Add EXIF writing and machine metadata support

spruce/orchestrator.py

@@ -14,11 +14,12 @@ from typing import Any
 class RunStats:
     """Accumulated counters for one or more machines."""
 
-    scans_fetched: int = 0      # metadata fetched from server this run
+    scans_fetched: int = 0      # scan detail page fetched (metadata), not tiles/mosaics
     scans_skipped: int = 0      # metadata.json already on disk; no HTTP request
-    scans_failed: int = 0       # fetch error or missing grid params
+    scans_failed: int = 0       # metadata fetch error or missing grid params
     metadata_written: int = 0   # new metadata.json files created
     mosaics_downloaded: int = 0
+    mosaics_failed: int = 0  # mosaic URL attempted but 0 bytes / HTTP error
     tiles_downloaded: int = 0
 
     def merge(self, other: "RunStats") -> None:
@@ -27,10 +28,12 @@ class RunStats:
         self.scans_failed += other.scans_failed
         self.metadata_written += other.metadata_written
         self.mosaics_downloaded += other.mosaics_downloaded
+        self.mosaics_failed += other.mosaics_failed
         self.tiles_downloaded += other.tiles_downloaded
 
 from tqdm import tqdm
 
+from spruce.exif import write_mosaic_exif
 from spruce.paths import machine_dir_name, tile_dest, mosaic_dest, _extract_date
 from spruce.progress import ProgressTracker, CsvWriter
 from spruce.session import MachineSession
@@ -50,6 +53,7 @@ def _download_mosaic(
     mosaic_path: Path,
     progress: ProgressTracker,
     machine: dict[str, Any],
+    config: dict[str, Any],
     dry_run: bool,
 ) -> bool:
     """Download the scan mosaic if not already done. Returns True if downloaded."""
@@ -62,6 +66,13 @@ def _download_mosaic(
     log.info("[%s] Downloading mosaic for scan %d …", machine["label"], scan_id)
     size = sess.download_file(url, mosaic_path)
     if size:
+        if config.get("write_exif", True):
+            mmeta: dict[str, Any] | None = config.get("machine_metadata", {}).get(
+                machine["label"]
+            )
+            write_mosaic_exif(
+                mosaic_path, scan_meta, machine, scan_id, mmeta
+            )
         progress.mark_done(url)
         progress.save()
         log.info(
@@ -283,10 +294,24 @@ def process_scan(
         mosaic_just_downloaded = False
     else:
         mosaic_just_downloaded = _download_mosaic(
-            sess, scan_meta, scan_id, mosaic_path, progress, machine, dry_run
+            sess,
+            scan_meta,
+            scan_id,
+            mosaic_path,
+            progress,
+            machine,
+            config,
+            dry_run,
         )
-    if mosaic_just_downloaded:
+    if not metadata_only and mosaic_just_downloaded:
         stats.mosaics_downloaded += 1
+    elif (
+        not metadata_only
+        and not dry_run
+        and not mosaic_already_done
+        and not mosaic_just_downloaded
+    ):
+        stats.mosaics_failed += 1
 
     # Write scan-level CSV row only if this scan hasn't been recorded before.
     if mosaic_already_done and not metadata_only: