From 5e9a597acd3a8a3b937026b54ca339dc9495cc30 Mon Sep 17 00:00:00 2001
From: Daniel Bauer <daniel.bauer@senckenberg.de>
Date: Wed, 15 Jan 2025 13:38:32 +0100
Subject: [PATCH] remove code duplication in dataset

---
 rocrate/model/dataset.py | 48 +++++++++++-----------------------------
 rocrate/rocrate.py       |  1 -
 2 files changed, 13 insertions(+), 36 deletions(-)

diff --git a/rocrate/model/dataset.py b/rocrate/model/dataset.py
index dfb9048c..976762ee 100644
--- a/rocrate/model/dataset.py
+++ b/rocrate/model/dataset.py
@@ -48,42 +48,21 @@ def format_id(self, identifier):
         return identifier.rstrip("/") + "/"
 
     def write(self, base_path):
-        out_path = Path(base_path) / self.id
-        if is_url(str(self.source)):
-            if self.validate_url and not self.fetch_remote:
-                with urlopen(self.source) as _:
-                    self._jsonld['sdDatePublished'] = iso_now()
-            if self.fetch_remote:
-                self.__get_parts(out_path)
-        else:
-            if self.source is None:
-                out_path.mkdir(parents=True, exist_ok=True)
-            else:
-                if not Path(self.source).exists():
-                    raise FileNotFoundError(
-                        errno.ENOENT, os.strerror(errno.ENOENT), str(self.source)
-                    )
-                out_path.mkdir(parents=True, exist_ok=True)
-                if not self.crate.source:
-                    self.crate._copy_unlisted(self.source, out_path)
-
-    def __get_parts(self, out_path):
-        out_path.mkdir(parents=True, exist_ok=True)
-        base = self.source.rstrip("/")
-        for entry in self._jsonld.get("hasPart", []):
-            try:
-                part = entry["@id"]
-            except KeyError:
-                continue
-            if is_url(part) or part.startswith("/"):
-                raise RuntimeError(f"'{self.source}': part '{part}' is not a relative path")
-            part_uri = f"{base}/{part}"
-            part_out_path = out_path / part
-            with urlopen(part_uri) as r, open(part_out_path, 'wb') as f:
-                shutil.copyfileobj(r, f)
+        out_path = Path(base_path)
+        
+        out_file_path, out_file = None, None
+        for path, chunk in self.stream():
+            if path != out_file_path:
+                if out_file:
+                    out_file.close()
+                out_file_path = out_path / path
+                out_file_path.parent.mkdir(parents=True, exist_ok=True) 
+                out_file = open(out_file_path, "wb")
+            out_file.write(chunk)
+        if out_file is not None:
+            out_file.close()
 
     def stream(self) -> Generator[tuple[str, bytes], None, None]:
-        # TODO code duplication from self.write. Refactor to use stream?
         out_path = Path(self.id)  # relative output path
         if is_url(str(self.source)):
             if self.validate_url and not self.fetch_remote:
@@ -107,7 +86,6 @@ def stream(self) -> Generator[tuple[str, bytes], None, None]:
                     errno.ENOENT, os.strerror(errno.ENOENT), str(self.source)
                 )
             if not self.crate.source:
-                # code copied from rocrate.__copy_unlisted
                 for root, _, files in os.walk(self.source):
                     root = Path(root)
                     for name in files:
diff --git a/rocrate/rocrate.py b/rocrate/rocrate.py
index 3e535a7b..df2c538a 100644
--- a/rocrate/rocrate.py
+++ b/rocrate/rocrate.py
@@ -476,7 +476,6 @@ def write_zip(self, out_path):
                 out_file.write(chunk)
         return out_path
     
-    # TODO use context manager? https://docs.python.org/3/library/contextlib.html
     def stream_zip(self):
         """ Creates a stream of bytes representing the RO-Crate as a ZIP file."""
         buffer = MemoryBuffer()