fix: remove logs in model_repository, add log to s3 storage, update v…

…ersion to 2.1.4 (#194)

h1st/model/repository/model_repository.py

@@ -410,7 +410,6 @@ def persist(self, model, version=None):
         # assert isinstance(model, Model)
         # TODO: use version format: v_20200714-1203
         version = version or str(ulid.new())
-        logger.info(f"Persisting model {model} version {version}")
 
         try:
             # serialize a model to a temporary folder and then clean up later
@@ -421,20 +420,17 @@ def persist(self, model, version=None):
 
             self._serder.serialize(model, serialized_dir)
             _tar_create(tar_file, serialized_dir)
-            logger.info(f'Created tar file {tar_file} for model {model} version {version}')
 
             with open(tar_file, mode="rb") as f:
                 self._storage.set_bytes(
                     self._get_key(model, version),
                     f.read(),
                 )
-                logger.info(f'Set bytes {self._get_key(model, version)} to storage from local {tar_file}')
 
                 self._storage.set_obj(
                     self._get_key(model, "latest"),
                     version,
                 )
-                logger.info(f'Set obj {self._get_key(model, "latest")} to storage from local {tar_file}')
 
                 model.version = version
         except Exception as e:

h1st/model/repository/storage/s3.py

@@ -2,14 +2,15 @@
 import cloudpickle
 import s3fs
 from h1st.model.repository.storage.base import Storage
+from loguru import logger
 
 
 class S3Storage(Storage):
     """
     Provide data storage on top of AWS S3
     """
 
-    def __init__(self, bucket_name: str = "", prefix: str = ""):
+    def __init__(self, bucket_name: str = None, prefix: str = None):
         """
         :param bucket_name: s3 bucket name to store data into
         :param prefix: s3 object prefix, leave blank to store at bucket root
@@ -25,6 +26,8 @@ def get_obj(self, name: str) -> Any:
         :param name: object name
         """
         key = self._to_key(name)
+        logger.info(f"---Loading obj {key} from S3")
+
         try:
             with self.fs.open(key, 'rb') as f:
                 return cloudpickle.load(f)
@@ -38,6 +41,8 @@ def get_bytes(self, name) -> bytes:
         :param name: object name
         """
         key = self._to_key(name)
+        logger.info(f"---Loading bytes {key} from S3")
+
         try:
             with self.fs.open(key, 'rb') as f:
                 return f.read()
@@ -52,6 +57,7 @@ def set_obj(self, name: str, value: Any) -> NoReturn:
         :param value: value in python object
         """
         key = self._to_key(name)
+        logger.info(f"---Saving obj {key} to S3, value {value}")
 
         with self.fs.open(key, 'wb') as f:
             return cloudpickle.dump(value, f)
@@ -64,6 +70,7 @@ def set_bytes(self, name: str, value: bytes) -> NoReturn:
         :param value: value in bytes
         """
         key = self._to_key(name)
+        logger.info(f"---Saving bytes {key} to S3, value {value}")
 
         with self.fs.open(key, 'wb') as f:
             f.write(value)

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "h1st"
-version = "2.1.3"
+version = "2.1.4"
 description = "Human-First AI (H1st)"
 authors = ["Aitomatic, Inc. <engineering@aitomatic.com>"]
 license = "Apache-2.0"