Merge branch 'master' of gitlab.utc.fr:crozatst/hdoc

hdoc_to_basex/.gitignore

+!input/.gitkeep
+!output/.gitkeep
\ No newline at end of file

hdoc_to_basex/basex/xquery/searchDocByAuthor.hq

@@ -3,11 +3,10 @@
 (: For example, $author := '^Baptiste Montangé$', to search for an exact name:)
 (: For example, $name := 'Montangé', to search for documents whose author named Coutant:)
 (: Remark : Accents in the authors' names have been taken care of :)
-
-<documents>{
-  let $name := 'Montangé'
+declare function local:searchDocByAuthor($name as xs:string, $docs as node()*) as node()*
+{
   let $name_noAcc := translate($name, 'áàâäéèêëíìîïóòôöúùûü','aaaaeeeeiiiioooouuuu')
-  for $doc in //document
+  for $doc in $docs
   return
     for $author in $doc/authors/author
     let $titre := $doc/titre
@@ -15,4 +14,7 @@
     where matches($author_noAcc, $name_noAcc)
     group by $titre (: Avoid duplications of documents by their titre:)
     return $doc
-}</documents>
\ No newline at end of file
+};
+
+(: Example :)
+local:searchDocByAuthor('Montangé', //document)
\ No newline at end of file

hdoc_to_basex/basex/xquery/searchDocByTitle.hq

@@ -3,9 +3,12 @@
 (: For example, $name := '^NF29_HdocEtherpad$', to search for an exact name:)
 (: For example, $name := 'NF29', to search for documents whose name contains 'NF29':)
 
-<documents>{
-  let $name := '^NF29_HdocEtherpad$'
-  for $doc in //document
+declare function local:searchDocByTitle($name as xs:string, $docs as node()*) as node()*
+{
+  for $doc in $docs
   where matches($doc/titre, $name, "i")
   return $doc
-}</documents>
\ No newline at end of file
+};
+
+(: Example :)
+local:searchDocByTitle('^NF29_HdocEtherpad$', //document)
\ No newline at end of file

wikipedia_to_hdoc/xslt/prepare_wiki_to_hdoc.xsl

@@ -37,5 +37,5 @@
     <xsl:template match="script"/>
 
     <!-- Ignoring divs that are not useful and that might interfere with the true xslt transformation -->
-    <xsl:template match="/html/body/div/div/div/div[@class!='mw-highlight mw-content-ltr']"/>
+    <xsl:template match="/html/body/div/div/div/div[@class!='mw-highlight mw-content-ltr' and not(descendant::table)]"/>
 </xsl:stylesheet>
\ No newline at end of file