New IDE debug configs for each step

tetras_extraction/macao_12/script/.vscode/launch.json

@@ -5,7 +5,28 @@
     "version": "0.2.0",
     "configurations": [
         {
-            "name": "Python Debugger: Current File",
+            "name": "Python: extract",
+            "type": "debugpy",
+            "request": "launch",
+            "program": "src/extract.py",
+            "console": "integratedTerminal"
+        },
+        {
+            "name": "Python: transform",
+            "type": "debugpy",
+            "request": "launch",
+            "program": "src/transform.py",
+            "console": "integratedTerminal"
+        },
+        {
+            "name": "Python: export",
+            "type": "debugpy",
+            "request": "launch",
+            "program": "src/export.py",
+            "console": "integratedTerminal"
+        },
+        {
+            "name": "Python: main",
             "type": "debugpy",
             "request": "launch",
             "program": "src/main.py",

tetras_extraction/macao_12/script/src/extract.py

 import filecmp
-from re import sub
 
 from lxml import etree
 from rdflib import RDFS, Graph, Literal, URIRef

tetras_extraction/macao_12/script/src/extract_mosetp.py

 import re
 import subprocess
-from os import path
 
 from rdflib import OWL, RDF, RDFS, Graph, Literal
 
@@ -46,7 +45,6 @@ def parse_mosetp(graph: Graph, filepath: str, id: str):
     :param filepath: path to the MosEtp file
     :param id: text identifier of the subsection
     """
-    filename = path.basename(filepath)
     # Prepare regex with capturing groups to match lines
     regex = re.compile(r'.*new PageContenu\("(.*)", "(.*)", "(.*)", ""\);')
     # The lines we need are fairly basic, grep is much faster

tetras_extraction/macao_12/script/src/transform.py

@@ -19,6 +19,7 @@ def construct_while(g: Graph, query: str):
     while True:
         res = g.query(query)
         if res.graph is not None and len(res.graph - g) > 0:
+            print(f"Constructed {len(res.graph - g)} triples")
             g += res.graph
         else:
             break
@@ -92,3 +93,7 @@ def main():
     graph.add((onto_uri, RDF.type, OWL.Ontology))
     graph.add((onto_uri, RDFS.label, Literal("macao-full")))
     graph.serialize(RDF_FULL_FILE, "turtle", base=NS)
+
+
+if __name__ == "__main__":
+    main()