Task 010 implement research manifest flow

2026-05-21 21:56:44 +03:00
parent 5b27d1a3ca
commit 7d4a79acb5
23 changed files with 1063 additions and 26 deletions
@@ -20,6 +20,8 @@ from src.domain.contracts import (
    PlanSummary,
    PublishingRules,
    PublishingStatus,
+    ResearchArtifactManifestSummary,
+    ResearchArtifactSummary,
    ArticleWorkflowStatus,
    WorkflowEventSummary,
    Role,
@@ -46,6 +48,7 @@ class BackendRepository:
        self.articles = ArticlesRepository(self)
        self.boundary_questions = BoundaryQuestionsRepository(self)
        self.article_plans = ArticlePlansRepository(self)
+        self.research_manifests = ResearchManifestsRepository(self)
        self.agent_jobs = AgentJobsRepository(self)
        self.script_config_versions = ScriptConfigVersionsRepository(self)
        self.script_config_version_events = ScriptConfigVersionAuditEventsRepository(self)
@@ -1005,6 +1008,121 @@ class ArticlePlansRepository:
        """


+class ResearchManifestsRepository:
+    def __init__(self, repository: BackendRepository) -> None:
+        self._repository = repository
+
+    def create(
+        self,
+        *,
+        article_id: UUID,
+        agent_job_id: UUID,
+        s3_prefix: str,
+        artifacts: list[JsonObject],
+        created_at: datetime,
+    ) -> ResearchArtifactManifestSummary:
+        manifest_id = uuid4()
+        placeholder = self._repository.placeholder()
+        json_cast = self._repository.json_cast()
+        source_urls = [artifact["source_url"] for artifact in artifacts]
+        object_keys = [artifact["object_key"] for artifact in artifacts]
+        content_hashes = [artifact["content_hash"] for artifact in artifacts]
+        artifact_types = [artifact["artifact_type"] for artifact in artifacts]
+        metadata_references = [
+            artifact.get("metadata", {}).get("metadata_object_key")
+            for artifact in artifacts
+        ]
+        with self._repository.connection() as connection:
+            connection.execute(
+                f"""
+                INSERT INTO research_run_manifests (
+                    id,
+                    article_id,
+                    agent_job_id,
+                    s3_prefix,
+                    source_urls,
+                    object_keys,
+                    content_hashes,
+                    artifact_types,
+                    metadata_references,
+                    artifacts,
+                    created_at
+                )
+                VALUES (
+                    {placeholder},
+                    {placeholder},
+                    {placeholder},
+                    {placeholder},
+                    {placeholder}{json_cast},
+                    {placeholder}{json_cast},
+                    {placeholder}{json_cast},
+                    {placeholder}{json_cast},
+                    {placeholder}{json_cast},
+                    {placeholder}{json_cast},
+                    {placeholder}
+                )
+                """,
+                (
+                    str(manifest_id),
+                    str(article_id),
+                    str(agent_job_id),
+                    s3_prefix,
+                    _json_value(source_urls),
+                    _json_value(object_keys),
+                    _json_value(content_hashes),
+                    _json_value(artifact_types),
+                    _json_value(metadata_references),
+                    _json_value(artifacts),
+                    _datetime_value(created_at),
+                ),
+            )
+
+        return self.get(manifest_id)
+
+    def list_for_article(self, article_id: UUID) -> list[ResearchArtifactManifestSummary]:
+        placeholder = self._repository.placeholder()
+        with self._repository.connection() as connection:
+            rows = connection.execute(
+                f"""
+                SELECT
+                    id,
+                    article_id,
+                    agent_job_id,
+                    s3_prefix,
+                    artifacts,
+                    created_at
+                FROM research_run_manifests
+                WHERE article_id = {placeholder}
+                ORDER BY created_at
+                """,
+                (str(article_id),),
+            ).fetchall()
+
+        return [_research_manifest_from_row(row) for row in rows]
+
+    def get(self, manifest_id: UUID) -> ResearchArtifactManifestSummary:
+        placeholder = self._repository.placeholder()
+        with self._repository.connection() as connection:
+            row = connection.execute(
+                f"""
+                SELECT
+                    id,
+                    article_id,
+                    agent_job_id,
+                    s3_prefix,
+                    artifacts,
+                    created_at
+                FROM research_run_manifests
+                WHERE id = {placeholder}
+                """,
+                (str(manifest_id),),
+            ).fetchone()
+
+        if row is None:
+            raise LookupError(f"Research manifest not found: {manifest_id}")
+        return _research_manifest_from_row(row)
+
+
 class AgentJobsRepository:
    def __init__(self, repository: BackendRepository) -> None:
        self._repository = repository
@@ -1715,6 +1833,20 @@ def _plan_summary_from_row(
    )


+def _research_manifest_from_row(row: Any) -> ResearchArtifactManifestSummary:
+    return ResearchArtifactManifestSummary(
+        id=_row_value(row, "id"),
+        article_id=_row_value(row, "article_id"),
+        agent_job_id=_row_value(row, "agent_job_id"),
+        s3_prefix=_row_value(row, "s3_prefix"),
+        artifacts=[
+            ResearchArtifactSummary(**artifact)
+            for artifact in _json_from_row(row, "artifacts")
+        ],
+        created_at=_row_value(row, "created_at"),
+    )
+
+
 def _agent_job_summary_from_row(row: Any) -> AgentJobSummary:
    return AgentJobSummary(
        id=_row_value(row, "id"),