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 +1,319 @@
+"""Unit-Tests für monorepo.knowledge.ingestion.config.
+
+Testet SourceConfig, PipelineSettings, Env-Var-Auflösung,
+Validierung und sources.yaml-Parsing.
+"""
+
+from __future__ import annotations
+
+import os
+from pathlib import Path
+from textwrap import dedent
+
+import pytest
+
+from monorepo.knowledge.ingestion.config import (
+    ConfigValidationError,
+    EnrichmentSettings,
+    GitSettings,
+    OCRSettings,
+    PipelineSettings,
+    SourceConfig,
+    load_sources_config,
+    resolve_env_vars,
+)
+
+
+# ---------------------------------------------------------------------------
+# SourceConfig Dataclass
+# ---------------------------------------------------------------------------
+
+
+class TestSourceConfig:
+    """Tests für die SourceConfig-Dataclass."""
+
+    def test_required_fields(self) -> None:
+        cfg = SourceConfig(type="confluence", name="Team Wiki")
+        assert cfg.type == "confluence"
+        assert cfg.name == "Team Wiki"
+
+    def test_defaults(self) -> None:
+        cfg = SourceConfig(type="jira", name="Sprint Issues")
+        assert cfg.params == {}
+        assert cfg.target_folder == ""
+        assert cfg.sync_frequency == "manual"
+        assert cfg.enabled is True
+
+    def test_full_config(self) -> None:
+        cfg = SourceConfig(
+            type="confluence",
+            name="DB Confluence",
+            params={"base_url": "https://confluence.bahn.de", "spaces": ["ACV2"]},
+            target_folder="references",
+            sync_frequency="daily",
+            enabled=True,
+        )
+        assert cfg.params["base_url"] == "https://confluence.bahn.de"
+        assert cfg.target_folder == "references"
+        assert cfg.sync_frequency == "daily"
+
+
+# ---------------------------------------------------------------------------
+# PipelineSettings Dataclass
+# ---------------------------------------------------------------------------
+
+
+class TestPipelineSettings:
+    """Tests für PipelineSettings und Unter-Dataclasses."""
+
+    def test_defaults(self) -> None:
+        settings = PipelineSettings()
+        assert settings.create_tasks is True
+        assert settings.enrichment.provider == "kiro"
+        assert settings.enrichment.confidence_threshold == 0.7
+        assert settings.ocr.provider == "kiro"
+        assert settings.ocr.languages == "deu+eng"
+        assert settings.git.auto_commit is True
+
+    def test_custom_settings(self) -> None:
+        settings = PipelineSettings(
+            create_tasks=False,
+            enrichment=EnrichmentSettings(provider="litellm", model="gpt-4"),
+            ocr=OCRSettings(provider="tesseract", languages="deu"),
+            git=GitSettings(auto_commit=False),
+        )
+        assert settings.create_tasks is False
+        assert settings.enrichment.provider == "litellm"
+        assert settings.enrichment.model == "gpt-4"
+        assert settings.ocr.provider == "tesseract"
+        assert settings.git.auto_commit is False
+
+
+# ---------------------------------------------------------------------------
+# resolve_env_vars
+# ---------------------------------------------------------------------------
+
+
+class TestResolveEnvVars:
+    """Tests für die Umgebungsvariablen-Auflösung."""
+
+    def test_no_vars(self) -> None:
+        assert resolve_env_vars("hello world") == "hello world"
+
+    def test_single_var(self, monkeypatch: pytest.MonkeyPatch) -> None:
+        monkeypatch.setenv("MY_TOKEN", "secret123")
+        assert resolve_env_vars("${MY_TOKEN}") == "secret123"
+
+    def test_var_in_string(self, monkeypatch: pytest.MonkeyPatch) -> None:
+        monkeypatch.setenv("HOST", "example.com")
+        assert resolve_env_vars("https://${HOST}/api") == "https://example.com/api"
+
+    def test_multiple_vars(self, monkeypatch: pytest.MonkeyPatch) -> None:
+        monkeypatch.setenv("USER", "admin")
+        monkeypatch.setenv("PASS", "secret")
+        result = resolve_env_vars("${USER}:${PASS}")
+        assert result == "admin:secret"
+
+    def test_unset_var_raises(self, monkeypatch: pytest.MonkeyPatch) -> None:
+        monkeypatch.delenv("NONEXISTENT_VAR_XYZ", raising=False)
+        with pytest.raises(ConfigValidationError) as exc_info:
+            resolve_env_vars("${NONEXISTENT_VAR_XYZ}")
+        assert "NONEXISTENT_VAR_XYZ" in str(exc_info.value)
+
+    def test_empty_string(self) -> None:
+        assert resolve_env_vars("") == ""
+
+    def test_var_with_empty_value(self, monkeypatch: pytest.MonkeyPatch) -> None:
+        monkeypatch.setenv("EMPTY_VAR", "")
+        assert resolve_env_vars("prefix_${EMPTY_VAR}_suffix") == "prefix__suffix"
+
+
+# ---------------------------------------------------------------------------
+# load_sources_config – Erfolgsfälle
+# ---------------------------------------------------------------------------
+
+
+class TestLoadSourcesConfig:
+    """Tests für load_sources_config."""
+
+    def test_basic_config(self, tmp_path: Path, monkeypatch: pytest.MonkeyPatch) -> None:
+        monkeypatch.setenv("CONF_TOKEN", "tok123")
+        config_file = tmp_path / "sources.yaml"
+        config_file.write_text(dedent("""\
+            version: "1.0"
+            sources:
+              - name: "Team Confluence"
+                type: confluence
+                enabled: true
+                sync_frequency: daily
+                params:
+                  base_url: "https://confluence.bahn.de"
+                  api_key: "${CONF_TOKEN}"
+                  spaces: ["ACV2"]
+                target_folder: references
+            settings:
+              create_tasks: true
+              enrichment:
+                provider: "kiro"
+                confidence_threshold: 0.8
+              ocr:
+                provider: "kiro"
+              git:
+                auto_commit: false
+        """), encoding="utf-8")
+
+        sources, settings = load_sources_config(config_file)
+
+        assert len(sources) == 1
+        src = sources[0]
+        assert src.type == "confluence"
+        assert src.name == "Team Confluence"
+        assert src.params["api_key"] == "tok123"
+        assert src.params["spaces"] == ["ACV2"]
+        assert src.target_folder == "references"
+        assert src.sync_frequency == "daily"
+        assert src.enabled is True
+
+        assert settings.create_tasks is True
+        assert settings.enrichment.provider == "kiro"
+        assert settings.enrichment.confidence_threshold == 0.8
+        assert settings.git.auto_commit is False
+
+    def test_multiple_sources(self, tmp_path: Path) -> None:
+        config_file = tmp_path / "sources.yaml"
+        config_file.write_text(dedent("""\
+            version: "1.0"
+            sources:
+              - name: "Source A"
+                type: file
+                params:
+                  path: "/data"
+              - name: "Source B"
+                type: jira
+                enabled: false
+                params:
+                  jql: "project = X"
+        """), encoding="utf-8")
+
+        sources, settings = load_sources_config(config_file)
+        assert len(sources) == 2
+        assert sources[0].name == "Source A"
+        assert sources[1].name == "Source B"
+        assert sources[1].enabled is False
+
+    def test_empty_sources_list(self, tmp_path: Path) -> None:
+        config_file = tmp_path / "sources.yaml"
+        config_file.write_text(dedent("""\
+            version: "1.0"
+            sources: []
+        """), encoding="utf-8")
+
+        sources, settings = load_sources_config(config_file)
+        assert sources == []
+        assert isinstance(settings, PipelineSettings)
+
+    def test_no_settings_block(self, tmp_path: Path) -> None:
+        config_file = tmp_path / "sources.yaml"
+        config_file.write_text(dedent("""\
+            version: "1.0"
+            sources:
+              - name: "Minimal"
+                type: file
+        """), encoding="utf-8")
+
+        sources, settings = load_sources_config(config_file)
+        assert len(sources) == 1
+        # Should use defaults
+        assert settings.create_tasks is True
+        assert settings.enrichment.provider == "kiro"
+
+    def test_file_not_found(self, tmp_path: Path) -> None:
+        with pytest.raises(FileNotFoundError):
+            load_sources_config(tmp_path / "nonexistent.yaml")
+
+
+# ---------------------------------------------------------------------------
+# load_sources_config – Validierungsfehler
+# ---------------------------------------------------------------------------
+
+
+class TestLoadSourcesConfigValidation:
+    """Tests für Validierungsfehler in load_sources_config."""
+
+    def test_missing_type(self, tmp_path: Path) -> None:
+        config_file = tmp_path / "sources.yaml"
+        config_file.write_text(dedent("""\
+            version: "1.0"
+            sources:
+              - name: "No Type"
+                params: {}
+        """), encoding="utf-8")
+
+        with pytest.raises(ConfigValidationError) as exc_info:
+            load_sources_config(config_file)
+        assert "type" in str(exc_info.value).lower()
+
+    def test_missing_name(self, tmp_path: Path) -> None:
+        config_file = tmp_path / "sources.yaml"
+        config_file.write_text(dedent("""\
+            version: "1.0"
+            sources:
+              - type: confluence
+                params: {}
+        """), encoding="utf-8")
+
+        with pytest.raises(ConfigValidationError) as exc_info:
+            load_sources_config(config_file)
+        assert "name" in str(exc_info.value).lower()
+
+    def test_unresolvable_env_var(self, tmp_path: Path, monkeypatch: pytest.MonkeyPatch) -> None:
+        monkeypatch.delenv("MISSING_SECRET_XYZ", raising=False)
+        config_file = tmp_path / "sources.yaml"
+        config_file.write_text(dedent("""\
+            version: "1.0"
+            sources:
+              - name: "Bad Source"
+                type: confluence
+                params:
+                  api_key: "${MISSING_SECRET_XYZ}"
+        """), encoding="utf-8")
+
+        with pytest.raises(ConfigValidationError) as exc_info:
+            load_sources_config(config_file)
+        assert "MISSING_SECRET_XYZ" in str(exc_info.value)
+
+    def test_multiple_errors_collected(self, tmp_path: Path, monkeypatch: pytest.MonkeyPatch) -> None:
+        monkeypatch.delenv("TOKEN_A", raising=False)
+        monkeypatch.delenv("TOKEN_B", raising=False)
+        config_file = tmp_path / "sources.yaml"
+        config_file.write_text(dedent("""\
+            version: "1.0"
+            sources:
+              - name: "Source 1"
+                type: confluence
+                params:
+                  key_a: "${TOKEN_A}"
+                  key_b: "${TOKEN_B}"
+        """), encoding="utf-8")
+
+        with pytest.raises(ConfigValidationError) as exc_info:
+            load_sources_config(config_file)
+        assert "TOKEN_A" in str(exc_info.value)
+        assert "TOKEN_B" in str(exc_info.value)
+
+    def test_missing_fields_and_env_vars(self, tmp_path: Path, monkeypatch: pytest.MonkeyPatch) -> None:
+        monkeypatch.delenv("GHOST_VAR", raising=False)
+        config_file = tmp_path / "sources.yaml"
+        config_file.write_text(dedent("""\
+            version: "1.0"
+            sources:
+              - type: jira
+                params:
+                  token: "${GHOST_VAR}"
+        """), encoding="utf-8")
+
+        with pytest.raises(ConfigValidationError) as exc_info:
+            load_sources_config(config_file)
+        # Both missing name and unresolvable env var
+        assert "name" in str(exc_info.value).lower()
+        assert "GHOST_VAR" in str(exc_info.value)