Trim any-llm development dependencies

[IceCodeNew]

Summary

• replace the any-llm-sdk[all] development extra with the providers and exception adapter used by this project
• keep compatibility tests meaningful when optional provider modules cannot be loaded
• regenerate uv.lock, removing unused provider SDK dependency trees

Why

The development environment installed every optional AnyLLM provider even though weather-briefing only supports DeepSeek, OpenAI, and OpenRouter at runtime. Slimming that dependency group made provider-wide unit tests fail while importing optional SDKs, so the tests now classify every loadable provider and retain the fixed default-header compatibility guard.

Impact

The dev and unittest environments resolve substantially fewer packages without changing production behavior or the Docker dependency group.

Validation

• prek run --all-files
• uv run --with pytest --with pytest-cov -- pytest --cov --cov-branch --cov-report=xml (1254 passed, line coverage 99.90%, branch coverage 99.66%)
• CodeRabbit CLI review (0 issues)

Summary by CodeRabbit

• Bug Fixes

  • Improved validation of provider compatibility and JSON response support.
  • Ensured compatibility checks accurately reflect providers available in the current environment.

• Tests

  • Expanded coverage for dynamically available providers.
  • Added verification for key provider integrations and factory classification.
  • Improved test support documentation and shared testing utilities.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro Plus

Run ID: c3ae1f00-87e1-4e7f-a1bc-02703fae0f3b

📥 Commits

Reviewing files that changed from the base of the PR and between c87b94c and ed89369.

⛔ Files ignored due to path filters (1)

• uv.lock is excluded by !**/*.lock

📒 Files selected for processing (5)

• pyproject.toml
• tests/__init__.py
• tests/any_llm_helpers.py
• tests/test_any_llm_compatibility.py
• tests/test_any_llm_provider.py

---

📝 Walkthrough

Walkthrough

Development dependencies now select specific AnyLLM provider extras. Shared test discovery filters providers by successful SDK imports, and compatibility and factory tests use that runtime-loadable provider set.

Changes

Runtime-loadable AnyLLM provider testing

Layer / File(s)	Summary
Runtime provider setup pyproject.toml, tests/__init__.py, tests/any_llm_helpers.py	The development dependency selects DeepSeek, OpenAI, OpenRouter, and Otari extras, while shared test support discovers importable providers.
Runtime compatibility assertions tests/test_any_llm_compatibility.py, tests/test_any_llm_provider.py	Compatibility checks and factory parametrization use loadable providers, with assertions for blacklist intersections and required development providers.

Estimated code review effort: 2 (Simple) | ~10 minutes

Possibly related PRs

• IceCodeNew/weather-briefing#41: Introduces related AnyLLM provider factory test coverage.
• IceCodeNew/weather-briefing#116: Changes provider extras that affect the runtime-loadable provider set.
• IceCodeNew/weather-briefing#118: Updates related AnyLLM compatibility assertions.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly matches the main change: reducing any-llm development dependencies.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
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.

✨ 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 streamline-dev

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

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

[IceCodeNew]

/agentic_review

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 99.86%. Comparing base (c87b94c) to head (ed89369).
✅ All tests successful. No failed tests found.

Additional details and impacted files

@@           Coverage Diff           @@
##           master     #122   +/-   ##
=======================================
  Coverage   99.86%   99.86%           
=======================================
  Files         115      116    +1     
  Lines       12327    12345   +18     
  Branches      736      737    +1     
=======================================
+ Hits        12310    12328   +18     
  Misses         12       12           
  Partials        5        5           

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

[qodo-code-review]

Code Review by Qodo

🐞 Bugs (0) 📘 Rule violations (0) 📜 Skill insights (0)

Context used

✅ Compliance rules (platform): 46 rules

1. Vacuous provider test ✓ Resolved 🐞 Bug ☼ Reliability

Description

test_factory_classifies_every_loadable_provider is parametrized from _loadable_providers(); when
every AnyLLM.get_provider_class(...) raises ImportError, the list is empty and pytest collects 0
cases, so provider classification is never exercised. This can yield a false-green test run in
environments where provider SDKs are absent/misinstalled.

Code

tests/test_any_llm_provider.py[R414-429]

+def _loadable_providers() -> list[str]:
+    result: list[str] = []
+    for provider in AnyLLM.get_supported_providers():
+        try:
+            AnyLLM.get_provider_class(provider)
+            result.append(provider)
+        except ImportError:
+            pass
+    return result
+
+
@pytest.mark.parametrize(
    "provider",
-    tuple(AnyLLM.get_supported_providers()),
+    _loadable_providers(),
)
-def test_factory_classifies_every_any_llm_provider(monkeypatch, provider: str) -> None:
+def test_factory_classifies_every_loadable_provider(monkeypatch, provider: str) -> None:

Relevance

⭐⭐⭐ High

Repo often strengthens tests to avoid false-green/flaky runs; empty-parametrize guard is a
deterministic reliability fix.

PR-#114
PR-#11
PR-#108

ⓘ Recommendations generated based on similar findings in past PRs

Evidence

_loadable_providers() catches and ignores ImportError, so it can produce an empty provider list;
pytest will then collect no cases for the parametrized test, leaving the classification behavior
untested.

tests/test_any_llm_provider.py[414-429]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

### Issue description
`test_factory_classifies_every_loadable_provider` is currently generated from `_loadable_providers()`. Because `_loadable_providers()` swallows `ImportError`, it can legitimately return an empty list, which causes pytest to collect **zero** cases for the parametrized test. That makes the suite pass without validating provider classification at all.

### Issue Context
This PR intentionally makes optional providers potentially unloadable (missing SDKs). The test should still fail loudly if **nothing** is loadable, otherwise it becomes a vacuous check.

### Fix Focus Areas
- tests/test_any_llm_provider.py[414-429]

### Suggested fix
Add an explicit guard so the suite fails when no providers are loadable. Options:
- Add a separate test like `test_any_llm_has_at_least_one_loadable_provider()` asserting `_loadable_providers()` is non-empty.
- Or, in `_loadable_providers()`, if the result is empty, `pytest.fail("No any-llm providers are loadable in this environment")`.
- If there are specific providers this project requires for tests (e.g., deepseek/openai/openrouter), assert that those are present in the loadable list.

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

2. Duplicated provider helper ✓ Resolved 🐞 Bug ⚙ Maintainability

Description

The provider-loadability filtering logic is duplicated in two test modules as separate
_loadable_providers() functions (with different return types), increasing the chance of future drift
in which providers each suite considers testable. A change to one helper could silently change test
coverage without updating the other.

Code

tests/test_any_llm_compatibility.py[R10-18]

+def _loadable_providers() -> set[str]:
+    result: set[str] = set()
+    for provider in AnyLLM.get_supported_providers():
+        try:
+            AnyLLM.get_provider_class(provider)
+            result.add(provider)
+        except ImportError:
+            pass
+    return result

Relevance

⭐⭐ Medium

Deduplicating test helpers is reasonable, but no close repo precedent; could be seen as optional
refactor.

PR-#114
PR-#108

ⓘ Recommendations generated based on similar findings in past PRs

Evidence

Both modules define their own _loadable_providers() with the same filtering algorithm, which is a
duplication introduced by this PR.

tests/test_any_llm_compatibility.py[10-18]
tests/test_any_llm_provider.py[414-422]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

### Issue description
Two independent `_loadable_providers()` implementations exist in separate test modules. This is duplicated logic and increases divergence risk.

### Issue Context
Both helpers iterate `AnyLLM.get_supported_providers()`, attempt `AnyLLM.get_provider_class(provider)`, and ignore `ImportError`, differing mainly in return type (`set[str]` vs `list[str]`).

### Fix Focus Areas
- tests/test_any_llm_compatibility.py[10-18]
- tests/test_any_llm_provider.py[414-422]

### Suggested fix
Create a shared helper (e.g., `tests/any_llm_test_utils.py` or `tests/conftest.py`) that returns a canonical type like `tuple[str, ...]` or `list[str]`. In call sites:
- Convert to `set(...)` where set operations are needed.
- Use the sequence directly for parametrization.
This keeps provider-filtering behavior consistent across the test suite.

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

[IceCodeNew]

/agentic_review

[qodo-code-review]

PR Summary by Qodo

Trim AnyLLM dev extras and make provider tests import-safe

⚙️ Configuration changes 🧪 Tests 🕐 20-40 Minutes

AI Description

• Slim dev AnyLLM extras to only required provider SDKs and adapter.
• Update provider-wide tests to only assert against importable (loadable) providers.
• Add a guard test ensuring dev deps still include runtime providers (DeepSeek/OpenAI/OpenRouter).

Diagram

graph TD
  A["pyproject.toml (dev extras)"] --> B["AnyLLM SDK (selected providers)"] --> C["tests/any_llm_helpers.py"] --> D["test_any_llm_compatibility.py"]
  C --> E["test_any_llm_provider.py"]

Loading

High-Level Assessment

The following are alternative approaches to this PR:

1. Use pytest.importorskip per provider module

• ➕ More explicit skip reasons per missing provider SDK
• ➕ Keeps individual tests readable when tied to a single provider
• ➖ Harder to keep global invariants (e.g., blacklist equality) meaningful
• ➖ More scattered conditional logic across test files

2. Introduce a dedicated 'test' extra separate from 'dev'

• ➕ Clarifies the minimal dependency set required to run the full test suite
• ➕ Allows an even slimmer 'dev' set if desired
• ➖ Extra packaging surface area to maintain (docs/CI wiring)
• ➖ Can confuse contributors about which extra to install locally

Recommendation: The current approach (central helper that filters to loadable providers and intersects blacklists accordingly) is the best fit for keeping provider-wide invariants non-vacuous while allowing a trimmed dev dependency set. Consider a separate 'test' extra only if you foresee multiple distinct local workflows (e.g., docs/dev vs. CI tests) needing different footprints.

Files changed (5) +35 / -14

Tests (4) +34 / -13

__init__.pyMake tests a package for shared imports +1/-0

Make tests a package for shared imports

• Adds an '__init__.py' so test helpers can be imported via 'tests.*' consistently across the suite.

tests/init.py

any_llm_helpers.pyAdd helper to list loadable AnyLLM providers +15/-0

Add helper to list loadable AnyLLM providers

• Introduces 'loadable_any_llm_providers()' which filters 'AnyLLM.get_supported_providers()' down to providers whose classes can be imported. Missing optional SDKs are handled by catching 'ImportError'.

tests/any_llm_helpers.py

test_any_llm_compatibility.pyMake compatibility assertions non-vacuous under trimmed deps +11/-11

Make compatibility assertions non-vacuous under trimmed deps

• Updates provider-compatibility tests to compute completion providers from the loadable provider set. Provider blacklists are intersected with loadable providers so equality/coverage assertions remain meaningful when optional SDKs are not installed.

tests/test_any_llm_compatibility.py

test_any_llm_provider.pyParametrize factory tests only over loadable providers +7/-2

Parametrize factory tests only over loadable providers

• Adds a guard test asserting the dev environment can load the runtime providers (deepseek/openai/openrouter). Changes parametrization to use only loadable providers and renames the test to reflect the new intent.

tests/test_any_llm_provider.py

Other (1) +1 / -1

pyproject.tomlTrim AnyLLM dev extras to used providers +1/-1

Trim AnyLLM dev extras to used providers

• Replaces the broad 'any-llm-sdk[all]' dev extra with a narrower set of provider extras used by this project (plus the adapter dependency). This reduces transitive SDK installs in development without affecting the docker/runtime dependency group.

pyproject.toml

[qodo-code-review]

Code review by qodo was updated up to the latest commit ed89369