Merge branch 'previews' into develop

9 files changed, 285 insertions, 9 deletions

diff --git a/bsie/reader/chain.py b/bsie/reader/chain.py
index 5e9e0d5..1dbc52b 100644
--- a/bsie/reader/chain.py
+++ b/bsie/reader/chain.py
@@ -73,16 +73,19 @@ class ReaderChain(base.Reader, typing.Generic[T_CONTENT]):
         return hash((super().__hash__(), self._children))
 
     def __call__(self, path: str) -> T_CONTENT:
-        raise_error = errors.UnsupportedFileFormatError
+        raise_error = False
         for child in self._children:
             try:
                 return child(path)
             except errors.UnsupportedFileFormatError:
+                # child cannot read the file, skip.
                 pass
             except errors.ReaderError:
-                # child cannot read the file, skip.
-                raise_error = errors.ReaderError # type: ignore [assignment] # mypy is confused
+                # child failed to read the file, skip.
+                raise_error = True
 
-        raise raise_error(path)
+        if raise_error:
+            raise errors.ReaderError(path)
+        raise errors.UnsupportedFileFormatError(path)
 
 ## EOF ##
diff --git a/bsie/reader/image/__init__.py b/bsie/reader/image/__init__.py
index 1f290b5..c5d2a2a 100644
--- a/bsie/reader/image/__init__.py
+++ b/bsie/reader/image/__init__.py
@@ -27,7 +27,6 @@ __all__: typing.Sequence[str] = (
 
 ## code ##
 
-# FIXME: Check if PIL.Image or PIL.Image.Image, or if version-dependent
 class Image(chain.ReaderChain[PIL.Image.Image]): # pylint: disable=too-few-public-methods
     """Read an image file."""
 
diff --git a/bsie/reader/image/_pillow.py b/bsie/reader/image/_pillow.py
index 3144509..5b2bdf2 100644
--- a/bsie/reader/image/_pillow.py
+++ b/bsie/reader/image/_pillow.py
@@ -27,7 +27,7 @@ __all__: typing.Sequence[str] = (
 class PillowImage(base.Reader):
     """Use PIL to read content of a variety of image file types."""
 
-    def __call__(self, path: str) -> PIL.Image:
+    def __call__(self, path: str) -> PIL.Image.Image:
         try:
             # open file with PIL
             return PIL.Image.open(path)
diff --git a/bsie/reader/image/_raw.py b/bsie/reader/image/_raw.py
index cd60453..257fdb3 100644
--- a/bsie/reader/image/_raw.py
+++ b/bsie/reader/image/_raw.py
@@ -32,17 +32,17 @@ class RawImage(base.Reader):
     """Use rawpy to read content of raw image file types."""
 
     # file matcher
-    match: filematcher.Matcher
+    _match: filematcher.Matcher
 
     # additional kwargs to rawpy's postprocess
-    rawpy_kwargs: typing.Dict[str, typing.Any]
+    _rawpy_kwargs: typing.Dict[str, typing.Any]
 
     def __init__(self, **rawpy_kwargs):
         match_rule = rawpy_kwargs.pop('file_match_rule', MATCH_RULE)
         self._match = filematcher.parse(match_rule)
         self._rawpy_kwargs = rawpy_kwargs
 
-    def __call__(self, path: str) -> PIL.Image:
+    def __call__(self, path: str) -> PIL.Image.Image:
         # perform quick checks first
         if not self._match(path):
             raise errors.UnsupportedFileFormatError(path)
diff --git a/bsie/reader/preview/__init__.py b/bsie/reader/preview/__init__.py
new file mode 100644
index 0000000..3e69a4a
--- /dev/null
+++ b/bsie/reader/preview/__init__.py
@@ -0,0 +1,39 @@
+"""
+
+Part of the bsie module.
+A copy of the license is provided with the project.
+Author: Matthias Baumgartner, 2022
+"""
+# imports
+import typing
+
+# external imports
+import PIL.Image
+
+# inner-module imports
+from .. import chain
+
+# constants
+_FILE_FORMAT_READERS: typing.Sequence[str] = (
+    # native image formats
+    __package__ + '._pillow.PillowPreviewReader',
+    __package__ + '._rawpy.RawpyPreviewReader',
+    # multiformat readers
+    __package__ + '._pg.PreviewGeneratorReader',
+    )
+
+# exports
+__all__: typing.Sequence[str] = (
+    'Preview',
+    )
+
+
+## code ##
+
+class Preview(chain.ReaderChain[typing.Callable[[int], PIL.Image.Image]]): # pylint: disable=too-few-public-methods
+    """Create a preview from a file."""
+
+    def __init__(self, cfg: typing.Optional[typing.Any] = None):
+        super().__init__(_FILE_FORMAT_READERS, cfg)
+
+## EOF ##
diff --git a/bsie/reader/preview/_pg.py b/bsie/reader/preview/_pg.py
new file mode 100644
index 0000000..097c513
--- /dev/null
+++ b/bsie/reader/preview/_pg.py
@@ -0,0 +1,86 @@
+"""
+
+Part of the bsie module.
+A copy of the license is provided with the project.
+Author: Matthias Baumgartner, 2022
+"""
+# standard imports
+from functools import partial
+import contextlib
+import io
+import os
+import shutil
+import tempfile
+import typing
+
+# external imports
+from preview_generator.manager import PreviewManager
+import PIL.Image
+
+# bsie imports
+from bsie.utils import errors
+
+# inner-module imports
+from .. import base
+
+# exports
+__all__: typing.Sequence[str] = (
+    'PreviewGeneratorReader',
+    )
+
+
+## code ##
+
+class PreviewGeneratorReader(base.Reader):
+    """Uses preview_generator to create previews for various data formats.
+    See `https://github.com/algoo/preview-generator`_ for details.
+    """
+
+    # PreviewManager instance.
+    _mngr: PreviewManager
+
+    # Set of mime types supported by PreviewManager.
+    _supported_mimetypes: typing.Set[str]
+
+    # PreviewManager cache.
+    _cache: str
+
+    # Determines whether the cache directory should be deleted after use.
+    _cleanup: bool
+
+    def __init__(self, cache: typing.Optional[str] = None):
+        # initialize cache directory
+        # TODO: initialize in memory, e.g., via PyFilesystem
+        if cache is None:
+            self._cache = tempfile.mkdtemp(prefix='bsie-preview-cache-')
+            self._cleanup = True
+        else:
+            self._cache = cache
+            self._cleanup = False
+        # create preview generator
+        with contextlib.redirect_stderr(io.StringIO()):
+            self._mngr = PreviewManager(self._cache, create_folder=True)
+        self._supported_mimetypes = set(self._mngr.get_supported_mimetypes())
+
+    def __del__(self):
+        if self._cleanup:
+            shutil.rmtree(self._cache, ignore_errors=True)
+
+    def __call__(self, path: str) -> typing.Callable[[int], PIL.Image.Image]:
+        if not os.path.exists(path):
+            raise errors.ReaderError(path)
+        if self._mngr.get_mimetype(path) not in self._supported_mimetypes:
+            raise errors.UnsupportedFileFormatError(path)
+        return partial(self._preview_callback, path)
+
+    def _preview_callback(self, path: str, max_side: int) -> PIL.Image.Image:
+        """Produce a jpeg preview of *path* with at most *max_side* side length."""
+        try:
+            # generate the preview
+            preview_path = self._mngr.get_jpeg_preview(path, width=max_side, height=max_side)
+            # open the preview and return
+            return PIL.Image.open(preview_path)
+        except Exception as err: # FIXME: less generic exception!
+            raise errors.ReaderError(path) from err
+
+## EOF ##
diff --git a/bsie/reader/preview/_pillow.py b/bsie/reader/preview/_pillow.py
new file mode 100644
index 0000000..174d509
--- /dev/null
+++ b/bsie/reader/preview/_pillow.py
@@ -0,0 +1,44 @@
+"""
+
+Part of the bsie module.
+A copy of the license is provided with the project.
+Author: Matthias Baumgartner, 2022
+"""
+# standard imports
+from functools import partial
+import typing
+
+# external imports
+import PIL.Image
+
+# bsie imports
+from bsie.utils import errors
+
+# inner-module imports
+from . import utils
+from .. import base
+
+# exports
+__all__: typing.Sequence[str] = (
+    'PillowPreviewReader',
+    )
+
+
+## code ##
+
+class PillowPreviewReader(base.Reader):
+    """Produce previews for image files using the Pillow library."""
+
+    def __call__(self, path: str) -> typing.Callable[[int], PIL.Image.Image]:
+        try:
+            # open file with PIL
+            img = PIL.Image.open(path)
+            # return callback
+            return partial(utils.resize, img)
+        except PIL.UnidentifiedImageError as err:
+            # failed to open, skip file
+            raise errors.UnsupportedFileFormatError(path) from err
+        except IOError as err:
+            raise errors.ReaderError(path) from err
+
+# EOF ##
diff --git a/bsie/reader/preview/_rawpy.py b/bsie/reader/preview/_rawpy.py
new file mode 100644
index 0000000..2c20a48
--- /dev/null
+++ b/bsie/reader/preview/_rawpy.py
@@ -0,0 +1,66 @@
+"""
+
+Part of the bsie module.
+A copy of the license is provided with the project.
+Author: Matthias Baumgartner, 2022
+"""
+# standard imports
+from functools import partial
+import typing
+
+# external imports
+import PIL.Image
+import rawpy
+
+# bsie imports
+from bsie.utils import errors, filematcher
+
+# inner-module imports
+from . import utils
+from .. import base
+
+# constants
+MATCH_RULE = 'mime={image/x-nikon-nef} | extension={nef}'
+
+# exports
+__all__: typing.Sequence[str] = (
+    'RawpyPreviewReader',
+    )
+
+
+## code ##
+
+class RawpyPreviewReader(base.Reader):
+    """Produce previews for raw image files using the rawpy library."""
+
+    # file matcher
+    _match: filematcher.Matcher
+
+    # additional kwargs to rawpy's postprocess
+    _rawpy_kwargs: typing.Dict[str, typing.Any]
+
+    def __init__(self, **rawpy_kwargs):
+        match_rule = rawpy_kwargs.pop('file_match_rule', MATCH_RULE)
+        self._match = filematcher.parse(match_rule)
+        self._rawpy_kwargs = rawpy_kwargs
+
+    def __call__(self, path: str) -> typing.Callable[[int], PIL.Image.Image]:
+        # perform quick checks first
+        if not self._match(path):
+            raise errors.UnsupportedFileFormatError(path)
+
+        try:
+            # open file with rawpy
+            ary = rawpy.imread(path).postprocess(**self._rawpy_kwargs)
+            # convert to PIL.Image
+            img = PIL.Image.fromarray(ary)
+            # return callback
+            return partial(utils.resize, img)
+
+        except (rawpy.LibRawFatalError, # pylint: disable=no-member # pylint doesn't find the errors
+                rawpy.NotSupportedError, # pylint: disable=no-member
+                rawpy.LibRawNonFatalError, # pylint: disable=no-member
+               ) as err:
+            raise errors.ReaderError(path) from err
+
+## EOF ##
diff --git a/bsie/reader/preview/utils.py b/bsie/reader/preview/utils.py
new file mode 100644
index 0000000..2ef1562
--- /dev/null
+++ b/bsie/reader/preview/utils.py
@@ -0,0 +1,39 @@
+"""
+
+Part of the tagit module.
+A copy of the license is provided with the project.
+Author: Matthias Baumgartner, 2022
+"""
+# standard imports
+import typing
+
+# external imports
+import PIL.Image
+
+# exports
+__all__: typing.Sequence[str] = (
+    'resize',
+    )
+
+
+## code ##
+
+def resize(
+        img: PIL.Image.Image,
+        max_size: int,
+        ) -> PIL.Image.Image:
+    """Resize an image to a given maximum side length."""
+    # determine target dimensions
+    ratio = img.width / img.height
+    if img.width > img.height:
+        width, height = max_size, round(max_size / ratio)
+    else:
+        width, height = round(ratio * max_size), max_size
+    # rescale and return
+    return img.resize(
+        (width, height),
+        resample=PIL.Image.Resampling.LANCZOS, # create high-quality image
+        reducing_gap=3.0, # optimize computation via fast size reduction
+        )
+
+## EOF ##