]> git.openstreetmap.org Git - chef.git/blobdiff - cookbooks/planet/templates/default/planetdump.erb
create XML RSS Feed for .torrent files (using xmlstarlet)
[chef.git] / cookbooks / planet / templates / default / planetdump.erb
index 17adfc4d11fbb55a4ae127ad2a21541c4fd184c0..d2da54bb74bd80b1f23b8ec99110662dcf3973f7 100644 (file)
@@ -81,24 +81,72 @@ function mk_torrent {
   web_dir="$3"
   name="${type}-${date}.osm.${format}"
   web_path="${web_dir}/${name}"
+  rss_web_dir="https://planet.openstreetmap.org/${web_dir}"
+  rss_file="${type}-${format}-rss.xml"
+  torrent_file="${name}.torrent"
+  torrent_url="${rss_web_dir}/${torrent_file}"
 
+  # create .torrent file
   mktorrent -l 22 ${name} \
      -a udp://tracker.opentrackr.org:1337 \
      -a udp://tracker.datacenterlight.ch:6969/announce,http://tracker.datacenterlight.ch:6969/announce \
      -a udp://tracker.torrent.eu.org:451 \
      -a udp://tracker-udp.gbitt.info:80/announce,http://tracker.gbitt.info/announce,https://tracker.gbitt.info/announce \
      -a http://retracker.local/announce \
+     -w https://planet.openstreetmap.org/${web_path} \
      -w https://ftp5.gwdg.de/pub/misc/openstreetmap/planet.openstreetmap.org/${web_path} \
      -w https://ftpmirror.your.org/pub/openstreetmap/${web_path} \
      -w https://mirror.init7.net/openstreetmap/${web_path} \
-     -w https://planet.openstreetmap.org/${web_path} \
      -w https://free.nchc.org.tw/osm.planet/${web_path} \
      -w https://ftp.fau.de/osm-planet/${web_path} \
      -w https://ftp.spline.de/pub/openstreetmap/${web_path} \
      -w https://osm.openarchive.site/${name} \
      -w https://downloads.opencagedata.com/planet/${name} \
-     -c "OpenStreetMap planet database dump, licensed under https://opendatacommons.org/licenses/odbl/ by OpenStreetMap contributors" \
-     -o ${name}.torrent
+     -w https://planet.osm-hr.org/${web_path} \
+     -c "OpenStreetMap ${type} data export, licensed under https://opendatacommons.org/licenses/odbl/ by OpenStreetMap contributors" \
+     -o ${torrent_file}
+
+  # create .xml global RSS headers if missing
+  torrent_time_rfc="`date -R -r ${torrent_file}`"
+  test -f "${rss_file}" || echo "<x/>" | xmlstarlet select --xml-decl --indent \
+       -N "atom=http://www.w3.org/2005/Atom" \
+       -N "dcterms=http://purl.org/dc/terms/" \
+       -N "content=http://purl.org/rss/1.0/modules/content/" \
+       --encode "UTF-8" \
+       --template \
+       --match / \
+       --elem "rss" --attr "version" --output "2.0" --break \
+       --elem "channel" \
+       --elem "title" --output "OpenStreetMap ${type} ${format} torrent RSS" --break \
+       --elem "link"  --output "${rss_web_dir}" --break \
+       --elem "atom:link" \
+               --attr "href" --output "${rss_web_dir}/${rss_file}" --break \
+               --attr "rel" --output "self" --break \
+               --attr "type" --output "application/rss+xml" --break \
+               --break \
+       --elem "description" --output "${type}.osm.${format}.torrent RSS feed" --break \
+       --elem "generator" --output "OpenStreetMap xmlstarlet powered shell script v1.0" --break \
+       --elem "language" --output "en" --break \
+       --elem "lastBuildDate" --output "${torrent_time_rfc}" \
+       > ${rss_file}
+
+  # add newly created .torrent file as new entry to .xml RSS feed, removing excess entries
+  xmlstarlet edit --inplace \
+       -s /rss/channel -t elem -n NEW_item  \
+       -s //NEW_item -t elem -n "title" -v "${torrent_file}" \
+       -s //NEW_item -t elem -n "guid" -v "${torrent_url}" \
+       -s //NEW_item -t elem -n "pubDate" -v "${torrent_time_rfc}" \
+       -s //NEW_item -t elem -n "category" -v "OpenStreetMap data" \
+       -s //NEW_item -t elem -n "link" -v "${torrent_url}" \
+       -s //NEW_item -t elem -n "enclosure" \
+               -s //NEW_item/enclosure -t attr -n "type" -v "application/x-bittorrent" \
+               -s //NEW_item/enclosure -t attr -n "url" -v "${torrent_url}" \
+       -s //NEW_item -t elem -n "description" -v "OpenStreetMap torrent ${torrent_file}" \
+       -s //NEW_item -t elem -n "comments" -v "Source: OpenStreetMap contributors, under ODbL 1.0 licence" \
+       -r //NEW_item -v "item" \
+       -u /rss/channel/lastBuildDate -v "${torrent_time_rfc}" \
+       -d '/rss/channel/item[position() <= count(/rss/channel/item) - 5]' \
+       ${rss_file}
 }
 
 # Function to install a dump in place
@@ -109,17 +157,23 @@ function install_dump {
   year="$4"
   name="${type}-${date}.osm.${format}"
   latest="${type}-latest.osm.${format}"
+  rss_file="${type}-${format}-rss.xml"
 
   md5sum "${name}" > "${name}.md5"
   mkdir -p "${dir}/${year}"
-  test -f "${name}.torrent" && mv "${name}.torrent" "${dir}/${year}"
   mv "${name}" "${name}.md5" "${dir}/${year}"
   ln -sf "${year:-.}/${name}" "${dir}/${latest}"
+  test -f "${name}.torrent" && mv "${name}.torrent" "${dir}/${year}" && ln -sf "${year:-.}/${name}.torrent" "${dir}/${latest}.torrent"
+  test -f "${rss_file}" && xmllint --noout "${rss_file}" && cp -f "${rss_file}" "${dir}/${year}"
   rm -f "${dir}/${latest}.md5"
   sed -e "s/${name}/${latest}/" "${dir}/${year}/${name}.md5" > "${dir}/${latest}.md5"
 }
 
 # Create *.torrent files
+mk_torrent "changesets" "bz2" "planet/${year}"
+mk_torrent "discussions" "bz2" "planet/${year}"
+mk_torrent "planet" "bz2" "planet/${year}"
+mk_torrent "history" "bz2" "planet/full-history/${year}"
 mk_torrent "planet" "pbf" "pbf"
 mk_torrent "history" "pbf" "pbf/full-history"
 
@@ -132,4 +186,12 @@ install_dump "planet" "pbf" "<%= node[:planet][:dump][:pbf_directory] %>"
 install_dump "history" "pbf" "<%= node[:planet][:dump][:pbf_history_directory] %>"
 
 # Remove pbf dumps older than 90 days
-find "<%= node[:planet][:dump][:pbf_directory] %>" "<%= node[:planet][:dump][:pbf_history_directory] %>" -maxdepth 1 -mindepth 1 -type f -mtime +90 \( -iname 'planet-*.pbf' -o -iname 'history-*.pbf' -o -iname 'planet-*.pbf.md5' -o -iname 'history-*.pbf.md5' -o -iname 'planet-*.pbf.torrent' -o -iname 'history-*.pbf.torrent' \) -delete
+find "<%= node[:planet][:dump][:pbf_directory] %>" "<%= node[:planet][:dump][:pbf_history_directory] %>" \
+     -maxdepth 1 -mindepth 1 -type f -mtime +90 \
+     \( \
+     -iname 'planet-*.pbf' \
+     -o -iname 'history-*.pbf' \
+     -o -iname 'planet-*.pbf.md5' \
+     -o -iname 'history-*.pbf.md5' \
+     \) \
+     -delete