hpr-tools/workflow/postshow.bash

#!/usr/bin/env bash
# Copyright Ken Fallon - Released into the public domain. http://creativecommons.org/publicdomain/ 
#============================================================

processing_dir="$HOME/tmp/hpr/processing" # The directory where the files will be copied to for processing

if [ ! -d "${processing_dir}" ]
then
  echo "ERROR: The application \"${this_program}\" is required but is not installed."
  exit 1
fi

###################
# Check that all the programs are installed

function is_installed () {
  for this_program in "$@"
  do
    if ! command -v ${this_program} 2>&1 >/dev/null
    then
      echo "ERROR: The application \"${this_program}\" is required but is not installed."
      exit 2
    fi
  done
}

is_installed awk base64 cat curl curl date detox eval ffprobe file find grep grep head jq jq kate magick mediainfo mv mv rsync rsync seamonkey sed sed sort sponge ssh touch touch wget

echo "Processing the next HPR Show in the queue"

###################
# Get the show
#
# Replaced METADATA_PROCESSED with SHOW_SUBMITTED
response=$( curl --silent --netrc-file ${HOME}/.netrc "https://hub.hackerpublicradio.org/cms/status.php" | \
grep ',SHOW_SUBMITTED,' | \
head -1 | \
sed 's/,/ /g' )

if [ -z "${response}" ]
then
  echo "INFO: There appear to be no more shows with the status \"SHOW_SUBMITTED\"."
  echo "Getting a list of all the reservations."
  curl --silent --netrc-file ${HOME}/.netrc "https://hub.hackerpublicradio.org/cms/status.php" | sort -n
  exit 3
fi

timestamp_epoc="$( echo ${response} | awk '{print $1}' )"
ep_num="$( echo ${response} | awk '{print $2}' )"
ep_date="$( echo ${response} | awk '{print $3}' )"
key="$( echo ${response} | awk '{print $4}' )"
status="$( echo ${response} | awk '{print $5}' )"
email="$( echo ${response} | awk '{print $6}' )"
email_unpadded="$( echo $email | sed 's/.nospam@nospam./@/g' )"

upload_dir="/home/hpr/upload/${timestamp_epoc}_${ep_num}_${ep_date}_${key}"
source_dir="hpr:${upload_dir}"
publish_dir="hpr:www/eps/hpr${ep_num}"
dest_dir="${timestamp_epoc}_${ep_num}_${ep_date}_${key}"

ssh hpr -t "detox -v ${upload_dir}/"

echo "INFO: Downloading hpr${ep_num} from ${email_unpadded}"
echo ""
echo rsync -ave ssh --partial --progress ${source_dir}/ ${processing_dir}/${dest_dir}/
rsync -ave ssh --partial --progress ${source_dir}/ ${processing_dir}/${dest_dir}/
echo ""
echo "INFO: Working directory is \"${processing_dir}/${dest_dir}/\""
echo ""

shownotes_json="${processing_dir}/${dest_dir}/shownotes.json"

if [ ! -s "${shownotes_json}" ]
then
  echo "ERROR: \"${shownotes_json}\" is missing"
  exit 4
fi

if [ "$( file "${shownotes_json}" | grep -ic "text" )" -eq 0 ]
then
  echo "ERROR: \"${shownotes_json}\" is not a text file"
  exit 5
fi

mv -v "${shownotes_json}" "${shownotes_json%.*}_origional.json"

jq '.' "${shownotes_json%.*}_origional.json" > "${shownotes_json}"

###################
# Get the media
#

remote_media="$( jq --raw-output '.metadata.url' "${shownotes_json}" )"

if [ -n "${remote_media}" ]
then
  echo "INFO: Fetching remote media from \"${remote_media}\""
  wget --timestamping  --directory-prefix="${processing_dir}/${dest_dir}/" "${remote_media}"
  if [ $? -ne 0 ]
  then
    echo "ERROR: Could not get the remote media"
    exit 6
  fi
fi

shownotes_html="${processing_dir}/${dest_dir}/shownotes.html"

jq --raw-output '.episode.Show_Notes' "${shownotes_json}" > "${shownotes_html}"

if [ ! -s "${shownotes_html}" ]
then
  echo "ERROR: \"${shownotes_html}\" is missing"
  exit 7
fi

# Process Shownotes
sed "s#>#>\n#g" "${shownotes_html}" | sponge "${shownotes_html}"

# Extract Images
## TODO Temp fix until https://repo.anhonesthost.net/HPR/hpr-tools/issues/3

image_count="1"

for image in $( grep --color=never --perl-regexp --only-matching 'data:image/[^;]*;base64,\K[a-zA-Z0-9+/=]*' "${shownotes_html}" )
do
  this_image="${processing_dir}/${dest_dir}/hpr${ep_num}_${image_count}"
  echo -n "$image" | base64 -di > ${this_image}
  this_ext="$( file --mime-type ${this_image} | awk -F '/' '{print $NF}' )"
  mv -v "${this_image}" "${this_image}.${this_ext}"
  this_width="$( mediainfo "${this_image}.${this_ext}" | grep Width | awk -F ': | pixels' '{print $2}' | sed 's/ //g' )"
  if [ "${this_width}" -gt "400" ]
  then
    magick "${this_image}.${this_ext}" -resize 400x "${this_image}_tn.${this_ext}"
  fi
  ((image_count=image_count+1))
done

for image in $( grep --color=never --perl-regexp --only-matching '<img.*src.*http.*>' "${shownotes_html}" | awk -F 'src=' '{print $2}' | awk -F '"' '{print $2}' )
do
  this_image="${processing_dir}/${dest_dir}/hpr${ep_num}_${image_count}"
  wget "${image}" --output-document=${this_image}
  this_ext="$( file --mime-type ${this_image} | awk -F '/' '{print $NF}' )"
  if [ ! -e "${this_image}.${this_ext}" ]
  then
    mv -v "${this_image%.*}" "${this_image}.${this_ext}"
  fi
  this_width="$( mediainfo "${this_image}.${this_ext}" | grep Width | awk -F ': | pixels' '{print $2}' | sed 's/ //g' )"
  if [ "${this_width}" -gt "400" ]
  then
    magick "${this_image}.${this_ext}" -resize 400x "${this_image}_tn.${this_ext}"
  fi
  ((image_count=image_count+1))
done

## TODO End Temp fix until https://repo.anhonesthost.net/HPR/hpr-tools/issues/3

ls -al "${processing_dir}/${dest_dir}/"

## Manually edit the shownotes to fix issues

kate "${shownotes_html}" >/dev/null 2>&1 &
# librewolf "${shownotes_html}" >/dev/null 2>&1 &
seamonkey "${shownotes_html}" >/dev/null 2>&1 &
# bluefish "${shownotes_html}" >/dev/null 2>&1 &

read -p "Does the metadata 'look ok ? (N|y) ? " -n 1 -r
echo    # (optional) move to a new line
if [[ ! $REPLY =~ ^[yY]$ ]]
then
  echo "skipping...."
  exit 8
fi

media=$( find "${processing_dir}/${dest_dir}/" -type f -exec file {} \; | grep -Ei 'audio|mpeg|video|MP4' | awk -F ': ' '{print $1}' )
if [ -z "${media}" ]
then
  echo "ERROR: Can't find any media in \"${processing_dir}/${dest_dir}/\""
  find "${processing_dir}/${dest_dir}/" -type f
  exit 9
fi

the_show_media=""

if [ "$( echo "${media}" | wc -l )" -ne 1 ]
then
  echo "Multiple files found. Which one do you want to use ?"
  select this_media in $( echo "${media}" )
  do 
    echo "INFO: You selected \"${this_media}\"."
    ls -al "${this_media}"
    the_show_media="${this_media}"
    break
  done
else
  echo "INFO: Selecting media as \"${media}\"."
  the_show_media="${media}"
fi

duration=$( \date -ud "1970-01-01 $( ffprobe -i "${the_show_media}" 2>&1| awk -F ': |, ' '/Duration:/ { print $2 }' )" +%s )
if [ $? -ne 0 ]
then
  echo 'ERROR: Invalid duration found in '\"${media}\" >&2
  exit 10
fi

###################
# Gather episode information
#

if [ "$( curl --silent --write-out '%{http_code}'  http://hackerpublicradio.org/say.php?id=${ep_num} --output /dev/null )" == 200 ]
then
  echo "ERROR: The Episode hpr${ep_num} has already been posted"
  exit 11
fi

if [ "$( jq --raw-output '.metadata.Episode_Number' ${shownotes_json} )" != "${ep_num}" ]
then
  echo "ERROR: The Episode_Number: \"${ep_num}\" was not found in \"${shownotes_json}\""
  exit 12
fi

if [ "$( jq --raw-output '.metadata.Episode_Date' ${shownotes_json} )" != "${ep_date}" ]
then
  echo "ERROR: The Episode_Date: \"${ep_date}\" was not found in \"${shownotes_json}\""
  exit 13
fi

if [ "$( jq --raw-output '.host.Host_Email' ${shownotes_json} )" != "${email_unpadded}" ]
then
  echo "ERROR: The Host_Email: \"${email_unpadded}\" was not found in \"${shownotes_json}\""
  exit 14
fi

###################
# Assemble the components
#
# https://newbedev.com/how-to-urlencode-data-for-curl-command/
hostid="$( jq --raw-output '.host.Host_ID' ${shownotes_json} | jq --slurp --raw-input @uri | sed -e 's/%0A"$//g' -e 's/^"//g'  )"
host_name="$( jq --raw-output '.host.Host_Name' ${shownotes_json} | jq --slurp --raw-input @uri | sed -e 's/%0A"$//g' -e 's/^"//g' )"
title=$( jq --raw-output '.episode.Title' ${shownotes_json} | jq --slurp --raw-input @uri | sed -e 's/%0A"$//g' -e 's/^"//g' )
summary="$( jq --raw-output '.episode.Summary' ${shownotes_json} | jq --slurp --raw-input @uri | sed -e 's/%0A"$//g' -e 's/^"//g'  )"
series_id="$( jq --raw-output '.episode.Series' ${shownotes_json} | jq --slurp --raw-input @uri | sed -e 's/%0A"$//g' -e 's/^"//g'  )"
series_name="$( jq --raw-output '.episode.Series_Name' ${shownotes_json} | jq --slurp --raw-input @uri | sed -e 's/%0A"$//g' -e 's/^"//g' )"
explicit="$( jq --raw-output '.episode.Explicit' ${shownotes_json} | jq --slurp --raw-input @uri | sed -e 's/%0A"$//g' -e 's/^"//g'  )"
episode_license="$( jq --raw-output '.episode.Show_License' ${shownotes_json}  | jq --slurp --raw-input @uri | sed -e 's/%0A"$//g' -e 's/^"//g' )"
tags="$( jq --raw-output '.episode.Tags' ${shownotes_json} | jq --slurp --raw-input @uri | sed -e 's/%0A"$//g' -e 's/^"//g' )"
host_license=$( jq --raw-output '.host.Host_License' ${shownotes_json} | jq --slurp --raw-input @uri | sed -e 's/%0A"$//g' -e 's/^"//g' )
host_profile=$( jq --raw-output '.host.Host_Profile' ${shownotes_json} | jq --slurp --raw-input @uri | sed -e 's/%0A"$//g' -e 's/^"//g' )

notes="$( cat "${shownotes_html}" | jq --slurp --raw-input @uri | sed -e 's/%0A"$//g' -e 's/^"//g'  )"

if [ $# -gt 0 ]
then
  declare -A hash
  for argument
  do
    if [[ $argument =~ ^[^=]+=.*$ ]]
    then
      this_key="${argument%=*}"                                 # "${} Kate format hack
      this_value="${argument#*=}"                               # "${} Kate format hack
      this_value="$( echo "${this_value}" | jq --slurp --raw-input @uri | sed -e 's/%0A"$//g' -e 's/^"//g'  )"
      eval "${this_key}=${this_value}"
      echo "INFO: Replacing \"${this_key}\" with \"${this_value}\"."
    fi
  done
fi

if [ "${hostid}" == '0' ]
then
  echo "ERROR: The hostid is 0. Create the host and use \"hostid=???\" to override"
  exit 15
fi

###################
# Post show to HPR
#

post_show_json="${processing_dir}/${dest_dir}/post_show.json"
post_show_response="${processing_dir}/${dest_dir}/post_show_response.txt"

echo "Sending:"
cat "${post_show}"
echo "key=${key}
ep_num=${ep_num}
ep_date=${ep_date}
email=${email}
title=${title}
duration=${duration}
summary=${summary}
series_id=${series_id}
series_name=${series_name}
explicit=${explicit}
episode_license=${episode_license}
tags=${tags}
hostid=${hostid}
host_name=${host_name}
host_license=${host_license}
host_profile=${host_profile}
notes=${notes}"

echo "{
  \"key\": \"${key}\",
  \"ep_num\": \"${ep_num}\",
  \"ep_date\": \"${ep_date}\",
  \"email\": \"${email}\",
  \"title\": \"${title}\",
  \"duration\": \"${duration}\",
  \"summary\": \"${summary}\",
  \"series_id\": \"${series_id}\",
  \"series_name\": \"${series_name}\",
  \"explicit\": \"${explicit}\",
  \"episode_license\": \"${episode_license}\",
  \"tags\": \"${tags}\",
  \"hostid\": \"${hostid}\",
  \"host_name\": \"${host_name}\",
  \"host_license\": \"${host_license}\",
  \"host_profile\": \"${host_profile}\",
  \"notes\": \"${notes}\"
}" | tee "${post_show_json}"

echo "INFO: Uploading processed files to \"${processing_dir}/${dest_dir}/\""
echo ""
echo rsync -ave ssh --partial --progress ${processing_dir}/${dest_dir}/ ${source_dir}/ 
rsync -ave ssh --partial --progress ${processing_dir}/${dest_dir}/ ${source_dir}/ 
echo ""

curl --netrc --include --request POST "https://hub.hackerpublicradio.org/cms/add_show_json.php" --header "Content-Type: application/json" --data-binary "@${post_show_json}"

if [ "$( curl --silent --write-out '%{http_code}'  http://hackerpublicradio.org/say.php?id=${ep_num} --output /dev/null )" != 200 ]
then
  echo "ERROR: The Episode hpr${ep_num} has not been posted"
  exit 16
fi

echo "INFO: Uploading completed files to \"${publish_dir}/\""
echo ""
echo "rsync -ave ssh --partial --progress \"${processing_dir}/${dest_dir}/hpr${ep_num}\"* \"${publish_dir}/\""
rsync -ave ssh --partial --progress "${processing_dir}/${dest_dir}/hpr${ep_num}"* "${publish_dir}/"
echo ""