Add EXIF writing and machine metadata support

tests/test_exif.py

@@ -0,0 +1,138 @@
+"""Tests for spruce.exif — mosaic EXIF injection."""
+
+import shutil
+from pathlib import Path
+
+import piexif
+import pytest
+from piexif import ExifIFD, GPSIFD, ImageIFD
+
+from spruce.exif import USER_COMMENT_ASCII, write_mosaic_exif
+
+FIXTURES = Path(__file__).parent / "fixtures"
+BLANK_JPEG = FIXTURES / "blank.jpg"
+
+
+def _dms_pair_to_float(num: int, den: int) -> float:
+    return num / den if den else 0.0
+
+
+def _dms_to_deg(
+    ref: bytes, dms: list[tuple[tuple[int, int], tuple[int, int], tuple[int, int]]]
+) -> float:
+    d0, m0, s0 = dms
+    deg = (
+        _dms_pair_to_float(d0[0], d0[1])
+        + _dms_pair_to_float(m0[0], m0[1]) / 60.0
+        + _dms_pair_to_float(s0[0], s0[1]) / 3600.0
+    )
+    if ref in (b"S", b"W"):
+        return -deg
+    return deg
+
+
+@pytest.fixture
+def tmp_jpeg(tmp_path: Path) -> Path:
+    dest = tmp_path / "mosaic.jpg"
+    shutil.copy(BLANK_JPEG, dest)
+    return dest
+
+
+@pytest.fixture
+def scan_meta() -> dict:
+    return {
+        "scan_id": 157743,
+        "name": "Plot 7 AMR18 Full tube scan",
+        "scan_time": "2024-06-28 11:00:00",
+        "user": "Joanne",
+        "nx": 103,
+        "ny": 328,
+        "dx": 3.01,
+        "dy": 2.26,
+        "end_x": 310.0,
+        "end_y": 740.0,
+    }
+
+
+@pytest.fixture
+def machine() -> dict:
+    return {"label": "BW1-7 [AMR-18]", "version": "3.0.0.18", "machine_id": "7"}
+
+
+def test_write_mosaic_exif_round_trip(tmp_jpeg: Path, scan_meta: dict, machine: dict):
+    assert write_mosaic_exif(
+        tmp_jpeg,
+        scan_meta,
+        machine,
+        157743,
+        None,
+    )
+    exif = piexif.load(str(tmp_jpeg))
+    assert exif["Exif"][ExifIFD.DateTimeOriginal] == b"2024:06:28 11:00:00"
+    assert (
+        exif["0th"][ImageIFD.ImageDescription]
+        == b"BW1-7 [AMR-18] scan 157743 (Plot 7 AMR18 Full tube scan)"
+    )
+    assert exif["0th"][ImageIFD.Artist] == b"Joanne"
+    assert exif["0th"][ImageIFD.Software] == b"RootView 3.0.0.18"
+    assert exif["0th"][ImageIFD.ProcessingSoftware] == b"spruce-scraper/1.0"
+    uc = exif["Exif"][ExifIFD.UserComment]
+    assert uc.startswith(USER_COMMENT_ASCII)
+    tail = uc[len(USER_COMMENT_ASCII) :]
+    assert b"SPRUCE scan 157743" in tail
+    assert b"103x328" in tail
+    assert b"see metadata.json" in tail
+    assert b"plot_number" not in tail
+    assert ImageIFD.XPKeywords not in exif["0th"]
+    assert "GPS" not in exif or not exif["GPS"]
+
+
+def test_gps_decode(tmp_jpeg: Path, scan_meta: dict, machine: dict):
+    mmeta = {
+        "latitude_wgs_84": 47.5047,
+        "longitude_wgs_84": -93.4530,
+        "elevation_masl": 418.0,
+    }
+    assert write_mosaic_exif(tmp_jpeg, scan_meta, machine, 157743, mmeta)
+    exif = piexif.load(str(tmp_jpeg))
+    gps = exif["GPS"]
+    lat = _dms_to_deg(gps[GPSIFD.GPSLatitudeRef], gps[GPSIFD.GPSLatitude])
+    lon = _dms_to_deg(gps[GPSIFD.GPSLongitudeRef], gps[GPSIFD.GPSLongitude])
+    assert abs(lat - 47.5047) < 1e-5
+    assert abs(lon - (-93.4530)) < 1e-5
+    alt = gps[GPSIFD.GPSAltitude]
+    assert alt[0] / alt[1] == pytest.approx(418.0)
+
+
+def test_no_gps_no_keywords_when_meta_none(
+    tmp_jpeg: Path, scan_meta: dict, machine: dict
+):
+    assert write_mosaic_exif(tmp_jpeg, scan_meta, machine, 157743, None)
+    exif = piexif.load(str(tmp_jpeg))
+    assert ImageIFD.XPKeywords not in exif["0th"]
+    assert "GPS" not in exif or not exif["GPS"]
+
+
+def test_xp_keywords_treatment(tmp_jpeg: Path, scan_meta: dict, machine: dict):
+    mmeta = {
+        "plot_number": 7,
+        "enclosure": False,
+        "temp_treatment": 2.25,
+        "co2_treatment": "ambient",
+    }
+    assert write_mosaic_exif(tmp_jpeg, scan_meta, machine, 157743, mmeta)
+    exif = piexif.load(str(tmp_jpeg))
+    raw = exif["0th"][ImageIFD.XPKeywords]
+    text = bytes(raw).decode("utf-16le").rstrip("\x00")
+    assert "SPRUCE" in text
+    assert "plot 7" in text
+    assert "no enclosure" in text
+    assert "temp" in text
+    assert "2.25" in text
+    assert "aCO2" in text
+    uc = exif["Exif"][ExifIFD.UserComment]
+    uct = uc[len(USER_COMMENT_ASCII) :].decode("ascii", errors="replace")
+    assert "plot_number 7" in uct
+    assert "enclosure no" in uct
+    assert "temp_treatment 2.25" in uct
+    assert "co2_treatment ambient" in uct

tests/test_recheck.py

@@ -143,7 +143,11 @@ def test_recheck_archive_skips_mosaic_urls(tmp_path):
     mosaic_url = "http://192.0.2.1:8011/RootView_Database/158374/mosaic.jpg"
     p.mark_done(mosaic_url)
     p.save()
+    # recheck_verifies a non-zero mosaic exists under */*/<scan_id>/mosaic.jpg
+    mpath = tmp_path / "M" / "2020-01-01" / "158374" / "mosaic.jpg"
+    mpath.parent.mkdir(parents=True)
+    mpath.write_bytes(b"\xff\xd8\xff\xd9")  # minimal JPEG soff + eoi
 
     removed = recheck_archive(tmp_path, p)
     assert removed == 0
-    assert p.is_done(mosaic_url)  # mosaics are never touched
+    assert p.is_done(mosaic_url)

tests/test_settings.py

@@ -60,6 +60,8 @@ def test_load_config_defaults(tmp_path):
     assert cfg["timeout"] == 60
     assert cfg["request_delay"] == 0.5
     assert cfg["output_dir"] == "archives"
+    assert cfg["write_exif"] is True
+    assert cfg["machine_metadata"] == {}
 
 
 def test_load_config_overrides(tmp_path):