Correct location for the new server

Added SQL rss feed, Fixed regex for image, and scp files to hpr server
Automating sql generation on db change
2025-09-15 20:15:59 +02:00 · 2025-09-11 17:50:31 +02:00 · 2025-09-10 21:23:37 +02:00 · 2025-09-08 09:49:49 +02:00 · 2025-08-29 12:21:07 +02:00
8 changed files with 348 additions and 10 deletions
--- a/workflow/hpr-get-and-transcribe.bash
+++ b/workflow/hpr-get-and-transcribe.bash
@@ -0,0 +1,35 @@
+#!/bin/bash
+
+rsync -av --partial --progress hpr:hub.hackerpublicradio.org/upload/ $HOME/tmp/hpr/processing/
+
+find $HOME/tmp/hpr/processing/*_*_????-??-??_* -type d | sort -t _ -k 2 | while read show_dir
+do
+  echo "${show_dir}"
+  if [ "$( find "${show_dir}" -type f -iname "*srt" | wc -l )" -eq "0" ]
+  then
+    cd "${show_dir}"
+    ls -haltr
+    find "${show_dir}/" -type f -exec file {} \; | grep -Ei 'audio|mpeg|video|MP4' | awk -F ': ' '{print $1}' | while read this_media
+    do
+      whisper --model tiny --language en --output_dir "${show_dir}" "${this_media}"
+    done
+    rsync -av --partial --progress "${show_dir}/" hpr:hub.hackerpublicradio.org/upload/$( basename "${show_dir}")/
+  fi
+done
+
+rsync -av --partial --progress hpr:hub.hackerpublicradio.org/reserve/ $HOME/tmp/hpr/reserve/
+
+find $HOME/tmp/hpr/reserve/*_*_* -type d | sort -t _ -k 2 | while read show_dir
+do
+  echo "${show_dir}"
+  if [ "$( find "${show_dir}" -type f -iname "*srt" | wc -l )" -eq "0" ]
+  then
+    cd "${show_dir}"
+    ls -haltr
+    find "${show_dir}/" -type f -exec file {} \; | grep -Ei 'audio|mpeg|video|MP4' | awk -F ': ' '{print $1}' | while read this_media
+    do
+      whisper --model tiny --language en --output_dir "${show_dir}" "${this_media}"
+    done
+    rsync -av --partial --progress "${show_dir}/" hpr:hub.hackerpublicradio.org/reserve/$( basename "${show_dir}")/
+  fi
+done
--- a/workflow/hpr-publish.bash
+++ b/workflow/hpr-publish.bash
@@ -0,0 +1,44 @@
+#!/usr/bin/env bash
+# Copyright Ken Fallon - Released into the public domain. http://creativecommons.org/publicdomain/ 
+#============================================================
+
+git_dir="$HOME/tmp/hpr/hpr_generator/sourcecode"
+if [ ! -d "${git_dir}/.git" ]
+then
+  git clone gitea@repo.anhonesthost.net:HPR/hpr_generator.git "${git_dir}"
+fi
+
+cd "${git_dir}"
+
+git pull
+
+# ssh hpr -t "ls -al /home/hpr/www/hpr.sql;md5sum /home/hpr/www/hpr.sql"
+# ssh hpr -t "/home/hpr/bin/hpr_db_backup.bash"
+# ssh hpr -t "ls -al /home/hpr/www/hpr.sql;md5sum /home/hpr/www/hpr.sql"
+
+./utils/update-hpr-db.sh
+if [ $? -ne 0 ]
+then
+  echo 'Terminating...' >&2
+  exit 1
+fi
+
+./site-generator --all --verbose
+if [ $? -ne 0 ]
+then
+  echo 'Terminating...' >&2
+  exit 1
+fi
+
+rsync -av --partial --progress "${git_dir}/public_html/" hpr:hackerpublicradio.org/public_html
+#rsync -av --partial --progress "${git_dir}/public_html/" hobbypublicradio.org:hobbypublicradio.org/
+
+cd $HOME/sourcecode/hpr/hpr_hub/
+git pull 
+cd $HOME/sourcecode/hpr/hpr_hub/sql
+split --hex-suffixes --lines=1000 --additional-suffix=.sql hpr.sql hpr-db-part-
+cd $HOME/sourcecode/hpr/hpr_hub/
+git add $HOME/sourcecode/hpr/hpr_hub/sql/hpr*sql 
+git commit -m "$(\date -u +%Y-%m-%d_%H-%M-%SZ_%A ) database changed"
+git push
+#xdg-open https://hackerpublicradio.org/
--- a/workflow/hpr_ccdn_stats.bash
+++ b/workflow/hpr_ccdn_stats.bash
@@ -0,0 +1,4 @@
+#!/bin/bash
+
+yesterday="$( \date -u +%Y-%m-%d -d yesterday)" 
+echo -e "${yesterday}\t$( grep -Ec "${yesterday}T.*Sending request to" /home/hpr/logs/naughty-ip.txt )" >> /home/hpr/hub/hpr_ccdn_stats.tsv
--- a/workflow/hpr_db_backup.bash
+++ b/workflow/hpr_db_backup.bash
@@ -0,0 +1,194 @@
+#!/bin/bash
+
+# 5,20,35,50 * * * * $HOME/userfiles/sql/hpr_db_backup.bash >> $HOME/userfiles/sql/cron.log 2>&1 &
+
+#TODO
+#add a rss feed with the latest updates. Linking to the changes on gittea.
+# run that every 5 minutes and then if there is a change sleep 5 and confirm there has been no change since.
+# Then download the db and regenerate the site.
+# While making sure to generate the site at least once a day 
+# check the skip-extended-insert export into of the sql into git, so that the small diffs show up.
+# generate a rss feed with the latest changes.
+# update all the other scripts reserve, stats etc.
+
+
+sql_save_dir="$HOME/userfiles/sql"
+credential_file="${sql_save_dir}/.my-hpr.cnf"
+last_update_txt="${sql_save_dir}/last_update.txt"
+hpr_full_sql="${sql_save_dir}/hpr_hpr_full.sql"
+hpr_site_sql="${sql_save_dir}/hpr.sql"
+full_mysqldump_sql="${sql_save_dir}/mysqldump.sql"
+hpr_database_file="hackerpublicradio.org/public_html/hpr.sql"
+
+sync_delay_seconds="300" # 5 minutes
+
+last_update_query="SELECT
+  update_time
+FROM
+  information_schema.tables tab
+WHERE
+  update_time > (current_timestamp() - interval 30 day)
+  AND table_type = 'BASE TABLE'
+  AND table_name not in ('reservations')
+  AND table_schema not in ('information_schema', 'sys', 'performance_schema', 'mysql')
+ORDER BY
+  update_time DESC
+LIMIT 1;"
+
+if [ ! -s "${credential_file}" ]
+then
+  echo "The file \"${credential_file}\" is missing"
+  exit 
+fi
+
+local_db_last_update_epoch="0"
+if [ -s "${last_update_txt}" ]
+then
+  echo "Found the last update file \"${last_update_txt}\""
+  local_db_last_update_iso8601="$( \date --utc --date="$( cat ${last_update_txt} )" +%Y-%m-%dT%H:%M:%SZ )"
+  local_db_last_update_epoch="$( \date --utc --date="$( cat ${last_update_txt} )" +%s )"
+  echo -e "Local DB update time is\t${local_db_last_update_iso8601} (${local_db_last_update_epoch})"
+fi
+
+live_db_last_update_iso8601="$( mysql --defaults-file="${credential_file}" --disable-column-names --batch --execute="${last_update_query}" | sed -e 's/ /T/g' -e 's/$/Z/g' )"
+if [ -z "${live_db_last_update_iso8601}" ]
+then
+  echo "The live db update time \"live_db_last_update_iso8601\" is missing"
+  exit 1
+fi
+
+live_db_last_update_epoch="$( \date --utc --date="${live_db_last_update_iso8601}" +%s )"
+if [ -z "${live_db_last_update_epoch}" ]
+then
+  echo "The live db update time \"live_db_last_update_epoch\" is missing"
+  exit 2
+fi
+
+echo -e "Live DB update time is\t${live_db_last_update_iso8601} (${live_db_last_update_epoch})"
+
+if [ "${local_db_last_update_epoch}" -eq "${live_db_last_update_epoch}" ]
+then
+  echo "No changes detected. Skipping export."
+  exit 0
+fi
+
+echo "Starting export full with -complete-insert."
+if [ -s "${hpr_full_sql}" ]
+then
+  hpr_full_sql_write_time_iso8601="$( \date --utc --date="$( ls -al --full-time "${hpr_full_sql}" | awk '{print $6, $7, $8}' )" +%Y-%m-%dT%H:%M:%SZ )"
+  hpr_full_sql_write_time_epoch="$( \date --utc --date="${hpr_full_sql_write_time_iso8601}" +%s )"
+  if [ -z "${hpr_full_sql_write_time_epoch}" ]
+  then
+    echo "The live db update time \"hpr_full_sql_write_time_epoch\" is missing"
+    exit 3
+  fi
+  echo -e "Full DB write time is\t${hpr_full_sql_write_time_iso8601} (${hpr_full_sql_write_time_epoch})"
+
+  hpr_full_sql_write_time_with_delay_epoch="$(( ${hpr_full_sql_write_time_epoch} + ${sync_delay_seconds} ))"
+  time_now_epoch="$( \date --utc +%s )"
+
+  if [ "${hpr_full_sql_write_time_with_delay_epoch}" -gt "${time_now_epoch}" ]
+  then
+    echo "Skipping export. The Database has been recently created \"${hpr_full_sql_write_time_iso8601}\". Try again after $( \date --utc --date="@${hpr_full_sql_write_time_with_delay_epoch}" +%Y-%m-%dT%H:%M:%SZ )."
+    exit 4
+  fi
+fi
+mysqldump --defaults-file="${credential_file}" --tz-utc --add-drop-database --extended-insert --complete-insert --skip-extended-insert --default-character-set=utf8 --single-transaction --skip-set-charset --databases hpr_hpr > "${hpr_full_sql}"
+tail "${hpr_full_sql}" | grep 'Dump completed on'
+
+echo "Starting export full for static site generation."
+if [ -s "${hpr_site_sql}" ]
+then
+  hpr_site_sql_write_time_iso8601="$( \date --utc --date="$( ls -al --full-time "${hpr_site_sql}" | awk '{print $6, $7, $8}' )" +%Y-%m-%dT%H:%M:%SZ )"
+  hpr_site_sql_write_time_epoch="$( \date --utc --date="${hpr_site_sql_write_time_iso8601}" +%s )"
+  if [ -z "${hpr_site_sql_write_time_epoch}" ]
+  then
+    echo "The live db update time \"hpr_site_sql_write_time_epoch\" is missing"
+    exit 5
+  fi
+  echo -e "Full DB write time is\t${hpr_site_sql_write_time_iso8601} (${hpr_site_sql_write_time_epoch})"
+
+  hpr_site_sql_write_time_with_delay_epoch="$(( ${hpr_site_sql_write_time_epoch} + ${sync_delay_seconds} ))"
+  time_now_epoch="$( \date --utc +%s )"
+
+  if [ "${hpr_site_sql_write_time_with_delay_epoch}" -gt "${time_now_epoch}" ]
+  then
+    echo "Skipping export. The Database has been recently created \"${hpr_site_sql_write_time_iso8601}\". Try again after $( \date --utc --date="@${hpr_site_sql_write_time_with_delay_epoch}" +%Y-%m-%dT%H:%M:%SZ )."
+    exit 6
+  fi
+fi
+mysqldump --defaults-file="${credential_file}" --tz-utc --add-drop-database --complete-insert --extended-insert --default-character-set=utf8 --single-transaction --skip-set-charset --databases hpr_hpr --ignore-table=hpr_hpr.reservations > "${hpr_site_sql}"
+tail "${hpr_site_sql}" | grep 'Dump completed on'
+
+echo "Starting export full for data recovery."
+if [ -s "${full_mysqldump_sql}" ]
+then
+  full_mysqldump_sql_write_time_iso8601="$( \date --utc --date="$( ls -al --full-time "${full_mysqldump_sql}" | awk '{print $6, $7, $8}' )" +%Y-%m-%dT%H:%M:%SZ )"
+  full_mysqldump_sql_write_time_epoch="$( \date --utc --date="${full_mysqldump_sql_write_time_iso8601}" +%s )"
+  if [ -z "${full_mysqldump_sql_write_time_epoch}" ]
+  then
+    echo "The live db update time \"full_mysqldump_sql_write_time_epoch\" is missing"
+    exit 5
+  fi
+  echo -e "Full DB write time is\t${full_mysqldump_sql_write_time_iso8601} (${full_mysqldump_sql_write_time_epoch})"
+
+  full_mysqldump_sql_write_time_with_delay_epoch="$(( ${full_mysqldump_sql_write_time_epoch} + ${sync_delay_seconds} ))"
+  time_now_epoch="$( \date --utc +%s )"
+
+  if [ "${full_mysqldump_sql_write_time_with_delay_epoch}" -gt "${time_now_epoch}" ]
+  then
+    echo "Skipping export. The Database has been recently created \"${full_mysqldump_sql_write_time_iso8601}\". Try again after $( \date --utc --date="@${full_mysqldump_sql_write_time_with_delay_epoch}" +%Y-%m-%dT%H:%M:%SZ )."
+    exit 6
+  fi
+fi
+mysqldump --defaults-file="${credential_file}" --tz-utc --add-drop-database  --databases hpr_hpr> "${full_mysqldump_sql}"
+tail "${full_mysqldump_sql}" | grep 'Dump completed on'
+
+echo "<?xml version=\"1.0\" encoding=\"UTF-8\"?>
+<rss xmlns:atom=\"https://www.w3.org/2005/Atom\" version=\"2.0\">
+  <channel>
+    <title>Hacker Public Radio ~ Database Feed</title>
+    <link>http://hackerpublicradio.org/about.html</link>
+    <description>This Feed provides information the latest version of the HPR database.</description>
+    <language>en-us</language>
+    <copyright>Creative Commons Attribution-ShareAlike 4.0 International (CC BY-SA 4.0) License</copyright>
+    <managingEditor>feedback.nospam@nospam.hackerpublicradio.org (HPR Feedback)</managingEditor>
+    <webMaster>admin.nospam@nospam.hackerpublicradio.org (HPR Webmaster)</webMaster>
+    <generator>https://repo.anhonesthost.net/HPR/hpr-tools/src/branch/main/workflow/hpr_db_backup.bash</generator>
+    <docs>https://www.rssboard.org/rss-specification</docs>
+    <ttl>15</ttl>
+    <image>
+      <url>http://hackerpublicradio.org/images/hpr_feed_small.png</url>
+      <title>Hacker Public Radio ~ Database Feed</title>
+      <link>http://hackerpublicradio.org/about.html</link>
+      <description>The Hacker Public Radio Old Microphone Logo</description>
+      <height>164</height>
+      <width>144</width>
+    </image>
+    <atom:link href=\"https://hackerpublicradio.org/hpr.sql.rss\" rel=\"self\" type=\"application/rss+xml\"/>
+    <pubDate>$( date --utc --rfc-email )</pubDate>
+    <item>
+      <title>Export of the Public mariadb SQL for ${live_db_last_update_iso8601}</title>
+      <author>admin.nospam@nospam.hackerpublicradio.org (Janitors)</author>
+      <link>http://hackerpublicradio.org/hpr.sql#${live_db_last_update_iso8601}</link>
+      <description/>
+      <pubDate>$( date --utc --rfc-email )</pubDate>
+      <enclosure url=\"http://hackerpublicradio.org/hpr.sql#${live_db_last_update_iso8601}\" length=\"$( ls -al "${hpr_site_sql}" | awk '{print $5}' )\" type=\"application/sql\"/>
+      <guid isPermaLink=\"false\">sha1sum:$( sha1sum "${hpr_site_sql}" | awk '{print $1}' ),md5sum:$( md5sum "${hpr_site_sql}" | awk '{print $1}' )</guid>      
+    </item>
+  </channel>
+</rss>" > "${hpr_site_sql}.rss"
+
+if [ $HOSTNAME = "whp01.cloud-hosting.io" ]
+then
+  cp -v "${hpr_site_sql}" "$HOME/${hpr_database_file}"
+  cp -v "${hpr_site_sql}.rss" "$HOME/${hpr_database_file}.rss"
+else
+  rsync -av --partial --progress ${hpr_site_sql} hpr:${hpr_database_file}
+  rsync -av --partial --progress ${hpr_site_sql}.rss hpr:${hpr_database_file}.rss
+fi
+
+echo "${live_db_last_update_iso8601}" > "${last_update_txt}"
+
+echo "Finished export of \"${live_db_last_update_iso8601}\""
+
--- a/workflow/mdb.bash
+++ b/workflow/mdb.bash
@@ -0,0 +1,4 @@
+#!/bin/bash
+pw=$(grep -E '^\$databasePassword = ' /home/hpr/php/credentials.php | awk -F "'" '{print $2}' )
+pw=${pw##* }
+mysql --host=localhost --user=hpr_hpr --password="$pw" hpr_hpr
--- a/workflow/process_episode.bash
+++ b/workflow/process_episode.bash
@@ -91,7 +91,7 @@ function program_checks() {
    done
  }

-  is_installed audio2image.bash awk base64 cat csvtojson curl date detox eval extract_images ffprobe file find grep head jq kate magick mediainfo mv realpath remove-image.pl rsync seamonkey sed sed sort sponge ssh touch touch wget hpr-check-ccdn-links
+  is_installed audio2image.bash awk base64 cat csvtojson curl date eval extract_images ffprobe ffmpeg file find grep head jq kate magick mediainfo mv realpath remove-image.pl rsync seamonkey sed sed sort sponge ssh touch touch wget hpr-check-ccdn-links

  for arg in $*
  do
@@ -518,10 +518,10 @@ function check_variable_is_correct() {
        then
          echo_error "The \"shownotes_json\" file is missing."
        fi
-        if [ "$( file --brief --mime-type "${shownotes_json}" | grep --count 'application/json' )" -ne "1" ]
-        then
-          echo_error "The \"shownotes_json\" variable has not a valid \"application/json\" mime type."
-        fi
+#         if [ "$( file --brief --mime-type "${shownotes_json}" | grep --count 'application/json' )" -ne "1" ]
+#         then
+#           echo_error "The \"shownotes_json\" variable has not a valid \"application/json\" mime type."
+#         fi
        jq '.' "${shownotes_json}"  >/dev/null 2>&1
        if [ $? -ne 0 ]
        then
@@ -637,7 +637,7 @@ function get_next_show_from_hpr_hub() {
  email="$( echo ${response} | awk '{print $6}' )"
  email_unpadded="$( echo $email | sed 's/.nospam@nospam./@/g' )"

-  hpr_upload_dir="/home/hpr/upload/${timestamp_epoc}_${ep_num}_${ep_date}_${key}"
+  hpr_upload_dir="hub.hackerpublicradio.org/upload/${timestamp_epoc}_${ep_num}_${ep_date}_${key}"
  source_dir="hpr:${hpr_upload_dir}"
  dest_dir="${timestamp_epoc}_${ep_num}_${ep_date}_${key}"
  working_dir="${processing_dir}/${dest_dir}"
@@ -647,7 +647,6 @@ function get_next_show_from_hpr_hub() {
  echo_debug "Downloading hpr${ep_num} from ${email_unpadded}"
  echo_debug ""
  echo_debug "rsync -ave ssh --partial --progress ${source_dir}/ ${working_dir}/"
-  ssh hpr -t "detox -v ${hpr_upload_dir}/"
  rsync -ave ssh --partial --progress ${source_dir}/ ${working_dir}/

 }
@@ -870,7 +869,7 @@ function extract_images_brute_force() {
  # Download referenced images
  image_count_external="1"

-  for image in $( grep --color=never --perl-regexp --only-matching '<img.*src.*http.*>' "${shownotes_html}" | awk -F 'src=' '{print $2}' | awk -F '"' '{print $2}' )
+  for image in $( grep --color=never --perl-regexp --only-matching '<img.*src.*http.*://.*>' "${shownotes_html}" | awk -F 'src=' '{print $2}' | awk -F '"' '{print $2}' )
  do
    this_image="${working_dir}/hpr${ep_num}_image_ext_${image_count_external}"
    wget "${image}" --output-document=${this_image}
@@ -1032,7 +1031,7 @@ function media_checks() {
  shownotes_srt="${media%.*}.srt"
  if [[ -z "${shownotes_srt}" || ! -s "${shownotes_srt}" ]]
  then
-    echo_error "Could not find the subtitles for media \"${media}\""
+    echo_error "Could not find the subtitles for media \"${media}\" in \"${shownotes_srt}\""
  fi
  
  #TODO fix close duration
@@ -1968,7 +1967,7 @@ function register_assets() {
 }

 #################################################
-# Register the assets with the hpr database
+# Copy the files to the origin server

 function copy_files_to_origin_server() {

@@ -2014,6 +2013,48 @@ function copy_files_to_origin_server() {
  done
 }

+#################################################
+# Copy the images to the hpr server
+
+function copy_files_to_hpr_server() {
+
+  echo_debug "Copying the files to the hpr server. copy_files_to_hpr_server()"
+
+  check_variable_is_correct working_dir ep_num
+
+  # Get the file list from the HPR db
+  if [ -z "${files_json}" ]
+  then
+    files_json="${working_dir}/hpr${ep_num}_files.json"
+  fi
+
+  if [ -s "${files_json}" ]
+  then
+    echo_debug "Removing stale \"files_json\" file \"${files_json}\"."
+    rm -v "${files_json}"
+  fi
+  
+  echo_debug "Getting the \"files_json\" file from the HPR site."
+  if [ "$( curl --silent --netrc --write-out '%{http_code}'  https://hub.hackerpublicradio.org/cms/assets.php?id=${ep_num} --output "${files_json}" )" != 200 ]
+  then
+    echo_error "The Episode hpr${ep_num} has not been posted"
+  fi
+  check_variable_is_correct files_json
+
+  ssh hpr -t "mkdir -v hackerpublicradio.org/public_html/eps/hpr${ep_num}"
+  
+  for this_file in $( jq --raw-output ".hpr${ep_num} | keys | @tsv" "${files_json}" )
+  do
+    skip_file="$( echo "$this_file" | grep --count --perl-regexp "hpr${ep_num}.flac|hpr${ep_num}.mp3|hpr${ep_num}.ogg|hpr${ep_num}.opus|hpr${ep_num}.spx|hpr${ep_num}.srt|hpr${ep_num}.txt|hpr${ep_num}.wav" )"
+    if [ "${skip_file}" -eq "1" ]
+    then
+      continue
+    fi
+    scp "${working_dir}/$this_file" hpr:hackerpublicradio.org/public_html/eps/hpr${ep_num}/
+  done
+
+}
+
 #################################################
 # Send the derived files to the server borg to be sent to borg

@@ -2335,6 +2376,8 @@ copy_files_to_origin_server

 check_variable_is_correct working_dir ep_num shownotes_edited

+copy_files_to_hpr_server
+
 copy_derived_files_to_borg

 create_item_on_the_internet_archive
--- a/workflow/run-speedtest.bash
+++ b/workflow/run-speedtest.bash
@@ -0,0 +1,6 @@
+#!/bin/bash
+
+# * * * * * /usr/local/bin/run-speedtest.bash >/dev/null 2>&1
+
+speedtest-cli --json | jq '.' > /var/www/html/speedtest.json
+chown apache:apache /var/www/html/speedtest.json
--- a/workflow/update-stats.bash
+++ b/workflow/update-stats.bash
@@ -0,0 +1,8 @@
+#!/bin/bash
+
+hub_dir="/home/hpr/hub"
+
+for format in txt csv json xml
+do 
+  curl --silent --netrc-file /home/hpr/.netrc https://hub.hackerpublicradio.org/cms/stats.php?format=${format} --output ${hub_dir}/stats.${format} 
+done
Author	SHA1	Message	Date
Ken Fallon	8efc6459c3	Correct location for the new server	2025-09-15 20:15:59 +02:00
Ken Fallon	2b4ef438b3	Added SQL rss feed, Fixed regex for image, and scp files to hpr server	2025-09-11 17:50:31 +02:00
Ken Fallon	16afecfb82	Automating sql generation on db change	2025-09-10 21:23:37 +02:00
Ken Fallon	27bb3be4af	Move HPR/hpr_hub#88	2025-09-08 09:49:49 +02:00
Ken Fallon	1a69042b24	added check for ffmpeg	2025-08-29 12:21:07 +02:00