#!/bin/bash

# List of media file extensions
mediaext='(mp4|avi|mkv|webm|mp3|m4a|flac)'
# Command for triggering cache
cache='curl -Ss --range 0-0 >/dev/null'
# Script for sed, to remove the first section from a log line
popsec='s/^[^|]*| //'

# Log format: 12/Aug/2023:hh:mm:ss <GMT offset> | <type>:<result> | 
#             <bytes sent> to <ip address> in <time> |
#             /path/to/thing?arg=stuff&nextarg

# Log lines are filtered and cut down to just the URI path
sedargs=(
  -e "$popsec"                # Remove first section (date and time)
  -e "/^GET:200/!d; $popsec"  # Filter for successful GET requests
  -e 's/?.*$//'               # Cut off arguments (? and beyond)
  -e "/\.$mediaext$/!d"       # Filter for media files
)
# Wait for new requests to show up in the log file
tail -n0 -F /var/log/nginx/access.log | sed "${sedargs[@]}" |
{ while read request; do

  # URI must be of the form: /media/collection/item/[.../]episode
  [[ $request =~ (/+[^/]+){4,} ]] || continue

  # Remove bytes sent
  request=${request#*to }
  # Extract client address
  client="http://${request%% *}"

  # Test the client for an HTTP server
  curl --head -sf $client >/dev/null || continue
  # Remove section, leaving URI path
  request=${request#*| }

  # Output from wget is filtered for media file URIs
  sedargs=(\
    -e '/^--/!d'              # Filter for lines beginning with --
    -e "/\.$mediaext$/!d"     # Filter for media files
    -e 's|^[^/]*//[^/]*||'    # Remove everything through http://hostname
  )
  # Get file index for request directory: /media/collection/item/[.../]
  wget -np -r --spider "http://localhost${request%/*}/" 2>&1 |
    sed "${sedargs[@]}" |
  { read first

    # Save the first URI, and see if it matches the request URI
    unset found
    [[ $first == $request ]] && found=1

    while read uri; do
      # Suggest to cache this URI if the previous URI matched
      [[ -z $found ]] || { $cache $client$uri; unset found; break; }
      # Check if this URI matches
      [[ $uri == $request ]] && found=1
    done
    # Suggest to cache the first URI if the final URI matched
    [[ -z $found ]] || $cache $client$first
  }
done; }

# tail shouldn't return unless there's a problem
exit 1