diff --git a/tetras_extraction/script/src/transform.py b/tetras_extraction/script/src/transform.py index aad92eb65d3158ab107800ca97104f5a255bff4d..10a43634dd23de4ba825534f320dead7dbc25c8e 100644 --- a/tetras_extraction/script/src/transform.py +++ b/tetras_extraction/script/src/transform.py @@ -98,6 +98,10 @@ def transform_html(graph: Graph): # l[1] = NS[prop+'_md'] # graph.add(tuple(l)) + # Add mannually edited surveyjs_json to the main graph + #for t in graph.triples((None, surveyjs_json, None)) : + + # Process all html content through Markitdown for prop in html_properties: for t in graph.triples((None, NS[prop], None)) : @@ -110,11 +114,15 @@ def transform_html(graph: Graph): l = list(t) l[2] = Literal(desc_md) l[1] = NS[prop+'_md'] + ################################################ + # Add manually eddited content to the main graph manual_edition_list = list(graph.triples((l[0], NS[prop+"_md_manual_edition"], None))) if len(manual_edition_list)>=1: t_manual_edition = manual_edition_list[0] l[2] = t_manual_edition[2] graph.add(tuple(l)) + ################################################# + def main():