Fix yaml loader (#963)

Author: mmcauliffe

montreal_forced_aligner/abc.py

@@ -1077,7 +1077,7 @@ def __init__(
 
     def load_mapping(self) -> None:
         with mfa_open(self.phone_mapping_path, "r") as f:
-            self.phone_remapping = yaml.load(f, Loader=yaml.Loader)
+            self.phone_remapping = yaml.load(f, Loader=yaml.SafeLoader)
         for key, values in self.phone_remapping.items():
             if not isinstance(values, list):
                 self.phone_remapping[key] = [values]

montreal_forced_aligner/command_line/utils.py

@@ -266,7 +266,7 @@ def validate_model_arg(name: str, model_type: str) -> MfaModel:
             raise click.BadParameter(str(FileArgumentNotFoundError(name)))
         if model_type == "dictionary" and name.suffix.lower() == ".yaml":
             with mfa_open(name, "r") as f:
-                data = yaml.load(f, Loader=yaml.Loader)
+                data = yaml.load(f, Loader=yaml.SafeLoader)
                 paths = sorted(set(data.values()))
                 for path in paths:
                     validate_model_arg(path, "dictionary")

montreal_forced_aligner/config.py

@@ -105,7 +105,7 @@ def load_command_history() -> List[Dict[str, Any]]:
     history = []
     if path.exists():
         with mfa_open(path, "r") as f:
-            history = yaml.load(f, Loader=yaml.Loader)
+            history = yaml.load(f, Loader=yaml.SafeLoader)
             if not history:
                 history = []
     history = [h for h in history if h["command"]]
@@ -267,7 +267,7 @@ def save(self) -> None:
     def load(self) -> None:
         """Load MFA configuration"""
         with mfa_open(self.config_path, "r") as f:
-            data = yaml.load(f, Loader=yaml.Loader)
+            data = yaml.load(f, Loader=yaml.SafeLoader)
         for k, v in data.items():
             if any(k.endswith(x) for x in ["_path", "_directory", "_dir"]):
                 data[k] = pathlib.Path(v)
@@ -288,7 +288,7 @@ def load_configuration():
     if not config_path.exists():
         return
     with mfa_open(config_path, "r") as f:
-        data = yaml.load(f, Loader=yaml.Loader)
+        data = yaml.load(f, Loader=yaml.SafeLoader)
     profiles = data.pop("profiles", {})
     if CURRENT_PROFILE_NAME == "global":
         update_configuration(data)

montreal_forced_aligner/corpus/remapper.py

@@ -58,7 +58,7 @@ def setup(self) -> None:
 
     def load_mapping(self) -> None:
         with mfa_open(self.phone_mapping_path, "r") as f:
-            data = yaml.load(f, Loader=yaml.Loader)
+            data = yaml.load(f, Loader=yaml.SafeLoader)
         for key, value in data.items():
             if isinstance(value, list):
                 value = value[0]

montreal_forced_aligner/dictionary/multispeaker.py

@@ -166,7 +166,7 @@ def load_phone_groups(self) -> None:
         """
         if self.phone_groups_path is not None and self.phone_groups_path.exists():
             with mfa_open(self.phone_groups_path) as f:
-                self._phone_groups = yaml.load(f, Loader=yaml.Loader)
+                self._phone_groups = yaml.load(f, Loader=yaml.SafeLoader)
                 if isinstance(self._phone_groups, list):
                     self._phone_groups = {k: v for k, v in enumerate(self._phone_groups)}
                 for k, v in self._phone_groups.items():
@@ -214,7 +214,7 @@ def load_phone_topologies(self) -> None:
             with mfa_open(self.topology_path) as f:
                 self._topologies = {
                     k: v
-                    for k, v in yaml.load(f, Loader=yaml.Loader).items()
+                    for k, v in yaml.load(f, Loader=yaml.SafeLoader).items()
                     if k in self.non_silence_phones
                 }
             found_phones = set(self._topologies.keys())
@@ -932,7 +932,7 @@ def load_phonological_rules(self) -> None:
         if not self.rules_path or not self.rules_path.exists():
             return
         with mfa_open(self.rules_path) as f:
-            rule_data = yaml.load(f, Loader=yaml.Loader)
+            rule_data = yaml.load(f, Loader=yaml.SafeLoader)
         with self.session() as session:
             if session.query(PhonologicalRule).first() is not None:
                 return

montreal_forced_aligner/helper.py

@@ -126,7 +126,7 @@ def load_configuration(config_path: typing.Union[str, Path]) -> typing.Dict[str,
         config_path = Path(config_path)
     with mfa_open(config_path, "r") as f:
         if config_path.suffix == ".yaml":
-            data = yaml.load(f, Loader=yaml.Loader)
+            data = yaml.load(f, Loader=yaml.SafeLoader)
         elif config_path.suffix == ".json":
             data = json.load(f)
     if not data:
@@ -544,7 +544,7 @@ def default(self, o: typing.Any) -> typing.Any:
 
 def load_evaluation_mapping(custom_mapping_path):
     with mfa_open(custom_mapping_path, "r") as f:
-        mapping = yaml.load(f, Loader=yaml.Loader)
+        mapping = yaml.load(f, Loader=yaml.SafeLoader)
     for k, v in mapping.items():
         if isinstance(v, str):
             mapping[k] = {v}

montreal_forced_aligner/models/classes.py

@@ -666,7 +666,7 @@ def meta(self) -> dict:
                 file_format = "yaml"
             with mfa_open(meta_path, "r") as f:
                 if file_format == "yaml":
-                    self._meta = yaml.load(f, Loader=yaml.Loader)
+                    self._meta = yaml.load(f, Loader=yaml.SafeLoader)
                 else:
                     self._meta = json.load(f)
         return self._meta
@@ -1159,7 +1159,7 @@ def meta(self) -> dict:
                     if format == "json":
                         self._meta = json.load(f)
                     else:
-                        self._meta = yaml.load(f, Loader=yaml.Loader)
+                        self._meta = yaml.load(f, Loader=yaml.SafeLoader)
             self._meta["phones"] = set(self._meta.get("phones", []))
             self._meta["graphemes"] = set(self._meta.get("graphemes", []))
             self._meta["evaluation"] = self._meta.get("evaluation", [])
@@ -1329,7 +1329,7 @@ def meta(self) -> dict:
                     if format == "json":
                         self._meta = json.load(f)
                     else:
-                        self._meta = yaml.load(f, Loader=yaml.Loader)
+                        self._meta = yaml.load(f, Loader=yaml.SafeLoader)
             self._meta["evaluation"] = self._meta.get("evaluation", [])
             self._meta["training"] = self._meta.get("training", [])
         return self._meta
@@ -1751,7 +1751,7 @@ def load_dictionary_paths(
         mapping = {}
         if self.is_multiple:
             with mfa_open(self.path, "r") as f:
-                data = yaml.load(f, Loader=yaml.Loader)
+                data = yaml.load(f, Loader=yaml.SafeLoader)
                 for speaker, path in data.items():
                     if path not in mapping:
                         if path != "nonnative":