Comments are now added to RDF graph

Basilisk/MACAO/macao_12/extracted/extract.py

@@ -116,7 +116,7 @@ def parse_manifest_rec(
 
 def main():
     g = create_graph()
-    extract_page.parse_page(SOURCE_DIR+"/contenu/pages/pg60.html", "pg60")
+    extract_page.parse_page(g, SOURCE_DIR + "/contenu/pages/pg112.html", "pg112")
     parse_manifest(g)
     export_graph(g)
 

Basilisk/MACAO/macao_12/extracted/extract_page.py

@@ -3,6 +3,7 @@ from pprint import pprint
 from typing import List
 
 from lxml import html
+from rdflib import Graph, Literal
 
 from constants import *
 
@@ -37,7 +38,7 @@ class Page:
 regex_comment = re.compile(r"(\D*)(\d*)")
 
 
-def parse_page(filepath: str, id: str):
+def parse_page(graph: Graph, filepath: str, id: str):
     page = Page()
     # Parse with lxml
     tree = html.parse(filepath)
@@ -58,13 +59,16 @@ def parse_page(filepath: str, id: str):
                     print(f"Comment, num={num}")
                     comment.num = int(num)
                     page.comments_misc.append(comment)
+                    graph.add((NS[id], NS["commentaireInfo"], Literal(comment.text)))
                 case ["divSugg", num]:
                     print(f"Suggestion, num={num}")
                     comment.num = int(num)
                     page.comments_sugg.append(comment)
+                    graph.add((NS[id], NS["commentaireSugg"], Literal(comment.text)))
                 case ["divCmtSucces", _]:
                     print(f"Succès")
                     page.comment_success = comment
+                    graph.add((NS[id], NS["commentaireSucces"], Literal(comment.text)))
                 case [other, _]:
                     print(f"other: {other}")
     pprint(page)