supabase / wrappers / 31848370433
81%

DEFAULT BRANCH: main
Ran
Jobs 1
Files 66
Run time 1min
Badge
Embed ▾
README BADGES
x

If you need to use a raster PNG badge, change the '.svg' to '.png' in the link

Markdown

Textile

RDoc

HTML

Rst
coverage: 80.969%. Remained the same
31848370433

push

github

web-flow 
feat: add Langfuse Wasm FDW to the catalog (#622)

* feat: add Langfuse Wasm FDW

Reads trace and observation data from Langfuse, an open source LLM
observability platform, so LLM spend can be attributed to rows in a local users
table in SQL.

Targets the Langfuse Public API rather than the ClickHouse store behind it,
which upstream documents as an unstable contract and which Langfuse Cloud users
cannot reach at all.

Supports select on `traces` and `observations`, import foreign schema, limit and
equality pushdown, and time bounds via fromTimestamp/toTimestamp (or
fromStartTime/toStartTime on observations — the column and params differ per
endpoint). Credentials resolve from a Vault secret name, a secret id, or a
plaintext option.

Usage and cost arrive as objects keyed by metric name and are flattened into
scalar columns so aggregates need no JSON extraction. The fallback chains are
load-bearing: `observations` returns calculatedTotalCost with no totalCost key,
and promptTokens rather than usageDetails.input.

Verified end to end on Supabase against a live Langfuse Cloud project.

* Inline format args to satisfy clippy

CI runs clippy with RUSTFLAGS="-D warnings", which promotes
uninlined_format_args to an error. Nine call sites used the positional form.

* Add smoke test for the Langfuse Wasm FDW

Mock responses are shaped so a regression fails rather than silently passing:

- Both endpoints report page/totalPages instead of a cursor, and the server is
  configured with page_size 1 against a two-page traces mock, so the test only
  sees both rows if pagination is followed.
- The traces mock returns nothing for any userId other than user-alice, so a
  dropped equality pushdown fails the assertion instead of falling back to
  unfiltered rows.
- The two observation rows spell usage and cost differently — one nests them
  under usageDetails/costDetails, the other only has the flat promptTokens and
  calculated*Cost forms. Both must map to the same colu... (continued)

9488 of 11718 relevant lines covered (80.97%)

70.59 hits per line

Jobs
ID Job ID Ran Files Coverage
1 31848370433.1 66
80.97
 GitHub Action Run
Source Files on build 31848370433