fix: specify encoding to utf-8 in read_text() and write_text() (#524)

src/so_vits_svc_fork/dataset.py

@@ -19,7 +19,7 @@ def __init__(self, hps: HParams, is_validation: bool = False):
             for x in Path(
                 hps.data.validation_files if is_validation else hps.data.training_files
             )
-            .read_text()
+            .read_text("utf-8")
             .splitlines()
         ]
         self.hps = hps

src/so_vits_svc_fork/gui.py

@@ -29,9 +29,11 @@ def play_audio(path: Path | str):
 
 
 def load_presets() -> dict:
-    defaults = json.loads(GUI_DEFAULT_PRESETS_PATH.read_text())
+    defaults = json.loads(GUI_DEFAULT_PRESETS_PATH.read_text("utf-8"))
     users = (
-        json.loads(GUI_PRESETS_PATH.read_text()) if GUI_PRESETS_PATH.exists() else {}
+        json.loads(GUI_PRESETS_PATH.read_text("utf-8"))
+        if GUI_PRESETS_PATH.exists()
+        else {}
     )
     # prioriy: defaults > users
     # order: defaults -> users

src/so_vits_svc_fork/preprocessing/preprocess_flist_config.py

@@ -52,32 +52,28 @@ def preprocess_config(
 
     LOG.info(f"Writing {train_list_path}")
     train_list_path.parent.mkdir(parents=True, exist_ok=True)
-    with train_list_path.open("w", encoding="utf-8") as f:
-        for fname in train:
-            wavpath = fname.as_posix()
-            f.write(wavpath + "\n")
+    train_list_path.write_text(
+        "\n".join([x.as_posix() for x in train]), encoding="utf-8"
+    )
 
     LOG.info(f"Writing {val_list_path}")
     val_list_path.parent.mkdir(parents=True, exist_ok=True)
-    with val_list_path.open("w", encoding="utf-8") as f:
-        for fname in val:
-            wavpath = fname.as_posix()
-            f.write(wavpath + "\n")
+    val_list_path.write_text("\n".join([x.as_posix() for x in val]), encoding="utf-8")
 
     LOG.info(f"Writing {test_list_path}")
     test_list_path.parent.mkdir(parents=True, exist_ok=True)
-    with test_list_path.open("w", encoding="utf-8") as f:
-        for fname in test:
-            wavpath = fname.as_posix()
-            f.write(wavpath + "\n")
+    test_list_path.write_text("\n".join([x.as_posix() for x in test]), encoding="utf-8")
 
     config = deepcopy(
         json.loads(
             (
-                CONFIG_TEMPLATE_DIR / f"{config_name}.json"
-                if not config_name.endswith(".json")
-                else config_name
-            ).read_text()
+                CONFIG_TEMPLATE_DIR
+                / (
+                    config_name
+                    if config_name.endswith(".json")
+                    else config_name + ".json"
+                )
+            ).read_text(encoding="utf-8")
         )
     )
     config["spk"] = spk_dict

src/so_vits_svc_fork/utils.py

@@ -377,7 +377,7 @@ def get_backup_hparams(
 
 
 def get_hparams(config_path: Path | str) -> HParams:
-    config = json.loads(Path(config_path).read_text())
+    config = json.loads(Path(config_path).read_text("utf-8"))
     hparams = HParams(**config)
     return hparams
 

tests/test_main.py

@@ -63,7 +63,7 @@ def test_train(self):
         from so_vits_svc_fork.train import train
 
         config_path = Path("tests/logs/44k/config.json")
-        config_json = json.loads(config_path.read_text())
+        config_json = json.loads(config_path.read_text("utf-8"))
         config_json["train"]["epochs"] = 1
-        config_path.write_text(json.dumps(config_json))
+        config_path.write_text(json.dumps(config_json), "utf-8")
         train(config_path, "tests/logs/44k")