diff --git a/tetras_extraction/macao_schema.ttl b/tetras_extraction/macao_schema.ttl
index dd1d3d8dd4a35536d35d779043f97ca7b75dcfe3..db69207ecafeace6034ce0dcf9cbd729d5dd4c19 100644
--- a/tetras_extraction/macao_schema.ttl
+++ b/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 ;
@@ -89,21 +93,51 @@
### http://www.semanticweb.org/eliott/ontologies/2024/4/macao/commentaireInfo
:commentaireInfo rdf:type owl:DatatypeProperty ;
- rdfs:domain :Activite ;
- rdfs:range rdf:XMLLiteral .
+ 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:range rdf:XMLLiteral .
+ 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 ;
- rdfs:range rdf:XMLLiteral .
+ rdfs:domain :Activite ,
+ :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
@@ -111,6 +145,10 @@
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
diff --git a/tetras_extraction/script/requirements.txt b/tetras_extraction/script/requirements.txt
index 7fbd5744524cc77f264ebd200d6ddb9518d811f3..f8f98c309a84e3274a3ac920658ed90f254bff3b 100644
--- a/tetras_extraction/script/requirements.txt
+++ b/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
diff --git a/tetras_extraction/script/src/transform.py b/tetras_extraction/script/src/transform.py
index 01e4900aa1848c86fb4c500651fe4dcba6bc0951..672a4262c7b041a8bb81b6f42a1370e32c6c5b4e 100644
--- a/tetras_extraction/script/src/transform.py
+++ b/tetras_extraction/script/src/transform.py
@@ -1,4 +1,6 @@
-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...")
diff --git a/tetras_extraction/script/templates/activite.rq b/tetras_extraction/script/templates/activite.rq
index 485db41652eb5985bd95dbb0004709a7d753bc3b..440a4387a7b1c73a6b8602b66ff45569e4c84749 100644
--- a/tetras_extraction/script/templates/activite.rq
+++ b/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
diff --git a/tetras_extraction/script/templates/qcu.rq b/tetras_extraction/script/templates/qcu.rq
index e55e895f2481983d6fe845bbb667420ac8b1155c..c6fa414eed84df23c779e1c89707608e937e8148 100644
--- a/tetras_extraction/script/templates/qcu.rq
+++ b/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 .
diff --git a/tetras_extraction/script/templates/quiz.rq b/tetras_extraction/script/templates/quiz.rq
index 95f4af238877b0c023aa8e5bd47b2825db4aa965..76b5aba7291fbec410c6ddb5c110dec55ed63395 100644
--- a/tetras_extraction/script/templates/quiz.rq
+++ b/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,