Store posting results to a mongodb

.env.example

@@ -33,3 +33,9 @@ CONTENT_AUTOMATION_STORAGE__S3__REGION_NAME=
 CONTENT_AUTOMATION_STORAGE__S3__ENDPOINT_URL=
 CONTENT_AUTOMATION_STORAGE__S3__PUBLIC_URL_BASE=
 CONTENT_AUTOMATION_STORAGE__S3__URL_EXPIRATION_SECONDS=3600
+
+# MongoDB publish persistence
+CONTENT_AUTOMATION_MONGODB__ENABLED=false
+CONTENT_AUTOMATION_MONGODB__CONNECTION_URI=
+CONTENT_AUTOMATION_MONGODB__DATABASE_NAME=content_automation
+CONTENT_AUTOMATION_MONGODB__COLLECTION_NAME=published_content

pyproject.toml

@@ -8,6 +8,7 @@ dependencies = [
     "boto3>=1.37.17",
     "google-api-python-client>=2.190.0",
     "google-auth-oauthlib>=1.2.4",
+    "pymongo>=4.15.2",
     "pydantic>=2.12.5",
     "pydantic-settings>=2.11.0",
     "pytest>=8.3.5",

src/content_automation/adapters/publish_store/__init__.py

@@ -0,0 +1,5 @@
+from __future__ import annotations
+
+from content_automation.adapters.publish_store.mongodb import MongoPublishedContentStore
+
+__all__ = ["MongoPublishedContentStore"]

src/content_automation/adapters/publish_store/mongodb.py

@@ -0,0 +1,22 @@
+from __future__ import annotations
+
+from pymongo import MongoClient
+
+from content_automation.interfaces import PublishedContentStore
+from content_automation.models import PublishedContentRecord
+
+
+class MongoPublishedContentStore(PublishedContentStore):
+    """MongoDB-backed persistence for publish operation records."""
+
+    def __init__(
+        self,
+        connection_uri: str,
+        database_name: str,
+        collection_name: str,
+    ) -> None:
+        self._client = MongoClient(connection_uri)
+        self._collection = self._client[database_name][collection_name]
+
+    def save(self, record: PublishedContentRecord) -> None:
+        self._collection.insert_one(record.model_dump(mode="json"))

src/content_automation/controller.py

@@ -1,7 +1,11 @@
 from __future__ import annotations
 
+from collections.abc import Mapping
+from typing import Any
+
 from content_automation.adapters.storage.base import StorageAdapterBase
-from content_automation.interfaces import SocialNetworkAdapter
+from content_automation.interfaces import PublishedContentStore, SocialNetworkAdapter
+from content_automation.models import PublishedContentRecord
 from content_automation.settings import AppSettings
 
 
@@ -13,22 +17,60 @@ class PublishController:
         settings: AppSettings,
         storage: StorageAdapterBase,
         social_adapters: dict[str, SocialNetworkAdapter],
+        published_content_store: PublishedContentStore | None = None,
     ) -> None:
         self._settings = settings
         self._storage = storage
         self._social_adapters = social_adapters
+        self._published_content_store = published_content_store
 
-    def publish(self, relative_path: str, caption: str) -> dict[str, str]:
+    def publish(self, relative_path: str, caption: str) -> dict[str, Any]:
         if not self._storage.exists(relative_path):
             raise FileNotFoundError(
                 f"Media file is not available in storage: {relative_path}"
             )
 
         media_url = self._storage.get_public_url(relative_path)
-        result: dict[str, str] = {}
+        result: dict[str, Any] = {}
         for network in self._settings.target_social_networks:
             adapter = self._social_adapters.get(network)
             if adapter is None:
                 raise ValueError(f"No adapter configured for network: {network}")
             result[network] = adapter.post_media(media_url=media_url, caption=caption)
+
+        if self._published_content_store is not None:
+            record = PublishedContentRecord(
+                relative_path=relative_path,
+                media_url=media_url,
+                caption=caption,
+                platform_responses=self._serialize_platform_responses(result),
+            )
+            self._published_content_store.save(record)
+
         return result
+
+    def _serialize_platform_responses(
+        self, responses: Mapping[str, Any]
+    ) -> dict[str, Any]:
+        serialized: dict[str, Any] = {}
+        for network, response in responses.items():
+            if network == "youtube":
+                serialized[network] = self._extract_youtube_video_id(response)
+                continue
+            serialized[network] = response
+        return serialized
+
+    @staticmethod
+    def _extract_youtube_video_id(response: Any) -> str:
+        if isinstance(response, str):
+            return response
+
+        if isinstance(response, Mapping):
+            video_id = response.get("id")
+            if isinstance(video_id, str) and video_id:
+                return video_id
+
+        raise ValueError(
+            "Unsupported YouTube publish response: expected a string ID "
+            "or payload containing an 'id' field."
+        )

src/content_automation/factories.py

@@ -1,11 +1,12 @@
 from __future__ import annotations
 
+from content_automation.adapters.publish_store.mongodb import MongoPublishedContentStore
 from content_automation.adapters.social.instagram import InstagramAdapter
 from content_automation.adapters.social.youtube import YouTubeAdapter
 from content_automation.adapters.storage.base import StorageAdapterBase
 from content_automation.adapters.storage.local import LocalFilesystemStorageAdapter
 from content_automation.adapters.storage.s3 import S3StorageAdapter
-from content_automation.interfaces import SocialNetworkAdapter
+from content_automation.interfaces import PublishedContentStore, SocialNetworkAdapter
 from content_automation.settings import AppSettings
 
 
@@ -49,3 +50,22 @@ def build_social_adapters(settings: AppSettings) -> dict[str, SocialNetworkAdapt
             expiry=settings.youtube.expiry or None,
         ),
     }
+
+
+def build_published_content_store(
+    settings: AppSettings,
+) -> PublishedContentStore | None:
+    if not settings.mongodb.enabled:
+        return None
+
+    if not settings.mongodb.connection_uri:
+        raise ValueError(
+            "CONTENT_AUTOMATION_MONGODB__CONNECTION_URI must be set when "
+            "MongoDB persistence is enabled."
+        )
+
+    return MongoPublishedContentStore(
+        connection_uri=settings.mongodb.connection_uri,
+        database_name=settings.mongodb.database_name,
+        collection_name=settings.mongodb.collection_name,
+    )

src/content_automation/interfaces.py

@@ -1,6 +1,8 @@
 from __future__ import annotations
 
-from typing import Protocol
+from typing import Any, Protocol
+
+from content_automation.models import PublishedContentRecord
 
 
 class SocialNetworkAdapter(Protocol):
@@ -8,5 +10,12 @@ class SocialNetworkAdapter(Protocol):
 
     name: str
 
-    def post_media(self, media_url: str, caption: str) -> str:
-        """Publish media and return an external post identifier."""
+    def post_media(self, media_url: str, caption: str) -> Any:
+        """Publish media and return a provider response payload."""
+
+
+class PublishedContentStore(Protocol):
+    """Contract for persisting publish operation metadata."""
+
+    def save(self, record: PublishedContentRecord) -> None:
+        """Persist a published content record."""

src/content_automation/main.py

@@ -3,7 +3,11 @@ from __future__ import annotations
 import argparse
 
 from content_automation.controller import PublishController
-from content_automation.factories import build_social_adapters, build_storage_adapter
+from content_automation.factories import (
+    build_published_content_store,
+    build_social_adapters,
+    build_storage_adapter,
+)
 from content_automation.settings import AppSettings
 
 
@@ -23,8 +27,12 @@ def main() -> None:
     settings = AppSettings()
     storage = build_storage_adapter(settings)
     adapters = build_social_adapters(settings)
+    published_content_store = build_published_content_store(settings)
     controller = PublishController(
-        settings=settings, storage=storage, social_adapters=adapters
+        settings=settings,
+        storage=storage,
+        social_adapters=adapters,
+        published_content_store=published_content_store,
     )
     publish_results = controller.publish(
         relative_path=args.relative_path, caption=args.caption

src/content_automation/models.py

@@ -0,0 +1,16 @@
+from __future__ import annotations
+
+from datetime import UTC, datetime
+from typing import Any
+
+from pydantic import BaseModel, Field
+
+
+class PublishedContentRecord(BaseModel):
+    """Stored metadata for a publish operation across social networks."""
+
+    relative_path: str
+    media_url: str
+    caption: str
+    platform_responses: dict[str, Any]
+    published_at: datetime = Field(default_factory=lambda: datetime.now(UTC))

src/content_automation/settings.py

@@ -48,9 +48,16 @@ class AppSettings(BaseSettings):
         local: LocalSettings = Field(default_factory=LocalSettings)
         s3: S3Settings = Field(default_factory=S3Settings)
 
+    class MongoDbSettings(BaseModel):
+        enabled: bool = False
+        connection_uri: str = ""
+        database_name: str = "content_automation"
+        collection_name: str = "published_content"
+
     target_social_networks: list[str] = Field(
         default_factory=lambda: ["instagram", "youtube"]
     )
     instagram: InstagramSettings = Field(default_factory=InstagramSettings)
     youtube: YoutubeSettings = Field(default_factory=YoutubeSettings)
     storage: StorageSettings = Field(default_factory=StorageSettings)
+    mongodb: MongoDbSettings = Field(default_factory=MongoDbSettings)

tests/test_controller.py

@@ -4,6 +4,7 @@ import pytest
 
 from content_automation.adapters.storage.base import StorageAdapterBase
 from content_automation.controller import PublishController
+from content_automation.models import PublishedContentRecord
 from content_automation.settings import AppSettings
 
 
@@ -29,6 +30,23 @@ class FakeAdapter:
         return f"{self.name}-post-id"
 
 
+class FakeAdapterWithResponsePayload:
+    def __init__(self, payload: object) -> None:
+        self.name = "adapter-with-payload"
+        self._payload = payload
+
+    def post_media(self, media_url: str, caption: str) -> object:
+        return self._payload
+
+
+class FakePublishedContentStore:
+    def __init__(self) -> None:
+        self.records: list[PublishedContentRecord] = []
+
+    def save(self, record: PublishedContentRecord) -> None:
+        self.records.append(record)
+
+
 def test_controller_publishes_to_all_configured_networks() -> None:
     settings = AppSettings.model_validate(
         {"target_social_networks": ["instagram", "youtube"]}
@@ -47,6 +65,47 @@ def test_controller_publishes_to_all_configured_networks() -> None:
     assert result == {"instagram": "instagram-post-id", "youtube": "youtube-post-id"}
 
 
+def test_controller_stores_published_content_with_youtube_id_only() -> None:
+    settings = AppSettings.model_validate(
+        {"target_social_networks": ["instagram", "youtube"]}
+    )
+    published_content_store = FakePublishedContentStore()
+    controller = PublishController(
+        settings=settings,
+        storage=FakeStorage(exists_result=True),
+        social_adapters={
+            "instagram": FakeAdapterWithResponsePayload(
+                {"id": "ig-123", "status": "ok"}
+            ),
+            "youtube": FakeAdapterWithResponsePayload(
+                {
+                    "id": "yt-456",
+                    "kind": "youtube#video",
+                    "snippet": {"title": "hello"},
+                }
+            ),
+        },
+        published_content_store=published_content_store,
+    )
+
+    result = controller.publish(relative_path="video.mp4", caption="hello")
+
+    assert result["youtube"] == {
+        "id": "yt-456",
+        "kind": "youtube#video",
+        "snippet": {"title": "hello"},
+    }
+    assert len(published_content_store.records) == 1
+    saved_record = published_content_store.records[0]
+    assert saved_record.relative_path == "video.mp4"
+    assert saved_record.caption == "hello"
+    assert saved_record.platform_responses["instagram"] == {
+        "id": "ig-123",
+        "status": "ok",
+    }
+    assert saved_record.platform_responses["youtube"] == "yt-456"
+
+
 def test_controller_raises_when_file_missing() -> None:
     settings = AppSettings.model_validate({"target_social_networks": ["youtube"]})
     controller = PublishController(

tests/test_publish_store.py

@@ -0,0 +1,115 @@
+from __future__ import annotations
+
+import pytest
+
+from content_automation.adapters.publish_store import MongoPublishedContentStore
+from content_automation.factories import build_published_content_store
+from content_automation.models import PublishedContentRecord
+from content_automation.settings import AppSettings
+
+
+class FakeCollection:
+    def __init__(self) -> None:
+        self.documents: list[dict[str, object]] = []
+
+    def insert_one(self, document: dict[str, object]) -> None:
+        self.documents.append(document)
+
+
+class FakeDatabase:
+    def __init__(self, collection: FakeCollection) -> None:
+        self._collection = collection
+
+    def __getitem__(self, collection_name: str) -> FakeCollection:
+        return self._collection
+
+
+class FakeMongoClient:
+    def __init__(self, connection_uri: str) -> None:
+        self.connection_uri = connection_uri
+        self.collection = FakeCollection()
+
+    def __getitem__(self, database_name: str) -> FakeDatabase:
+        return FakeDatabase(self.collection)
+
+
+def test_mongo_published_content_store_saves_record(monkeypatch) -> None:
+    fake_client = FakeMongoClient("mongodb://example:27017")
+
+    monkeypatch.setattr(
+        "content_automation.adapters.publish_store.mongodb.MongoClient",
+        lambda connection_uri: fake_client,
+    )
+
+    store = MongoPublishedContentStore(
+        connection_uri="mongodb://example:27017",
+        database_name="content_automation",
+        collection_name="published_content",
+    )
+
+    store.save(
+        PublishedContentRecord(
+            relative_path="media/video.mp4",
+            media_url="https://cdn.example.com/media/video.mp4",
+            caption="caption",
+            platform_responses={
+                "instagram": {"id": "ig-1"},
+                "youtube": "yt-1",
+            },
+        )
+    )
+
+    assert len(fake_client.collection.documents) == 1
+    assert fake_client.collection.documents[0]["relative_path"] == "media/video.mp4"
+    assert fake_client.collection.documents[0]["platform_responses"] == {
+        "instagram": {"id": "ig-1"},
+        "youtube": "yt-1",
+    }
+
+
+def test_build_published_content_store_returns_none_when_disabled() -> None:
+    settings = AppSettings.model_validate({"mongodb": {"enabled": False}})
+
+    store = build_published_content_store(settings)
+
+    assert store is None
+
+
+def test_build_published_content_store_requires_connection_uri() -> None:
+    settings = AppSettings.model_validate({"mongodb": {"enabled": True}})
+
+    with pytest.raises(ValueError):
+        build_published_content_store(settings)
+
+
+def test_build_published_content_store_returns_mongo_store(monkeypatch) -> None:
+    class FakeMongoStore:
+        def __init__(
+            self,
+            connection_uri: str,
+            database_name: str,
+            collection_name: str,
+        ) -> None:
+            self.connection_uri = connection_uri
+            self.database_name = database_name
+            self.collection_name = collection_name
+
+    monkeypatch.setattr("content_automation.factories.MongoPublishedContentStore", FakeMongoStore)
+
+    settings = AppSettings.model_validate(
+        {
+            "mongodb": {
+                "enabled": True,
+                "connection_uri": "mongodb://localhost:27017",
+                "database_name": "content_automation",
+                "collection_name": "published_content",
+            }
+        }
+    )
+
+    store = build_published_content_store(settings)
+
+    assert isinstance(store, FakeMongoStore)
+    assert store.connection_uri == "mongodb://localhost:27017"
+    assert store.database_name == "content_automation"
+    assert store.collection_name == "published_content"

tests/test_settings.py

@@ -9,6 +9,15 @@ def test_settings_parse_nested_env(monkeypatch) -> None:
     monkeypatch.setenv("CONTENT_AUTOMATION_YOUTUBE__USE_RESUMABLE_UPLOAD", "true")
     monkeypatch.setenv("CONTENT_AUTOMATION_STORAGE__BACKEND", "s3")
     monkeypatch.setenv("CONTENT_AUTOMATION_STORAGE__S3__BUCKET_NAME", "bucket-a")
+    monkeypatch.setenv("CONTENT_AUTOMATION_MONGODB__ENABLED", "true")
+    monkeypatch.setenv(
+        "CONTENT_AUTOMATION_MONGODB__CONNECTION_URI",
+        "mongodb://localhost:27017",
+    )
+    monkeypatch.setenv(
+        "CONTENT_AUTOMATION_MONGODB__DATABASE_NAME",
+        "content-automation",
+    )
 
     settings = AppSettings()
 
@@ -17,3 +26,6 @@ def test_settings_parse_nested_env(monkeypatch) -> None:
     assert settings.youtube.use_resumable_upload is True
     assert settings.storage.backend == "s3"
     assert settings.storage.s3.bucket_name == "bucket-a"
+    assert settings.mongodb.enabled is True
+    assert settings.mongodb.connection_uri == "mongodb://localhost:27017"
+    assert settings.mongodb.database_name == "content-automation"