feat: add support for ARW, DNG, CR2 raw (#140)

This pull request for the issue #15 will introduce:

1. Support for RAW file processors.

  - [x] ARW
  - [ ] DNG
- yurij: doesn't actually work with DNG produced by Lightroom, disabled
in this PR
  - [x] CR2

The other raw format should work by adding the extension in the
IMAGE_REGEX constant but the below formats are not yet tested.

  -  CR3
  -  CRW

2. If a RAW file and a JPG file with the same name exist in the same
directory, the logic will prioritize processing the JPG file and ignore
the corresponding RAW file.

---------

Co-authored-by: Yurij Mikhalevich <[email protected]>

pyproject.toml

@@ -32,6 +32,7 @@ torchvision = [
   { version = "==0.17.2+cpu", source = "pytorch-cpu", markers = "sys_platform == 'linux' and platform_machine != 'aarch64'" }
 ]
 tqdm = "^4.65.0"
+rawpy = "^0.23.2"
 
 [tool.poetry.group.dev.dependencies]
 pycodestyle = ">=2.7,<3.0"

rclip/const.py

@@ -4,3 +4,8 @@
 IS_MACOS = sys.platform == 'darwin'
 IS_LINUX = sys.platform.startswith('linux')
 IS_WINDOWS = sys.platform == 'win32' or sys.platform == 'cygwin'
+
+# these images are always processed
+IMAGE_EXT = ["jpg", "jpeg", "png", "webp"]
+# RAW images are processed only if there is no processed image alongside it
+IMAGE_RAW_EXT = ["arw", "cr2"]

rclip/main.py

@@ -11,6 +11,7 @@
 from PIL import Image, ImageFile
 
 from rclip import db, fs, model
+from rclip.const import IMAGE_EXT, IMAGE_RAW_EXT
 from rclip.utils.preview import preview
 from rclip.utils.snap import check_snap_permissions, is_snap
 from rclip.utils import helpers
@@ -41,7 +42,7 @@ def is_image_meta_equal(image: db.Image, meta: ImageMeta) -> bool:
 
 class RClip:
   EXCLUDE_DIRS_DEFAULT = ['@eaDir', 'node_modules', '.git']
-  IMAGE_REGEX = re.compile(r'^.+\.(jpe?g|png|webp)$', re.I)
+  IMAGE_REGEX = re.compile(f'^.+\\.({"|".join([*IMAGE_EXT, *IMAGE_RAW_EXT])})$', re.I)
   DB_IMAGES_BEFORE_COMMIT = 50_000
 
   class SearchResult(NamedTuple):
@@ -67,7 +68,7 @@ def _index_files(self, filepaths: List[str], metas: List[ImageMeta]):
     filtered_paths: List[str] = []
     for path in filepaths:
       try:
-        image = Image.open(path)
+        image = helpers.read_image(path)
         images.append(image)
         filtered_paths.append(path)
       except PIL.UnidentifiedImageError as ex:
@@ -88,6 +89,18 @@ def _index_files(self, filepaths: List[str], metas: List[ImageMeta]):
         vector=vector.tobytes()
       ), commit=False)
 
+  def _does_processed_image_exist_for_raw(self, raw_path: str) -> bool:
+    """Check if there is a processed image alongside the raw one; doesn't support mixed-case extensions,
+    e.g. it won't detect the .JpG image, but will detect .jpg or .JPG"""
+
+    image_path = os.path.splitext(raw_path)[0]
+    for ext in IMAGE_EXT:
+      if os.path.isfile(image_path + "." + ext):
+        return True
+      if os.path.isfile(image_path + "." + ext.upper()):
+        return True
+    return False
+
   def ensure_index(self, directory: str):
     print(
       'checking images in the current directory for changes;'
@@ -113,7 +126,13 @@ def update_total_images(count: int):
       metas: List[ImageMeta] = []
       for entry in fs.walk(directory, self._exclude_dir_regex, self.IMAGE_REGEX):
         filepath = entry.path
-        image = self._db.get_image(filepath=filepath)
+
+        file_ext = helpers.get_file_extension(filepath)
+        if file_ext in IMAGE_RAW_EXT and self._does_processed_image_exist_for_raw(filepath):
+          images_processed += 1
+          pbar.update()
+          continue
+
         try:
           meta = get_image_meta(entry)
         except Exception as ex:
@@ -125,6 +144,7 @@ def update_total_images(count: int):
         images_processed += 1
         pbar.update()
 
+        image = self._db.get_image(filepath=filepath)
         if image and is_image_meta_equal(image, meta):
           self._db.remove_indexing_flag(filepath, commit=False)
           continue

rclip/utils/helpers.py

@@ -4,11 +4,13 @@
 import textwrap
 from PIL import Image, UnidentifiedImageError
 import re
+import numpy as np
+import rawpy
 import requests
 import sys
 from importlib.metadata import version
 
-from rclip.const import IS_LINUX, IS_MACOS, IS_WINDOWS
+from rclip.const import IMAGE_RAW_EXT, IS_LINUX, IS_MACOS, IS_WINDOWS
 
 
 MAX_DOWNLOAD_SIZE_BYTES = 50_000_000
@@ -186,15 +188,29 @@ def download_image(url: str) -> Image.Image:
   return img
 
 
+def get_file_extension(path: str) -> str:
+  return os.path.splitext(path)[1].lower()[1:]
+
+
+def read_raw_image_file(path: str):
+  raw = rawpy.imread(path)
+  rgb = raw.postprocess()
+  return Image.fromarray(np.array(rgb))
+
+
 def read_image(query: str) -> Image.Image:
   path = remove_prefix(query, 'file://')
   try:
-    img = Image.open(path)
+    file_ext = get_file_extension(path)
+    if file_ext in IMAGE_RAW_EXT:
+      image = read_raw_image_file(path)
+    else:
+      image = Image.open(path)
   except UnidentifiedImageError as e:
     # by default the filename on the UnidentifiedImageError is None
     e.filename = path
     raise e
-  return img
+  return image
 
 
 def is_http_url(path: str) -> bool:

rclip/utils/preview.py

@@ -3,6 +3,8 @@
 import os
 from PIL import Image
 
+from rclip.utils.helpers import read_image
+
 
 def _get_start_sequence():
   term_env_var = os.getenv('TERM')
@@ -19,7 +21,7 @@ def _get_end_sequence():
 
 
 def preview(filepath: str, img_height_px: int):
-  with Image.open(filepath) as img:
+  with read_image(filepath) as img:
     if img_height_px >= img.height:
       width_px, height_px = img.width, img.height
     else: