Merge pull request #387 from andreidenissov-cog/feature/386

Implement local filesystem option for studioml experiment storage/database.

docs/local_filesystem_setup.rst

@@ -0,0 +1,45 @@
+Setting up experiment storage and database in local filesystem
+==============================================================
+
+This page describes how to setup studioml to use
+local filesystem for storing experiment artifacts and meta-data.
+With this option, there is no need to setup any external
+connection to S3/Minio/GCS etc.
+
+StudioML configuration
+--------------------
+
+::
+
+      "studio_ml_config": {
+
+          ...
+
+          "database": {
+                    "type": "local",
+                    "endpoint": SOME_DB_LOCAL_PATH,
+                    "bucket": DB_BUCKET_NAME,
+                    "authentication": "none"
+          },
+          "storage": {
+                    "type": "local",
+                    "endpoint": SOME_ARTIFACTS_LOCAL_PATH,
+                    "bucket": ARTIFACTS_BUCKET_NAME,
+          }
+
+          ...
+      }
+
+
+With StudioML database type set to "local",
+all experiment meta-data will be stored locally under
+directory: SOME_DB_LOCAL_PATH/DB_BUCKET_NAME.
+Similarly, with storage type set to "local",
+all experiment artifacts will be stored locally under
+directory: SOME_ARTIFACTS_LOCAL_PATH/ARTIFACTS_BUCKET_NAME.
+
+Note: if you are using "local" mode, it is recommended to use it
+for both storage and database configuration.
+But it's technically possible to mix, for example, local storage configuration
+and S3-based database configuration etc.
+

studio/artifact_store.py

@@ -1,8 +1,8 @@
 from .firebase_artifact_store import FirebaseArtifactStore
 from .gcloud_artifact_store import GCloudArtifactStore
+from .local_artifact_store import LocalArtifactStore
 from .s3_artifact_store import S3ArtifactStore
 
-
 def get_artifact_store(config, blocking_auth=True, verbose=10):
     if config['type'].lower() == 'firebase':
         return FirebaseArtifactStore(
@@ -11,5 +11,7 @@ def get_artifact_store(config, blocking_auth=True, verbose=10):
         return GCloudArtifactStore(config, verbose=verbose)
     elif config['type'].lower() == 's3':
         return S3ArtifactStore(config, verbose=verbose)
+    elif config['type'].lower() == 'local':
+        return LocalArtifactStore(config, verbose=verbose)
     else:
         raise ValueError('Unknown storage type: ' + config['type'])

studio/local_artifact_store.py

@@ -0,0 +1,64 @@
+import calendar
+import os
+import shutil
+
+from .tartifact_store import TartifactStore
+
+class LocalArtifactStore(TartifactStore):
+    def __init__(self, config,
+                 bucket_name=None,
+                 verbose=10,
+                 measure_timestamp_diff=False,
+                 compression=None):
+
+        if compression is None:
+            compression = config.get('compression')
+
+        self.endpoint = config.get('endpoint', '~')
+        self.store_root = os.path.realpath(os.path.expanduser(self.endpoint))
+        if not os.path.exists(self.store_root) \
+            or not os.path.isdir(self.store_root):
+            raise ValueError()
+
+        self.bucket = bucket_name
+        if self.bucket is None:
+            self.bucket = config.get('bucket')
+        self.store_root = os.path.join(self.store_root, self.bucket)
+        self._ensure_path_dirs_exist(self.store_root)
+
+        super(LocalArtifactStore, self).__init__(
+            measure_timestamp_diff,
+            compression=compression,
+            verbose=verbose)
+
+    def _ensure_path_dirs_exist(self, path):
+        dirs = os.path.dirname(path)
+        os.makedirs(dirs, mode = 0o777, exist_ok = True)
+
+    def _upload_file(self, key, local_path):
+        target_path = os.path.join(self.store_root, key)
+        self._ensure_path_dirs_exist(target_path)
+        shutil.copyfile(local_path, target_path)
+
+    def _download_file(self, key, local_path, bucket=None):
+        source_path = os.path.join(self.store_root, key)
+        self._ensure_path_dirs_exist(local_path)
+        shutil.copyfile(source_path, local_path)
+
+    def _delete_file(self, key):
+        os.remove(os.path.join(self.store_root, key))
+
+    def _get_file_url(self, key, method='GET'):
+        return str(os.path.join(self.store_root, key))
+
+    def _get_file_post(self, key):
+        return str(os.path.join(self.store_root, key))
+
+    def _get_file_timestamp(self, key):
+            return None
+
+    def get_qualified_location(self, key):
+        return 'file:/' + self.store_root + '/' + key
+
+    def get_bucket(self):
+        return self.bucket

studio/local_db_provider.py

@@ -0,0 +1,51 @@
+import os
+import json
+
+from .keyvalue_provider import KeyValueProvider
+from .local_artifact_store import LocalArtifactStore
+
+class LocalDbProvider(KeyValueProvider):
+
+    def __init__(self, config, blocking_auth=True, verbose=10, store=None):
+        self.config = config
+        self.bucket = config.get('bucket', 'studioml-meta')
+
+        self.endpoint = config.get('endpoint', '~')
+        self.db_root = os.path.realpath(os.path.expanduser(self.endpoint))
+        if not os.path.exists(self.db_root) \
+            or not os.path.isdir(self.db_root):
+            raise ValueError("Local DB root {} doesn't exist or not a directory!".format(self.db_root))
+
+        self.bucket = config.get('bucket')
+        self.db_root = os.path.join(self.db_root, self.bucket)
+        self._ensure_path_dirs_exist(self.db_root)
+
+        super(LocalDbProvider, self).__init__(
+            config,
+            blocking_auth,
+            verbose,
+            store)
+
+    def _ensure_path_dirs_exist(self, path):
+        dirs = os.path.dirname(path)
+        os.makedirs(dirs, mode = 0o777, exist_ok = True)
+
+    def _get(self, key, shallow=False):
+        file_name = os.path.join(self.db_root, key)
+        if not os.path.exists(file_name):
+            return None
+        with open(file_name) as infile:
+            result = json.load(infile)
+        return result
+
+    def _delete(self, key):
+        file_name = os.path.join(self.db_root, key)
+        if os.path.exists(file_name):
+            os.remove(file_name)
+
+    def _set(self, key, value):
+        file_name = os.path.join(self.db_root, key)
+        self._ensure_path_dirs_exist(file_name)
+        with open(file_name, 'w') as outfile:
+            json.dump(value, outfile)
+

studio/model.py

@@ -14,6 +14,8 @@
 from .artifact_store import get_artifact_store
 from .http_provider import HTTPProvider
 from .firebase_provider import FirebaseProvider
+from .local_artifact_store import LocalArtifactStore
+from .local_db_provider import LocalDbProvider
 from .s3_provider import S3Provider
 from .gs_provider import GSProvider
 from .model_setup import setup_model
@@ -102,6 +104,16 @@ def get_db_provider(config=None, blocking_auth=True):
                           blocking_auth=blocking_auth)
         artifact_store = db_provider.get_artifact_store()
 
+    elif db_config['type'].lower() == 'local':
+        if artifact_store is None:
+            artifact_store = LocalArtifactStore(db_config, "storage", verbose)
+
+        db_provider = LocalDbProvider(db_config,
+                          verbose=verbose,
+                          store=artifact_store,
+                          blocking_auth=blocking_auth)
+        artifact_store = db_provider.get_artifact_store()
+
     else:
         _model_setup = None
         raise ValueError('Unknown type of the database ' + db_config['type'])