Initial commit

Add spruce scraper with CLI, session management, parsers, progress tracking,
recheck logic, and test suite. Includes example config and README.

tests/test_settings.py

@@ -0,0 +1,91 @@
+"""Tests for spruce.settings — config loading and worker clamping."""
+
+import logging
+
+import pytest
+import yaml
+
+from spruce.settings import (
+    MAX_SAFE_WORKERS,
+    _clamp_workers,
+    load_config,
+)
+
+
+# ---------------------------------------------------------------------------
+# _clamp_workers
+# ---------------------------------------------------------------------------
+
+
+def test_clamp_workers_below_limit():
+    assert _clamp_workers(2) == 2
+
+
+def test_clamp_workers_at_limit():
+    assert _clamp_workers(MAX_SAFE_WORKERS) == MAX_SAFE_WORKERS
+
+
+def test_clamp_workers_above_limit_caps(caplog):
+    with caplog.at_level(logging.WARNING):
+        result = _clamp_workers(MAX_SAFE_WORKERS + 1)
+    assert result == MAX_SAFE_WORKERS
+    assert "exceeds the safe limit" in caplog.text
+
+
+def test_clamp_workers_zero():
+    assert _clamp_workers(0) == 0
+
+
+# ---------------------------------------------------------------------------
+# load_config
+# ---------------------------------------------------------------------------
+
+
+def _write_config(tmp_path, **overrides):
+    base = {
+        "username": "testuser",
+        "password": "testpass",
+    }
+    base.update(overrides)
+    path = tmp_path / "config.yaml"
+    path.write_text(yaml.dump(base))
+    return str(path)
+
+
+def test_load_config_defaults(tmp_path):
+    path = _write_config(tmp_path)
+    cfg = load_config(path)
+    assert cfg["base_url"] == "http://205.149.147.131:8010/"
+    assert cfg["workers"] == 2
+    assert cfg["timeout"] == 60
+    assert cfg["request_delay"] == 0.5
+    assert cfg["output_dir"] == "archives"
+
+
+def test_load_config_overrides(tmp_path):
+    path = _write_config(tmp_path, workers=3, output_dir="my_archives")
+    cfg = load_config(path)
+    assert cfg["workers"] == 3
+    assert cfg["output_dir"] == "my_archives"
+
+
+def test_load_config_caps_workers(tmp_path, caplog):
+    path = _write_config(tmp_path, workers=MAX_SAFE_WORKERS + 2)
+    with caplog.at_level(logging.WARNING):
+        cfg = load_config(path)
+    assert cfg["workers"] == MAX_SAFE_WORKERS
+    assert "exceeds the safe limit" in caplog.text
+
+
+def test_load_config_missing_username_exits(tmp_path):
+    path = tmp_path / "config.yaml"
+    path.write_text(yaml.dump({"password": "x"}))
+    with pytest.raises(SystemExit):
+        load_config(str(path))
+
+
+def test_load_config_missing_password_exits(tmp_path):
+    path = tmp_path / "config.yaml"
+    path.write_text(yaml.dump({"username": "x"}))
+    with pytest.raises(SystemExit):
+        load_config(str(path))