New settings to filter directories and files with pattern matching (ref #63 #80).

docs/changelog.rst

@@ -22,6 +22,8 @@ Released on 2014-xx-xx.
   ``albums_sort_reverse``, ``medias_sort_attr``, ``medias_sort_reverse``.
 - New setting ``autorotate_images`` to disable autorotation of images, and warn
   about the incompatibility between autorotation and EXIF copy.
+- New settings ``ignore_directories`` and ``ignore_files`` to filter
+  directories and files with pattern matching.
 
 Version 0.6.0
 ~~~~~~~~~~~~~

sigal/gallery.py

@@ -24,6 +24,7 @@
 from __future__ import absolute_import, print_function
 
 import codecs
+import fnmatch
 import logging
 import markdown
 import multiprocessing
@@ -393,19 +394,42 @@ class Gallery(object):
         albums = self.albums = {}
         src_path = self.settings['source']
 
+        ignore_dirs = settings['ignore_directories']
+        ignore_files = settings['ignore_files']
+
         for path, dirs, files in os.walk(src_path, followlinks=True,
                                          topdown=False):
             relpath = os.path.relpath(path, src_path)
 
+            # Test if the directory match the ignore_dirs settings
+            if ignore_dirs and any(fnmatch.fnmatch(relpath, ignore)
+                                   for ignore in ignore_dirs):
+                self.logger.info('Ignoring %s', relpath)
+                continue
+
+            # Remove files that match the ignore_files settings
+            if ignore_files:
+                files_path = {join(relpath, f) for f in files}
+                for ignore in ignore_files:
+                    files_path -= set(fnmatch.filter(files_path, ignore))
+
+                self.logger.debug('Files before filtering: %r', files)
+                files = [os.path.split(f)[1] for f in files_path]
+                self.logger.debug('Files after filtering: %r', files)
+
+            # Remove sub-directories that have been ignored in a previous
+            # iteration (as topdown=False, sub-directories are processed before
+            # their parent
             for d in dirs[:]:
                 path = join(relpath, d) if relpath != '.' else d
                 if path not in albums.keys():
                     dirs.remove(d)
 
-            album = Album(relpath, self.settings, dirs, files, self)
+            album = Album(relpath, settings, dirs, files, self)
 
             if not album.medias and not album.albums:
                 self.logger.info('Skip empty album: %r', album)
+                # TODO delete empty directories
             else:
                 albums[relpath] = album
 

sigal/settings.py

@@ -28,6 +28,7 @@ from pprint import pformat
 
 from .compat import PY2
 
+
 _DEFAULT_CONFIG = {
     'adjust_options': {'color': 1.0, 'brightness': 1.0,
                        'contrast': 1.0, 'sharpness': 1.0},
@@ -38,6 +39,8 @@ _DEFAULT_CONFIG = {
     'destination': '_build',
     'files_to_copy': (),
     'google_analytics': '',
+    'ignore_directories': [],
+    'ignore_files': [],
     'img_ext_list': ['.jpg', '.jpeg', '.JPG', '.JPEG', '.png'],
     'img_processor': 'ResizeToFit',
     'img_size': (640, 480),

sigal/templates/sigal.conf.py

@@ -71,6 +71,12 @@ thumb_size = (280, 210)
 # Reverse sort for medias
 # medias_sort_reverse = False
 
+# Filter directories and files.
+# The settings take a list of patterns matched with the fnmatch module:
+# http://docs.python.org/2/library/fnmatch.html
+ignore_directories = []
+ignore_files = []
+
 # Jpeg options
 # jpg_options = {'quality': 85,
 #                'optimize': True,

tests/test_gallery.py

@@ -197,3 +197,18 @@ def test_empty_dirs(settings):
     gal = Gallery(settings, ncpu=1)
     assert 'empty' not in gal.albums
     assert 'dir1/empty' not in gal.albums
+
+
+def test_ignores(settings, tmpdir):
+    tmp = str(tmpdir)
+    settings['destination'] = tmp
+    settings['ignore_directories'] = ['*test2']
+    settings['ignore_files'] = ['dir2/Hubble*', '*.png']
+    gal = Gallery(settings, ncpu=1)
+    gal.build()
+
+    assert 'test2' not in os.listdir(join(tmp, 'dir1'))
+    assert 'archlinux-kiss-1024x640.png' not in os.listdir(
+        join(tmp, 'dir1', 'test1'))
+    assert 'Hubble Interacting Galaxy NGC 5257.jpg' not in os.listdir(
+        join(tmp, 'dir2'))