Merge pull request #362 from Glandos/performance_one_open_for_metadata

Performance: Read all metadata at once

sigal/gallery.py

@@ -40,8 +40,7 @@ from urllib.parse import quote as url_quote
 from click import get_terminal_size, progressbar
 
 from . import image, signals, video
-from .image import (get_exif_data, get_exif_tags, get_iptc_data, get_size,
-                    process_image)
+from .image import get_exif_tags, get_image_metadata, get_size, process_image
 from .settings import get_thumb
 from .utils import (Devnull, cached_property, check_or_create_dir, copy,
                     get_mime, is_valid_html5_video, read_markdown,
@@ -84,7 +83,10 @@ class Media:
         self.thumb_path = join(settings['destination'], path, self.thumb_name)
 
         self.logger = logging.getLogger(__name__)
+
+        self.file_metadata = None
         self._get_metadata()
+
         # default: title is the filename
         if not self.title:
             self.title = self.filename
@@ -196,32 +198,25 @@ class Image(Media):
 
     def _get_metadata(self):
         super()._get_metadata()
+        self.file_metadata = get_image_metadata(self.src_path)
+
         # If a title or description hasn't been obtained by other means, look
         #  for the information in IPTC fields
         if self.title and self.description:
             # Nothing to do - we already have title and description
             return
 
-        try:
-            iptc_data = get_iptc_data(self.src_path)
-        except Exception as e:
-            self.logger.warning('Could not read IPTC data from %s: %s',
-                                self.src_path, e)
-        else:
-            if not self.title and iptc_data.get('title'):
-                self.title = iptc_data['title']
-            if not self.description and iptc_data.get('description'):
-                self.description = iptc_data['description']
+        iptc_data = self.file_metadata['iptc']
+        if not self.title and iptc_data.get('title'):
+            self.title = iptc_data['title']
+        if not self.description and iptc_data.get('description'):
+            self.description = iptc_data['description']
 
     @cached_property
     def raw_exif(self):
         """If not `None`, contains the raw EXIF tags."""
-        try:
-            return (get_exif_data(self.src_path)
-                    if self.ext in ('.jpg', '.jpeg') else None)
-        except Exception as e:
-            self.logger.warning('Could not read EXIF data from %s: %s',
-                                self.src_path, e)
+        if self.ext in ('.jpg', '.jpeg'):
+            return self.file_metadata['exif']
 
     @cached_property
     def size(self):
@@ -460,7 +455,7 @@ class Album:
                     # fallback to the size of src_path if dst_path is missing
                     size = f.size
                     if size is None:
-                        size = get_size(f.src_path)
+                        size = f.file_metadata['size']
 
                     if size['width'] > size['height']:
                         self._thumbnail = (url_quote(self.name) + '/' +

sigal/image.py

@@ -56,6 +56,10 @@ def _has_exif_tags(img):
 
 
 def _read_image(file_path):
+    # The image is already opened
+    if isinstance(file_path, PILImage.Image):
+        return file_path
+
     with warnings.catch_warnings(record=True) as caught_warnings:
         im = PILImage.open(file_path)
 
@@ -203,10 +207,7 @@ def get_size(file_path):
         logger.error("Could not read size of %s due to %r", file_path, e)
     else:
         width, height = im.size
-        return {
-            'width': width,
-            'height': height
-        }
+        return {'width': width, 'height': height}
 
 
 def get_exif_data(filename):
@@ -271,6 +272,34 @@ def get_iptc_data(filename):
     return iptc_data
 
 
+def get_image_metadata(filename):
+    logger = logging.getLogger(__name__)
+    exif, iptc, size = {}, {}, {}
+
+    try:
+        img = _read_image(filename)
+    except Exception as e:
+        logger.error('Could not open image %s metadata: %s', filename, e)
+    else:
+        try:
+            if os.path.splitext(filename)[1].lower() in ('.jpg', '.jpeg'):
+                exif = get_exif_data(img)
+        except Exception as e:
+            logger.warning('Could not read EXIF data from %s: %s', filename, e)
+
+        try:
+            iptc = get_iptc_data(img)
+        except Exception as e:
+            logger.warning('Could not read IPTC data from %s: %s', filename, e)
+
+        try:
+            size = get_size(img)
+        except Exception as e:
+            logger.warning('Could not read size from %s: %s', filename, e)
+
+    return {'exif': exif, 'iptc': iptc, 'size': size}
+
+
 def dms_to_degrees(v):
     """Convert degree/minute/second to decimal degrees."""