loop.sh

#!/usr/bin/env bash

# Optional check for tmux
if [ "$TMUX" == '' ]; then
    echo 'Not running inside tmux. Abort.'
    exit 1
fi

# Stop execution, if something goes wrong
set -e

# If you changed the default values using command line arguments,
# you should change the values here too.
clone_directory_path='cloned'
lock_directory_path='locks'
qresult_file='qresults'

# Run until undocumented GitHub API limit (max. 20 pages)
for i in {0..19}; do
    # Manual garbage collection
    rm -rf "$clone_directory_path" "$lock_directory_path" "$qresult_file"

    # Save the current page offset for debugging purposes
    # The number may be used to continue the loop, if something should crash
    echo "$i" > current_page_offset

    # $(nproc) + 1 = $(nproc) (or os.cpu_count()) pull requests + default branch

    # Query some repositories
    ./query.py "$qresult_file" \
        --repo-page-offset "$i" \
        --report-threshold "$(($(nproc) + 1))"

    # Run detector
    xargs -P "$(nproc)" -I {} \
        ./detect.py --report-threshold "$(($(nproc) + 1))" \
        '{}' < "$qresult_file"
done