nanobot: 2026-09-10 12:33:37

skills/wiki/tests/test_wiki_config.py

@@ -0,0 +1,165 @@
+"""Scope precedence (paths -> include -> exclude) and config validation."""
+
+import sys
+from pathlib import Path
+
+import pytest
+
+SCRIPTS = Path(__file__).parent.parent / "scripts"
+sys.path.insert(0, str(SCRIPTS))
+
+import wiki_config
+from wiki_config import ConfigError, SourceConfig, load_config, source_root
+
+WORKSPACE_YAML = """
+embedding:
+  endpoint: http://nvidia.hell:11434/
+  model: qwen3-embedding:0.6b
+  dims: 1024
+  batch: 32
+  keep_alive: -1
+  query_prefix: "Instruct: task\\nQuery: "
+
+sources:
+  index:
+    kind: git
+    url: git@git.fnet.cz:lachtan/index.git
+    paths: ["**"]
+    include: ["*.md"]
+    exclude:
+      - "**/node_modules/**"
+      - "**/vendor/**"
+
+  workspace:
+    kind: workspace
+    paths:
+      - "notes/**"
+      - "develop/**"
+    include: ["*.md"]
+    exclude:
+      - "**/inbox/**"
+      - "develop/history.md"
+"""
+
+
+def _write(tmp_path, text):
+    path = tmp_path / "config.yaml"
+    path.write_text(text, encoding="utf-8")
+    return path
+
+
+def _source(
+    source_id: str = "ws",
+    kind: str = "workspace",
+    url: str | None = None,
+    paths: tuple[str, ...] = ("**",),
+    include: tuple[str, ...] = ("*.md",),
+    exclude: tuple[str, ...] = (),
+) -> SourceConfig:
+    return SourceConfig(source_id=source_id, kind=kind, url=url, paths=paths, include=include, exclude=exclude)
+
+
+def test_loads_embedding_and_sources(tmp_path):
+    config = load_config(_write(tmp_path, WORKSPACE_YAML))
+    assert config.embedding.endpoint == "http://nvidia.hell:11434"  # trailing slash trimmed
+    assert config.embedding.model == "qwen3-embedding:0.6b"
+    assert config.embedding.dims == 1024
+    assert config.embedding.keep_alive == -1
+    assert config.embedding.query_prefix.endswith("Query: ")
+    assert [s.source_id for s in config.sources] == ["index", "workspace"]
+    index = config.source("index")
+    assert index is not None and index.kind == "git"
+    assert config.source("nope") is None
+
+
+def test_string_keep_alive_is_rejected():
+    """Ollama answers HTTP 400 to a string "-1" — catch it in config, not at runtime."""
+    with pytest.raises(ConfigError, match="keep_alive"):
+        wiki_config._parse_embedding({"endpoint": "http://x", "model": "m", "dims": 1024, "keep_alive": "-1"})
+
+
+def test_missing_config_file(tmp_path):
+    with pytest.raises(ConfigError, match="missing config"):
+        load_config(tmp_path / "nope.yaml")
+
+
+def test_git_source_needs_url(tmp_path):
+    yaml_text = WORKSPACE_YAML.replace("    url: git@git.fnet.cz:lachtan/index.git\n", "")
+    with pytest.raises(ConfigError, match="need a url"):
+        load_config(_write(tmp_path, yaml_text))
+
+
+def test_unknown_kind_is_rejected(tmp_path):
+    yaml_text = WORKSPACE_YAML.replace("    kind: workspace", "    kind: mirror")
+    with pytest.raises(ConfigError, match="kind must be one of"):
+        load_config(_write(tmp_path, yaml_text))
+
+
+def test_paths_is_required(tmp_path):
+    yaml_text = WORKSPACE_YAML.replace('    paths: ["**"]\n', "")
+    with pytest.raises(ConfigError, match="`paths` is required"):
+        load_config(_write(tmp_path, yaml_text))
+
+
+def test_paths_whitelist_gates_everything():
+    source = _source(paths=("notes/**", "develop/**"))
+    assert source.covers("notes/a.md")
+    assert source.covers("notes/deep/nested/a.md")
+    assert source.covers("develop/knowledge.md")
+    # Not on the whitelist -> does not exist for the index.
+    assert not source.covers("tmp/a.md")
+    assert not source.covers("skills/wiki/SKILL.md")
+    assert not source.covers("AGENTS.md")
+
+
+def test_include_filters_extensions():
+    source = _source(paths=("**",))
+    assert source.covers("notes/a.md")
+    assert not source.covers("notes/main.py")
+    assert not source.covers("memory/history.jsonl")
+    assert not source.covers("assets/photo.png")
+
+
+def test_exclude_wins_over_paths_and_include():
+    source = _source(paths=("notes/**", "develop/**"), exclude=("**/inbox/**", "develop/history.md"))
+    assert not source.covers("notes/inbox/raw.md")
+    assert not source.covers("develop/history.md")
+    assert source.covers("develop/knowledge.md")
+    assert source.covers("notes/notes.md")
+
+
+def test_double_star_matches_whole_repo():
+    source = _source(paths=("**",), exclude=("**/node_modules/**", "**/vendor/**"))
+    assert source.covers("README.md")
+    assert source.covers("japan/tokyo/metro.md")
+    assert not source.covers("node_modules/pkg/README.md")
+    assert not source.covers("web/vendor/lib/CHANGELOG.md")
+
+
+def test_glob_star_does_not_cross_a_slash():
+    source = _source(paths=("notes/*",))
+    assert source.covers("notes/a.md")
+    assert not source.covers("notes/deep/a.md")
+
+
+def test_covers_dir_prunes_the_walk():
+    source = _source(paths=("notes/**", "develop/**"), exclude=("**/inbox/**",))
+    assert source.covers_dir("")
+    assert source.covers_dir("notes")
+    assert source.covers_dir("notes/deep")
+    assert source.covers_dir("develop")
+    assert not source.covers_dir("tmp")
+    assert not source.covers_dir("notes/inbox")
+
+
+def test_covers_dir_never_prunes_a_double_star_source():
+    source = _source(paths=("**",), exclude=("**/node_modules/**",))
+    assert source.covers_dir("anything/deep")
+    assert not source.covers_dir("app/node_modules")
+
+
+def test_source_root_derives_clone_path_from_id():
+    assert source_root(_source(source_id="travel", kind="git", url="git@x:y.git")) == (
+        wiki_config.REMOTE_DIR / "travel"
+    )
+    assert source_root(_source(kind="workspace")) == wiki_config.WORKSPACE