feat: Added warnings for potentially unreachable 3D files (3D paths pt. 1)

colorizer_data/utils.py

@@ -776,3 +776,32 @@ def _get_frame_count_from_3d_source(source: str) -> int:
     # Attempt to read the image to get info (such as length)
     img = BioImage(source)
     return int(img.dims.T)
+
+
+def is_url(source: str) -> bool:
+    """
+    Checks if a source string is an HTTP(S) URL.
+    """
+    return source.startswith("http://") or source.startswith("https://")
+
+
+def check_file_source(name: str, source: str | None, outpath: pathlib.Path):
+    """
+    Logs warnings for missing or unreachable file sources.
+    """
+    if source is None:
+        logging.error(f"{name} is undefined.")
+    elif not is_url(source):
+        # Check for absolute paths, parent paths, or missing files/folders.
+        if os.path.isabs(source):
+            logging.error(
+                f"{name} must be a relative path inside the dataset directory or an HTTP(S) URL. Received: '{source}'"
+            )
+        elif ".." in pathlib.Path(source).parts:
+            logging.warning(
+                f"{name} should not contain parent directory references ('..'), as it may fail to load in certain deploy environments. Received: '{source}'"
+            )
+        elif not os.path.exists(outpath / source):
+            logging.warning(
+                f"{name} path could not be found. Please check that it exists. Received: '{source}'"
+            )

colorizer_data/writer.py

@@ -22,6 +22,7 @@
     DEFAULT_FRAME_PREFIX,
     DEFAULT_FRAME_SUFFIX,
     _get_frame_count_from_3d_source,
+    check_file_source,
     cast_feature_to_info_type,
     copy_remote_or_local_file,
     generate_frame_paths,
@@ -711,3 +712,16 @@ def validate_dataset(
                     + " or add an offset if your frame numbers do not start at 0."
                     + " You may also need to generate the list of frames yourself if your dataset is skipping frames."
                 )
+
+        # Check that frames3d sources are reachable
+        if "frames3d" in self.manifest:
+            frames3d_metadata = Frames3dMetadata.from_dict(self.manifest["frames3d"])
+            source = frames3d_metadata.source
+            check_file_source("3D frames source", source, self.outpath)
+            # Validate backdrops
+            if frames3d_metadata.backdrops is not None:
+                for i in range(len(frames3d_metadata.backdrops)):
+                    backdrop_source = frames3d_metadata.backdrops[i].source
+                    check_file_source(
+                        f"3D frames backdrop {i} source", backdrop_source, self.outpath
+                    )