2026-05-05 05:50:01 +02:00
|
|
|
"""ExtractionService coverage — pattern-based message classification."""
|
|
|
|
|
|
|
|
|
|
from __future__ import annotations
|
|
|
|
|
|
2026-05-06 21:02:31 +02:00
|
|
|
from typing import ClassVar
|
2026-05-05 05:50:01 +02:00
|
|
|
from uuid import uuid4
|
|
|
|
|
|
2026-05-06 21:02:31 +02:00
|
|
|
import anthropic as anthropic_mod
|
2026-05-05 05:50:01 +02:00
|
|
|
import pytest
|
2026-05-06 21:02:31 +02:00
|
|
|
from roboco.llm import ToonAdapter
|
2026-05-05 05:50:01 +02:00
|
|
|
from roboco.models import MessageType
|
|
|
|
|
from roboco.models.extraction import (
|
|
|
|
|
ExtractionConfig,
|
|
|
|
|
ExtractionContext,
|
|
|
|
|
)
|
|
|
|
|
from roboco.services.extraction import ExtractionPipeline, ExtractionService
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
def _ctx(content: str) -> ExtractionContext:
|
|
|
|
|
return ExtractionContext(
|
|
|
|
|
content=content,
|
|
|
|
|
agent_id=uuid4(),
|
|
|
|
|
channel_id=uuid4(),
|
|
|
|
|
session_id=uuid4(),
|
|
|
|
|
group_id=uuid4(),
|
|
|
|
|
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
@pytest.fixture
|
|
|
|
|
def svc() -> ExtractionService:
|
|
|
|
|
return ExtractionService()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
# ---------------------------------------------------------------------------
|
|
|
|
|
# Empty / short content
|
|
|
|
|
# ---------------------------------------------------------------------------
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
|
|
|
async def test_extract_returns_empty_for_short_content(
|
|
|
|
|
svc: ExtractionService,
|
|
|
|
|
) -> None:
|
|
|
|
|
result = await svc.extract(_ctx("hi"))
|
|
|
|
|
assert result.messages == []
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
|
|
|
async def test_extract_handles_empty_segments(
|
|
|
|
|
svc: ExtractionService,
|
|
|
|
|
) -> None:
|
|
|
|
|
"""Whitespace-only segments are skipped."""
|
|
|
|
|
result = await svc.extract(_ctx("\n\n \n\n \n\n"))
|
|
|
|
|
assert result.messages == []
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
# ---------------------------------------------------------------------------
|
|
|
|
|
# Pattern classification
|
|
|
|
|
# ---------------------------------------------------------------------------
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
|
|
|
async def test_classifies_reasoning(svc: ExtractionService) -> None:
|
|
|
|
|
result = await svc.extract(
|
|
|
|
|
_ctx("I'm thinking about how to solve this problem efficiently.")
|
|
|
|
|
)
|
|
|
|
|
assert len(result.messages) >= 1
|
|
|
|
|
assert MessageType.REASONING in result.types_extracted
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
|
|
|
async def test_classifies_dialogue(svc: ExtractionService) -> None:
|
|
|
|
|
result = await svc.extract(_ctx("Hey, can someone help me debug this issue?"))
|
|
|
|
|
assert len(result.messages) >= 1
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
|
|
|
async def test_classifies_decision(svc: ExtractionService) -> None:
|
|
|
|
|
result = await svc.extract(
|
|
|
|
|
_ctx("Decision: I will use the async pattern for this case.")
|
|
|
|
|
)
|
|
|
|
|
assert len(result.messages) >= 1
|
|
|
|
|
assert MessageType.DECISION in result.types_extracted
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
|
|
|
async def test_classifies_action(svc: ExtractionService) -> None:
|
|
|
|
|
result = await svc.extract(_ctx("Starting the deployment process now."))
|
|
|
|
|
assert len(result.messages) >= 1
|
|
|
|
|
assert MessageType.ACTION in result.types_extracted
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
|
|
|
async def test_classifies_blocker(svc: ExtractionService) -> None:
|
|
|
|
|
result = await svc.extract(_ctx("Blocked: waiting for QA to review the PR."))
|
|
|
|
|
assert len(result.messages) >= 1
|
|
|
|
|
assert MessageType.BLOCKER in result.types_extracted
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
|
|
|
async def test_classifies_technical_with_code_block(
|
|
|
|
|
svc: ExtractionService,
|
|
|
|
|
) -> None:
|
|
|
|
|
content = "Here is the implementation:\n\n```python\ndef foo(): pass\n```"
|
|
|
|
|
result = await svc.extract(_ctx(content))
|
|
|
|
|
assert len(result.messages) >= 1
|
|
|
|
|
assert MessageType.TECHNICAL in result.types_extracted
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
|
|
|
async def test_unmatched_defaults_to_reasoning(svc: ExtractionService) -> None:
|
|
|
|
|
"""Content without recognizable patterns falls back to REASONING."""
|
|
|
|
|
result = await svc.extract(_ctx("xyzzy plugh fnord arglebargle"))
|
|
|
|
|
assert len(result.messages) >= 1
|
|
|
|
|
# Default classification is REASONING.
|
|
|
|
|
assert any(m.type == MessageType.REASONING for m in result.messages)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
# ---------------------------------------------------------------------------
|
|
|
|
|
# Segmentation
|
|
|
|
|
# ---------------------------------------------------------------------------
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
|
|
|
async def test_extract_splits_on_double_newlines(
|
|
|
|
|
svc: ExtractionService,
|
|
|
|
|
) -> None:
|
|
|
|
|
content = "First paragraph here.\n\nSecond paragraph here."
|
|
|
|
|
result = await svc.extract(_ctx(content))
|
2026-05-06 21:02:31 +02:00
|
|
|
_PARAS = 2
|
|
|
|
|
assert len(result.messages) == _PARAS
|
2026-05-05 05:50:01 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
|
|
|
async def test_extract_keeps_code_blocks_intact(svc: ExtractionService) -> None:
|
|
|
|
|
content = (
|
2026-05-06 00:32:52 +02:00
|
|
|
"Some explanation.\n\n```\nline1\nline2\nline3\n```\n\nSome more explanation."
|
2026-05-05 05:50:01 +02:00
|
|
|
)
|
|
|
|
|
result = await svc.extract(_ctx(content))
|
2026-05-06 00:32:52 +02:00
|
|
|
code_segments = [m for m in result.messages if m.content.startswith("```")]
|
2026-05-05 05:50:01 +02:00
|
|
|
assert len(code_segments) >= 1
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
# ---------------------------------------------------------------------------
|
|
|
|
|
# Config respect
|
|
|
|
|
# ---------------------------------------------------------------------------
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
|
|
|
async def test_extract_respects_max_segments() -> None:
|
|
|
|
|
svc = ExtractionService(ExtractionConfig(max_segments_per_buffer=2))
|
|
|
|
|
content = "First.\n\nSecond.\n\nThird.\n\nFourth.\n\nFifth."
|
2026-05-06 21:02:31 +02:00
|
|
|
_PARAS = 2
|
2026-05-05 05:50:01 +02:00
|
|
|
result = await svc.extract(_ctx(content))
|
2026-05-06 21:02:31 +02:00
|
|
|
assert len(result.messages) <= _PARAS
|
2026-05-05 05:50:01 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
|
|
|
async def test_extract_respects_min_content_length() -> None:
|
|
|
|
|
svc = ExtractionService(ExtractionConfig(min_content_length=100))
|
|
|
|
|
result = await svc.extract(_ctx("Short message only."))
|
|
|
|
|
assert result.messages == []
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
# ---------------------------------------------------------------------------
|
|
|
|
|
# Result metadata
|
|
|
|
|
# ---------------------------------------------------------------------------
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
|
|
|
async def test_extract_populates_confidence_scores(
|
|
|
|
|
svc: ExtractionService,
|
|
|
|
|
) -> None:
|
|
|
|
|
result = await svc.extract(_ctx("I'm thinking carefully about this."))
|
|
|
|
|
assert result.confidence_scores
|
|
|
|
|
for score in result.confidence_scores.values():
|
|
|
|
|
assert 0.0 <= score <= 1.0
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
|
|
|
async def test_extract_records_pattern_matches(
|
|
|
|
|
svc: ExtractionService,
|
|
|
|
|
) -> None:
|
|
|
|
|
result = await svc.extract(_ctx("Decision: going with option A."))
|
|
|
|
|
assert result.pattern_matches # Non-empty.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
# ---------------------------------------------------------------------------
|
|
|
|
|
# Pipeline
|
|
|
|
|
# ---------------------------------------------------------------------------
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
|
|
|
async def test_pipeline_invokes_callback() -> None:
|
|
|
|
|
pipeline = ExtractionPipeline()
|
|
|
|
|
received: list = []
|
|
|
|
|
|
|
|
|
|
async def on_message(msg) -> None:
|
|
|
|
|
received.append(msg)
|
|
|
|
|
|
|
|
|
|
pipeline.on_message(on_message)
|
|
|
|
|
result = await pipeline.process_buffer(
|
|
|
|
|
_ctx("First message here.\n\nSecond message here.")
|
|
|
|
|
)
|
|
|
|
|
assert result.message_count >= 1
|
|
|
|
|
assert len(received) == result.message_count
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
|
|
|
async def test_pipeline_swallows_callback_errors() -> None:
|
|
|
|
|
"""Callback failure should not abort the pipeline."""
|
|
|
|
|
pipeline = ExtractionPipeline()
|
|
|
|
|
|
2026-05-06 21:02:31 +02:00
|
|
|
async def bad_callback(_msg) -> None:
|
2026-05-05 05:50:01 +02:00
|
|
|
raise RuntimeError("boom")
|
|
|
|
|
|
|
|
|
|
pipeline.on_message(bad_callback)
|
|
|
|
|
# Should complete without raising despite callback error.
|
|
|
|
|
result = await pipeline.process_buffer(_ctx("Hello there.\n\nGoodbye."))
|
|
|
|
|
assert result is not None
|
2026-05-06 21:02:31 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
# ---------------------------------------------------------------------------
|
|
|
|
|
# Mentions extraction (line 209)
|
|
|
|
|
# ---------------------------------------------------------------------------
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
|
|
|
async def test_extract_logs_mentions_found(svc: ExtractionService) -> None:
|
|
|
|
|
"""When @mentions are present, the debug branch fires (line 209)."""
|
|
|
|
|
content = "@be-dev-1 can you take a look at this code please?"
|
|
|
|
|
result = await svc.extract(_ctx(content))
|
|
|
|
|
assert len(result.messages) >= 1
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
|
|
|
async def test_extract_skips_empty_segments_from_segmenter(
|
|
|
|
|
svc: ExtractionService, monkeypatch: pytest.MonkeyPatch
|
|
|
|
|
) -> None:
|
|
|
|
|
"""If _segment_content yields a whitespace-only segment, it's skipped.
|
|
|
|
|
|
|
|
|
|
Triggers the defensive `if not segment.strip(): continue` (line 193).
|
|
|
|
|
"""
|
|
|
|
|
monkeypatch.setattr(
|
|
|
|
|
svc, "_segment_content", lambda _content: [" ", "Real content here."]
|
|
|
|
|
)
|
|
|
|
|
result = await svc.extract(_ctx("Long enough content to bypass min length."))
|
|
|
|
|
# Only the non-empty segment becomes a message.
|
|
|
|
|
assert len(result.messages) == 1
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
# ---------------------------------------------------------------------------
|
|
|
|
|
# extract_with_llm (lines 323-404)
|
|
|
|
|
# ---------------------------------------------------------------------------
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
|
|
|
async def test_extract_with_llm_falls_back_on_error(
|
|
|
|
|
svc: ExtractionService, monkeypatch: pytest.MonkeyPatch
|
|
|
|
|
) -> None:
|
|
|
|
|
"""If the Anthropic call fails, falls back to pattern-based extract."""
|
|
|
|
|
|
|
|
|
|
class _BadClient:
|
|
|
|
|
def __init__(self, **_kwargs: object) -> None:
|
|
|
|
|
self.messages = self
|
|
|
|
|
|
|
|
|
|
async def create(self, **_kwargs: object) -> object:
|
|
|
|
|
raise RuntimeError("API down")
|
|
|
|
|
|
|
|
|
|
monkeypatch.setattr(anthropic_mod, "AsyncAnthropic", _BadClient)
|
|
|
|
|
result = await svc.extract_with_llm(_ctx("I'm thinking about this."))
|
|
|
|
|
# Pattern fallback path executes; messages list is not empty.
|
|
|
|
|
assert result is not None
|
|
|
|
|
assert len(result.messages) >= 1
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
|
|
|
async def test_extract_with_llm_parses_response(
|
|
|
|
|
svc: ExtractionService, monkeypatch: pytest.MonkeyPatch
|
|
|
|
|
) -> None:
|
|
|
|
|
"""Mock a successful LLM response and verify it gets parsed."""
|
|
|
|
|
|
|
|
|
|
class _Block:
|
|
|
|
|
text = "json-or-toon"
|
|
|
|
|
|
|
|
|
|
class _Resp:
|
|
|
|
|
content: ClassVar = [_Block()]
|
|
|
|
|
|
|
|
|
|
class _OkClient:
|
|
|
|
|
def __init__(self, **_kwargs: object) -> None:
|
|
|
|
|
self.messages = self
|
|
|
|
|
|
|
|
|
|
async def create(self, **_kwargs: object) -> _Resp:
|
|
|
|
|
return _Resp()
|
|
|
|
|
|
|
|
|
|
def _decode_dicts(_self: ToonAdapter, _text: str) -> list[dict[str, object]]:
|
|
|
|
|
return [
|
|
|
|
|
{"type": "reasoning", "content": "thinking", "confidence": 0.9},
|
|
|
|
|
{"type": "action", "content": "doing", "confidence": 0.95},
|
|
|
|
|
]
|
|
|
|
|
|
|
|
|
|
monkeypatch.setattr(anthropic_mod, "AsyncAnthropic", _OkClient)
|
|
|
|
|
monkeypatch.setattr(ToonAdapter, "decode", _decode_dicts)
|
|
|
|
|
|
|
|
|
|
result = await svc.extract_with_llm(_ctx("Some content for the LLM."))
|
|
|
|
|
assert result is not None
|
|
|
|
|
_COUNT = 2
|
|
|
|
|
assert len(result.messages) == _COUNT
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
|
|
|
async def test_extract_with_llm_handles_non_dict_segments(
|
|
|
|
|
svc: ExtractionService, monkeypatch: pytest.MonkeyPatch
|
|
|
|
|
) -> None:
|
|
|
|
|
"""When the toon decoder returns non-dict segments, fallback type."""
|
|
|
|
|
|
|
|
|
|
class _Block:
|
|
|
|
|
text = "raw response"
|
|
|
|
|
|
|
|
|
|
class _Resp:
|
|
|
|
|
content: ClassVar = [_Block()]
|
|
|
|
|
|
|
|
|
|
class _OkClient:
|
|
|
|
|
def __init__(self, **_kwargs: object) -> None:
|
|
|
|
|
self.messages = self
|
|
|
|
|
|
|
|
|
|
async def create(self, **_kwargs: object) -> _Resp:
|
|
|
|
|
return _Resp()
|
|
|
|
|
|
|
|
|
|
# Patch toon to yield strings instead of dicts.
|
|
|
|
|
def _decode_strings(_self: ToonAdapter, _text: str) -> list[str]:
|
|
|
|
|
return ["plain string segment", "another"]
|
|
|
|
|
|
|
|
|
|
monkeypatch.setattr(anthropic_mod, "AsyncAnthropic", _OkClient)
|
|
|
|
|
monkeypatch.setattr(ToonAdapter, "decode", _decode_strings)
|
|
|
|
|
|
|
|
|
|
result = await svc.extract_with_llm(_ctx("Some content for the LLM."))
|
|
|
|
|
assert result is not None
|
|
|
|
|
_COUNT = 2
|
|
|
|
|
assert len(result.messages) == _COUNT
|