        # echo "there are $(ls -l $thfivecatDIR |
        #     wc -l) files of category35"
    
        # test also for "0"
        if grep -w "0" $FILE;
        then
            echo "category 0 found"
            mv $i $zerocatDIR
        fi

    elif grep "0" $FILE;         # elif grep -w "0" $FILE;
    then                         # ... to avoid 108!
        echo "category 0 found"
        mv $i $zerocatDIR
    else
        echo "doesn't belong to any category"
    fi
done

As you can see... I don't use the "sha-bang" (#!/bin/bash) because
I run the script with "bash grep_labels.sh".
A warning... I didn't test my solution.
I wouldn't use a script but a one-liner expression using find/exec...
Bye.

[desp_to_learn]

thank you a million fiomba, it is now working half way to my intention.
now my question is how do i get it to place the documents into their category according to the second label?
for example;
if there are two labels say
0 35
it should place it to the 35 label directory.(the script i have now will place it to directory 0label)
if there is only one label say
0
it should place it to the 0 label directory.
Also since i have a long list of different labels how can i use "case statement" to make my code clean. i tried but the script wouldnt run at all.OR is there any idea how to make it more read-friendly??
thank you

[matonb]

If your happy that the first line will always contain the labels you need to filter, awk maybe a solution for you for example:

[<user>@centos5 ~]$ echo "0" | awk '{ print $NF }'
0
[<user>@centos5 ~]$ echo "0 35" | awk '{ print $NF }'
35

$NF basically prints the last field in the line.

So something along the lines of

for i in $(ls -1 ./features/); do
  folder=$(head -n1 $i | awk '{ print $NF }');
  mv ./features/$i category$folder/.
done;

[fiomba]

Hi desp_to_learn,
... always the same! very thrifty when you clarify...
So (as I have understood) there are only 2 cases:

- "0" --> single label to be moved in 0 label dir
- "0" and "35" --> multi label to be moved in 35 label dir

If it is so... (I'm not sure...):

# grep_labels.sh
# This script reads the labels of documents and
#     put them in their approriate label directory
for i in ./features/*;
do
    head -n 1 $i > HeadFile.txt  
    FILE=HeadFile.txt
    thfivecatDIR=category35   # multi  label
    zerocatDIR=category0      # single label 

    if grep -w "0" $FILE; 
    then 
        # test multi label (test also for "35")
        if grep -w "35" $FILE;
        then
            echo "category 35 found... multi label!"
            mv $i $thfivecatDIR
        else
            echo "single label"
            mv $i $zerocatDIR
        fi
    else
        echo "doesn't belong to any category"
    fi
done

The usual warning... I didn't test my solution.
About the "case statement" you have only to google "bash case esac", but
I suggest to stick to the structure if/elif/elif..../else (well indented)
to simulate the "case"...
Bye.

[matonb]

Your also probably better off using find instead of ls for example:

find ./features -type f -printf "%f\n"

Will return the list of "regular" files in ./features ( directories exlcuded ) the printf %f\n returns only the filename with the directory part striped off.