feat(copy): add exclude_from parameter for rsync-based copy in Singularity builds

Author: allansp84

hpccm/primitives/copy.py

@@ -66,6 +66,17 @@ class copy(object):
 
     src: A file, or a list of files, to copy
 
+    exclude_from: String or list of strings.  One or more filenames
+    containing rsync-style exclude patterns (e.g., `.apptainerignore`).
+    Only used when building for Singularity or Apptainer.  If specified,
+    the copy operation is emitted in the `%setup` section using
+    `rsync --exclude-from=<file>` rather than the standard `%files`
+    copy directive.  This enables selective exclusion of files and
+    directories during the image build, for example to omit large data
+    files, caches, or temporary artifacts.  Multiple exclusion files may
+    be provided as a list or tuple.  The default is an empty list
+    (Singularity specific).
+
     # Examples
 
     ```python
@@ -80,6 +91,10 @@ class copy(object):
     copy(files={'a': '/tmp/a', 'b': '/opt/b'})
     ```
 
+    ```python
+    copy(src='.', dest='/opt/app', exclude_from='.apptainerignore')
+    ```
+
     """
 
     def __init__(self, **kwargs):
@@ -96,6 +111,14 @@ def __init__(self, **kwargs):
         self._post = kwargs.get('_post', '')  # Singularity specific
         self.__src = kwargs.get('src', '')
 
+        ef = kwargs.get('exclude_from', None)
+        if ef is None:
+            self.__exclude_from = []
+        elif isinstance(ef, (list, tuple)):
+            self.__exclude_from = list(ef)
+        else:
+            self.__exclude_from = [ef]
+
         if self._mkdir and self._post:
             logging.error('_mkdir and _post are mutually exclusive!')
             self._post = False # prefer _mkdir
@@ -179,6 +202,10 @@ def __str__(self):
                 else:
                     logging.warning(msg)
 
+            # If exclusion list is defined, switch to rsync copy method
+            if self.__exclude_from:
+                logging.info('copy: using rsync with exclude-from %s', self.__exclude_from)
+
             # Format:
             # %files
             #     src1 dest
@@ -211,6 +238,13 @@ def __str__(self):
                 dest = pair['dest']
                 src = pair['src']
 
+                # Use rsync if exclusion file provided and not multi-stage copy
+                if self.__exclude_from and not self.__from:
+                    excl_opts = ' '.join('--exclude-from={}'.format(x) for x in self.__exclude_from)
+                    pre.append('    mkdir -p ${{SINGULARITY_ROOTFS}}{0}'.format(dest))
+                    pre.append('    rsync -av {0} {1}/ ${{SINGULARITY_ROOTFS}}{2}/'.format(excl_opts, src, dest))
+                    continue
+
                 if self._post:
                     dest = '/'
 

test/test_copy.py

@@ -255,3 +255,38 @@ def test_from_temp_staging(self):
         """Singularity files from previous stage in tmp"""
         c = copy(_from='base', src='foo', dest='/var/tmp/foo')
         self.assertEqual(str(c), '%files from base\n    foo /var/tmp/foo')
+
+    @singularity
+    def test_exclude_from_single_singularity(self):
+        """rsync-based copy with exclude_from (single source)"""
+        c = copy(src='.', dest='/opt/app', exclude_from='.apptainerignore')
+        recipe = str(c)
+        self.assertIn('%setup', recipe)
+        self.assertIn('rsync -av', recipe)
+        self.assertIn('--exclude-from=.apptainerignore', recipe)
+        # Allow trailing %files section but ensure rsync setup comes first
+        self.assertTrue(recipe.strip().startswith('%setup'),
+                        "Expected rsync setup section to appear first")
+
+    @singularity
+    def test_exclude_from_multiple_singularity(self):
+        """rsync-based copy with multiple exclude_from files"""
+        c = copy(src='data', dest='/opt/data',
+                 exclude_from=['.ignore1', '.ignore2'])
+        recipe = str(c)
+        self.assertIn('%setup', recipe)
+        self.assertIn('rsync -av', recipe)
+        self.assertIn('--exclude-from=.ignore1', recipe)
+        self.assertIn('--exclude-from=.ignore2', recipe)
+        # Ensure setup section appears before %files
+        self.assertTrue(recipe.strip().startswith('%setup'),
+                        "Expected rsync setup section to appear first")
+
+    @docker
+    def test_exclude_from_docker_ignored(self):
+        """exclude_from ignored in Docker context"""
+        c = copy(src='.', dest='/opt/app', exclude_from='.apptainerignore')
+        recipe = str(c)
+        self.assertIn('COPY', recipe)
+        self.assertNotIn('rsync', recipe)
+        self.assertNotIn('%setup', recipe)