Extend image metadata

lib/galaxy/datatypes/images.py

@@ -5,12 +5,23 @@
 import base64
 import json
 import logging
-from typing import Optional
+from typing import (
+    Any,
+    Dict,
+    List,
+    Optional,
+)
 
 import mrcfile
 import numpy as np
 import tifffile
 
+try:
+    import PIL
+    import PIL.Image
+except ImportError:
+    PIL = None  # type: ignore[assignment, unused-ignore]
+
 from galaxy.datatypes.binary import Binary
 from galaxy.datatypes.metadata import (
     FileParameter,
@@ -50,6 +61,70 @@ class Image(data.Data):
     edam_format = "format_3547"
     file_ext = ""
 
+    MetadataElement(
+        name="axes",
+        desc="Axes of the image data",
+        readonly=True,
+        visible=True,
+        optional=True,
+    )
+
+    MetadataElement(
+        name="dtype",
+        desc="Data type of the image pixels or voxels",
+        readonly=True,
+        visible=True,
+        optional=True,
+    )
+
+    MetadataElement(
+        name="num_unique_values",
+        desc="Number of unique values in the image data (e.g., should be 2 for binary images)",
+        readonly=True,
+        visible=True,
+        optional=True,
+    )
+
+    MetadataElement(
+        name="width",
+        desc="Width of the image (in pixels)",
+        readonly=True,
+        visible=True,
+        optional=True,
+    )
+
+    MetadataElement(
+        name="height",
+        desc="Height of the image (in pixels)",
+        readonly=True,
+        visible=True,
+        optional=True,
+    )
+
+    MetadataElement(
+        name="channels",
+        desc="Number of channels of the image",
+        readonly=True,
+        visible=True,
+        optional=True,
+    )
+
+    MetadataElement(
+        name="depth",
+        desc="Depth of the image (number of slices)",
+        readonly=True,
+        visible=True,
+        optional=True,
+    )
+
+    MetadataElement(
+        name="frames",
+        desc="Number of frames in the image sequence (number of time steps)",
+        readonly=True,
+        visible=True,
+        optional=True,
+    )
+
     def __init__(self, **kwd):
         super().__init__(**kwd)
         self.image_formats = [self.file_ext.upper()]
@@ -73,6 +148,33 @@ def handle_dataset_as_image(self, hda: DatasetProtocol) -> str:
             base64_image_data = base64.b64encode(f.read()).decode("utf-8")
         return f"![{name}](data:image/{self.file_ext};base64,{base64_image_data})"
 
+    def set_meta(
+        self, dataset: DatasetProtocol, overwrite: bool = True, metadata_tmp_files_dir: Optional[str] = None, **kwd
+    ) -> None:
+        """
+        Try to populate the metadata of the image using a generic image loading library (pillow), if available.
+
+        If an image has two axes, they are assumed to be ``YX``. If an image has three axes, they are assumed to be ``YXC``.
+        """
+        if PIL is not None:
+            try:
+                with PIL.Image.open(dataset.get_file_name()) as im:
+                    im_arr = np.array(im)
+                    dataset.metadata.dtype = str(im_arr.dtype)
+                    dataset.metadata.num_unique_values = str(len(np.unique(im)))
+                    dataset.metadata.width = str(im_arr.shape[1])
+                    dataset.metadata.height = str(im_arr.shape[0])
+                    dataset.metadata.depth = "0"
+                    dataset.metadata.frames = "0"
+                    if im_arr.ndim == 2:
+                        dataset.metadata.axes = "YX"
+                        dataset.metadata.channels = "0"
+                    elif im_arr.ndim == 3:
+                        dataset.metadata.axes = "YXC"
+                        dataset.metadata.channels = str(im_arr.shape[2])
+            except PIL.UnidentifiedImageError:
+                pass
+
 
 class Jpg(Image):
     edam_format = "format_3579"
@@ -104,6 +206,9 @@ class Tiff(Image):
     def set_meta(
         self, dataset: DatasetProtocol, overwrite: bool = True, metadata_tmp_files_dir: Optional[str] = None, **kwd
     ) -> None:
+        """
+        Populate the metadata of the TIFF image using the tifffile library.
+        """
         spec_key = "offsets"
         offsets_file = dataset.metadata.offsets
         if not offsets_file:
@@ -112,10 +217,56 @@ def set_meta(
             )
         with tifffile.TiffFile(dataset.get_file_name()) as tif:
             offsets = [page.offset for page in tif.pages]
+
+            # Aggregate a list of values for each metadata field (one value for each series of the TIFF file)
+            metadata: Dict[str, List[Any]] = {
+                key: []
+                for key in [
+                    "axes",
+                    "dtype",
+                    "width",
+                    "height",
+                    "channels",
+                    "depth",
+                    "frames",
+                    "num_unique_values",
+                ]
+            }
+            for series in tif.series:
+
+                # Determine the metadata values that should be generally available
+                metadata["axes"].append(series.axes.upper())
+                metadata["dtype"].append(series.dtype)
+
+                # Determine the metadata values that require reading the image data
+                try:
+                    im_arr = series.asarray()
+                except ValueError:  # Occurs if the compression of the TIFF file is unsupported
+                    im_arr = None
+                if im_arr is not None:
+                    axes = metadata["axes"][-1].replace("S", "C")
+                    metadata["width"].append(Tiff._get_axis_size(im_arr, axes, "X"))
+                    metadata["height"].append(Tiff._get_axis_size(im_arr, axes, "Y"))
+                    metadata["channels"].append(Tiff._get_axis_size(im_arr, axes, "C"))
+                    metadata["depth"].append(Tiff._get_axis_size(im_arr, axes, "Z"))
+                    metadata["frames"].append(Tiff._get_axis_size(im_arr, axes, "T"))
+                    metadata["num_unique_values"].append(len(np.unique(im_arr)))
+
+            # Populate the metadata fields based on the values determined above
+            for key, values in metadata.items():
+                if len(values) > 0:
+                    setattr(dataset.metadata, key, ",".join(str(value) for value in values))
+
+        # Populate the "offsets" file and metadata field
         with open(offsets_file.get_file_name(), "w") as f:
             json.dump(offsets, f)
         dataset.metadata.offsets = offsets_file
 
+    @staticmethod
+    def _get_axis_size(im_arr: "np.typing.NDArray", axes: str, axis: str) -> int:
+        idx = axes.find(axis)
+        return im_arr.shape[idx] if idx >= 0 else 0
+
 
 class OMETiff(Tiff):
     file_ext = "ome.tiff"

test/functional/tools/sample_tool_conf.xml

@@ -149,6 +149,7 @@
   <tool file="top_level_data.xml" />
   <tool file="validation_hdf5.xml" />
   <tool file="validation_image.xml"/>
+  <tool file="validation_image_metadata.xml"/>
   <tool file="validation_zip.xml" />
   <tool file="validation_tar.xml" />
   <tool file="validation_tar_gz.xml" />

test/functional/tools/validation_image_metadata.xml

@@ -0,0 +1,154 @@
+<tool id="validation_image_metadata" name="validation_image_metadata" version="1.0">
+    <macros>
+        <!-- Tests are not executed if they have no output checks, thus define a dummy check -->
+        <xml name="test-output">
+            <output name="output">
+                <assert_contents>
+                    <has_size value="0" delta="0" />
+                </assert_contents>
+            </output>
+        </xml>
+   </macros>
+    <command><![CDATA[
+      cat /dev/null > testfile
+    ]]></command>
+    <inputs>
+        <!-- Checks for "axes" metadata -->
+        <param name="input_axes_yx" type="data" format="data" optional="true">
+            <validator type="dataset_metadata_equal" metadata_name="axes" value="YX" />
+        </param>
+        <param name="input_axes_yxc" type="data" format="data" optional="true">
+            <validator type="dataset_metadata_equal" metadata_name="axes" value="YXC" />
+        </param>
+        <param name="input_axes_zcyx" type="data" format="data" optional="true">
+            <validator type="dataset_metadata_equal" metadata_name="axes" value="ZCYX" />
+        </param>
+        <!-- Checks for "dtype" metadata -->
+        <param name="input_dtype_uint8" type="data" format="data" optional="true">
+            <validator type="dataset_metadata_equal" metadata_name="dtype" value="uint8" />
+        </param>
+        <param name="input_dtype_uint16" type="data" format="data" optional="true">
+            <validator type="dataset_metadata_equal" metadata_name="dtype" value="uint16" />
+        </param>
+        <param name="input_dtype_float64" type="data" format="data" optional="true">
+            <validator type="dataset_metadata_equal" metadata_name="dtype" value="float64" />
+        </param>
+        <!-- Checks for "num_unique_values" metadata -->
+        <param name="input_num_unique_values_1" type="data" format="data" optional="true">
+            <validator type="dataset_metadata_equal" metadata_name="num_unique_values" value="1" />
+        </param>
+        <param name="input_num_unique_values_2" type="data" format="data" optional="true">
+            <validator type="dataset_metadata_equal" metadata_name="num_unique_values" value="2" />
+        </param>
+        <param name="input_num_unique_values_618" type="data" format="data" optional="true">
+            <validator type="dataset_metadata_equal" metadata_name="num_unique_values" value="618" />
+        </param>
+        <!-- Checks for "width" metadata -->
+        <param name="input_width_16" type="data" format="data" optional="true">
+            <validator type="dataset_metadata_equal" metadata_name="width" value="16" />
+        </param>
+        <param name="input_width_32" type="data" format="data" optional="true">
+            <validator type="dataset_metadata_equal" metadata_name="width" value="32" />
+        </param>
+        <!-- Checks for "height" metadata -->
+        <param name="input_height_8" type="data" format="data" optional="true">
+            <validator type="dataset_metadata_equal" metadata_name="height" value="8" />
+        </param>
+        <param name="input_height_32" type="data" format="data" optional="true">
+            <validator type="dataset_metadata_equal" metadata_name="height" value="32" />
+        </param>
+        <!-- Checks for "channels" metadata -->
+        <param name="input_channels_0" type="data" format="data" optional="true">
+            <validator type="dataset_metadata_equal" metadata_name="channels" value="0" />
+        </param>
+        <param name="input_channels_2" type="data" format="data" optional="true">
+            <validator type="dataset_metadata_equal" metadata_name="channels" value="2" />
+        </param>
+        <param name="input_channels_3" type="data" format="data" optional="true">
+            <validator type="dataset_metadata_equal" metadata_name="channels" value="3" />
+        </param>
+        <!-- Checks for "depth" metadata -->
+        <param name="input_depth_0" type="data" format="data" optional="true">
+            <validator type="dataset_metadata_equal" metadata_name="depth" value="0" />
+        </param>
+        <param name="input_depth_25" type="data" format="data" optional="true">
+            <validator type="dataset_metadata_equal" metadata_name="depth" value="25" />
+        </param>
+        <!-- Checks for "frames" metadata -->
+        <param name="input_frames_0" type="data" format="data" optional="true">
+            <validator type="dataset_metadata_equal" metadata_name="frames" value="0" />
+        </param>
+        <param name="input_frames_5" type="data" format="data" optional="true">
+            <validator type="dataset_metadata_equal" metadata_name="frames" value="5" />
+        </param>
+        <!-- Checks for TIFF files with multiple series -->
+        <param name="input_multi_series_tiff" type="data" format="data" optional="true">
+            <validator type="dataset_metadata_equal" metadata_name="axes" value="YXS,YX" />
+            <validator type="dataset_metadata_equal" metadata_name="dtype" value="uint8,uint16" />
+            <validator type="dataset_metadata_equal" metadata_name="num_unique_values" value="2,255" />
+            <validator type="dataset_metadata_equal" metadata_name="width" value="32,256" />
+            <validator type="dataset_metadata_equal" metadata_name="height" value="32,256" />
+            <validator type="dataset_metadata_equal" metadata_name="channels" value="3,0" />
+            <validator type="dataset_metadata_equal" metadata_name="depth" value="0,0" />
+            <validator type="dataset_metadata_equal" metadata_name="frames" value="0,0" />
+        </param>
+        <!-- Checks for unsupported file formats -->
+        <param name="input_unsupported_tiff_compression" type="data" format="data" optional="true">
+            <!-- If the compression of a TIFF is unsupported, the fields "axes" and "dtype" should still be there -->
+            <validator type="dataset_metadata_equal" metadata_name="axes" value="YX" />
+            <validator type="dataset_metadata_equal" metadata_name="dtype" value="bool" />
+            <!-- The other fields should be missing -->
+            <validator type="dataset_metadata_equal" metadata_name="num_unique_values" value="" />
+            <validator type="dataset_metadata_equal" metadata_name="width" value="" />
+            <validator type="dataset_metadata_equal" metadata_name="height" value="" />
+            <validator type="dataset_metadata_equal" metadata_name="channels" value="" />
+            <validator type="dataset_metadata_equal" metadata_name="depth" value="" />
+            <validator type="dataset_metadata_equal" metadata_name="frames" value="" />
+        </param>
+    </inputs>
+    <outputs>
+        <data name="output" format="data" />
+    </outputs>
+    <tests>
+        <!-- Tests with TIFF files -->
+        <test>
+            <param name="input_axes_yx" value="im1_uint8.tif" />
+            <param name="input_axes_zcyx" value="im6_uint8.tif" />
+            <param name="input_dtype_uint8" value="im6_uint8.tif" />
+            <param name="input_dtype_uint16" value="im8_uint16.tif" />
+            <param name="input_dtype_float64" value="im4_float.tif" />
+            <param name="input_num_unique_values_2" value="im3_b.tif" />
+            <param name="input_num_unique_values_618" value="im4_float.tif" />
+            <param name="input_width_16" value="im7_uint8.tif" /><!-- axes: ZYX -->
+            <param name="input_width_32" value="im3_b.tif" /><!-- axes: YXS -->
+            <param name="input_height_8" value="im7_uint8.tif" /><!-- axes: ZYX -->
+            <param name="input_height_32" value="im3_b.tif" /><!-- axes: YXS -->
+            <param name="input_channels_0" value="im1_uint8.tif" />
+            <param name="input_channels_2" value="im5_uint8.tif" /><!-- axes: CYX -->
+            <param name="input_channels_3" value="im3_b.tif" /><!-- axes: YXS -->
+            <param name="input_depth_0" value="im1_uint8.tif" /><!-- axes: YXS -->
+            <param name="input_depth_25" value="im7_uint8.tif" /><!-- axes: ZYX -->
+            <param name="input_frames_0" value="im1_uint8.tif" /><!-- axes: YXS -->
+            <param name="input_frames_5" value="im8_uint16.tif" /><!-- axes: TYX -->
+            <param name="input_multi_series_tiff" value="im9_multiseries.tif" />
+            <param name="input_unsupported_tiff_compression" value="1.tiff" />
+            <expand macro="test-output" />
+        </test>
+        <!-- Tests with PNG files -->
+        <test>
+            <param name="input_axes_yx" value="im1_uint8.png" />
+            <param name="input_axes_yxc" value="im3_a.png" />
+            <param name="input_dtype_uint8" value="im1_uint8.png" />
+            <param name="input_num_unique_values_1" value="im2_a.png" />
+            <param name="input_num_unique_values_2" value="im2_b.png" />
+            <param name="input_width_32" value="im2_b.png" />
+            <param name="input_height_32" value="im2_b.png" />
+            <param name="input_channels_0" value="im1_uint8.png" />
+            <param name="input_channels_3" value="im3_a.png" />
+            <param name="input_depth_0" value="im1_uint8.png" />
+            <param name="input_frames_0" value="im1_uint8.png" />
+            <expand macro="test-output" />
+        </test>
+        <!-- End of tests -->
+    </tests>
+</tool>