Initial commit

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

tests/test_paths.py

@@ -0,0 +1,102 @@
+"""Tests for spruce.paths — pure path helpers, no network."""
+
+from pathlib import Path
+
+import pytest
+from spruce.paths import _extract_date, machine_dir_name, mosaic_dest, tile_dest
+
+
+MACHINE = {"label": "BW3-20 [AMR-26]", "machine_id": "26"}
+SCAN_META = {"scan_time": "2024-07-29 05:00", "nx": 103, "ny": 328}
+TILE = {"scan_id": 158374, "row_index": 0, "col_index": 2}
+
+
+# ---------------------------------------------------------------------------
+# machine_dir_name
+# ---------------------------------------------------------------------------
+
+
+def test_machine_dir_name_sanitises_brackets():
+    name = machine_dir_name({"label": "BW3-20 [AMR-26]"})
+    # Brackets and spaces replaced by underscores
+    assert "[" not in name
+    assert "]" not in name
+    assert " " not in name
+
+
+def test_machine_dir_name_no_leading_trailing_underscores():
+    name = machine_dir_name({"label": "BW3-20 [AMR-26]"})
+    assert not name.startswith("_")
+    assert not name.endswith("_")
+
+
+def test_machine_dir_name_stable():
+    # Same label should always produce the same dir name
+    assert machine_dir_name(MACHINE) == machine_dir_name(MACHINE)
+
+
+# ---------------------------------------------------------------------------
+# _extract_date
+# ---------------------------------------------------------------------------
+
+
+def test_extract_date_standard():
+    assert _extract_date("2024-07-29 05:00:00") == "2024-07-29"
+
+
+def test_extract_date_date_only():
+    assert _extract_date("2024-07-29") == "2024-07-29"
+
+
+def test_extract_date_no_date():
+    assert _extract_date("no date here") == "unknown"
+
+
+def test_extract_date_empty():
+    assert _extract_date("") == "unknown"
+
+
+# ---------------------------------------------------------------------------
+# tile_dest
+# ---------------------------------------------------------------------------
+
+
+def test_tile_dest_structure(tmp_path):
+    dest = tile_dest(tmp_path, MACHINE, SCAN_META, TILE)
+    parts = dest.parts
+    assert str(TILE["scan_id"]) in parts
+    assert "tiles" in parts
+    assert dest.suffix == ".jpg"
+    assert "tile_r" in dest.name
+
+
+def test_tile_dest_zero_padded(tmp_path):
+    # ny=328 → row index needs 3 digits; nx=103 → col index needs 3 digits
+    dest = tile_dest(tmp_path, MACHINE, SCAN_META, TILE)
+    # row_index=0 padded to 3 digits (max is 327) → "000"
+    assert "tile_r000_c" in dest.name
+
+
+def test_tile_dest_contains_date(tmp_path):
+    dest = tile_dest(tmp_path, MACHINE, SCAN_META, TILE)
+    assert "2024-07-29" in str(dest)
+
+
+# ---------------------------------------------------------------------------
+# mosaic_dest
+# ---------------------------------------------------------------------------
+
+
+def test_mosaic_dest_filename(tmp_path):
+    dest = mosaic_dest(tmp_path, MACHINE, SCAN_META, 158374)
+    assert dest.name == "mosaic.jpg"
+
+
+def test_mosaic_dest_contains_scan_id(tmp_path):
+    dest = mosaic_dest(tmp_path, MACHINE, SCAN_META, 158374)
+    assert "158374" in str(dest)
+
+
+def test_mosaic_dest_contains_date(tmp_path):
+    dest = mosaic_dest(tmp_path, MACHINE, SCAN_META, 158374)
+    assert "2024-07-29" in str(dest)