I need a file converter script

I need to convert a file that looks like this:

Fri Apr 14 15:42:02 UTC 2023

MemTotal: 65039504 kB 
MemFree: 41010436 kB 
MemAvailable: 45100588 kB

Fri Apr 14 16:35:01 UTC 2023

MemTotal: 65039504 kB 
MemFree: 40409508 kB 
MemAvailable: 44902852 kB

Fri Apr 14 16:36:01 UTC 2023

MemTotal: 65039504 kB 
MemFree: 40411232 kB 
MemAvailable: 44905376 kB

To something that looks like this:

15:42:02,65039504,41010436,45100588 
16:35:01,65039504,40409508,44902852 
16:36:01,65039504,40411232,44905376

Here's the script that I came up with:

#!/bin/bash
set -x

export TIME
export MEMTOTAL
export MEMFREE
export MEMAVAIL
export FILE="./SMALL-SAMPLE.txt"

while read -a LINE from $FILE
do
    WORD1=${LINE[0]}
    WORD2=${LINE[1]}
    WORD3=${LINE[2]}
    WORD4=${LINE[3]}
    WORD5=${LINE[4]}
    WORD6=${LINE[5]}
    WORD7=${LINE[6]}
    case $WORD1 in
        "Fri")
           TIME=$WORD4
           ;;
        "MemTotal")
           MEMTOTAL=$WORD2
           ;;
        "MemFree")
           MEMFREE=$WORD2
           ;;
        "MemAvailable")
          MEMAVAIL=$WORD2
          ;;
       *)continue;;
    esac
    LINEOUT="$TIME,$MEMTOTAL,$MEMFREE,$MEMAVAIL"
    echo $LINEOUT
done < $FILE

Here's the output:

15:42:02,,, 
16:35:01,,, 
16:36:01,,,

I've got a rookie mistake hidden in that script somewhere...any ideas about why I cannot get my data in?

Because TMTOWTDI, a shorter Perl version:

<file perl -anE '
    push @a, $F[3]||$F[1]||();
    say join",",splice@a if @a>3;
'

• -n makes Perl run the script for each record/line.
• -a turns on autosplit of records (by whitespace) into array @F
• Append to @a the first defined of $F[3] (timestamp) or $F[1] (value). If neither is defined, appending () is a no-op.
• If @a has 4 elements, print them, and truncate the array.

Your code has a few errors and stylistic issues.

• As @barmar stated, you should only print a line of output when you see the MemAvailable line (and so you also don't need the */continue clause)
• read ... from $FILE doesn't mean what you may think it does.
• You are not matching the trailing colon.
• You should normally quote variable use to avoid unintended word-splitting, globbing, etc (although that shouldn't happen here).
• You shouldn't use all-caps variable names - they are reserved for the system.
• There is no great advantage to defining new variables that only get used once.
• No need to export any variables.

file="./SMALL-SAMPLE.txt"

while read -a line
do
   case "${line[0]}" in
      Fri)
         time=${line[3]}
         ;;
      MemTotal:)
         memtotal=${line[1]}
         ;;
      MemFree:)
         memfree=${line[1]}
         ;;
      MemAvailable:)
         memavail=${line[1]}
         echo "$time,$memtotal,$memfree,$memavail"
         ;;
   esac
done < "$file"

One approach would be:

1. Remove all empty lines (with sed '/^$/d' $FILE)

2. Read four lines at once each iteration (repeating read for each one)

3. Extract the desired field using cut command
   <!-- -->

   $ cat script.sh
   #!/bin/bash

   FILE="./SMALL-SAMPLE.txt"

   while read ltime;
   read lmemt;
   read lmemf;
   read lmema;
   do
   TIME=$(echo $ltime | cut -d ' ' -f 4)
   MEMT=$(echo $lmemt | cut -d ' ' -f 2)
   MEMF=$(echo $lmemf | cut -d ' ' -f 2)
   MEMA=$(echo $lmema | cut -d ' ' -f 2)

   echo "$TIME,$MEMT,$MEMF,$MEMA"
   

   done < <(sed '/^$/d' $FILE)

Testing:

$ ./script.sh 
15:42:02,65039504,41010436,45100588
16:35:01,65039504,40409508,44902852
16:36:01,65039504,40411232,44905376

perl -ne '
    if (/:[^ ]/) { print /\d+:\d+:\d+/g, "," }
    elsif (/: /) { print /(\d+)/, $. % 6 == 5 ? "\n" : "," }
    ' -- file

• -n reads the input line by line and runs the code for each line;
• if the line contains a colon followed by non-space, print the part that contains three groups of digits separated by colons, and a comma;
• otherwise, if the line contains a colon followed by a space, print the digits it contains, followed by a newline on every last line of a group, otherwise a comma.

You forgot the trailing colons in your case choices ("MemTotal:" instead of "MemTotal"). But there are much simpler and much faster solutions (bash loops are slow).

Example with awk (tested with GNU awk and the awk that comes with macOS Ventura):

$ awk -v RS= -v OFS=, '/^Mem/ {print t,$2,$5,$8;next} {t=$4}' file
15:42:02,65039504,41010436,45100588
16:35:01,65039504,40409508,44902852
16:36:01,65039504,40411232,44905376

Explanations:

• -v RS= sets the record separator to empty lines.
• -v OFS=, sets the output field separator to commas.
• /^Mem/ {print t,$2,$5,$8;next} applies to records starting with Mem, prints the value of variable t and fields 2, 5 and 8 (the 3 sizes in the record), and goes to the next record.
• {t=$4} stores the fourth field (time) in variable t.

echo 'Fri Apr 14 15:42:02 UTC 2023

MemTotal: 65039504 kB
MemFree: 41010436 kB
MemAvailable: 45100588 kB

Fri Apr 14 16:35:01 UTC 2023

MemTotal: 65039504 kB
MemFree: 40409508 kB
MemAvailable: 44902852 kB

Fri Apr 14 16:36:01 UTC 2023

MemTotal: 65039504 kB
MemFree: 40411232 kB
MemAvailable: 44905376 kB' | 

> nawk '(NF = NF)^(ORS = NR % 4 ? "," : "\n")' RS='\n+'
> OFS= FS='^(([^: ]+ )+|[^ :]+: )| [[:alpha:]]+.+$'

• more succinct ::
  > gawk '(ORS=/v/?RS:",")^!(NF+=OFS=_)' FS='^([^:]+|[^ ]+) | [?-|].+$'

• extreme compression ::
  > mawk 'NF&&/v/*($__==/: /?","$2:$4)'

  15:42:02,65039504,41010436,45100588
  16:35:01,65039504,40409508,44902852
  16:36:01,65039504,40411232,44905376