Fix training data handling (#58)

- fixed the issue reported in #56 
- improves unzipping if no to be compiled qasm files are present

Co-authored-by: Lukas Burgholzer <[email protected]>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

Author: 3 people

src/mqt/predictor/ml/Predictor.py

@@ -141,22 +141,22 @@ def generate_compiled_circuits(
         if target_path is None:
             target_path = str(ml.helper.get_path_training_circuits_compiled())
 
-        source_circuits_list = []
-
-        for file in Path(source_path).iterdir():
-            if "qasm" in str(file):
-                source_circuits_list.append(str(file))
-
         path_zip = Path(source_path) / "mqtbench_training_samples.zip"
-        if len(source_circuits_list) == 0 and path_zip.exists():
+        if (
+            not any(file.suffix == ".qasm" for file in Path(source_path).iterdir())
+            and path_zip.exists()
+        ):
             path_zip = str(path_zip)
             import zipfile
 
             with zipfile.ZipFile(path_zip, "r") as zip_ref:
                 zip_ref.extractall(source_path)
 
-        if not Path(source_path).is_dir():
-            Path(source_path).mkdir()
+        Path(target_path).mkdir(exist_ok=True)
+
+        source_circuits_list = [
+            file.name for file in Path(source_path).iterdir() if file.suffix == ".qasm"
+        ]
 
         Parallel(n_jobs=-1, verbose=100)(
             delayed(self.compile_all_circuits_for_qc)(
@@ -174,7 +174,7 @@ def generate_trainingdata_from_qasm_files(
 
         Keyword arguments:
         source_path -- path to file
-        target_directory -- path to directory for compiled circuit
+        target_path -- path to directory for compiled circuit
 
         Return values:
         training_data_ML_aggregated -- training data
@@ -194,7 +194,7 @@ def generate_trainingdata_from_qasm_files(
 
         results = Parallel(n_jobs=-1, verbose=100)(
             delayed(self.generate_training_sample)(
-                str(filename), source_path, target_path
+                str(filename.name), source_path, target_path
             )
             for filename in Path(source_path).iterdir()
         )

tests/ml/test_predictor_ml.py

@@ -81,6 +81,7 @@ def test_generate_compiled_circuits():
     qasm_path = Path("compiled_test.qasm")
     qc.qasm(filename=str(qasm_path))
     predictor.generate_compiled_circuits(source_path, str(target_path))
+    assert any(file.suffix == ".qasm" for file in target_path.iterdir())
 
     training_sample, circuit_name, scores = predictor.generate_training_sample(
         str(qasm_path), source_path, target_path