Problem with utilizing a "while loop" and subsequently processing data in a bash script

Question

I'm trying to create a script, which should read a video folder and create a list of video files to be processed by ffprobe to identify the codec. Videos NOT processed with a specific codec (in this case HEVC) should be put in new list for further processing by ffmpeg.

I created a very rudimentary script, but hit a brick wall at a point where the variable ffprobe_input needs to be changed in order to be passed as the next input for ffprobe.

Also, even if this part of the script was working, I'm puzzled as to how to create the filtered list of files after the ffprobe processing, since the only output is a single word, ex: hevc or x264.

The actual script is below, alongside with my notes, which should be more descriptive, also in the notes are some of the ways I tried to make things work.

This is the intended use of the script: ./script.sh -p /path\ to\ videos

#!/bin/bash
#Read path (-p) input and exit on error.
while getopts p: flag
    do
        case "${flag}" in
            p) vpath=${OPTARG};;
            *) echo "usage: $0 [-p]" >&2
            exit 1 ;;
        esac
    done
#Now we echo the path for neatness
    echo -e "Selected root video path: $vpath";
#Check if the path is valid. The path must be escaped. Cd into the folder and execute: printf "%q\n" "$(pwd)"
    [ -d "$vpath" ] && echo "Directory $vpath exists." || echo "Error: Directory $vpath does not exist. Tip: make sure the spaces are escaped in folder names, ex: ===video\ folder===."
#Prepare a list of video files with full escaped paths,ready for ffprobe/ffmpeg input.
    find "$vpath" -type f ( -iname ".mkv" -o -iname ".mp4" -o -iname "*.avi" ) | sed 's/ /\ /g' >> full_list.txt
#read the total number of lines from full_list.txt
    nrl_total="$(wc -l full_list.txt | grep -Eo "[0-9]{0,7}")"
    echo -e "There are a total of $nrl_total videos for further processing."
#read line number and pass to $ffprobe_input
nrl=($(seq 1 "$nrl_total"))
nrl={1..$nrl_total..1}
for $nlr in {1..$nrl_total..1}; do
nrl=({1..$nrl_total..1})
filename='full_list.txt'
nrl=1
while read line; do
    echo "$nrl"
    nrl=$((n+1))
#done < $filename
#ffprobe_input="$(sed -n 1p full_list.txt)" Use line number in "p" attribute, ex: 1p.
ffprobe_input="$(sed -n 1p full_list.txt)"
ffprobe_input=&quot;$(sed -n &quot;$nrl&quot;p full_list.txt)&quot;


#Now pass the input to ffprobe to determine if the videos are HEVC or not. Output is single word, ex: hevc or x264.
    eval ffprobe -v error -select_streams v:0 -show_entries stream=codec_name -of default=noprint_wrappers=1:nokey=1 -i "$ffprobe_input"
done < $filename
rm full_list.txt

Answer 1

Assuming your filenames don't contain newlines, you don't need to mangle them in any way. The output from file has one line per filename, so just store it and loop over the resulting file:

> non-hevc.txt        # clear the output list
find "$vpath" -type f \( -iname "*.mkv" -o -iname "*.mp4" -o -iname "*.avi" \) \
 > full_list.txt
while IFS= read -r file; do 
    result=$(ffprobe -v error -select_streams v:0 -show_entries \
             stream=codec_name -of default=noprint_wrappers=1:nokey=1 -i "$file")
    if [ "$result" != hevc ]; then
        echo "$file" >> non-hevc.txt
    fi
done < full_list.txt
rm -f full_list.txt

Here, the output of ffprobe is captured with the command substitution $(...) and stored to result, which we then look at.

I don't see any reason for the dance with sed -n "$nrl"p inside the loop reading the filename list, since read already reads the same line. We do need IFS= and -r to not mangle the input, though.

There's also no reason to escape any whitespace with backslashes, the quoted expansion of "$file" passes the contents of the variable as-is to the command. Undoing the escaping would also be difficult, when you use eval, it processes a lot of other stuff too, and would barf on e.g. parenthesis.

Not sure if you want to append the output of find to whatever full_list.txt already contained, or recreate the list. Since we process the list immediately, it seem to me to make more sense to ignore any old contents.

Note that like terdon comments, you don't strictly need the intermediate file to store the list of filenames. You could do just find ... | while IFS= read file, do ..., or with process substitution in Bash/ksh/zsh while IFS= read file, do ... done < <(find ...). The difference between the two matters if you want to set variables inside the while loop, see: Why is my variable local in one 'while read' loop, but not in another seemingly similar loop?