1st attempt : clean up audio <script> tags

tetras_extraction/script/src/transform.py

 from rdflib import OWL, RDF, Graph
+from lxml import html
 
 from common import *
 
@@ -29,6 +30,26 @@ def construct_while(g: Graph, query: str):
         pass
 
 
+# 1st attempt : clean up audio <script> tags
+def transform_html(graph: Graph):
+    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
+
+
+
 def main():
     # Load graph
     graph = Graph()
@@ -36,6 +57,8 @@ 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...")