test fixes

[TejasGhatte]

Summary

This PR delivers a collection of fixes and improvements across the Gemini provider, Anthropic provider, Bedrock Responses API, Azure config redaction, and migration test infrastructure for v1.5.4.

Changes

• Gemini – tool schema passthrough via ParametersJSONSchema: Tool parameter schemas are now forwarded to Gemini using the parametersJsonSchema wire key instead of the structured parameters field. This removes the previous schema-transformation layer (union-type expansion, anyOf rewriting, sibling-field stripping, nullable injection) and passes the raw JSON Schema through unchanged. As a result, array-typed unions, sibling fields alongside anyOf, and description fields are all preserved. The propertyOrdering field is no longer emitted separately; property order is maintained by the underlying OrderedMap serialization.
• Gemini – tool response role corrected to "user": Function/tool response content blocks were being emitted with role "model"; they are now correctly emitted with role "user" across both the Chat and Responses API paths.
• Gemini – structured output + tools conflict: When both tools and a JSON response format are present for Gemini 2.5, responseJsonSchema is now also dropped (previously only responseMimeType was dropped).
• Anthropic – stop reason normalization: end_turn → stop, tool_use → tool_calls, max_tokens → length to align with the normalized Bifrost stop-reason vocabulary. Tests updated accordingly.
• Anthropic – computer-use tool version mapping: text_editor_20250124/str_replace_editor is now upgraded to text_editor_20250728/str_replace_based_edit_tool for claude-sonnet-4-5 models. Test names and expectations updated to reflect the corrected behavior.
• Bedrock – Responses API hasToolUse detection: Replaced the content-block scan (which checked for unmatched toolUse blocks) with a direct check on bifrostResp.Output for ResponsesMessageTypeFunctionCall, making the detection more reliable and consistent with the Responses API data model.
• Azure config redaction: Fixed a panic/incorrect redaction when AzureKeyConfig.Endpoint is not sourced from an environment variable. The endpoint is now only redacted when IsFromEnv() is true; otherwise the original value is preserved as-is.
• JSON parser plugin test: Added missing Params with json_object format to the Responses stream end-to-end test to properly exercise the parser plugin.
• Migration tests – v1.5.4 columns: Added dynamic column update blocks for three new v1.5.4 migrations — governance_virtual_key_provider_configs.blacklisted_models, governance_virtual_keys.created_by_user_id, and logs.inc_number — for both PostgreSQL and SQLite paths. Also added azure_api_version to the list of dropped columns on config_keys for snapshot comparison.

Type of change

• Bug fix
• Feature
• Refactor
• Documentation
• Chore/CI

Affected areas

• Core (Go)
• Transports (HTTP)
• Providers/Integrations
• Plugins
• UI (React)
• Docs

How to test

go test ./core/providers/gemini/...
go test ./core/providers/anthropic/...
go test ./core/providers/bedrock/...
go test ./framework/configstore/...
go test ./plugins/jsonparser/...

For migration tests, run the migration test workflow against a PostgreSQL and SQLite target and verify that v1.5.4 column additions and the azure_api_version column drop are handled without snapshot mismatches.

Breaking changes

• Yes
• No

The Gemini tool schema wire format changes from parameters to parametersJsonSchema. Clients or tests that assert on the exact wire key or rely on the previous union-type/anyOf rewriting behavior will need to be updated. The Anthropic stop reason values (end_turn, tool_use, max_tokens) are replaced with normalized values (stop, tool_calls, length); any downstream code matching on the raw Anthropic strings will need to be updated.

Security considerations

The Azure endpoint redaction fix ensures that plain (non-env-var) endpoint values are not incorrectly processed through the Redacted() path, preventing potential nil-pointer panics and ensuring the correct value is returned in config responses.

Checklist

• I read docs/contributing/README.md and followed the guidelines
• I added/updated tests where appropriate
• I updated documentation where needed
• I verified builds succeed (Go and UI)
• I verified the CI pipeline passes locally if applicable

[TejasGhatte]

• test fixes #3761 👈 (View in Graphite)
• dev

This stack of pull requests is managed by Graphite. Learn more about stacking.

[coderabbitai]

📝 Walkthrough

Summary by CodeRabbit

• Bug Fixes

  • Improved tool-use detection and stop-reason mapping across providers.
  • Refined handling of tool-response grouping and structured output for Gemini.

• Behavior Changes

  • Bedrock “nova” pathways now allow web search and code-execution tools.
  • Azure endpoint redaction now preserves literal endpoints unless sourced from env.

• Tests

  • Expanded and adjusted tests for provider tool conversions and schema validation.

• Chores

  • Updated migration-test data/backfill logic for new schema columns.

Walkthrough

This PR updates test expectations and infrastructure across multiple provider implementations. Migration test generation now populates v1.5.4 schema columns with conditional guards, and snapshot validation excludes intentionally dropped columns. Provider tests are refactored to align with updated tool parameter serialization, stop-reason mappings, and role attribution in Gemini, Anthropic, and Bedrock providers. Two production changes update Bedrock tool-use detection logic and Azure endpoint redaction handling.

Changes

Provider Test Updates and Migration Infrastructure

Layer / File(s)	Summary
Migration test infrastructure for v1.5.4 schema population .github/workflows/scripts/run-migration-tests.sh	Dynamic UPDATE statements populate newly added v1.5.4 columns (blacklisted_models, created_by_user_id, inc_number) with schema-existence guards for both PostgreSQL and SQLite; PostgreSQL snapshot validation extends dropped-column exclusion list to include azure_api_version.
Bedrock/Anthropic provider feature flags and tests core/internal/llmtests/provider_feature_support_test.go, core/providers/anthropic/types.go, core/providers/anthropic/utils.go	Adds Bedrock nova-specific feature flags (WebSearchNova, CodeExecNova) and updates provider feature/support and pipeline tests to reflect Responses-path nova capability flags.
Anthropic stop-reason and tool-remap tests core/providers/anthropic/compaction_test.go, core/providers/anthropic/utils_test.go, core/providers/anthropic/requestbuilder_test.go	Updates Anthropic stop-reason mapping expectations and tool-remapping test cases; adds Bedrock web_search allow-case and narrows typed-path tool-validation to reject web_fetch.
Bedrock tool-use detection for stop-reason derivation core/providers/bedrock/responses.go	ToBedrockConverseResponse now derives hasToolUse by iterating bifrostResp.Output and detecting FunctionCall message types instead of scanning assembled message content.
Gemini ParametersJSONSchema parsing and schema assertions core/providers/gemini/gemini_test.go	Adds parseToolParams/getSchemaProperty test helpers and reworks Gemini tool-conversion and Responses API tests to validate ParametersJSONSchema passthrough as generic maps: required fields, nested objects/arrays (including empty items), validation constraints, anyOf unions with sibling fields, and property ordering via serialized properties.
Gemini union-type wire-format assertions core/providers/gemini/uniontype_test.go	Wire-level tests now require union-typed type:[...] arrays be preserved in serialized JSON and forbid emitting nullable/anyOf for those cases.
Gemini tool/function-response role attribution core/providers/gemini/gemini_test.go	Updates grouped tool/function-response expectations so gemini.Content.Role is "user" instead of "model" across parallel and mixed grouping tests.
Azure endpoint redaction logic in configuration framework/configstore/clientconfig.go	ProviderConfig.Redacted() conditionally redacts Azure Endpoint only when sourced from environment variables; literal endpoints are preserved unchanged.
JSON parser plugin streaming format configuration plugins/jsonparser/plugin_test.go	Updates TestJsonParserPluginResponsesStreamEndToEnd to configure request-level Params.Text.Format.Type = "json_object" for streaming JSON-object output expectations.

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~25 minutes

Possibly related PRs

• maximhq/bifrost#3671: Both PRs modify .github/workflows/scripts/run-migration-tests.sh to extend the migration-test faker’s schema-guarded/dynamic UPDATE logic for PostgreSQL/SQLite (including governance_virtual_keys and logs columns).
• maximhq/bifrost#3633: Related Gemini structured-output changes; both PRs adjust Gemini tool+structured-output behavior and tests around dropping ResponseMIMEType/ResponseJSONSchema when tools are present.
• maximhq/bifrost#3520: Both PRs change Gemini union-type test expectations to verify parametersJsonSchema passthrough/preserved "type":[...] behavior rather than normalizing into nullable/anyOf forms.

Suggested reviewers

• danpiths
• akshaydeo

"A rabbit hops through test-lined fields of green,
Where schemas mend and endpoints go unseen,
Roles tumble from model down to user light,
Tools whisper JSON through the streaming night,
I nibble bugs and dance—tests pass in sight." 🐇✨

🚥 Pre-merge checks | ✅ 3 | ❌ 2

❌ Failed checks (2 warnings)

Check name	Status	Explanation	Resolution
Title check	⚠️ Warning	The title 'test fixes' is vague and generic, failing to summarize the substantial changes across multiple providers and infrastructure components.	Revise the title to reflect the primary changes, e.g., 'Gemini schema passthrough, Anthropic normalization, Bedrock tool detection, and v1.5.4 migrations'.
Docstring Coverage	⚠️ Warning	Docstring coverage is 79.17% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (3 passed)

Check name	Status	Explanation
Description check	✅ Passed	The description is comprehensive and well-structured, covering all required sections: summary, changes, type of change, affected areas, testing instructions, breaking changes, security considerations, and a complete checklist.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch 05-26-test_fixes

Warning

There were issues while running some tools. Please review the errors and either fix the tool's configuration or disable the tool if it's a critical failure.

🔧 golangci-lint (2.12.2)

level=error msg="[linters_context] typechecking error: pattern ./...: directory prefix . does not contain main module or its selected dependencies"

Warning

Review ran into problems

🔥 Problems

Stopped waiting for pipeline failures after 30000ms. One of your pipelines takes longer than our 30000ms fetch window to run, so review may not consider pipeline-failure results for inline comments if any failures occurred after the fetch window. Increase the timeout if you want to wait longer or run a @coderabbit review after the pipeline has finished.

---

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.

---

tejas ghatte seems not to be a GitHub user. You need a GitHub account to be able to sign the CLA. If you have already a GitHub account, please add the email address used for this commit to your account.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}

[greptile-apps]

Confidence Score: 5/5

This PR is safe to merge; all behavioural changes are backed by matching test updates and the underlying logic is consistent with the existing data-model conventions.

Every fix is localised and well-understood: the Azure endpoint panic fix is guarded correctly, the Bedrock hasToolUse rewrite is consistent with how server-managed tool types are distinguished in the rest of the file, the Gemini passthrough and role corrections are tested end-to-end, and the migration script additions follow the established pattern for log-store tables. No unguarded code paths or data-model mismatches were found.

No files require special attention.

Important Files Changed

Filename	Overview
core/providers/bedrock/responses.go	Replaces the unmatched-toolUse content-block scan with a direct check on bifrostResp.Output for ResponsesMessageTypeFunctionCall; correct because server-managed tools produce WebSearchCall/CodeInterpreterCall types, not FunctionCall.
framework/configstore/clientconfig.go	Fixes nil-pointer panic in Azure endpoint redaction by guarding the Redacted() call with IsFromEnv(); plain endpoint values are now passed through unchanged.
core/providers/anthropic/types.go	Adds WebSearchNova and CodeExecNova fields to ProviderFeatureSupport and sets them for Bedrock instead of the Chat/Converse-scoped WebSearch/CodeExecution flags.
core/providers/anthropic/utils.go	ValidateToolsForProvider updated to accept web_search/code_interpreter when either the Chat or Nova (Responses-path) feature flag is set.
core/providers/gemini/gemini_test.go	Tests updated to ParametersJSONSchema passthrough and corrected tool-response role; property-ordering assertions weakened to presence-only checks (previously noted).
.github/workflows/scripts/run-migration-tests.sh	Adds v1.5.4 column blocks for PostgreSQL and SQLite; logs.inc_number follows the established unconditional-emission pattern for log-store tables (consistent with v1.5.3 log columns).

_{Reviews (2): Last reviewed commit: "test fixes" | Re-trigger Greptile}

[coderabbitai]

Actionable comments posted: 1

🧹 Nitpick comments (1)

core/providers/gemini/gemini_test.go (1)

735-739: ⚡ Quick win

Assert that empty items stays an object, not just a present key.

assert.Contains(..., "items") still passes for "items": null or any other non-object value. These regressions would slip through even though the edge case here is specifically preserving an empty schema object. Decode items as map[string]interface{} and assert it is empty.

Also applies to: 2609-2612

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@core/providers/gemini/gemini_test.go` around lines 735 - 739, The test
currently only asserts the "items" key exists (using assert.Contains) which
allows non-object values like null; update the assertion to decode
dataProp["items"] into a map[string]interface{} (using a type assertion or json
decoding) and assert that the resulting map is non-nil and has length 0 to
guarantee it's an empty object; apply the same change to the other occurrences
mentioned (the similar assertions around lines 2609-2612) and keep references to
parseToolParams, getSchemaProperty, and the local variable dataProp to locate
the checks.

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@core/providers/gemini/gemini_test.go`:
- Around line 435-445: The tests currently lose property ordering because
parseToolParams marshals then unmarshals into map[string]interface{}, which
discards key order; change parseToolParams (and the PropertyOrdering tests) to
preserve ordering by returning the marshaled JSON bytes (or
json.RawMessage/string) from json.Marshal(fd.ParametersJSONSchema) instead of
unmarshalling to a map, and update the assertions to compare the marshaled bytes
(or normalized compact string) against the expected ordered JSON so the tests
verify exact key order; references: function parseToolParams and the
PropertyOrdering test blocks that currently call it.

---

Nitpick comments:
In `@core/providers/gemini/gemini_test.go`:
- Around line 735-739: The test currently only asserts the "items" key exists
(using assert.Contains) which allows non-object values like null; update the
assertion to decode dataProp["items"] into a map[string]interface{} (using a
type assertion or json decoding) and assert that the resulting map is non-nil
and has length 0 to guarantee it's an empty object; apply the same change to the
other occurrences mentioned (the similar assertions around lines 2609-2612) and
keep references to parseToolParams, getSchemaProperty, and the local variable
dataProp to locate the checks.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

Run ID: 9b0c2dcf-8e4d-42f9-8b86-6560a01b3b5b

📥 Commits

Reviewing files that changed from the base of the PR and between aacdd98 and 03e8779.

📒 Files selected for processing (8)

• .github/workflows/scripts/run-migration-tests.sh
• core/providers/anthropic/compaction_test.go
• core/providers/anthropic/utils_test.go
• core/providers/bedrock/responses.go
• core/providers/gemini/gemini_test.go
• core/providers/gemini/uniontype_test.go
• framework/configstore/clientconfig.go
• plugins/jsonparser/plugin_test.go

[coderabbitai]

Actionable comments posted: 1

♻️ Duplicate comments (1)

core/providers/gemini/gemini_test.go (1)

435-445: ⚠️ Potential issue | 🟠 Major | ⚡ Quick win

Property-ordering tests currently can’t verify ordering.

parseToolParams converts schema into map[string]interface{}, which drops key order before the ordering tests run. The assertions at Line 1015 and Line 1063 only prove presence, not order preservation.

Proposed minimal fix

-func parseToolParams(t *testing.T, fd *gemini.FunctionDeclaration) map[string]interface{} {
+func parseToolParamsRaw(t *testing.T, fd *gemini.FunctionDeclaration) []byte {
 	t.Helper()
 	require.NotNil(t, fd.ParametersJSONSchema, "ParametersJSONSchema must be set")
 	raw, err := json.Marshal(fd.ParametersJSONSchema)
 	require.NoError(t, err)
-	var m map[string]interface{}
-	require.NoError(t, json.Unmarshal(raw, &m))
-	return m
+	return raw
+}
+
+func parseToolParams(t *testing.T, fd *gemini.FunctionDeclaration) map[string]interface{} {
+	t.Helper()
+	raw := parseToolParamsRaw(t, fd)
+	var m map[string]interface{}
+	require.NoError(t, json.Unmarshal(raw, &m))
+	return m
}

-params := parseToolParams(t, fd)
-props, ok := params["properties"].(map[string]interface{})
-require.True(t, ok, "parameters must have properties")
-assert.Len(t, props, 3)
-assert.Contains(t, props, "chain_of_thought")
-assert.Contains(t, props, "answer")
-assert.Contains(t, props, "citations")
+raw := string(parseToolParamsRaw(t, fd))
+idxChain := strings.Index(raw, `"chain_of_thought"`)
+idxAnswer := strings.Index(raw, `"answer"`)
+idxCitations := strings.Index(raw, `"citations"`)
+require.NotEqual(t, -1, idxChain)
+require.NotEqual(t, -1, idxAnswer)
+require.NotEqual(t, -1, idxCitations)
+assert.True(t, idxChain < idxAnswer && idxAnswer < idxCitations, "property order must be preserved")

Also applies to: 1015-1021, 1063-1075

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@core/providers/gemini/gemini_test.go` around lines 435 - 445, The tests lose
property ordering because parseToolParams currently returns a map
(map[string]interface{}) which drops key order; change parseToolParams (and its
callers) to return an ordered representation (e.g., []string of property names
in the schema) by marshaling fd.ParametersJSONSchema and using json.NewDecoder
to stream/iterate tokens and collect object keys in their original order
(specifically iterate into the top-level "properties" object and append each
property key to a slice). Update tests that assert ordering to consume the new
ordered slice instead of a map.

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@core/providers/bedrock/responses.go`:
- Around line 2595-2597: The tool-use detection currently only sets hasToolUse
when msg.Type equals schemas.ResponsesMessageTypeFunctionCall; update the check
in the loop over bifrostResp.Output to also consider
schemas.ResponsesMessageTypeWebSearchCall and
schemas.ResponsesMessageTypeCodeInterpreterCall (e.g., treat any of those three
types as tool use), so hasToolUse is true for any of those message types and the
stop reason will be correctly set to "tool_use". Ensure you change the
conditional that references ResponsesMessageTypeFunctionCall to check for all
three constants (or use a small switch/lookup) where hasToolUse is assigned.

---

Duplicate comments:
In `@core/providers/gemini/gemini_test.go`:
- Around line 435-445: The tests lose property ordering because parseToolParams
currently returns a map (map[string]interface{}) which drops key order; change
parseToolParams (and its callers) to return an ordered representation (e.g.,
[]string of property names in the schema) by marshaling fd.ParametersJSONSchema
and using json.NewDecoder to stream/iterate tokens and collect object keys in
their original order (specifically iterate into the top-level "properties"
object and append each property key to a slice). Update tests that assert
ordering to consume the new ordered slice instead of a map.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

Run ID: fe348e7e-6405-45db-ace1-79d015046933

📥 Commits

Reviewing files that changed from the base of the PR and between 03e8779 and 519e4f7.

📒 Files selected for processing (12)

• .github/workflows/scripts/run-migration-tests.sh
• core/internal/llmtests/provider_feature_support_test.go
• core/providers/anthropic/compaction_test.go
• core/providers/anthropic/requestbuilder_test.go
• core/providers/anthropic/types.go
• core/providers/anthropic/utils.go
• core/providers/anthropic/utils_test.go
• core/providers/bedrock/responses.go
• core/providers/gemini/gemini_test.go
• core/providers/gemini/uniontype_test.go
• framework/configstore/clientconfig.go
• plugins/jsonparser/plugin_test.go

✅ Files skipped from review due to trivial changes (1)

• core/providers/anthropic/compaction_test.go

[akshaydeo]

Merge activity

• May 26, 3:29 PM UTC: A user started a stack merge that includes this pull request via Graphite.
• May 26, 3:30 PM UTC: @akshaydeo merged this pull request with Graphite.