Merge branch 'daxid_html2md' into 'main'

Daxid html2md

See merge request !3

tetras_extraction/macao_schema.ttl

@@ -43,6 +43,13 @@
           rdfs:range :MacaoObject .
 
 
+###  http://www.semanticweb.org/eliott/ontologies/2024/4/macao/contientActivite
+:contientActivite rdf:type owl:ObjectProperty ;
+                  rdfs:subPropertyOf :contient ;
+                  rdfs:domain :SousPartie ;
+                  rdfs:range :Activite .
+
+
 ###  http://www.semanticweb.org/eliott/ontologies/2024/4/macao/contientModule
 :contientModule rdf:type owl:ObjectProperty ;
                 rdfs:subPropertyOf :contient ;
@@ -50,13 +57,6 @@
                 rdfs:range :Module .
 
 
-###  http://www.semanticweb.org/eliott/ontologies/2024/4/macao/contientPage
-:contientActivite rdf:type owl:ObjectProperty ;
-              rdfs:subPropertyOf :contient ;
-              rdfs:domain :SousPartie ;
-              rdfs:range :Activite .
-
-
 ###  http://www.semanticweb.org/eliott/ontologies/2024/4/macao/contientSousPartie
 :contientSousPartie rdf:type owl:ObjectProperty ;
                     rdfs:subPropertyOf :contientModule ;
@@ -81,6 +81,10 @@
 #    Data properties
 #################################################################
 
+###  http://www.semanticweb.org/eliott/ontologies/2024/4/macao#html_md
+:html_md rdf:type owl:DatatypeProperty .
+
+
 ###  http://www.semanticweb.org/eliott/ontologies/2024/4/macao/cheminFichier
 :cheminFichier rdf:type owl:DatatypeProperty ;
                rdfs:domain :MacaoRessource ;
@@ -89,28 +93,62 @@
 
 ###  http://www.semanticweb.org/eliott/ontologies/2024/4/macao/commentaireInfo
 :commentaireInfo rdf:type owl:DatatypeProperty ;
-                 rdfs:domain :Activite ;
+                 rdfs:domain :Activite .
+
+
+###  http://www.semanticweb.org/eliott/ontologies/2024/4/macao/commentaireInfo_html
+:commentaireInfo_html rdf:type owl:DatatypeProperty ;
+                      rdfs:subPropertyOf :commentaireInfo ;
                       rdfs:range rdf:XMLLiteral .
 
 
+###  http://www.semanticweb.org/eliott/ontologies/2024/4/macao/commentaireInfo_md
+:commentaireInfo_md rdf:type owl:DatatypeProperty ;
+                    rdfs:subPropertyOf :commentaireInfo .
+
+
 ###  http://www.semanticweb.org/eliott/ontologies/2024/4/macao/commentaireSucces
 :commentaireSucces rdf:type owl:DatatypeProperty ;
-                   rdfs:domain :Activite ;
+                   rdfs:domain :Activite .
+
+
+###  http://www.semanticweb.org/eliott/ontologies/2024/4/macao/commentaireSucces_html
+:commentaireSucces_html rdf:type owl:DatatypeProperty ;
+                        rdfs:subPropertyOf :commentaireSucces ;
                         rdfs:range rdf:XMLLiteral .
 
 
+###  http://www.semanticweb.org/eliott/ontologies/2024/4/macao/commentaireSucces_md
+:commentaireSucces_md rdf:type owl:DatatypeProperty ;
+                      rdfs:subPropertyOf :commentaireSucces .
+
+
 ###  http://www.semanticweb.org/eliott/ontologies/2024/4/macao/commentaireSugg
 :commentaireSugg rdf:type owl:DatatypeProperty ;
                  rdfs:domain :Activite ,
-                             :Reponse ;
+                             :Reponse .
+
+
+###  http://www.semanticweb.org/eliott/ontologies/2024/4/macao/commentaireSugg_html
+:commentaireSugg_html rdf:type owl:DatatypeProperty ;
+                      rdfs:subPropertyOf :commentaireSugg ;
                       rdfs:range rdf:XMLLiteral .
 
 
+###  http://www.semanticweb.org/eliott/ontologies/2024/4/macao/commentaireSugg_md
+:commentaireSugg_md rdf:type owl:DatatypeProperty ;
+                    rdfs:subPropertyOf :commentaireSugg .
+
+
 ###  http://www.semanticweb.org/eliott/ontologies/2024/4/macao/correct
 :correct rdf:type owl:DatatypeProperty ;
          rdfs:range xsd:boolean .
 
 
+###  http://www.semanticweb.org/eliott/ontologies/2024/4/macao/description_md
+:description_md rdf:type owl:DatatypeProperty .
+
+
 ###  http://www.semanticweb.org/eliott/ontologies/2024/4/macao/html
 :html rdf:type owl:DatatypeProperty ;
       rdfs:range rdf:XMLLiteral .
@@ -135,6 +173,11 @@
 #    Classes
 #################################################################
 
+###  http://www.semanticweb.org/eliott/ontologies/2024/4/macao/Activite
+:Activite rdf:type owl:Class ;
+          rdfs:subClassOf :MacaoContenu .
+
+
 ###  http://www.semanticweb.org/eliott/ontologies/2024/4/macao/Cours
 :Cours rdf:type owl:Class ;
        rdfs:subClassOf :Activite .
@@ -208,11 +251,6 @@
         rdfs:subClassOf :MacaoContenu .
 
 
-###  http://www.semanticweb.org/eliott/ontologies/2024/4/macao/Page
-:Activite rdf:type owl:Class ;
-      rdfs:subClassOf :MacaoContenu .
-
-
 ###  http://www.semanticweb.org/eliott/ontologies/2024/4/macao/Reponse
 :Reponse rdf:type owl:Class ;
          rdfs:subClassOf :MacaoContenu .
@@ -241,4 +279,4 @@
 ] .
 
 
-###  Generated by the OWL API (version 4.5.26.2023-07-17T20:34:13Z) https://github.com/owlcs/owlapi
+###  Generated by the OWL API (version 4.5.29.2024-05-13T12:11:03Z) https://github.com/owlcs/owlapi

tetras_extraction/script/requirements.txt

@@ -10,3 +10,4 @@ types-beautifulsoup4==4.12.0.20240511
 types-html5lib==1.1.11.20240228
 types-lxml==2024.4.14
 typing_extensions==4.12.1
+pandoc

tetras_extraction/script/src/transform.py

-from rdflib import OWL, RDF, Graph
+from rdflib import OWL, RDF, Graph, Literal
+from lxml import html
+import pandoc
 
 from common import *
 
@@ -29,6 +31,41 @@ def construct_while(g: Graph, query: str):
         pass
 
 
+def transform_html(graph: Graph):
+    html_properties = ['commentaireInfo', 'commentaireSucces', 'commentaireSugg', 'html', 'description']
+
+
+    # 1st attempt : clean up audio <script> tags
+    """
+    for t in graph.triples((None, NS['description'], None)):
+        desc_str = t[2]
+        tree = html.fragment_fromstring(desc_str)
+        for script in tree.findall(".//script"):
+            # `HtmlElement.drop_tree()` removes an element along with its
+            # children and text, however it has an interesting feature :
+            # the tail text is not removed, but instead joined to the previous
+            # sibling or parent automatically.
+            # This means that when we want to replace an element with string,
+            # we only need to prepend the string to the tail, and it will be
+            # inserted in the right place
+            script.tail = script.text + script.tail
+            script.drop_tree()
+        pass
+    """
+
+    # Process all html content through Pandoc
+    for prop in html_properties:
+        for t in graph.triples((None, NS[prop], None))  :
+            desc_str = t[2]
+            desc_doc = pandoc.read(desc_str, format="html")
+            desc_md = pandoc.write(desc_doc, format="markdown")
+            l = list(t)
+            l[2] = Literal(desc_md)
+            l[1] = NS[prop+'_md']
+            graph.add(tuple(l))
+
+
+
 def main():
     # Load graph
     graph = Graph()
@@ -36,6 +73,8 @@ def main():
     graph.parse(RDF_SCHEMA_FILE)
     graph.parse(RDF_CONTENT_FILE)
 
+    transform_html(graph)
+
     # Apply property 'subClassOf' transitively, except on the "fake" class
     # hierarchy based on MacaoRoot
     log.info("Adding transitive subclasses...")

tetras_extraction/script/templates/activite.rq

@@ -23,7 +23,7 @@ where {
     ?act a :Activite .
     ?act :id ?id .
     ?act :titre ?title .
-    ?act :description ?desc .
+    ?act :description_md ?desc .
     optional {
         # Turn the page index into a Hugo weight: increment and add a zero, to
         # leave room for adding new pages in-between later

tetras_extraction/script/templates/qcu.rq

@@ -31,7 +31,7 @@ where {
     optional {
         ?qcu :aReponse ?choice .
         ?choice :index ?choice_index .
-        ?choice :html ?choice_html
+        ?choice :html_md ?choice_html
     }
     optional {
         ?qcu :aReponse ?correct_choice .

tetras_extraction/script/templates/quiz.rq

@@ -39,10 +39,10 @@ where {
     ?quiz a :Exercice .
     ?quiz :id ?id .
     optional {
-        ?quiz :commentaireSucces ?correct_comment .
+        ?quiz :commentaireSucces_md ?correct_comment .
     }
     optional {
-        ?quiz :commentaireSugg ?incorrect_comment .
+        ?quiz :commentaireSugg_md ?incorrect_comment .
     }
 }
 # FIXME: some exercises have multiple incorrect comments, almost identical,