Add ingestion_agent Spark source for the ingestion agent's read API#184
Open
yyoli-db wants to merge 3 commits into
Open
Add ingestion_agent Spark source for the ingestion agent's read API#184yyoli-db wants to merge 3 commits into
yyoli-db wants to merge 3 commits into
Conversation
Introduces a new `ingestion_agent` Python Data Source format that dispatches on an `operation` option, returning DataFrames for list_objects, preview_table, get_object_metadata, validate_connection, and list_operations — the read-side surface specified in `experimental/yong-li/docs/ingestion-agent-api-design.md`. Defaults derive from the existing LakeflowConnect surface, so every existing connector gains the agent API automatically. Connectors that need richer responses (hierarchical listing, source-native sampling, extra metadata, custom prefixed operations) implement the optional `SupportsIngestionAgent` mixin. The `lakeflow_connect` format is unchanged; `ingestion_agent` is registered alongside it. Co-authored-by: Isaac
Refactor source-specific operations into AgentOperation classes, so adding a new op is a class + one entry on a source's agent_operations() map — no framework edits required. Mirrors the Scala AgentOperation design in databricks-eng/universe#1935135. - Add AgentOperation ABC (name, description, kind, schema / resolve_schema, pull) to the interface module. - Reframe the five built-ins as AgentOperation subclasses that consult the existing SupportsIngestionAgent per-method hooks before falling back to LakeflowConnect-derived defaults. - Source-defined operations override built-ins with the same name; list_operations reports both built-ins and source plug-ins. - Framework auto-appends _meta and converts pull() exceptions to an error row for kind=metadata; data-kind ops pass through with the source's natural schema. - Relax data-column nullability for metadata-kind ops so the framework's _meta-only error row validates against the schema. Co-authored-by: Isaac
Lower the cost of adding a source-specific ingestion-agent operation to a single decorated method. Compared to the AgentOperation subclass + agent_operations() map entry, the decorator path: - drops the subclass wrapper — name/description/schema/kind go on the decorator, the method body *is* the pull; - removes the agent_operations() override — the default implementation walks the class for @agent_operation-decorated attributes and wraps each one in an AgentOperation; - still composes with the class-based path: override agent_operations() and call super() to mix in stateful subclass operations alongside decorated ones. Class-based AgentOperation is kept for ops with state or dynamic schemas; the framework dispatches both paths identically. Co-authored-by: Isaac
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Introduces a new
ingestion_agentPython Data Source format that dispatches on anoperationoption, returning DataFrames for list_objects, preview_table, get_object_metadata, validate_connection, and list_operations — the read-side surface specified inexperimental/yong-li/docs/ingestion-agent-api-design.md.Defaults derive from the existing LakeflowConnect surface, so every existing connector gains the agent API automatically. Connectors that need richer responses (hierarchical listing, source-native sampling, extra metadata, custom prefixed operations) implement the optional
SupportsIngestionAgentmixin. Thelakeflow_connectformat is unchanged;ingestion_agentis registered alongside it.Co-authored-by: Isaac