Skip to content

dbt integration: FileSource generates invalid hardcoded paths #5874

@YassinNouh21

Description

@YassinNouh21

Context

PR #5827 added support for FileSource data sources in dbt integration.

Problem

The FileSource implementation generates hardcoded paths that likely don't exist:

# In mapper.py:240-242
return FileSource(
    name=f"{model.name}_source",
    path=f"/data/{model.name}.parquet",  # Hardcoded, likely invalid
    # ...
)

This path /data/{model.name}.parquet is unlikely to match actual file locations.

Impact

  • Generated FileSource objects won't work without manual editing
  • Confusing user experience
  • No clear documentation on how to fix the paths

Proposed Solutions

Option 1: Make path configurable

feast dbt import -m manifest.json -e driver_id --data-source-type file --file-path-template "/my/data/{model}.parquet"

Option 2: Read from dbt meta

Support custom path in dbt model meta:

# dbt model
models:
  - name: driver_stats
    meta:
      feast:
        file_path: "/data/warehouse/driver_stats.parquet"

Option 3: Require explicit path mapping file

feast dbt import -m manifest.json -e driver_id --file-path-config paths.yaml

Recommendation

Implement Option 1 (path template) as it's simplest and covers most use cases. Consider Option 2 for advanced scenarios.

Related

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions