feat: implement knowledge management system (spec complete, all 53 tasks done)

2026-07-01 14:14:16 +02:00
parent e4256d3222
commit 8ac664d33d
89 changed files with 19759 additions and 72 deletions
@@ -0,0 +1,542 @@
+"""Tests for EnrichmentAgent – KI-gestützte Anreicherung.
+
+Requirements: 9.1, 9.2, 9.3, 9.4, 9.5, 9.7
+"""
+
+from __future__ import annotations
+
+import pytest
+
+from monorepo.knowledge.ingestion.enrichment import (
+    CONFIDENCE_THRESHOLD,
+    ActionItem,
+    EnrichmentAgent,
+    EnrichmentResult,
+    _compute_action_item_hash,
+    _detect_category,
+    _extract_action_items,
+    _extract_people,
+    _extract_projects,
+    _extract_title,
+    _slugify,
+    generate_action_items_section,
+    generate_wiki_links,
+)
+
+
+# ---------------------------------------------------------------------------
+# _slugify tests
+# ---------------------------------------------------------------------------
+
+
+class TestSlugify:
+    """Tests for _slugify helper."""
+
+    def test_simple_name(self) -> None:
+        assert _slugify("Max Müller") == "max-mueller"
+
+    def test_umlaut_handling(self) -> None:
+        assert _slugify("André Knie") == "andre-knie"
+
+    def test_sharp_s(self) -> None:
+        assert _slugify("Hans Straße") == "hans-strasse"
+
+    def test_multiple_spaces(self) -> None:
+        assert _slugify("Max  Müller") == "max-mueller"
+
+    def test_leading_trailing_spaces(self) -> None:
+        assert _slugify("  Max Müller  ") == "max-mueller"
+
+    def test_single_word(self) -> None:
+        assert _slugify("Monorepo") == "monorepo"
+
+    def test_special_characters(self) -> None:
+        assert _slugify("Projekt (v2.0)") == "projekt-v2-0"
+
+
+# ---------------------------------------------------------------------------
+# _compute_action_item_hash tests
+# ---------------------------------------------------------------------------
+
+
+class TestComputeActionItemHash:
+    """Tests for _compute_action_item_hash."""
+
+    def test_deterministic(self) -> None:
+        h1 = _compute_action_item_hash("Do the thing")
+        h2 = _compute_action_item_hash("Do the thing")
+        assert h1 == h2
+
+    def test_case_insensitive(self) -> None:
+        h1 = _compute_action_item_hash("Do the thing")
+        h2 = _compute_action_item_hash("do the thing")
+        assert h1 == h2
+
+    def test_trims_whitespace(self) -> None:
+        h1 = _compute_action_item_hash("Do the thing")
+        h2 = _compute_action_item_hash("  Do the thing  ")
+        assert h1 == h2
+
+    def test_different_descriptions_differ(self) -> None:
+        h1 = _compute_action_item_hash("Do thing A")
+        h2 = _compute_action_item_hash("Do thing B")
+        assert h1 != h2
+
+    def test_returns_16_chars(self) -> None:
+        h = _compute_action_item_hash("some task")
+        assert len(h) == 16
+
+
+# ---------------------------------------------------------------------------
+# _detect_category tests
+# ---------------------------------------------------------------------------
+
+
+class TestDetectCategory:
+    """Tests for _detect_category."""
+
+    def test_meeting_keywords(self) -> None:
+        content = "Meeting Protokoll vom 15.01.2025\nTeilnehmer: Max, Anna"
+        category, confidence = _detect_category(content)
+        assert category == "meeting"
+        assert confidence >= 0.5
+
+    def test_decision_keywords(self) -> None:
+        content = "Entscheidung: Wir haben beschlossen, die API umzubauen."
+        category, _ = _detect_category(content)
+        assert category == "decision"
+
+    def test_project_keywords(self) -> None:
+        content = "Projekt Roadmap Q1 2025\nMeilenstein 1: Release Sprint 3"
+        category, _ = _detect_category(content)
+        assert category == "project"
+
+    def test_reference_keywords(self) -> None:
+        content = "Dokumentation: How-To Guide für den Deployment-Prozess"
+        category, _ = _detect_category(content)
+        assert category == "reference"
+
+    def test_link_detection(self) -> None:
+        content = "https://docs.python.org/3/library/re.html"
+        category, confidence = _detect_category(content)
+        assert category == "link"
+        assert confidence >= 0.8
+
+    def test_unknown_content_defaults_inbox(self) -> None:
+        content = "Einfach nur ein Gedanke."
+        category, confidence = _detect_category(content)
+        assert category == "inbox"
+        assert confidence < CONFIDENCE_THRESHOLD
+
+    def test_source_type_hint(self) -> None:
+        content = "Some content about things"
+        category, _ = _detect_category(content, source_type="jira")
+        assert category == "project"
+
+
+# ---------------------------------------------------------------------------
+# _extract_people tests
+# ---------------------------------------------------------------------------
+
+
+class TestExtractPeople:
+    """Tests for _extract_people."""
+
+    def test_at_mention(self) -> None:
+        content = "Assigned to @Max Müller for review."
+        people = _extract_people(content)
+        names = [n for n, _ in people]
+        assert any("Max" in n for n in names)
+
+    def test_name_pattern(self) -> None:
+        content = "Besprochen mit André Knie und Hans Schmidt."
+        people = _extract_people(content)
+        names = [n for n, _ in people]
+        assert any("André Knie" in n for n in names)
+        assert any("Hans Schmidt" in n for n in names)
+
+    def test_excludes_stop_names(self) -> None:
+        content = "Meeting Notes und Action Items besprochen."
+        people = _extract_people(content)
+        names = [n for n, _ in people]
+        assert "Action Items" not in names
+        assert "Meeting Notes" not in names
+
+    def test_empty_content(self) -> None:
+        people = _extract_people("")
+        assert people == []
+
+    def test_at_mention_high_confidence(self) -> None:
+        content = "Feedback von @André eingereicht."
+        people = _extract_people(content)
+        assert any(conf >= 0.9 for _, conf in people)
+
+
+# ---------------------------------------------------------------------------
+# _extract_projects tests
+# ---------------------------------------------------------------------------
+
+
+class TestExtractProjects:
+    """Tests for _extract_projects."""
+
+    def test_explicit_project_reference(self) -> None:
+        content = "Betrifft Projekt Orchestrator-Monorepo."
+        projects = _extract_projects(content)
+        names = [n for n, _ in projects]
+        assert any("Orchestrator-Monorepo" in n for n in names)
+
+    def test_english_project_reference(self) -> None:
+        content = "This relates to Project Knowledge-Pipeline."
+        projects = _extract_projects(content)
+        names = [n for n, _ in projects]
+        assert any("Knowledge-Pipeline" in n for n in names)
+
+    def test_existing_wiki_link(self) -> None:
+        content = "See [[projects/my-project]] for details."
+        projects = _extract_projects(content)
+        names = [n for n, _ in projects]
+        assert "my-project" in names
+
+    def test_empty_content(self) -> None:
+        projects = _extract_projects("")
+        assert projects == []
+
+
+# ---------------------------------------------------------------------------
+# _extract_action_items tests
+# ---------------------------------------------------------------------------
+
+
+class TestExtractActionItems:
+    """Tests for _extract_action_items."""
+
+    def test_checkbox_pattern(self) -> None:
+        content = "Tasks:\n- [ ] Finish the report\n- [ ] Send email"
+        items = _extract_action_items(content)
+        assert len(items) == 2
+        assert items[0].description == "Finish the report"
+        assert items[1].description == "Send email"
+
+    def test_todo_pattern(self) -> None:
+        content = "TODO: Review PR #42\nSome other text"
+        items = _extract_action_items(content)
+        assert len(items) == 1
+        assert "Review PR #42" in items[0].description
+
+    def test_action_pattern(self) -> None:
+        content = "ACTION: Schedule follow-up meeting"
+        items = _extract_action_items(content)
+        assert len(items) == 1
+        assert "Schedule follow-up meeting" in items[0].description
+
+    def test_arrow_pattern(self) -> None:
+        content = "→ Create Jira ticket for this"
+        items = _extract_action_items(content)
+        assert len(items) == 1
+        assert "Create Jira ticket" in items[0].description
+
+    def test_deduplication(self) -> None:
+        content = "- [ ] Do the thing\n- [ ] Do the thing"
+        items = _extract_action_items(content)
+        assert len(items) == 1
+
+    def test_empty_content(self) -> None:
+        items = _extract_action_items("")
+        assert items == []
+
+    def test_assignee_detection(self) -> None:
+        content = "- [ ] Review document @André"
+        items = _extract_action_items(content)
+        assert len(items) == 1
+        assert items[0].assignee == "André"
+
+    def test_deadline_detection(self) -> None:
+        content = "- [ ] Submit report bis 2025-03-15"
+        items = _extract_action_items(content)
+        assert len(items) == 1
+        assert items[0].deadline == "2025-03-15"
+
+    def test_hash_auto_computed(self) -> None:
+        content = "TODO: Something important"
+        items = _extract_action_items(content)
+        assert items[0].hash != ""
+        assert len(items[0].hash) == 16
+
+
+# ---------------------------------------------------------------------------
+# _extract_title tests
+# ---------------------------------------------------------------------------
+
+
+class TestExtractTitle:
+    """Tests for _extract_title."""
+
+    def test_markdown_heading(self) -> None:
+        content = "# My Document Title\n\nSome content here."
+        assert _extract_title(content) == "My Document Title"
+
+    def test_h2_heading(self) -> None:
+        content = "## Second Level Heading\n\nContent."
+        assert _extract_title(content) == "Second Level Heading"
+
+    def test_first_line_fallback(self) -> None:
+        content = "Just some text without a heading\nMore text."
+        title = _extract_title(content)
+        assert "Just some text" in title
+
+    def test_empty_content(self) -> None:
+        assert _extract_title("") == "Untitled"
+
+    def test_whitespace_only(self) -> None:
+        assert _extract_title("   \n\n  ") == "Untitled"
+
+    def test_long_first_line_truncated(self) -> None:
+        content = "A" * 200
+        title = _extract_title(content)
+        assert len(title) <= 80
+
+
+# ---------------------------------------------------------------------------
+# generate_wiki_links tests
+# ---------------------------------------------------------------------------
+
+
+class TestGenerateWikiLinks:
+    """Tests for generate_wiki_links."""
+
+    def test_people_links(self) -> None:
+        links = generate_wiki_links(["Max Müller", "André Knie"], [])
+        assert links["people"] == [
+            "[[people/max-mueller]]",
+            "[[people/andre-knie]]",
+        ]
+
+    def test_project_links(self) -> None:
+        links = generate_wiki_links([], ["Knowledge Pipeline"])
+        assert links["projects"] == ["[[projects/knowledge-pipeline]]"]
+
+    def test_empty_lists(self) -> None:
+        links = generate_wiki_links([], [])
+        assert links["people"] == []
+        assert links["projects"] == []
+
+    def test_combined(self) -> None:
+        links = generate_wiki_links(["Hans Schmidt"], ["Orchestrator"])
+        assert "[[people/hans-schmidt]]" in links["people"]
+        assert "[[projects/orchestrator]]" in links["projects"]
+
+
+# ---------------------------------------------------------------------------
+# generate_action_items_section tests
+# ---------------------------------------------------------------------------
+
+
+class TestGenerateActionItemsSection:
+    """Tests for generate_action_items_section."""
+
+    def test_empty_list(self) -> None:
+        assert generate_action_items_section([]) == ""
+
+    def test_single_item(self) -> None:
+        items = [ActionItem(description="Do the thing")]
+        section = generate_action_items_section(items)
+        assert "## Action Items" in section
+        assert "- [ ] Do the thing" in section
+
+    def test_item_with_assignee(self) -> None:
+        items = [ActionItem(description="Review code", assignee="Max")]
+        section = generate_action_items_section(items)
+        assert "@Max" in section
+
+    def test_item_with_deadline(self) -> None:
+        items = [ActionItem(description="Submit report", deadline="2025-03-15")]
+        section = generate_action_items_section(items)
+        assert "bis 2025-03-15" in section
+
+    def test_multiple_items(self) -> None:
+        items = [
+            ActionItem(description="Task A"),
+            ActionItem(description="Task B"),
+            ActionItem(description="Task C"),
+        ]
+        section = generate_action_items_section(items)
+        assert section.count("- [ ]") == 3
+
+
+# ---------------------------------------------------------------------------
+# EnrichmentAgent tests
+# ---------------------------------------------------------------------------
+
+
+class TestEnrichmentAgent:
+    """Tests for EnrichmentAgent class."""
+
+    @pytest.fixture
+    def agent(self) -> EnrichmentAgent:
+        return EnrichmentAgent(provider="kiro")
+
+    def test_default_provider_is_kiro(self) -> None:
+        agent = EnrichmentAgent()
+        assert agent.provider == "kiro"
+
+    def test_provider_available_for_kiro(self) -> None:
+        agent = EnrichmentAgent(provider="kiro")
+        assert agent._provider_available is True
+
+    def test_provider_unavailable_for_litellm(self) -> None:
+        agent = EnrichmentAgent(provider="litellm")
+        assert agent._provider_available is False
+
+    def test_enrich_empty_content(self, agent: EnrichmentAgent) -> None:
+        result = agent.enrich("")
+        assert result.title == "Untitled"
+        assert result.category == "inbox"
+        assert result.confidence == 0.0
+
+    def test_enrich_meeting_content(self, agent: EnrichmentAgent) -> None:
+        content = """# Meeting Protokoll 2025-01-15
+
+Teilnehmer: André Knie, Max Müller
+
+## Agenda
+1. Sprint Review
+2. Nächste Schritte
+
+## Action Items
+- [ ] André erstellt Jira-Ticket
+- [ ] Max reviewed den PR bis 2025-01-20
+"""
+        result = agent.enrich(content)
+        assert result.title == "Meeting Protokoll 2025-01-15"
+        assert result.category == "meeting"
+        assert result.confidence > 0.5
+        assert len(result.action_items) >= 2
+        assert result.enrichment_pending is False
+
+    def test_enrich_detects_people(self, agent: EnrichmentAgent) -> None:
+        content = "Besprochen mit André Knie und Hans Schmidt im Meeting."
+        result = agent.enrich(content)
+        assert len(result.people) >= 1
+
+    def test_enrich_detects_projects(self, agent: EnrichmentAgent) -> None:
+        content = "Betrifft Projekt Orchestrator-Monorepo und die Pipeline."
+        result = agent.enrich(content)
+        assert len(result.projects) >= 1
+
+    def test_enrich_confidence_threshold(self, agent: EnrichmentAgent) -> None:
+        """Entities below confidence threshold are excluded."""
+        content = "Some text mentioning @Max in a clear mention."
+        result = agent.enrich(content)
+        # All returned people must have passed confidence threshold
+        # (tested indirectly: if people are returned, they passed)
+        for person in result.people:
+            assert isinstance(person, str)
+            assert len(person) >= 2
+
+    def test_enrich_pending_when_provider_unavailable(self) -> None:
+        agent = EnrichmentAgent(provider="litellm")
+        result = agent.enrich("Some content to analyze")
+        assert result.enrichment_pending is True
+        assert result.category == "inbox"
+
+    def test_enrich_returns_tags(self, agent: EnrichmentAgent) -> None:
+        content = "# Python API Documentation\n\nUsing Docker and Kubernetes for deployment."
+        result = agent.enrich(content)
+        assert "python" in result.tags or "docker" in result.tags or "kubernetes" in result.tags
+
+    def test_enrich_link_category(self, agent: EnrichmentAgent) -> None:
+        content = "https://docs.python.org/3/library/re.html"
+        result = agent.enrich(content)
+        assert result.category == "link"
+
+
+# ---------------------------------------------------------------------------
+# classify_relevance tests
+# ---------------------------------------------------------------------------
+
+
+class TestClassifyRelevance:
+    """Tests for classify_relevance."""
+
+    @pytest.fixture
+    def agent(self) -> EnrichmentAgent:
+        return EnrichmentAgent(provider="kiro")
+
+    def test_empty_content_zero(self, agent: EnrichmentAgent) -> None:
+        assert agent.classify_relevance("") == 0.0
+
+    def test_whitespace_only_zero(self, agent: EnrichmentAgent) -> None:
+        assert agent.classify_relevance("   ") == 0.0
+
+    def test_short_content_low_score(self, agent: EnrichmentAgent) -> None:
+        score = agent.classify_relevance("Hi")
+        assert score <= 0.3
+
+    def test_structured_content_higher_score(self, agent: EnrichmentAgent) -> None:
+        content = """# Important Document
+
+This is a longer document with:
+- List items
+- More structure
+
+## Section Two
+
+Some [linked content](https://example.com) here.
+
+TODO: Do something important
+"""
+        score = agent.classify_relevance(content)
+        assert score >= 0.6
+
+    def test_score_between_zero_and_one(self, agent: EnrichmentAgent) -> None:
+        content = "Medium length content with some words and information."
+        score = agent.classify_relevance(content)
+        assert 0.0 <= score <= 1.0
+
+
+# ---------------------------------------------------------------------------
+# ActionItem dataclass tests
+# ---------------------------------------------------------------------------
+
+
+class TestActionItem:
+    """Tests for ActionItem dataclass."""
+
+    def test_auto_hash_on_creation(self) -> None:
+        item = ActionItem(description="Do something")
+        assert item.hash != ""
+        assert len(item.hash) == 16
+
+    def test_preserves_explicit_hash(self) -> None:
+        item = ActionItem(description="Do something", hash="custom_hash_val!")
+        assert item.hash == "custom_hash_val!"
+
+    def test_optional_fields_default_none(self) -> None:
+        item = ActionItem(description="Task")
+        assert item.assignee is None
+        assert item.deadline is None
+
+
+# ---------------------------------------------------------------------------
+# EnrichmentResult dataclass tests
+# ---------------------------------------------------------------------------
+
+
+class TestEnrichmentResult:
+    """Tests for EnrichmentResult dataclass."""
+
+    def test_defaults(self) -> None:
+        result = EnrichmentResult(title="Test", category="inbox")
+        assert result.tags == []
+        assert result.people == []
+        assert result.projects == []
+        assert result.action_items == []
+        assert result.confidence == 0.0
+        assert result.enrichment_pending is False
+
+    def test_enrichment_pending_flag(self) -> None:
+        result = EnrichmentResult(
+            title="Test", category="inbox", enrichment_pending=True
+        )
+        assert result.enrichment_pending is True