creating global context and refactoring tmp_dir logic

src/auto_archiver/core/__init__.py

@@ -1,6 +1,7 @@
 from .media import Media
 from .metadata import Metadata
 from .step import Step
+from .context import ArchivingContext
 
 # cannot import ArchivingOrchestrator/Config to avoid circular dep
 # from .orchestrator import ArchivingOrchestrator

src/auto_archiver/core/context.py

@@ -0,0 +1,38 @@
+
+class ArchivingContext:
+    """
+    Singleton context class.
+    ArchivingContext._get_instance() to retrieve it if needed
+    otherwise just 
+    ArchivingContext.set(key, value)
+    and 
+    ArchivingContext.get(key, default)
+    """
+    _instance = None
+
+    def __init__(self):
+        self.configs = {}
+
+    @staticmethod
+    def get_instance():
+        if ArchivingContext._instance is None:
+            ArchivingContext._instance = ArchivingContext()
+        return ArchivingContext._instance
+
+    @staticmethod
+    def set(key, value):
+        ArchivingContext.get_instance().configs[key] = value
+
+    @staticmethod
+    def get(key: str, default=None):
+        return ArchivingContext.get_instance().configs.get(key, default)
+
+    # ---- custom getters/setters for widely used context values
+
+    @staticmethod
+    def set_tmp_dir(tmp_dir: str):
+        ArchivingContext.get_instance().configs["tmp_dir"] = tmp_dir
+
+    @staticmethod
+    def get_tmp_dir() -> str:
+        return ArchivingContext.get_instance().configs.get("tmp_dir")

src/auto_archiver/core/media.py

@@ -6,8 +6,9 @@ from dataclasses import dataclass, field
 from dataclasses_json import dataclass_json
 import mimetypes
 
-# annotation order matters
-@dataclass_json
+
+
+@dataclass_json # annotation order matters
 @dataclass
 class Media:
     filename: str
@@ -40,3 +41,13 @@ class Media:
 
     def is_video(self) -> bool:
         return self.mimetype.startswith("video")
+
+    def is_audio(self) -> bool:
+        return self.mimetype.startswith("audio")
+
+    def store(self):
+        """
+        either stores this media entry and all its media descendants
+        or returns if that process is already completed
+        """
+        pass

src/auto_archiver/core/metadata.py

@@ -3,13 +3,12 @@ from __future__ import annotations
 from ast import List, Set
 from typing import Any, Union, Dict
 from dataclasses import dataclass, field
-from dataclasses_json import dataclass_json
+from dataclasses_json import dataclass_json, config
 import datetime
 from urllib.parse import urlparse
 from dateutil.parser import parse as parse_dt
 from .media import Media
 
-
 # annotation order matters
 @dataclass_json
 @dataclass
@@ -17,10 +16,14 @@ class Metadata:
     status: str = "no archiver"
     _processed_at: datetime = field(default_factory=datetime.datetime.utcnow)
     metadata: Dict[str, Any] = field(default_factory=dict)
-    tmp_keys: Set[str] = field(default_factory=set, repr=False, metadata={"exclude": True})  # keys that are not to be saved in DBs
     media: List[Media] = field(default_factory=list)
     rearchivable: bool = True  # defaults to true, archivers can overwrite
 
+    # properties below are excluded from JSON representation
+    tmp_keys: Set[str] = field(default_factory=set, repr=False, metadata=config(exclude=True))
+    # tmp_metadata: Dict[str, Any] = field(default_factory=dict, repr=False, metadata=config(exclude=True)) # contains internal properties not to be leaked when .to_json/repr/str is called
+
+
     def merge(self: Metadata, right: Metadata, overwrite_left=True) -> Metadata:
         """
         merges two Metadata instances, will overwrite according to overwrite_left flag
@@ -93,12 +96,6 @@ class Metadata:
     def get_title(self) -> str:
         return self.get("title")
 
-    def set_tmp_dir(self, tmp_dir: str) -> Metadata:
-        return self.set("tmp_dir", tmp_dir, True)
-
-    def get_tmp_dir(self) -> str:
-        return self.get("tmp_dir")
-
     def set_timestamp(self, timestamp: datetime.datetime) -> Metadata:
         if type(timestamp) == str:
             timestamp = parse_dt(timestamp)
@@ -144,3 +141,7 @@ class Metadata:
             {k: v for k, v in self.metadata.items() if k not in self.tmp_keys},
             **{"processed_at": self._processed_at}
         )
+
+    def __str__(self) -> str:
+        return self.__repr__()
+        

src/auto_archiver/core/orchestrator.py

@@ -2,6 +2,8 @@ from __future__ import annotations
 from ast import List
 from typing import Union
 
+from .context import ArchivingContext
+
 from ..archivers import Archiver
 from ..feeders import Feeder
 from ..formatters import Formatter
@@ -23,6 +25,7 @@ class ArchivingOrchestrator:
         self.archivers: List[Archiver] = config.archivers
         self.databases: List[Database] = config.databases
         self.storages: List[Storage] = config.storages
+        ArchivingContext.set("storages", self.storages)
 
         for a in self.archivers: a.setup()
 
@@ -33,7 +36,7 @@ class ArchivingOrchestrator:
     def feed_item(self, item: Metadata) -> Metadata:
         try:
             with tempfile.TemporaryDirectory(dir="./") as tmp_dir:
-                item.set_tmp_dir(tmp_dir)
+                ArchivingContext.set_tmp_dir(tmp_dir)
                 return self.archive(item)
         except KeyboardInterrupt:
             # catches keyboard interruptions to do a clean exit