crate-with-subcrate creation: support zipped output

Author: simleo

rocrate/rocrate.py

@@ -652,6 +652,10 @@ def _stream_zip(self, chunk_size=8192, out_path=None):
             while chunk := buffer.read(chunk_size):
                 yield chunk
 
+    def _all_streams(self, chunk_size=8192):
+        for writeable_entity in self.data_entities + self.default_entities:
+            yield from writeable_entity.stream(chunk_size=chunk_size)
+
     def add_workflow(
             self, source=None, dest_path=None, fetch_remote=False, validate_url=False, properties=None,
             main=False, lang="cwl", lang_version=None, gen_cwl=False, cls=ComputationalWorkflow,
@@ -924,6 +928,12 @@ def write(self, base_path):
         if self.crate.mode == Mode.CREATE:
             self.get_crate().write(base_path / unquote(self.id))
 
+    def stream(self, chunk_size=8192):
+        yield from super().stream(chunk_size=chunk_size)
+        if self.crate.mode == Mode.CREATE:
+            for path, chunk in self.get_crate()._all_streams(chunk_size=chunk_size):
+                yield os.path.join(unquote(self.id), path), chunk
+
 
 def make_workflow_rocrate(workflow_path, wf_type, include_files=[],
                           fetch_remote=False, cwl=None, diagram=None):

test/test_write.py

@@ -634,7 +634,7 @@ def test_write_subcrate(test_data_dir, tmpdir, to_zip):
     crate = ROCrate(test_data_dir / "crate_with_subcrates", load_subcrates=True)
     out_path = tmpdir / "ro_crate_out"
     if to_zip:
-        zip_path = tmpdir / 'ro_crate_out.crate.zip'
+        zip_path = tmpdir / 'ro_crate_out.zip'
         crate.write_zip(zip_path)
         with zipfile.ZipFile(zip_path, "r") as zf:
             zf.extractall(out_path)
@@ -651,7 +651,8 @@ def test_write_subcrate(test_data_dir, tmpdir, to_zip):
     assert (out_path / "subcrate" / "subsubcrate" / "ro-crate-metadata.json").is_file()
 
 
-def test_subcrates_creation(test_data_dir, tmpdir):
+@pytest.mark.parametrize("to_zip", [False, True])
+def test_subcrates_creation(test_data_dir, tmpdir, to_zip):
     crate = ROCrate()
     crate.add_file(test_data_dir / "sample_file.txt")
     subcrate = crate.add_subcrate(dest_path="subcrate/")
@@ -671,7 +672,14 @@ def test_subcrates_creation(test_data_dir, tmpdir):
     assert subcrate_crate.get("subsubcrate/setup.cfg") is subsubf
 
     out_path = tmpdir / "ro_crate_out"
-    crate.write(out_path)
+    if to_zip:
+        zip_path = tmpdir / 'ro_crate_out.zip'
+        crate.write_zip(zip_path)
+        with zipfile.ZipFile(zip_path, "r") as zf:
+            zf.extractall(out_path)
+    else:
+        crate.write(out_path)
+
     assert (out_path / "ro-crate-metadata.json").is_file()
     assert (out_path / "sample_file.txt").is_file()
     assert (out_path / "subcrate" / "ro-crate-metadata.json").is_file()