Merge pull request #227 from dracor-org/225-sitelinks-update

Update Wikipedia link count on play ingest

modules/api.xqm

@@ -422,6 +422,7 @@ function api:corpora-post-json($data) {
       util:log-system-out($corpus),
       xmldb:create-collection($config:data-root, $name),
       xmldb:create-collection($config:metrics-root, $name),
+      xmldb:create-collection($config:sitelinks-root, $name),
       xmldb:create-collection($config:rdf-root, $name),
       xmldb:store($tei-dir, "corpus.xml", $corpus),
       $json
@@ -626,6 +627,7 @@ function api:delete-corpus($corpusname, $auth) {
         (
           xmldb:remove($config:data-root || "/" || $corpusname),
           xmldb:remove($config:metrics-root || "/" || $corpusname),
+          xmldb:remove($config:sitelinks-root || "/" || $corpusname),
           xmldb:remove($config:rdf-root || "/" || $corpusname),
           map {
             "message": "corpus deleted",

modules/metrics.xqm

@@ -15,25 +15,54 @@ declare namespace trigger = "http://exist-db.org/xquery/trigger";
 declare namespace output = "http://www.w3.org/2010/xslt-xquery-serialization";
 declare namespace tei = "http://www.tei-c.org/ns/1.0";
 
+(:~
+ : Query sitelinks for given Wikidata ID and store them to the sitelinks
+ : collection.
+ :
+ : @param $id Wikidata ID
+ : @param $corpus Corpus name
+:)
+declare function metrics:update-sitelinks(
+  $id as xs:string,
+  $corpus as xs:string
+) {
+  if ($id) then
+    let $resource := $id || '.xml'
+    let $collection := $config:sitelinks-root || '/' || $corpus
+    let $log := util:log-system-out('querying sitelinks for ' || $resource)
+    let $sitelinks := <sitelinks id="{$id}" updated="{current-dateTime()}">{
+      for $uri in wd:get-sitelinks($id)
+      return <uri>{$uri}</uri>
+    }</sitelinks>
+    return (
+      $sitelinks,
+      xmldb:store($collection, $resource, $sitelinks)
+    )
+  else ()
+};
+
+(:~
+ : Update sitelinks for Wikidata ID of play with given url.
+ :
+ : @param $url Path to TEI file
+:)
+declare function metrics:update-sitelinks($url as xs:string) {
+  let $p := dutil:filepaths($url)
+  let $doc:= dutil:get-doc($p?corpusname, $p?playname)
+  let $id := dutil:get-play-wikidata-id($doc/tei:TEI)
+  return metrics:update-sitelinks($id, $p?corpusname)
+};
+
 (:~
  : Collect sitelinks for each play in a given corpus from wikidata and store
  : them to the sitelinks collection
  :
  : @param $corpus Corpus name
 :)
 declare function metrics:collect-sitelinks($corpus as xs:string) {
-  let $log := util:log('info', 'collecting sitelinks for corpus ' || $corpus)
-  let $sitelinks-col := xmldb:create-collection(
-    "/", $config:sitelinks-root || '/' || $corpus
-  )
+  util:log-system-out('collecting sitelinks for corpus ' || $corpus),
   for $id in dutil:get-play-wikidata-ids($corpus)
-  let $resource := $id || '.xml'
-  let $log := util:log('info', 'querying sitelinks for ' || $resource)
-  let $sitelinks := <sitelinks id="{$id}" updated="{current-dateTime()}">{
-    for $uri in wd:get-sitelinks($id)
-    return <uri>{$uri}</uri>
-  }</sitelinks>
-  return xmldb:store($sitelinks-col, $resource, $sitelinks)
+  return metrics:update-sitelinks($id, $corpus)
 };
 
 (:~

modules/trigger.xqm

@@ -16,6 +16,7 @@ declare function trigger:after-create-document($url as xs:anyURI) {
     (
       util:log-system-out("running CREATION TRIGGER for " || $url),
       metrics:update($url),
+      metrics:update-sitelinks($url),
       drdf:update($url)
     )
   else (
@@ -28,6 +29,7 @@ declare function trigger:after-update-document($url as xs:anyURI) {
     (
       util:log-system-out("running UPDATE TRIGGER for " || $url),
       metrics:update($url),
+      metrics:update-sitelinks($url),
       drdf:update($url)
     )
   else (
@@ -38,7 +40,9 @@ declare function trigger:after-update-document($url as xs:anyURI) {
 declare function trigger:before-delete-document($url as xs:anyURI) {
   if (doc($url)/tei:TEI) then
     let $paths := dutil:filepaths($url)
+    let $id := dutil:get-play-wikidata-id(doc($url)/tei:TEI)
     return try {
+      if ($id) then xmldb:remove($paths?collections?sitelinks, $id || '.xml') else (),
       xmldb:remove($paths?collections?metrics, $paths?filename),
       xmldb:remove($paths?collections?rdf, $paths?playname || ".rdf.xml")
     } catch * {

modules/util.xqm

@@ -30,6 +30,7 @@ declare function dutil:filepaths ($url as xs:string) as map() {
     "collections": map {
       "tei": $config:data-root || "/" || $corpusname,
       "metrics": $config:metrics-root || "/" || $corpusname,
+      "sitelinks": $config:sitelinks-root || "/" || $corpusname,
       "rdf": $config:rdf-root || "/" || $corpusname
     },
     "files": map {