Zarr Tile Sink

sources/zarr/large_image_source_zarr/__init__.py

@@ -1,19 +1,23 @@
 import math
 import os
+import shutil
+import tempfile
 import threading
+import uuid
 from importlib.metadata import PackageNotFoundError
 from importlib.metadata import version as _importlib_version
+from pathlib import Path
 
 import numpy as np
 import packaging.version
 import zarr
 
 import large_image
 from large_image.cache_util import LruCacheMetaclass, methodcache
-from large_image.constants import TILE_FORMAT_NUMPY, SourcePriority
+from large_image.constants import NEW_IMAGE_PATH_FLAG, TILE_FORMAT_NUMPY, SourcePriority
 from large_image.exceptions import TileSourceError, TileSourceFileNotFoundError
 from large_image.tilesource import FileTileSource
-from large_image.tilesource.utilities import nearPowerOfTwo
+from large_image.tilesource.utilities import _imageToNumpy, nearPowerOfTwo
 
 try:
     __version__ = _importlib_version(__name__)
@@ -52,6 +56,13 @@
         """
         super().__init__(path, **kwargs)
 
+        if str(path).startswith(NEW_IMAGE_PATH_FLAG):
+            self._initNew(path, **kwargs)
+        else:
+            self._initOpen(**kwargs)
+        self._tileLock = threading.RLock()
+
+    def _initOpen(self, **kwargs):
         self._largeImagePath = str(self._getLargeImagePath())
         self._zarr = None
         if not os.path.isfile(self._largeImagePath) and '//:' not in self._largeImagePath:
@@ -79,7 +90,43 @@
         except Exception:
             msg = 'File cannot be opened -- not an OME NGFF file or understandable zarr file.'
             raise TileSourceError(msg)
-        self._tileLock = threading.RLock()
+
+    def _initNew(self, path, **kwargs):
+        """
+        Initialize the tile class for creating a new image.
+        """
+        self._tempfile = tempfile.NamedTemporaryFile(suffix=path)
+        self._zarr_store = zarr.SQLiteStore(self._tempfile.name)
+        self._zarr = zarr.open(self._zarr_store, mode='w')
+        # Make unpickleable
+        self._unpickleable = True
+        self._largeImagePath = None
+        self.sizeX = self.sizeY = self.levels = 0
+        self.tileWidth = self.tileHeight = self._tileSize
+        self._frames = [0]
+        self._cacheValue = str(uuid.uuid4())
+        self._output = None
+        self._editable = True
+        self._bandRanges = None
+        self._addLock = threading.RLock()
+        self._framecount = 0
+        self._mm_x = 0
+        self._mm_y = 0
+
+    def __del__(self):
+        try:
+            self._zarr.close()
+            self._tempfile.close()
+        except Exception:
+            pass
+
+    def _checkEditable(self):
+        """
+        Raise an exception if this is not an editable image.
+        """
+        if not self._editable:
+            msg = 'Not an editable image'
+            raise TileSourceError(msg)
 
     def _getGeneralAxes(self, arr):
         """
@@ -396,6 +443,9 @@
 
     @methodcache()
     def getTile(self, x, y, z, pilImageAllowed=False, numpyAllowed=False, **kwargs):
+        if self._levels is None:
+            self._validateZarr()
+
         frame = self._getFrame(**kwargs)
         self._xyzInRange(x, y, z, frame, self._framecount)
         x0, y0, x1, y1, step = self._xyzToCorners(x, y, z)
@@ -438,6 +488,204 @@
         return self._outputTile(tile, TILE_FORMAT_NUMPY, x, y, z,
                                 pilImageAllowed, numpyAllowed, **kwargs)
 
+    def addTile(self, tile, x=0, y=0, mask=None, axes=None, **kwargs):
+        """
+        Add a numpy or image tile to the image, expanding the image as needed
+        to accommodate it.  Note that x and y can be negative.  If so, the
+        output image (and internal memory access of the image) will act as if
+        the 0, 0 point is the most negative position.  Cropping is applied
+        after this offset.
+
+        :param tile: a numpy array, PIL Image, or a binary string
+            with an image.  The numpy array can have 2 or 3 dimensions.
+        :param x: location in destination for upper-left corner.
+        :param y: location in destination for upper-left corner.
+        :param mask: a 2-d numpy array (or 3-d if the last dimension is 1).
+            If specified, areas where the mask is false will not be altered.
+        :param axes: a string or list of strings specifying the names of axes
+            in the same order as the tile dimensions
+        :param kwargs: start locations for any additional axes
+        """
+        # TODO: improve band bookkeeping
+
+        placement = {
+            'x': x,
+            'y': y,
+            **kwargs,
+        }
+        if axes is None:
+            if len(tile.shape) == 2:
+                axes = 'yx'
+            elif len(tile.shape) == 3:
+                axes = 'yxs'
+            else:
+                axes = ''
+        if isinstance(axes, str):
+            axes = axes.lower()
+        elif isinstance(axes, list):
+            axes = [x.lower() for x in axes]
+        else:
+            err = 'Invalid type for axes. Must be str or list[str].'
+            raise ValueError(err)
+
+        self._checkEditable()
+        tile, mode = _imageToNumpy(tile)
+        self._levels = None  # reset zarr validation
+
+        while len(tile.shape) < len(axes):
+            tile = np.expand_dims(tile, axis=0)
+
+        current_arrays = dict(self._zarr.arrays())
+        if 'root' in current_arrays:
+            root = current_arrays['root']
+        else:
+            root = self._zarr.create_dataset('root', data=tile)
+
+        new_dims = {
+            a: max(
+                root.shape[i],
+                placement.get(
+                    a,
+                    0) +
+                tile.shape[i]) for i,
+            a in enumerate(axes)}
+        root_data = np.pad(
+            root,
+            [(0, d - root.shape[i]) for i, d in enumerate(new_dims.values())],
+            mode='empty',
+        )
+
+        tile_data = np.pad(
+            tile,
+            [(placement.get(a, 0), d - placement.get(a, 0) - tile.shape[i])
+             for i, (a, d) in enumerate(new_dims.items())],
+            mode='empty',
+        )
+
+        if mask is None:
+            mask = np.ones(tile.shape[:-1])
+        while len(mask.shape) < len(tile_data.shape):
+            mask = np.expand_dims(mask, axis=-1)
+            mask = np.repeat(mask, tile_data.shape[len(mask.shape) - 1], axis=-1)
+        mask_data = np.pad(
+            mask,
+            [(placement.get(a, 0), d - placement.get(a, 0) - mask.shape[i])
+             for i, (a, d) in enumerate(new_dims.items())],
+            mode='constant',
+            constant_values=0,
+        )
+        mask_data = mask_data.astype(bool)
+        np.copyto(root_data, tile_data, where=mask_data)
+
+        with self._addLock:
+            # This will rechunk data when necessary, according to new shape
+            root = self._zarr.create_dataset('root', data=root_data, overwrite=True)
+
+            # Edit OME metadata
+            self._zarr.attrs.update({
+                'multiscales': [{
+                    'version': '0.5-dev',
+                    'axes': [{
+                        'name': a,
+                        'type': 'space' if a in ['x', 'y'] else 'other',
+                    } for a in axes],
+                    'datasets': [{'path': 0}],
+                }],
+                'omero': {'version': '0.5-dev'},
+            })
+
+    @property
+    def crop(self):
+        """
+        Crop only applies to the output file, not the internal data access.
+
+        It consists of x, y, w, h in pixels.
+        """
+        return getattr(self, '_crop', None)
+
+    @crop.setter
+    def crop(self, value):
+        self._checkEditable()
+        if value is None:
+            self._crop = None
+            return
+        x, y, w, h = value
+        x = int(x)
+        y = int(y)
+        w = int(w)
+        h = int(h)
+        if x < 0 or y < 0 or w <= 0 or h <= 0:
+            msg = 'Crop must have non-negative x, y and positive w, h'
+            raise TileSourceError(msg)
+        self._crop = (x, y, w, h)
+
+    def write(
+        self,
+        path,
+        lossy=True,
+        alpha=True,
+        overwriteAllowed=True,
+    ):
+        """
+        Output the current image to a file.
+
+        :param path: output path.
+        :param lossy: if false, emit a lossless file.
+        :param alpha: True if an alpha channel is allowed.
+        :param overwriteAllowed: if False, raise an exception if the output
+            path exists.
+        """
+        # TODO: compute half, quarter, etc. resolutions
+        if not overwriteAllowed and os.path.exists(path):
+            raise TileSourceError('Output path exists (%s).' % str(path))
+
+        self._validateZarr()
+        suffix = Path(path).suffix
+        data_file = self._tempfile
+        data_store = self._zarr_store
+
+        if self.crop:
+            x, y, w, h = self.crop
+            current_arrays = dict(self._zarr.arrays())
+            # create new temp storage for cropped data
+            data_file = tempfile.NamedTemporaryFile()
+            data_store = zarr.SQLiteStore(data_file.name)
+            cropped_zarr = zarr.open(data_store, mode='w')
+            for arr_name in current_arrays:
+                arr = np.array(current_arrays[arr_name])
+                cropped_arr = arr.take(
+                    indices=range(x, x + w),
+                    axis=self._axes.get('x'),
+                ).take(
+                    indices=range(y, y + h),
+                    axis=self._axes.get('y'),
+                )
+                cropped_zarr.create_dataset(arr_name, data=cropped_arr, overwrite=True)
+                cropped_zarr.attrs.update(self._zarr.attrs)
+
+        data_file.flush()
+
+        if suffix in ['.db', '.sqlite']:
+            shutil.copy2(data_file.name, path)
+
+        elif suffix == '.zip':
+            zip_store = zarr.storage.ZipStore(path)
+            zarr.copy_store(data_store, zip_store)
+            zip_store.close()
+
+        elif suffix == '.zarr':
+            dir_store = zarr.storage.DirectoryStore(path)
+            zarr.copy_store(data_store, dir_store)
+            dir_store.close()
+
+        else:
+            from large_image_converter import convert
+
+            convert(data_file.name, path, overwrite=overwriteAllowed)
+
+        if self.crop:
+            data_file.close()
+
 
 def open(*args, **kwargs):
     """
@@ -451,3 +699,11 @@
     Check if an input can be read by the module class.
     """
     return ZarrFileTileSource.canRead(*args, **kwargs)
+
+
+def new(*args, **kwargs):
+    """
+    Create a new image, collecting the results from patches of numpy arrays or
+    smaller images.
+    """
+    return ZarrFileTileSource(NEW_IMAGE_PATH_FLAG + str(uuid.uuid4()), *args, **kwargs)