chore: refactor repo; rename main.py into strings2things; modularize

Author: marftn

justfile

@@ -57,7 +57,7 @@ test *args:
 
 # Run an executable.
 run *args:
-    uv run cli "$@"
+    uv run --env-file=.env cli "$@"
 
 # Run the Jupyter notebook.
 notebook *args:

pyproject.toml

@@ -1,5 +1,5 @@
 [project]
-name = "main.py"
+name = "strings2things"
 authors = [
   { name = "Robin Franken", email = "robin.franken@epfl.ch" },
   { name = "Martin Fontanet", email = "martin.fontanet@epfl.ch" },
@@ -22,7 +22,7 @@ dev = [
 ]
 
 [project.scripts]
-cli = 'main.py.cli:main'
+cli = 'strings2things.cli:main'
 
 [build-system]
 requires = ["hatchling"]

src/main.py/__pycache__/format.cpython-312.pyc

@@ -0,0 +1,74 @@
+from strings2things.format import append_input_term
+from strings2things.sparql import enumeration_query, find_matches_query, find_predicate_query
+from pyfuzon.matcher import TermMatcher
+import os
+import json
+import rdflib
+
+MATCH_THRESHOLD = 0.8
+
+def main():
+    knowledge_graph_path = os.getenv("KNOWLEDGE_GRAPH_PATH")
+    ontologies_path = os.getenv("ONTOLOGIES_PATH")
+
+    onto = rdflib.Graph()
+    onto.parse(ontologies_path)
+
+    # SPARQLwrapper
+    dataset = rdflib.Dataset()
+
+    #FIXME: Is `data` used?
+    data = dataset.graph("https://imaging-plaza.epfl.ch/finalGraph")
+    data.parse(knowledge_graph_path)
+    # Load the knowledge graph
+
+    # todo filter down ontology to only get triples related to enumerations
+
+    # Filter down ontology to only get triples related to enumerations
+
+    enumeration_results = onto.query(enumeration_query)
+    # Create a new graph to store the enumeration triples
+    enumeration_graph = rdflib.Graph()
+
+
+    # Add the results of the CONSTRUCT query to the new graph
+    for triple in enumeration_results.graph:
+        enumeration_graph.add(triple)
+
+    #FIXME: Is `enum` used?
+    enum = dataset.graph("https://imaging-plaza.epfl.ch/ontology#enums")
+    enum.parse(data=enumeration_graph.serialize(format="turtle"), format="turtle")
+
+    results = dataset.query(find_matches_query)
+
+    # Create a new graph to store the constructed triples
+    constructed_graph = rdflib.Graph()
+
+    # Add the results of the CONSTRUCT query to the new graph
+    for triple in results.graph:
+        constructed_graph.add(triple)
+
+
+    matcher = TermMatcher.from_files([ontologies_path])
+
+    inputdict = {}
+    for term in constructed_graph.query(find_predicate_query):
+        searchterm = term[0]
+        predicate = term[1]
+        if sorted(matcher.score(searchterm), reverse=True)[0] / len(searchterm) > MATCH_THRESHOLD:
+            suggestedterm = matcher.top(searchterm, 1)[0]
+            print(suggestedterm.uri)
+            append_input_term(inputdict, str(searchterm), str(predicate), suggestedterm.uri)
+        else :
+            append_input_term(inputdict, str(searchterm), str(predicate), None)
+
+    json_input = json.dumps(inputdict)
+
+    print(json_input)
+
+# TODO: create enums list
+# TODO: call LLM
+
+
+if __name__ == "__main__":
+    main()