From 91b3d6e7b8453ce51def696d97861346d799119b Mon Sep 17 00:00:00 2001
From: Laurent <laurent.thomas@embl.de>
Date: Tue, 2 Dec 2025 14:23:54 +0100
Subject: [PATCH 01/25] first attempt

---
 rocrate/model/__init__.py |  2 ++
 rocrate/model/subcrate.py | 45 +++++++++++++++++++++++++++++++++++++++
 rocrate/rocrate.py        | 27 ++++++++++++++++++++---
 3 files changed, 71 insertions(+), 3 deletions(-)
 create mode 100644 rocrate/model/subcrate.py

diff --git a/rocrate/model/__init__.py b/rocrate/model/__init__.py
index c2ed0e7..452ef8d 100644
--- a/rocrate/model/__init__.py
+++ b/rocrate/model/__init__.py
@@ -42,6 +42,7 @@
 from .person import Person
 from .root_dataset import RootDataset
 from .softwareapplication import SoftwareApplication
+from .subcrate import Subcrate
 from .testdefinition import TestDefinition
 from .testinstance import TestInstance
 from .preview import Preview
@@ -63,6 +64,7 @@
     "Preview",
     "RootDataset",
     "SoftwareApplication",
+    "Subcrate",
     "TestDefinition",
     "TestInstance",
     "TestService",
diff --git a/rocrate/model/subcrate.py b/rocrate/model/subcrate.py
new file mode 100644
index 0000000..7c1fdae
--- /dev/null
+++ b/rocrate/model/subcrate.py
@@ -0,0 +1,45 @@
+
+from ..model.dataset import Dataset
+from ..rocrate import ROCrate
+
+
+class Subcrate(Dataset):
+
+    def __init__(self, crate, source=None, dest_path=None, fetch_remote=False,
+                 validate_url=False, record_size=False):
+        """
+        This is a data-entity to represent a nested RO-Crate inside another RO-Crate.        
+        
+        :param crate: The parent crate
+        :param source: The relative path to the subcrate, or its URL
+        :param dest_path: Description
+        :param fetch_remote: Description
+        :param validate_url: Description
+        :param properties: Description
+        :param record_size: Description
+        """
+        super().__init__(crate, source, dest_path, fetch_remote,
+                         validate_url, properties=None, record_size=record_size)
+        
+        self.subcrate = None
+    
+    def load_subcrate(self):
+        """
+        Load the nested RO-Crate from the source path or URL.
+        
+        Populates the `subcrate` attribute with the loaded ROCrate instance,
+        and updates the JSON-LD representation accordingly.
+        """
+        if self.subcrate is not None:
+            return self.subcrate
+        
+        self.subcrate = ROCrate(self.source)
+
+        self._jsonld = self.subcrate.get("hasParts", default={})
+
+    def __getitem__(self, key):
+        
+        if self.subcrate is None:
+            self.load_subcrate()
+
+        return super().__getitem__(key)
\ No newline at end of file
diff --git a/rocrate/rocrate.py b/rocrate/rocrate.py
index d36750e..c0d95a2 100644
--- a/rocrate/rocrate.py
+++ b/rocrate/rocrate.py
@@ -49,6 +49,7 @@
     Preview,
     RootDataset,
     SoftwareApplication,
+    Subcrate,
     TestDefinition,
     TestInstance,
     TestService,
@@ -80,10 +81,24 @@ def pick_type(json_entity, type_map, fallback=None):
     except KeyError:
         raise ValueError(f'entity {json_entity["@id"]!r} has no @type')
     types = {_.strip() for _ in set(t if isinstance(t, list) else [t])}
+    
+    entity_class = None
     for name, c in type_map.items():
         if name in types:
-            return c
-    return fallback
+            entity_class = c
+            break
+    
+    if not entity_class:
+        return fallback
+    
+    if entity_class is Dataset:
+        
+        # TODO find a better way to check the profile
+        if json_entity.get("conformsTo", "").startswith("https://w3id.org"):
+            # Subcrate are a specific case of dataset
+            return Subcrate
+    
+        return Dataset
 
 
 def get_version(metadata_properties):
@@ -193,9 +208,15 @@ def __add_parts(self, parts, entities, source):
             entity = entities.pop(id_)
             assert id_ == entity.pop('@id')
             cls = pick_type(entity, type_map, fallback=DataEntity)
-            if cls is DataEntity:
+            
+            if cls is Subcrate:
+                instance = Subcrate(self, source / unquote(id_))
+
+            elif cls is DataEntity:
                 instance = DataEntity(self, identifier=id_, properties=entity)
+            
             else:
+                # cls is either a File or a Dataset (Directory)
                 if is_url(id_):
                     instance = cls(self, id_, properties=entity)
                 else:

From 8203025ba4630a0c55a23c0caedc4f978bee3307 Mon Sep 17 00:00:00 2001
From: Laurent <laurent.thomas@embl.de>
Date: Tue, 2 Dec 2025 15:35:41 +0100
Subject: [PATCH 02/25] move the Subcrate to rocrate main class

---
 rocrate/model/__init__.py |  2 --
 rocrate/model/subcrate.py | 45 ------------------------------
 rocrate/rocrate.py        | 58 ++++++++++++++++++++++++++++++++++++---
 3 files changed, 54 insertions(+), 51 deletions(-)
 delete mode 100644 rocrate/model/subcrate.py

diff --git a/rocrate/model/__init__.py b/rocrate/model/__init__.py
index 452ef8d..c2ed0e7 100644
--- a/rocrate/model/__init__.py
+++ b/rocrate/model/__init__.py
@@ -42,7 +42,6 @@
 from .person import Person
 from .root_dataset import RootDataset
 from .softwareapplication import SoftwareApplication
-from .subcrate import Subcrate
 from .testdefinition import TestDefinition
 from .testinstance import TestInstance
 from .preview import Preview
@@ -64,7 +63,6 @@
     "Preview",
     "RootDataset",
     "SoftwareApplication",
-    "Subcrate",
     "TestDefinition",
     "TestInstance",
     "TestService",
diff --git a/rocrate/model/subcrate.py b/rocrate/model/subcrate.py
deleted file mode 100644
index 7c1fdae..0000000
--- a/rocrate/model/subcrate.py
+++ /dev/null
@@ -1,45 +0,0 @@
-
-from ..model.dataset import Dataset
-from ..rocrate import ROCrate
-
-
-class Subcrate(Dataset):
-
-    def __init__(self, crate, source=None, dest_path=None, fetch_remote=False,
-                 validate_url=False, record_size=False):
-        """
-        This is a data-entity to represent a nested RO-Crate inside another RO-Crate.        
-        
-        :param crate: The parent crate
-        :param source: The relative path to the subcrate, or its URL
-        :param dest_path: Description
-        :param fetch_remote: Description
-        :param validate_url: Description
-        :param properties: Description
-        :param record_size: Description
-        """
-        super().__init__(crate, source, dest_path, fetch_remote,
-                         validate_url, properties=None, record_size=record_size)
-        
-        self.subcrate = None
-    
-    def load_subcrate(self):
-        """
-        Load the nested RO-Crate from the source path or URL.
-        
-        Populates the `subcrate` attribute with the loaded ROCrate instance,
-        and updates the JSON-LD representation accordingly.
-        """
-        if self.subcrate is not None:
-            return self.subcrate
-        
-        self.subcrate = ROCrate(self.source)
-
-        self._jsonld = self.subcrate.get("hasParts", default={})
-
-    def __getitem__(self, key):
-        
-        if self.subcrate is None:
-            self.load_subcrate()
-
-        return super().__getitem__(key)
\ No newline at end of file
diff --git a/rocrate/rocrate.py b/rocrate/rocrate.py
index c0d95a2..aaada38 100644
--- a/rocrate/rocrate.py
+++ b/rocrate/rocrate.py
@@ -49,7 +49,6 @@
     Preview,
     RootDataset,
     SoftwareApplication,
-    Subcrate,
     TestDefinition,
     TestInstance,
     TestService,
@@ -93,12 +92,19 @@ def pick_type(json_entity, type_map, fallback=None):
     
     if entity_class is Dataset:
         
+        # Check if the dataset is a Subcrate 
+        # i.e it has a conformsTo entry matching a RO-Crate profile
         # TODO find a better way to check the profile
-        if json_entity.get("conformsTo", "").startswith("https://w3id.org"):
-            # Subcrate are a specific case of dataset
-            return Subcrate
+        if list_profiles := json_entity.get("conformsTo", []):
+            
+            for profile_ref in as_list(list_profiles):
+                if profile_ref.get("@id", "").startswith("https://w3id.org/ro/crate/"):
+                    return Subcrate
     
         return Dataset
+    
+    else:
+        return entity_class
 
 
 def get_version(metadata_properties):
@@ -803,6 +809,50 @@ def __validate_suite(self, suite):
         return suite
 
 
+class Subcrate(Dataset):
+
+    def __init__(self, crate, source=None, dest_path=None, fetch_remote=False,
+                 validate_url=False, record_size=False):
+        """
+        This is a data-entity to represent a nested RO-Crate inside another RO-Crate.        
+        
+        :param crate: The parent crate
+        :param source: The relative path to the subcrate, or its URL
+        :param dest_path: Description
+        :param fetch_remote: Description
+        :param validate_url: Description
+        :param properties: Description
+        :param record_size: Description
+        """
+        super().__init__(crate, source, dest_path, fetch_remote,
+                         validate_url, properties=None, record_size=record_size)
+        
+        self.subcrate = None
+    
+    def load_subcrate(self):
+        """
+        Load the nested RO-Crate from the source path or URL.
+        
+        Populates the `subcrate` attribute with the loaded ROCrate instance,
+        and updates the JSON-LD representation accordingly.
+        """
+        if self.subcrate is None:
+            self.subcrate = ROCrate(self.source)
+            if list_parts := self.subcrate.root_dataset.get("hasPart"):
+                self._jsonld["hasPart"] = list_parts
+
+    def __getitem__(self, key):
+        
+        if self.subcrate is None:
+            self.load_subcrate()
+
+        return super().__getitem__(key)
+    
+    def as_jsonld(self):
+        if self.subcrate is None:
+            self.load_subcrate()
+        return super().as_jsonld()
+    
 def make_workflow_rocrate(workflow_path, wf_type, include_files=[],
                           fetch_remote=False, cwl=None, diagram=None):
     wf_crate = ROCrate()

From 25b56d3511aecd5b0b37a3331043ede3f0942e04 Mon Sep 17 00:00:00 2001
From: Laurent <laurent.thomas@embl.de>
Date: Tue, 2 Dec 2025 15:48:19 +0100
Subject: [PATCH 03/25] use get_norm_value instead of get + as_list

---
 rocrate/rocrate.py | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/rocrate/rocrate.py b/rocrate/rocrate.py
index aaada38..f907796 100644
--- a/rocrate/rocrate.py
+++ b/rocrate/rocrate.py
@@ -95,10 +95,10 @@ def pick_type(json_entity, type_map, fallback=None):
         # Check if the dataset is a Subcrate 
         # i.e it has a conformsTo entry matching a RO-Crate profile
         # TODO find a better way to check the profile
-        if list_profiles := json_entity.get("conformsTo", []):
+        if list_profiles := get_norm_value(json_entity, "conformsTo"):
             
-            for profile_ref in as_list(list_profiles):
-                if profile_ref.get("@id", "").startswith("https://w3id.org/ro/crate/"):
+            for profile_ref in list_profiles:
+                if profile_ref.startswith("https://w3id.org/ro/crate/"):
                     return Subcrate
     
         return Dataset

From c21e0e0bda088585254149efd51bf751ec06206a Mon Sep 17 00:00:00 2001
From: Laurent <laurent.thomas@embl.de>
Date: Tue, 2 Dec 2025 16:00:22 +0100
Subject: [PATCH 04/25] add support for get with rocrate

---
 rocrate/rocrate.py | 16 +++++++++++-----
 1 file changed, 11 insertions(+), 5 deletions(-)

diff --git a/rocrate/rocrate.py b/rocrate/rocrate.py
index f907796..7ba4017 100644
--- a/rocrate/rocrate.py
+++ b/rocrate/rocrate.py
@@ -833,8 +833,8 @@ def load_subcrate(self):
         """
         Load the nested RO-Crate from the source path or URL.
         
-        Populates the `subcrate` attribute with the loaded ROCrate instance,
-        and updates the JSON-LD representation accordingly.
+        This adds an attribute "hasPart" to the `subcrate` with the entities from the nested RO-Crate, 
+        updating the JSON-LD representation accordingly.
         """
         if self.subcrate is None:
             self.subcrate = ROCrate(self.source)
@@ -842,17 +842,23 @@ def load_subcrate(self):
                 self._jsonld["hasPart"] = list_parts
 
     def __getitem__(self, key):
-        
         if self.subcrate is None:
             self.load_subcrate()
+        
+        if key in self._jsonld:
+            # e.g the "original" entity keys such as id or type
+            return super().__getitem__(key)
+
+        # look into the subcrate entities
+        return self.subcrate.get(key)
 
-        return super().__getitem__(key)
-    
     def as_jsonld(self):
         if self.subcrate is None:
             self.load_subcrate()
         return super().as_jsonld()
     
+
+    
 def make_workflow_rocrate(workflow_path, wf_type, include_files=[],
                           fetch_remote=False, cwl=None, diagram=None):
     wf_crate = ROCrate()

From 0dbdb152eff326df4597483ce18a8d64fc54860f Mon Sep 17 00:00:00 2001
From: Laurent <laurent.thomas@embl.de>
Date: Tue, 2 Dec 2025 16:21:37 +0100
Subject: [PATCH 05/25] add simple tests

---
 test/test-data/crate_with_subcrate/file.txt   |  1 +
 .../ro-crate-metadata.json                    | 38 +++++++++++++++++++
 .../subcrate/ro-crate-metadata.json           | 29 ++++++++++++++
 .../crate_with_subcrate/subcrate/subfile.txt  |  1 +
 test/test_read.py                             | 29 +++++++++++++-
 5 files changed, 97 insertions(+), 1 deletion(-)
 create mode 100644 test/test-data/crate_with_subcrate/file.txt
 create mode 100644 test/test-data/crate_with_subcrate/ro-crate-metadata.json
 create mode 100644 test/test-data/crate_with_subcrate/subcrate/ro-crate-metadata.json
 create mode 100644 test/test-data/crate_with_subcrate/subcrate/subfile.txt

diff --git a/test/test-data/crate_with_subcrate/file.txt b/test/test-data/crate_with_subcrate/file.txt
new file mode 100644
index 0000000..7b4d68d
--- /dev/null
+++ b/test/test-data/crate_with_subcrate/file.txt
@@ -0,0 +1 @@
+empty
\ No newline at end of file
diff --git a/test/test-data/crate_with_subcrate/ro-crate-metadata.json b/test/test-data/crate_with_subcrate/ro-crate-metadata.json
new file mode 100644
index 0000000..46aff37
--- /dev/null
+++ b/test/test-data/crate_with_subcrate/ro-crate-metadata.json
@@ -0,0 +1,38 @@
+{
+    "@context": "https://w3id.org/ro/crate/1.1/context",
+    "@graph": [
+        {
+            "@id": "./",
+            "@type": "Dataset",
+            "name": "Top-level crate with subcrate",
+            "description": "A RO-Crate containing a subcrate",
+            "license": "https://spdx.org/licenses/MIT.html",
+            "datePublished": "2025-12-02T08:39:54+00:00",
+            "hasPart": [
+                    {"@id": "file.txt"},
+                    {"@id":"subcrate/"}
+            ]
+        },
+        {
+            "@id": "ro-crate-metadata.json",
+            "@type": "CreativeWork",
+            "about": {
+                "@id": "./"
+            },
+            "conformsTo": {
+                "@id": "https://w3id.org/ro/crate/1.1"
+            }
+        },
+        {
+            "@id": "file.txt",
+            "@type": "File"
+        },
+        {
+            "@id": "subcrate/",
+            "@type": "Dataset",
+            "conformsTo": {
+                "@id": "https://w3id.org/ro/crate/"
+            }
+        }
+    ]
+}
\ No newline at end of file
diff --git a/test/test-data/crate_with_subcrate/subcrate/ro-crate-metadata.json b/test/test-data/crate_with_subcrate/subcrate/ro-crate-metadata.json
new file mode 100644
index 0000000..b56eba0
--- /dev/null
+++ b/test/test-data/crate_with_subcrate/subcrate/ro-crate-metadata.json
@@ -0,0 +1,29 @@
+{
+    "@context": "https://w3id.org/ro/crate/1.1/context",
+    "@graph": [
+        {
+            "@id": "./",
+            "@type": "Dataset",
+            "datePublished": "2025-12-02T08:39:54+00:00",
+            "hasPart": [
+                
+                    {"@id": "subfile.txt"}
+                
+            ]
+        },
+        {
+            "@id": "ro-crate-metadata.json",
+            "@type": "CreativeWork",
+            "about": {
+                "@id": "./"
+            },
+            "conformsTo": {
+                "@id": "https://w3id.org/ro/crate/1.1"
+            }
+        },
+        {
+            "@id": "subfile.txt",
+            "@type": "File"
+        }
+    ]
+}
\ No newline at end of file
diff --git a/test/test-data/crate_with_subcrate/subcrate/subfile.txt b/test/test-data/crate_with_subcrate/subcrate/subfile.txt
new file mode 100644
index 0000000..7b4d68d
--- /dev/null
+++ b/test/test-data/crate_with_subcrate/subcrate/subfile.txt
@@ -0,0 +1 @@
+empty
\ No newline at end of file
diff --git a/test/test_read.py b/test/test_read.py
index 359da13..0563ce8 100644
--- a/test/test_read.py
+++ b/test/test_read.py
@@ -27,7 +27,8 @@
 import zipfile
 from pathlib import Path
 
-from rocrate.rocrate import ROCrate
+from rocrate import model
+from rocrate.rocrate import ROCrate, Subcrate
 from rocrate.model import DataEntity, ContextEntity, File, Dataset
 
 _URL = ('https://raw.githubusercontent.com/ResearchObject/ro-crate-py/master/'
@@ -191,6 +192,32 @@ def test_bad_crate(test_data_dir, tmpdir):
     with pytest.raises(ValueError):
         ROCrate(crate_dir)
 
+def load_crate_with_subcrate(test_data_dir):
+    return ROCrate(test_data_dir / "crate_with_subcrate")
+
+def test_crate_with_subcrate(test_data_dir):
+    
+    main_crate = load_crate_with_subcrate(test_data_dir)
+    
+    subcrate = main_crate.get("subcrate")
+    assert isinstance(subcrate, Subcrate)
+
+    # check that at this point, we have not yet loaded the subcrate
+    assert subcrate._jsonld == subcrate._empty()
+    assert "hasPart" not in subcrate
+    
+    # check lazy loading by accessing an entity from the subcrate
+    assert isinstance(subcrate.get("subfile.txt"), model.file.File)
+
+    # reload the crate to "reset" the state to unloaded
+    main_crate = load_crate_with_subcrate(test_data_dir)
+    subcrate = main_crate.get("subcrate")
+    
+    # as_jsonld should trigger loading of the subcrate
+    assert subcrate.as_jsonld() != subcrate._empty()
+    assert "hasPart" in subcrate
+    assert len(subcrate["hasPart"]) == 1
+
 
 @pytest.mark.parametrize("override", [False, True])
 def test_init(test_data_dir, tmpdir, helpers, override):

From ce2a0247977f7334e7e539238a30e845ed9d5911 Mon Sep 17 00:00:00 2001
From: Laurent <laurent.thomas@embl.de>
Date: Tue, 2 Dec 2025 17:01:12 +0100
Subject: [PATCH 06/25] handle url for subcrates

---
 rocrate/rocrate.py | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/rocrate/rocrate.py b/rocrate/rocrate.py
index 7ba4017..5184ebf 100644
--- a/rocrate/rocrate.py
+++ b/rocrate/rocrate.py
@@ -216,8 +216,11 @@ def __add_parts(self, parts, entities, source):
             cls = pick_type(entity, type_map, fallback=DataEntity)
             
             if cls is Subcrate:
-                instance = Subcrate(self, source / unquote(id_))
-
+                if is_url(id_):
+                    instance = Subcrate(self, id_)
+                else:
+                    instance = Subcrate(self, source / unquote(id_))
+                
             elif cls is DataEntity:
                 instance = DataEntity(self, identifier=id_, properties=entity)
             

From fbe0eeeee3c7e9dbd677254e09b7729f1105d2e8 Mon Sep 17 00:00:00 2001
From: Laurent <laurent.thomas@embl.de>
Date: Wed, 3 Dec 2025 09:44:31 +0100
Subject: [PATCH 07/25] add flag parse_subcrate

---
 rocrate/rocrate.py | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

diff --git a/rocrate/rocrate.py b/rocrate/rocrate.py
index 5184ebf..85399f7 100644
--- a/rocrate/rocrate.py
+++ b/rocrate/rocrate.py
@@ -74,7 +74,7 @@ def is_data_entity(entity):
     return DATA_ENTITY_TYPES.intersection(as_list(entity.get("@type", [])))
 
 
-def pick_type(json_entity, type_map, fallback=None):
+def pick_type(json_entity, type_map, fallback=None, parse_subcrate=False):
     try:
         t = json_entity["@type"]
     except KeyError:
@@ -95,7 +95,7 @@ def pick_type(json_entity, type_map, fallback=None):
         # Check if the dataset is a Subcrate 
         # i.e it has a conformsTo entry matching a RO-Crate profile
         # TODO find a better way to check the profile
-        if list_profiles := get_norm_value(json_entity, "conformsTo"):
+        if parse_subcrate and (list_profiles := get_norm_value(json_entity, "conformsTo")):
             
             for profile_ref in list_profiles:
                 if profile_ref.startswith("https://w3id.org/ro/crate/"):
@@ -117,7 +117,7 @@ def get_version(metadata_properties):
 
 class ROCrate():
 
-    def __init__(self, source=None, gen_preview=False, init=False, exclude=None, version=DEFAULT_VERSION):
+    def __init__(self, source=None, gen_preview=False, init=False, exclude=None, version=DEFAULT_VERSION, parse_subcrate=False):
         self.mode = None
         self.source = source
         self.exclude = exclude
@@ -142,6 +142,7 @@ def __init__(self, source=None, gen_preview=False, init=False, exclude=None, ver
             source = self.__read(source, gen_preview=gen_preview)
         # in the zip case, self.source is the extracted dir
         self.source = source
+        self.parse_subcrate = parse_subcrate
 
     def __init_from_tree(self, top_dir, gen_preview=False, version=DEFAULT_VERSION):
         top_dir = Path(top_dir)
@@ -213,7 +214,7 @@ def __add_parts(self, parts, entities, source):
                     continue
             entity = entities.pop(id_)
             assert id_ == entity.pop('@id')
-            cls = pick_type(entity, type_map, fallback=DataEntity)
+            cls = pick_type(entity, type_map, fallback=DataEntity, parse_subcrate=self.parse_subcrate)
             
             if cls is Subcrate:
                 if is_url(id_):

From bd02ee2c0b349d20c55d5d351be96431b87cfd85 Mon Sep 17 00:00:00 2001
From: Laurent <laurent.thomas@embl.de>
Date: Wed, 3 Dec 2025 11:58:25 +0100
Subject: [PATCH 08/25] fix missing flag parse_rocrate

---
 rocrate/rocrate.py | 19 +++++++++++--------
 test/test_read.py  |  2 +-
 2 files changed, 12 insertions(+), 9 deletions(-)

diff --git a/rocrate/rocrate.py b/rocrate/rocrate.py
index 85399f7..0cbcdea 100644
--- a/rocrate/rocrate.py
+++ b/rocrate/rocrate.py
@@ -117,10 +117,16 @@ def get_version(metadata_properties):
 
 class ROCrate():
 
-    def __init__(self, source=None, gen_preview=False, init=False, exclude=None, version=DEFAULT_VERSION, parse_subcrate=False):
+    def __init__(self,
+                 source=None,
+                 gen_preview=False,
+                 init=False, exclude=None,
+                 version=DEFAULT_VERSION,
+                 parse_subcrate=False):
         self.mode = None
         self.source = source
         self.exclude = exclude
+        self.parse_subcrate = parse_subcrate
         self.__entity_map = {}
         # TODO: add this as @base in the context? At least when loading
         # from zip
@@ -142,7 +148,6 @@ def __init__(self, source=None, gen_preview=False, init=False, exclude=None, ver
             source = self.__read(source, gen_preview=gen_preview)
         # in the zip case, self.source is the extracted dir
         self.source = source
-        self.parse_subcrate = parse_subcrate
 
     def __init_from_tree(self, top_dir, gen_preview=False, version=DEFAULT_VERSION):
         top_dir = Path(top_dir)
@@ -818,20 +823,18 @@ class Subcrate(Dataset):
     def __init__(self, crate, source=None, dest_path=None, fetch_remote=False,
                  validate_url=False, record_size=False):
         """
-        This is a data-entity to represent a nested RO-Crate inside another RO-Crate.        
+        Data-entity representing a subcrate inside another RO-Crate.        
         
         :param crate: The parent crate
         :param source: The relative path to the subcrate, or its URL
-        :param dest_path: Description
-        :param fetch_remote: Description
-        :param validate_url: Description
-        :param properties: Description
-        :param record_size: Description
         """
         super().__init__(crate, source, dest_path, fetch_remote,
                          validate_url, properties=None, record_size=record_size)
         
         self.subcrate = None
+        """
+        A ROCrate instance allowing access to the nested RO-Crate.
+        """
     
     def load_subcrate(self):
         """
diff --git a/test/test_read.py b/test/test_read.py
index 0563ce8..f7dcf41 100644
--- a/test/test_read.py
+++ b/test/test_read.py
@@ -193,7 +193,7 @@ def test_bad_crate(test_data_dir, tmpdir):
         ROCrate(crate_dir)
 
 def load_crate_with_subcrate(test_data_dir):
-    return ROCrate(test_data_dir / "crate_with_subcrate")
+    return ROCrate(test_data_dir / "crate_with_subcrate", parse_subcrate=True)
 
 def test_crate_with_subcrate(test_data_dir):
     

From bf6f276c7377f81de38243b8902ee3de6ff5c5c1 Mon Sep 17 00:00:00 2001
From: Laurent <laurent.thomas@embl.de>
Date: Wed, 3 Dec 2025 11:59:43 +0100
Subject: [PATCH 09/25] add subcrate_entities property

---
 rocrate/rocrate.py | 5 +++++
 test/test_read.py  | 1 +
 2 files changed, 6 insertions(+)

diff --git a/rocrate/rocrate.py b/rocrate/rocrate.py
index 0cbcdea..7efef1a 100644
--- a/rocrate/rocrate.py
+++ b/rocrate/rocrate.py
@@ -270,6 +270,11 @@ def contextual_entities(self):
                 if not isinstance(e, (RootDataset, Metadata, Preview))
                 and not hasattr(e, "write")]
 
+    @property
+    def subcrate_entities(self):
+        return [e for e in self.__entity_map.values()
+                if isinstance(e, Subcrate)]
+
     @property
     def name(self):
         return self.root_dataset.get('name')
diff --git a/test/test_read.py b/test/test_read.py
index f7dcf41..e6db9ef 100644
--- a/test/test_read.py
+++ b/test/test_read.py
@@ -201,6 +201,7 @@ def test_crate_with_subcrate(test_data_dir):
     
     subcrate = main_crate.get("subcrate")
     assert isinstance(subcrate, Subcrate)
+    assert main_crate.subcrate_entities == [subcrate]
 
     # check that at this point, we have not yet loaded the subcrate
     assert subcrate._jsonld == subcrate._empty()

From fd1fd53c4883dd09a6d8806e18a16aa6a9a6d9bc Mon Sep 17 00:00:00 2001
From: Laurent <laurent.thomas@embl.de>
Date: Wed, 3 Dec 2025 14:09:52 +0100
Subject: [PATCH 10/25] support get of subcrate entity from top crate

also fix flake8 with precommit
---
 rocrate/rocrate.py | 56 ++++++++++++++++++++++++++++------------------
 test/test_read.py  | 22 ++++++++++++------
 2 files changed, 49 insertions(+), 29 deletions(-)

diff --git a/rocrate/rocrate.py b/rocrate/rocrate.py
index 7efef1a..88a1cc6 100644
--- a/rocrate/rocrate.py
+++ b/rocrate/rocrate.py
@@ -80,29 +80,29 @@ def pick_type(json_entity, type_map, fallback=None, parse_subcrate=False):
     except KeyError:
         raise ValueError(f'entity {json_entity["@id"]!r} has no @type')
     types = {_.strip() for _ in set(t if isinstance(t, list) else [t])}
-    
+
     entity_class = None
     for name, c in type_map.items():
         if name in types:
             entity_class = c
             break
-    
+
     if not entity_class:
         return fallback
-    
+
     if entity_class is Dataset:
-        
-        # Check if the dataset is a Subcrate 
+
+        # Check if the dataset is a Subcrate
         # i.e it has a conformsTo entry matching a RO-Crate profile
-        # TODO find a better way to check the profile
+        # TODO find a better way to check the profiles?
         if parse_subcrate and (list_profiles := get_norm_value(json_entity, "conformsTo")):
-            
+
             for profile_ref in list_profiles:
                 if profile_ref.startswith("https://w3id.org/ro/crate/"):
                     return Subcrate
-    
+
         return Dataset
-    
+
     else:
         return entity_class
 
@@ -220,16 +220,16 @@ def __add_parts(self, parts, entities, source):
             entity = entities.pop(id_)
             assert id_ == entity.pop('@id')
             cls = pick_type(entity, type_map, fallback=DataEntity, parse_subcrate=self.parse_subcrate)
-            
+
             if cls is Subcrate:
                 if is_url(id_):
                     instance = Subcrate(self, id_)
                 else:
                     instance = Subcrate(self, source / unquote(id_))
-                
+
             elif cls is DataEntity:
                 instance = DataEntity(self, identifier=id_, properties=entity)
-            
+
             else:
                 # cls is either a File or a Dataset (Directory)
                 if is_url(id_):
@@ -238,6 +238,8 @@ def __add_parts(self, parts, entities, source):
                     instance = cls(self, source / unquote(id_), id_, properties=entity)
             self.add(instance)
             if instance.type == "Dataset":
+                # for Subcrate, type is currently Dataset too,
+                # but the hasPart is not populated yet only once accssing a subcrate element (lazy loading)
                 self.__add_parts(as_list(entity.get("hasPart", [])), entities, source)
 
     def __read_contextual_entities(self, entities):
@@ -407,7 +409,18 @@ def _get_root_jsonld(self):
 
     def dereference(self, entity_id, default=None):
         canonical_id = self.resolve_id(entity_id)
-        return self.__entity_map.get(canonical_id, default)
+
+        if canonical_id in self.__entity_map:
+            return self.__entity_map[canonical_id]
+
+        for subcrate_entity in self.subcrate_entities:
+            # check if the entity_id might be within a subcrate
+            if entity_id.startswith(subcrate_entity.id):
+                entity_id_in_subcrate = entity_id.replace(subcrate_entity.id, "")
+                return subcrate_entity.get(entity_id_in_subcrate, default=default)
+
+        # fallback
+        return default
 
     get = dereference
 
@@ -828,24 +841,24 @@ class Subcrate(Dataset):
     def __init__(self, crate, source=None, dest_path=None, fetch_remote=False,
                  validate_url=False, record_size=False):
         """
-        Data-entity representing a subcrate inside another RO-Crate.        
-        
+        Data-entity representing a subcrate inside another RO-Crate.
+
         :param crate: The parent crate
         :param source: The relative path to the subcrate, or its URL
         """
         super().__init__(crate, source, dest_path, fetch_remote,
                          validate_url, properties=None, record_size=record_size)
-        
+
         self.subcrate = None
         """
         A ROCrate instance allowing access to the nested RO-Crate.
         """
-    
+
     def load_subcrate(self):
         """
         Load the nested RO-Crate from the source path or URL.
-        
-        This adds an attribute "hasPart" to the `subcrate` with the entities from the nested RO-Crate, 
+
+        This adds an attribute "hasPart" to the `subcrate` with the entities from the nested RO-Crate,
         updating the JSON-LD representation accordingly.
         """
         if self.subcrate is None:
@@ -856,7 +869,7 @@ def load_subcrate(self):
     def __getitem__(self, key):
         if self.subcrate is None:
             self.load_subcrate()
-        
+
         if key in self._jsonld:
             # e.g the "original" entity keys such as id or type
             return super().__getitem__(key)
@@ -868,9 +881,8 @@ def as_jsonld(self):
         if self.subcrate is None:
             self.load_subcrate()
         return super().as_jsonld()
-    
 
-    
+
 def make_workflow_rocrate(workflow_path, wf_type, include_files=[],
                           fetch_remote=False, cwl=None, diagram=None):
     wf_crate = ROCrate()
diff --git a/test/test_read.py b/test/test_read.py
index e6db9ef..b7f3d3d 100644
--- a/test/test_read.py
+++ b/test/test_read.py
@@ -192,13 +192,15 @@ def test_bad_crate(test_data_dir, tmpdir):
     with pytest.raises(ValueError):
         ROCrate(crate_dir)
 
+
 def load_crate_with_subcrate(test_data_dir):
     return ROCrate(test_data_dir / "crate_with_subcrate", parse_subcrate=True)
 
+
 def test_crate_with_subcrate(test_data_dir):
-    
+
     main_crate = load_crate_with_subcrate(test_data_dir)
-    
+
     subcrate = main_crate.get("subcrate")
     assert isinstance(subcrate, Subcrate)
     assert main_crate.subcrate_entities == [subcrate]
@@ -206,18 +208,24 @@ def test_crate_with_subcrate(test_data_dir):
     # check that at this point, we have not yet loaded the subcrate
     assert subcrate._jsonld == subcrate._empty()
     assert "hasPart" not in subcrate
-    
+
     # check lazy loading by accessing an entity from the subcrate
-    assert isinstance(subcrate.get("subfile.txt"), model.file.File)
+    subfile_entity = subcrate.get("subfile.txt")
+    assert isinstance(subfile_entity, model.file.File)
+
+    # check access from the top-level crate works too
+    assert main_crate.get("subcrate/subfile.txt") is subfile_entity
+
+    # Check the subfile entity is listed under hasPart of the subcrate
+    assert "hasPart" in subcrate
+    assert subcrate["hasPart"] == [subfile_entity]
 
     # reload the crate to "reset" the state to unloaded
     main_crate = load_crate_with_subcrate(test_data_dir)
     subcrate = main_crate.get("subcrate")
-    
+
     # as_jsonld should trigger loading of the subcrate
     assert subcrate.as_jsonld() != subcrate._empty()
-    assert "hasPart" in subcrate
-    assert len(subcrate["hasPart"]) == 1
 
 
 @pytest.mark.parametrize("override", [False, True])

From 991664101deb86da24e1efffacd152f5ab9996c5 Mon Sep 17 00:00:00 2001
From: Laurent <laurent.thomas@embl.de>
Date: Wed, 3 Dec 2025 14:26:17 +0100
Subject: [PATCH 11/25] load_subcrate as hidden function

---
 rocrate/rocrate.py | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/rocrate/rocrate.py b/rocrate/rocrate.py
index 88a1cc6..33aec3c 100644
--- a/rocrate/rocrate.py
+++ b/rocrate/rocrate.py
@@ -854,7 +854,7 @@ def __init__(self, crate, source=None, dest_path=None, fetch_remote=False,
         A ROCrate instance allowing access to the nested RO-Crate.
         """
 
-    def load_subcrate(self):
+    def _load_subcrate(self):
         """
         Load the nested RO-Crate from the source path or URL.
 
@@ -868,7 +868,7 @@ def load_subcrate(self):
 
     def __getitem__(self, key):
         if self.subcrate is None:
-            self.load_subcrate()
+            self._load_subcrate()
 
         if key in self._jsonld:
             # e.g the "original" entity keys such as id or type
@@ -879,7 +879,7 @@ def __getitem__(self, key):
 
     def as_jsonld(self):
         if self.subcrate is None:
-            self.load_subcrate()
+            self._load_subcrate()
         return super().as_jsonld()
 
 

From 22b34f277e14dddeb897ac5560fa9422dc8a8cfd Mon Sep 17 00:00:00 2001
From: Laurent <laurent.thomas@embl.de>
Date: Wed, 3 Dec 2025 14:26:28 +0100
Subject: [PATCH 12/25] add get_entities to subcrate

---
 rocrate/rocrate.py | 5 +++++
 test/test_read.py  | 2 ++
 2 files changed, 7 insertions(+)

diff --git a/rocrate/rocrate.py b/rocrate/rocrate.py
index 33aec3c..422c163 100644
--- a/rocrate/rocrate.py
+++ b/rocrate/rocrate.py
@@ -882,6 +882,11 @@ def as_jsonld(self):
             self._load_subcrate()
         return super().as_jsonld()
 
+    def get_entities(self):
+        if self.subcrate is None:
+            self._load_subcrate()
+        return self.subcrate.get_entities()
+
 
 def make_workflow_rocrate(workflow_path, wf_type, include_files=[],
                           fetch_remote=False, cwl=None, diagram=None):
diff --git a/test/test_read.py b/test/test_read.py
index b7f3d3d..5eb26a8 100644
--- a/test/test_read.py
+++ b/test/test_read.py
@@ -220,6 +220,8 @@ def test_crate_with_subcrate(test_data_dir):
     assert "hasPart" in subcrate
     assert subcrate["hasPart"] == [subfile_entity]
 
+    assert len(subcrate.get_entities()) == 3  # root dataset, metadata.json, subfile
+
     # reload the crate to "reset" the state to unloaded
     main_crate = load_crate_with_subcrate(test_data_dir)
     subcrate = main_crate.get("subcrate")

From bf7e0eaa908c88ec5b96c5c28282feb9c0679b1d Mon Sep 17 00:00:00 2001
From: Laurent <laurent.thomas@embl.de>
Date: Wed, 3 Dec 2025 14:30:04 +0100
Subject: [PATCH 13/25] add Subcrate to test_data_entities

---
 test/test_model.py | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/test/test_model.py b/test/test_model.py
index a86c25f..26b13d2 100644
--- a/test/test_model.py
+++ b/test/test_model.py
@@ -29,7 +29,7 @@
 from pathlib import Path
 
 import pytest
-from rocrate.rocrate import ROCrate
+from rocrate.rocrate import ROCrate, Subcrate
 from rocrate.model import (
     DataEntity,
     File,
@@ -103,10 +103,11 @@ def test_data_entities(test_data_dir):
     crate = ROCrate()
     file_ = crate.add(File(crate, test_data_dir / 'sample_file.txt'))
     dataset = crate.add(Dataset(crate, test_data_dir / 'test_add_dir'))
+    subcrate = crate.add(Subcrate(crate, test_data_dir / 'crate-1.0'))
     data_entity = crate.add(DataEntity(crate, '#mysterious'))
-    assert set(crate.data_entities) == {file_, dataset, data_entity}
+    assert set(crate.data_entities) == {file_, dataset, subcrate, data_entity}
     part_ids = set(_["@id"] for _ in crate.root_dataset._jsonld["hasPart"])
-    assert set(_.id for _ in (file_, dataset, data_entity)) <= part_ids
+    assert set(_.id for _ in (file_, dataset, subcrate, data_entity)) <= part_ids
 
 
 @pytest.mark.skipif(sys.platform == "darwin", reason="CI sometimes fails on macOS")

From 9e9a4cf56007b24d101ab70ef897ed96ddda3032 Mon Sep 17 00:00:00 2001
From: Laurent <laurent.thomas@embl.de>
Date: Wed, 3 Dec 2025 15:11:38 +0100
Subject: [PATCH 14/25] fix issue with nested crates

---
 rocrate/rocrate.py                            | 11 ++++++--
 .../subcrate/ro-crate-metadata.json           | 14 +++++++---
 .../subcrate/subsubcrate/deepfile.txt         |  1 +
 .../subsubcrate/ro-crate-metadata.json        | 27 +++++++++++++++++++
 test/test_read.py                             |  9 ++++---
 5 files changed, 53 insertions(+), 9 deletions(-)
 create mode 100644 test/test-data/crate_with_subcrate/subcrate/subsubcrate/deepfile.txt
 create mode 100644 test/test-data/crate_with_subcrate/subcrate/subsubcrate/ro-crate-metadata.json

diff --git a/rocrate/rocrate.py b/rocrate/rocrate.py
index 422c163..3dcba09 100644
--- a/rocrate/rocrate.py
+++ b/rocrate/rocrate.py
@@ -414,9 +414,16 @@ def dereference(self, entity_id, default=None):
             return self.__entity_map[canonical_id]
 
         for subcrate_entity in self.subcrate_entities:
+
             # check if the entity_id might be within a subcrate
+            # i.e entity_id would start with a subcrate id e.g subcrate/subfile.txt
             if entity_id.startswith(subcrate_entity.id):
-                entity_id_in_subcrate = entity_id.replace(subcrate_entity.id, "")
+
+                # replace id of subcrate to use get in the subcrate
+                # subcrate/subfile.txt --> subfile.txt
+                # dont use replace, as it could replace in the middle of the id
+                entity_id_in_subcrate = entity_id[len(subcrate_entity.id):]
+
                 return subcrate_entity.get(entity_id_in_subcrate, default=default)
 
         # fallback
@@ -862,7 +869,7 @@ def _load_subcrate(self):
         updating the JSON-LD representation accordingly.
         """
         if self.subcrate is None:
-            self.subcrate = ROCrate(self.source)
+            self.subcrate = ROCrate(self.source, parse_subcrate=True)  # would load further nested RO-Crate
             if list_parts := self.subcrate.root_dataset.get("hasPart"):
                 self._jsonld["hasPart"] = list_parts
 
diff --git a/test/test-data/crate_with_subcrate/subcrate/ro-crate-metadata.json b/test/test-data/crate_with_subcrate/subcrate/ro-crate-metadata.json
index b56eba0..2bd736f 100644
--- a/test/test-data/crate_with_subcrate/subcrate/ro-crate-metadata.json
+++ b/test/test-data/crate_with_subcrate/subcrate/ro-crate-metadata.json
@@ -6,9 +6,8 @@
             "@type": "Dataset",
             "datePublished": "2025-12-02T08:39:54+00:00",
             "hasPart": [
-                
-                    {"@id": "subfile.txt"}
-                
+                    {"@id": "subfile.txt"},
+                    {"@id": "subsubcrate/"}
             ]
         },
         {
@@ -24,6 +23,13 @@
         {
             "@id": "subfile.txt",
             "@type": "File"
+        },
+        {
+            "@id": "subsubcrate/",
+            "@type": "Dataset",
+            "conformsTo": {
+                "@id": "https://w3id.org/ro/crate/"
+            }
         }
     ]
-}
\ No newline at end of file
+}
diff --git a/test/test-data/crate_with_subcrate/subcrate/subsubcrate/deepfile.txt b/test/test-data/crate_with_subcrate/subcrate/subsubcrate/deepfile.txt
new file mode 100644
index 0000000..c6cac69
--- /dev/null
+++ b/test/test-data/crate_with_subcrate/subcrate/subsubcrate/deepfile.txt
@@ -0,0 +1 @@
+empty
diff --git a/test/test-data/crate_with_subcrate/subcrate/subsubcrate/ro-crate-metadata.json b/test/test-data/crate_with_subcrate/subcrate/subsubcrate/ro-crate-metadata.json
new file mode 100644
index 0000000..1b5390f
--- /dev/null
+++ b/test/test-data/crate_with_subcrate/subcrate/subsubcrate/ro-crate-metadata.json
@@ -0,0 +1,27 @@
+{
+    "@context": "https://w3id.org/ro/crate/1.1/context",
+    "@graph": [
+        {
+            "@id": "./",
+            "@type": "Dataset",
+            "datePublished": "2025-12-02T08:39:54+00:00",
+            "hasPart": [
+                    {"@id": "deepfile.txt"}
+            ]
+        },
+        {
+            "@id": "ro-crate-metadata.json",
+            "@type": "CreativeWork",
+            "about": {
+                "@id": "./"
+            },
+            "conformsTo": {
+                "@id": "https://w3id.org/ro/crate/1.1"
+            }
+        },
+        {
+            "@id": "deepfile.txt",
+            "@type": "File"
+        }
+    ]
+}
diff --git a/test/test_read.py b/test/test_read.py
index 5eb26a8..07f3214 100644
--- a/test/test_read.py
+++ b/test/test_read.py
@@ -216,11 +216,14 @@ def test_crate_with_subcrate(test_data_dir):
     # check access from the top-level crate works too
     assert main_crate.get("subcrate/subfile.txt") is subfile_entity
 
-    # Check the subfile entity is listed under hasPart of the subcrate
+    # check with another nested rocrate
+    assert isinstance(main_crate.get("subcrate/subsubcrate/deepfile.txt"), model.file.File)
+
+    # Check the hasPart of the subcrate lists the file and the subsubcrate
     assert "hasPart" in subcrate
-    assert subcrate["hasPart"] == [subfile_entity]
+    assert len(subcrate["hasPart"]) == 2
 
-    assert len(subcrate.get_entities()) == 3  # root dataset, metadata.json, subfile
+    assert len(subcrate.get_entities()) == 4  # root dataset, metadata.json, subfile, subsubcrate
 
     # reload the crate to "reset" the state to unloaded
     main_crate = load_crate_with_subcrate(test_data_dir)

From 21556c8a57d08a274af574a8d7f5f49ea791c6c1 Mon Sep 17 00:00:00 2001
From: Laurent <laurent.thomas@embl.de>
Date: Thu, 4 Dec 2025 14:30:53 +0100
Subject: [PATCH 15/25] keep conformsTo in Subcrate

---
 rocrate/rocrate.py | 38 ++++++++++++++++++++++++--------------
 test/test_read.py  |  9 ++++++++-
 2 files changed, 32 insertions(+), 15 deletions(-)

diff --git a/rocrate/rocrate.py b/rocrate/rocrate.py
index 3dcba09..d2e5d02 100644
--- a/rocrate/rocrate.py
+++ b/rocrate/rocrate.py
@@ -209,6 +209,14 @@ def __read_data_entities(self, entities, source, gen_preview):
         self.__add_parts(parts, entities, source)
 
     def __add_parts(self, parts, entities, source):
+        """
+        Add entities to the crate from a list of entities id and Entity object.
+
+        :param self: Description
+        :param parts: a list of dicts (one dict per entity) in the form {@id : "entity_id"}
+        :param entities: a dict with the full list of entities information as in the hasPart of the root dataset of the crate.
+        :param source: Description
+        """
         type_map = OrderedDict((_.__name__, _) for _ in subclasses(FileOrDir))
         for ref in parts:
             id_ = ref['@id']
@@ -222,10 +230,11 @@ def __add_parts(self, parts, entities, source):
             cls = pick_type(entity, type_map, fallback=DataEntity, parse_subcrate=self.parse_subcrate)
 
             if cls is Subcrate:
+
                 if is_url(id_):
-                    instance = Subcrate(self, id_)
+                    instance = Subcrate(self, source=id_, properties=entity)
                 else:
-                    instance = Subcrate(self, source / unquote(id_))
+                    instance = Subcrate(self, source=source / unquote(id_), properties=entity)
 
             elif cls is DataEntity:
                 instance = DataEntity(self, identifier=id_, properties=entity)
@@ -239,7 +248,7 @@ def __add_parts(self, parts, entities, source):
             self.add(instance)
             if instance.type == "Dataset":
                 # for Subcrate, type is currently Dataset too,
-                # but the hasPart is not populated yet only once accssing a subcrate element (lazy loading)
+                # but the hasPart is not populated yet only once accessing a subcrate element (lazy loading)
                 self.__add_parts(as_list(entity.get("hasPart", [])), entities, source)
 
     def __read_contextual_entities(self, entities):
@@ -846,7 +855,7 @@ def __validate_suite(self, suite):
 class Subcrate(Dataset):
 
     def __init__(self, crate, source=None, dest_path=None, fetch_remote=False,
-                 validate_url=False, record_size=False):
+                 validate_url=False, properties=None, record_size=False):
         """
         Data-entity representing a subcrate inside another RO-Crate.
 
@@ -854,11 +863,12 @@ def __init__(self, crate, source=None, dest_path=None, fetch_remote=False,
         :param source: The relative path to the subcrate, or its URL
         """
         super().__init__(crate, source, dest_path, fetch_remote,
-                         validate_url, properties=None, record_size=record_size)
+                         validate_url, properties=properties, record_size=record_size)
 
-        self.subcrate = None
+        self._subcrate = None
         """
         A ROCrate instance allowing access to the nested RO-Crate.
+        The nested RO-Crate is loaded on first access to any of its attribute.
         """
 
     def _load_subcrate(self):
@@ -868,13 +878,13 @@ def _load_subcrate(self):
         This adds an attribute "hasPart" to the `subcrate` with the entities from the nested RO-Crate,
         updating the JSON-LD representation accordingly.
         """
-        if self.subcrate is None:
-            self.subcrate = ROCrate(self.source, parse_subcrate=True)  # would load further nested RO-Crate
-            if list_parts := self.subcrate.root_dataset.get("hasPart"):
+        if self._subcrate is None:
+            self._subcrate = ROCrate(self.source, parse_subcrate=True)  # would load further nested RO-Crate
+            if list_parts := self._subcrate.root_dataset.get("hasPart"):
                 self._jsonld["hasPart"] = list_parts
 
     def __getitem__(self, key):
-        if self.subcrate is None:
+        if self._subcrate is None:
             self._load_subcrate()
 
         if key in self._jsonld:
@@ -882,17 +892,17 @@ def __getitem__(self, key):
             return super().__getitem__(key)
 
         # look into the subcrate entities
-        return self.subcrate.get(key)
+        return self._subcrate.get(key)
 
     def as_jsonld(self):
-        if self.subcrate is None:
+        if self._subcrate is None:
             self._load_subcrate()
         return super().as_jsonld()
 
     def get_entities(self):
-        if self.subcrate is None:
+        if self._subcrate is None:
             self._load_subcrate()
-        return self.subcrate.get_entities()
+        return self._subcrate.get_entities()
 
 
 def make_workflow_rocrate(workflow_path, wf_type, include_files=[],
diff --git a/test/test_read.py b/test/test_read.py
index 07f3214..bed09a6 100644
--- a/test/test_read.py
+++ b/test/test_read.py
@@ -205,8 +205,15 @@ def test_crate_with_subcrate(test_data_dir):
     assert isinstance(subcrate, Subcrate)
     assert main_crate.subcrate_entities == [subcrate]
 
+    # Check the subcrate kept the conformsTo attribute from the original Dataset entity
+    assert "conformsTo" in subcrate
+
     # check that at this point, we have not yet loaded the subcrate
-    assert subcrate._jsonld == subcrate._empty()
+    # e.g the json ld should just have id, type and conformsTo
+    jsonld = subcrate._jsonld
+    jsonld.pop("conformsTo")
+
+    assert jsonld == subcrate._empty()
     assert "hasPart" not in subcrate
 
     # check lazy loading by accessing an entity from the subcrate

From d98b22f798d8b7e831cb8c014a36d72b5c524174 Mon Sep 17 00:00:00 2001
From: Laurent <laurent.thomas@embl.de>
Date: Thu, 4 Dec 2025 16:47:44 +0100
Subject: [PATCH 16/25] use getter for inner crate

also prevents directly accessing items listed in subcrate under hasPart e.g subcrate.get("subfile.txt")
---
 rocrate/model/entity.py |  1 +
 rocrate/rocrate.py      | 61 +++++++++++++++++++++++++++++------------
 test/test_read.py       |  8 ++++--
 3 files changed, 49 insertions(+), 21 deletions(-)

diff --git a/rocrate/model/entity.py b/rocrate/model/entity.py
index 473a7eb..b9acc16 100644
--- a/rocrate/model/entity.py
+++ b/rocrate/model/entity.py
@@ -43,6 +43,7 @@ def __init__(self, crate, identifier=None, properties=None):
                 if name.startswith("@"):
                     self._jsonld[name] = value
                 else:
+                    # this will call the __setitem__ method defined below
                     self[name] = value
 
     @property
diff --git a/rocrate/rocrate.py b/rocrate/rocrate.py
index d2e5d02..d9224ed 100644
--- a/rocrate/rocrate.py
+++ b/rocrate/rocrate.py
@@ -433,7 +433,7 @@ def dereference(self, entity_id, default=None):
                 # dont use replace, as it could replace in the middle of the id
                 entity_id_in_subcrate = entity_id[len(subcrate_entity.id):]
 
-                return subcrate_entity.get(entity_id_in_subcrate, default=default)
+                return subcrate_entity.get_crate().get(entity_id_in_subcrate, default=default)
 
         # fallback
         return default
@@ -865,44 +865,69 @@ def __init__(self, crate, source=None, dest_path=None, fetch_remote=False,
         super().__init__(crate, source, dest_path, fetch_remote,
                          validate_url, properties=properties, record_size=record_size)
 
-        self._subcrate = None
+        self._crate = None
         """
         A ROCrate instance allowing access to the nested RO-Crate.
         The nested RO-Crate is loaded on first access to any of its attribute.
+        This attribute should not be confused with the crate attribute, which is a reference to the parent crate.
+        Caller should rather use the get_crate() method to access the nested RO-Crate.
         """
 
+    def get_crate(self):
+        """
+        Return the RO-Crate object referenced by this subcrate.
+        """
+        if self._crate is None:
+            self._load_subcrate()
+
+        return self._crate
+
     def _load_subcrate(self):
         """
         Load the nested RO-Crate from the source path or URL.
 
-        This adds an attribute "hasPart" to the `subcrate` with the entities from the nested RO-Crate,
-        updating the JSON-LD representation accordingly.
+        This populates the attribute `hasPart` of the `Subcrate` entity,
+        with the data entities listed under the `root_dataset["hasPart"]` of the nested RO-Crate.
+        If the nested RO-crate does not list any part und `hasPart`,
+        then the `hasPart` attribute of the `Subcrate` entity will be an empty list.
+        """
+        if self._crate is None:
+            # parse_subcrate=True to load further nested RO-Crate (on-demand / lazily too)
+            self._crate = ROCrate(self.source, parse_subcrate=True)
+
+            # Note : assigning to hasPart keeps only the dict with id:entity not the actual entities
+            # such that when retrieving something from hasPart one was getting a string not an entity
+            self["hasPart"] = self._crate.root_dataset.get("hasPart", [])
+
+    def _get_parts_subcrate_root(self):
         """
-        if self._subcrate is None:
-            self._subcrate = ROCrate(self.source, parse_subcrate=True)  # would load further nested RO-Crate
-            if list_parts := self._subcrate.root_dataset.get("hasPart"):
-                self._jsonld["hasPart"] = list_parts
+        Get the list of data entities listed under the `root_dataset["hasPart"]` of the nested RO-Crate.
+
+        This will load the nested RO-Crate if not already loaded.
+
+        :return: A list of data entities of the nested RO-Crate,
+        or an empty list if the nested RO-Crate does not list any part.
+        """
+
+        return self.get_crate().root_dataset.get("hasPart", [])
 
     def __getitem__(self, key):
-        if self._subcrate is None:
-            self._load_subcrate()
 
-        if key in self._jsonld:
-            # e.g the "original" entity keys such as id or type
-            return super().__getitem__(key)
+        if key == "hasPart":
+            return self._get_parts_subcrate_root()
 
-        # look into the subcrate entities
-        return self._subcrate.get(key)
+        else:
+            return super().__getitem__(key)
 
     def as_jsonld(self):
-        if self._subcrate is None:
+        if self._crate is None:
             self._load_subcrate()
         return super().as_jsonld()
 
     def get_entities(self):
-        if self._subcrate is None:
+        if self._crate is None:
             self._load_subcrate()
-        return self._subcrate.get_entities()
+        return self._crate.get_entities()
 
 
 def make_workflow_rocrate(workflow_path, wf_type, include_files=[],
diff --git a/test/test_read.py b/test/test_read.py
index bed09a6..709669c 100644
--- a/test/test_read.py
+++ b/test/test_read.py
@@ -217,11 +217,13 @@ def test_crate_with_subcrate(test_data_dir):
     assert "hasPart" not in subcrate
 
     # check lazy loading by accessing an entity from the subcrate
-    subfile_entity = subcrate.get("subfile.txt")
-    assert isinstance(subfile_entity, model.file.File)
+    list_subcrate_parts = subcrate.get("hasPart", [])
+    assert len(list_subcrate_parts) == 2  # subfile.txt and subsubcrate/
+    assert isinstance(list_subcrate_parts[0], DataEntity)
+    assert "subfile.txt" in [e.id for e in list_subcrate_parts]
 
     # check access from the top-level crate works too
-    assert main_crate.get("subcrate/subfile.txt") is subfile_entity
+    assert main_crate.get("subcrate/subfile.txt") in list_subcrate_parts
 
     # check with another nested rocrate
     assert isinstance(main_crate.get("subcrate/subsubcrate/deepfile.txt"), model.file.File)

From 629bcfe5eeb2ce28684b6598103dd24b58dcb1ee Mon Sep 17 00:00:00 2001
From: Laurent <laurent.thomas@embl.de>
Date: Thu, 4 Dec 2025 17:41:09 +0100
Subject: [PATCH 17/25] remove get_entities from Subcrate

---
 rocrate/rocrate.py | 5 -----
 1 file changed, 5 deletions(-)

diff --git a/rocrate/rocrate.py b/rocrate/rocrate.py
index d9224ed..27b743c 100644
--- a/rocrate/rocrate.py
+++ b/rocrate/rocrate.py
@@ -924,11 +924,6 @@ def as_jsonld(self):
             self._load_subcrate()
         return super().as_jsonld()
 
-    def get_entities(self):
-        if self._crate is None:
-            self._load_subcrate()
-        return self._crate.get_entities()
-
 
 def make_workflow_rocrate(workflow_path, wf_type, include_files=[],
                           fetch_remote=False, cwl=None, diagram=None):

From 771247db0508781254cb2aab1306b96a04846d51 Mon Sep 17 00:00:00 2001
From: Laurent <laurent.thomas@embl.de>
Date: Thu, 4 Dec 2025 17:43:30 +0100
Subject: [PATCH 18/25] remove subcrate.get_entities from tests

---
 test/test_read.py | 2 --
 1 file changed, 2 deletions(-)

diff --git a/test/test_read.py b/test/test_read.py
index 709669c..315be15 100644
--- a/test/test_read.py
+++ b/test/test_read.py
@@ -232,8 +232,6 @@ def test_crate_with_subcrate(test_data_dir):
     assert "hasPart" in subcrate
     assert len(subcrate["hasPart"]) == 2
 
-    assert len(subcrate.get_entities()) == 4  # root dataset, metadata.json, subfile, subsubcrate
-
     # reload the crate to "reset" the state to unloaded
     main_crate = load_crate_with_subcrate(test_data_dir)
     subcrate = main_crate.get("subcrate")

From 13682081e0ef82456c31db5fd892aa79add26363 Mon Sep 17 00:00:00 2001
From: Laurent <laurent.thomas@embl.de>
Date: Fri, 5 Dec 2025 11:21:29 +0100
Subject: [PATCH 19/25] implement crate writing

---
 rocrate/rocrate.py | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)

diff --git a/rocrate/rocrate.py b/rocrate/rocrate.py
index 27b743c..a77e270 100644
--- a/rocrate/rocrate.py
+++ b/rocrate/rocrate.py
@@ -21,6 +21,7 @@
 # limitations under the License.
 
 import errno
+from typing import cast
 import uuid
 import zipfile
 import atexit
@@ -873,14 +874,14 @@ def __init__(self, crate, source=None, dest_path=None, fetch_remote=False,
         Caller should rather use the get_crate() method to access the nested RO-Crate.
         """
 
-    def get_crate(self):
+    def get_crate(self) -> ROCrate:
         """
         Return the RO-Crate object referenced by this subcrate.
         """
         if self._crate is None:
             self._load_subcrate()
 
-        return self._crate
+        return cast(ROCrate, self._crate)
 
     def _load_subcrate(self):
         """
@@ -924,6 +925,10 @@ def as_jsonld(self):
             self._load_subcrate()
         return super().as_jsonld()
 
+    def write(self, base_path):
+        self.get_crate().write(base_path / self.id)
+        # TODO check with URL
+
 
 def make_workflow_rocrate(workflow_path, wf_type, include_files=[],
                           fetch_remote=False, cwl=None, diagram=None):

From 3c77ad255401fbeaade0bf2b815f2830611f52be Mon Sep 17 00:00:00 2001
From: Laurent <laurent.thomas@embl.de>
Date: Tue, 9 Dec 2025 15:42:22 +0100
Subject: [PATCH 20/25] add test writing the subcrate

---
 test/test_write.py | 12 ++++++++++++
 1 file changed, 12 insertions(+)

diff --git a/test/test_write.py b/test/test_write.py
index 3cf0422..9b3cd4f 100644
--- a/test/test_write.py
+++ b/test/test_write.py
@@ -627,6 +627,18 @@ def test_write_zip_nested_dest(tmpdir, helpers):
     assert (unpack_path / "subdir" / "a b" / "j k" / "l m.txt").is_file()
 
 
+def test_write_subcrate(test_data_dir, tmpdir):
+    """Read the test crate with subcrate and write it to a new location.
+    Check that the subcrate contents are correctly written."""
+    crate = ROCrate(test_data_dir / "crate_with_subcrate")
+    crate.write(tmpdir / "ro_crate_out")
+
+    assert (tmpdir / "ro_crate_out" / "subcrate" / "ro-crate-metadata.json").is_file()
+    assert (tmpdir / "ro_crate_out" / "subcrate" / "subfile.txt").is_file()
+
+    assert (tmpdir / "ro_crate_out" / "subcrate" / "subsubcrate" / "deepfile.txt").is_file()
+
+
 @pytest.mark.parametrize("version", ["1.0", "1.1", "1.2"])
 def test_write_version(tmpdir, helpers, version):
     basename = helpers.LEGACY_METADATA_FILE_NAME if version == "1.0" else helpers.METADATA_FILE_NAME

From dfe45eb8fd493c5ecb146a80c6b30793138fac9b Mon Sep 17 00:00:00 2001
From: simleo <simleo@crs4.it>
Date: Thu, 11 Dec 2025 12:27:12 +0100
Subject: [PATCH 21/25] test_write_subcrate: activate parse_subcrate

---
 test/test_write.py | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/test/test_write.py b/test/test_write.py
index 9b3cd4f..d82b2c2 100644
--- a/test/test_write.py
+++ b/test/test_write.py
@@ -630,13 +630,17 @@ def test_write_zip_nested_dest(tmpdir, helpers):
 def test_write_subcrate(test_data_dir, tmpdir):
     """Read the test crate with subcrate and write it to a new location.
     Check that the subcrate contents are correctly written."""
-    crate = ROCrate(test_data_dir / "crate_with_subcrate")
+    crate = ROCrate(test_data_dir / "crate_with_subcrate", parse_subcrate=True)
     crate.write(tmpdir / "ro_crate_out")
 
+    assert (tmpdir / "ro_crate_out" / "file.txt").is_file()
+    assert (tmpdir / "ro_crate_out" / "ro-crate-metadata.json").is_file()
+
     assert (tmpdir / "ro_crate_out" / "subcrate" / "ro-crate-metadata.json").is_file()
     assert (tmpdir / "ro_crate_out" / "subcrate" / "subfile.txt").is_file()
 
     assert (tmpdir / "ro_crate_out" / "subcrate" / "subsubcrate" / "deepfile.txt").is_file()
+    assert (tmpdir / "ro_crate_out" / "subcrate" / "subsubcrate" / "ro-crate-metadata.json").is_file()
 
 
 @pytest.mark.parametrize("version", ["1.0", "1.1", "1.2"])

From 6ea62fe7d592043fa209cf7aca6e7e851583ab8c Mon Sep 17 00:00:00 2001
From: simleo <simleo@crs4.it>
Date: Thu, 11 Dec 2025 16:32:51 +0100
Subject: [PATCH 22/25] don't modify the main crate's jsonld when loading a
 subcrate

---
 rocrate/rocrate.py | 34 ----------------------------------
 test/test_read.py  | 38 ++++++++++++++++----------------------
 2 files changed, 16 insertions(+), 56 deletions(-)

diff --git a/rocrate/rocrate.py b/rocrate/rocrate.py
index a77e270..9e0e5dc 100644
--- a/rocrate/rocrate.py
+++ b/rocrate/rocrate.py
@@ -886,45 +886,11 @@ def get_crate(self) -> ROCrate:
     def _load_subcrate(self):
         """
         Load the nested RO-Crate from the source path or URL.
-
-        This populates the attribute `hasPart` of the `Subcrate` entity,
-        with the data entities listed under the `root_dataset["hasPart"]` of the nested RO-Crate.
-        If the nested RO-crate does not list any part und `hasPart`,
-        then the `hasPart` attribute of the `Subcrate` entity will be an empty list.
         """
         if self._crate is None:
             # parse_subcrate=True to load further nested RO-Crate (on-demand / lazily too)
             self._crate = ROCrate(self.source, parse_subcrate=True)
 
-            # Note : assigning to hasPart keeps only the dict with id:entity not the actual entities
-            # such that when retrieving something from hasPart one was getting a string not an entity
-            self["hasPart"] = self._crate.root_dataset.get("hasPart", [])
-
-    def _get_parts_subcrate_root(self):
-        """
-        Get the list of data entities listed under the `root_dataset["hasPart"]` of the nested RO-Crate.
-
-        This will load the nested RO-Crate if not already loaded.
-
-        :return: A list of data entities of the nested RO-Crate,
-        or an empty list if the nested RO-Crate does not list any part.
-        """
-
-        return self.get_crate().root_dataset.get("hasPart", [])
-
-    def __getitem__(self, key):
-
-        if key == "hasPart":
-            return self._get_parts_subcrate_root()
-
-        else:
-            return super().__getitem__(key)
-
-    def as_jsonld(self):
-        if self._crate is None:
-            self._load_subcrate()
-        return super().as_jsonld()
-
     def write(self, base_path):
         self.get_crate().write(base_path / self.id)
         # TODO check with URL
diff --git a/test/test_read.py b/test/test_read.py
index 315be15..0d694ae 100644
--- a/test/test_read.py
+++ b/test/test_read.py
@@ -27,7 +27,6 @@
 import zipfile
 from pathlib import Path
 
-from rocrate import model
 from rocrate.rocrate import ROCrate, Subcrate
 from rocrate.model import DataEntity, ContextEntity, File, Dataset
 
@@ -206,38 +205,33 @@ def test_crate_with_subcrate(test_data_dir):
     assert main_crate.subcrate_entities == [subcrate]
 
     # Check the subcrate kept the conformsTo attribute from the original Dataset entity
-    assert "conformsTo" in subcrate
+    assert subcrate.get("conformsTo") == "https://w3id.org/ro/crate/"
 
     # check that at this point, we have not yet loaded the subcrate
-    # e.g the json ld should just have id, type and conformsTo
-    jsonld = subcrate._jsonld
-    jsonld.pop("conformsTo")
+    assert subcrate._crate is None
 
-    assert jsonld == subcrate._empty()
-    assert "hasPart" not in subcrate
+    # check access from the top-level crate
+    subfile = main_crate.get("subcrate/subfile.txt")
+    assert isinstance(subfile, File)
 
-    # check lazy loading by accessing an entity from the subcrate
-    list_subcrate_parts = subcrate.get("hasPart", [])
-    assert len(list_subcrate_parts) == 2  # subfile.txt and subsubcrate/
-    assert isinstance(list_subcrate_parts[0], DataEntity)
-    assert "subfile.txt" in [e.id for e in list_subcrate_parts]
-
-    # check access from the top-level crate works too
-    assert main_crate.get("subcrate/subfile.txt") in list_subcrate_parts
+    # check that the above dereferencing triggered lazy loading
+    assert isinstance(subcrate._crate, ROCrate)
+    assert subfile.id == "subfile.txt"
+    assert subfile.crate is not main_crate
+    assert subfile.crate is subcrate._crate
 
     # check with another nested rocrate
-    assert isinstance(main_crate.get("subcrate/subsubcrate/deepfile.txt"), model.file.File)
-
-    # Check the hasPart of the subcrate lists the file and the subsubcrate
-    assert "hasPart" in subcrate
-    assert len(subcrate["hasPart"]) == 2
+    assert isinstance(main_crate.get("subcrate/subsubcrate/deepfile.txt"), File)
 
     # reload the crate to "reset" the state to unloaded
     main_crate = load_crate_with_subcrate(test_data_dir)
     subcrate = main_crate.get("subcrate")
+    assert subcrate._crate is None
 
-    # as_jsonld should trigger loading of the subcrate
-    assert subcrate.as_jsonld() != subcrate._empty()
+    # get_crate() should trigger loading of the subcrate
+    nested_crate = subcrate.get_crate()
+    assert isinstance(nested_crate, ROCrate)
+    assert subcrate._crate is nested_crate
 
 
 @pytest.mark.parametrize("override", [False, True])

From 284c9edd4c4a34595e4b8907d4e9f89d33f7aaf0 Mon Sep 17 00:00:00 2001
From: simleo <simleo@crs4.it>
Date: Fri, 12 Dec 2025 08:35:56 +0100
Subject: [PATCH 23/25] no trailing slash in generic ro-crate profile, as per
 the spec

---
 rocrate/rocrate.py                                            | 2 +-
 test/test-data/crate_with_subcrate/ro-crate-metadata.json     | 4 ++--
 .../crate_with_subcrate/subcrate/ro-crate-metadata.json       | 2 +-
 test/test_read.py                                             | 2 +-
 4 files changed, 5 insertions(+), 5 deletions(-)

diff --git a/rocrate/rocrate.py b/rocrate/rocrate.py
index 9e0e5dc..97ce542 100644
--- a/rocrate/rocrate.py
+++ b/rocrate/rocrate.py
@@ -99,7 +99,7 @@ def pick_type(json_entity, type_map, fallback=None, parse_subcrate=False):
         if parse_subcrate and (list_profiles := get_norm_value(json_entity, "conformsTo")):
 
             for profile_ref in list_profiles:
-                if profile_ref.startswith("https://w3id.org/ro/crate/"):
+                if profile_ref.startswith("https://w3id.org/ro/crate"):
                     return Subcrate
 
         return Dataset
diff --git a/test/test-data/crate_with_subcrate/ro-crate-metadata.json b/test/test-data/crate_with_subcrate/ro-crate-metadata.json
index 46aff37..ae5a7eb 100644
--- a/test/test-data/crate_with_subcrate/ro-crate-metadata.json
+++ b/test/test-data/crate_with_subcrate/ro-crate-metadata.json
@@ -31,8 +31,8 @@
             "@id": "subcrate/",
             "@type": "Dataset",
             "conformsTo": {
-                "@id": "https://w3id.org/ro/crate/"
+                "@id": "https://w3id.org/ro/crate"
             }
         }
     ]
-}
\ No newline at end of file
+}
diff --git a/test/test-data/crate_with_subcrate/subcrate/ro-crate-metadata.json b/test/test-data/crate_with_subcrate/subcrate/ro-crate-metadata.json
index 2bd736f..86273d1 100644
--- a/test/test-data/crate_with_subcrate/subcrate/ro-crate-metadata.json
+++ b/test/test-data/crate_with_subcrate/subcrate/ro-crate-metadata.json
@@ -28,7 +28,7 @@
             "@id": "subsubcrate/",
             "@type": "Dataset",
             "conformsTo": {
-                "@id": "https://w3id.org/ro/crate/"
+                "@id": "https://w3id.org/ro/crate"
             }
         }
     ]
diff --git a/test/test_read.py b/test/test_read.py
index 0d694ae..abe0c62 100644
--- a/test/test_read.py
+++ b/test/test_read.py
@@ -205,7 +205,7 @@ def test_crate_with_subcrate(test_data_dir):
     assert main_crate.subcrate_entities == [subcrate]
 
     # Check the subcrate kept the conformsTo attribute from the original Dataset entity
-    assert subcrate.get("conformsTo") == "https://w3id.org/ro/crate/"
+    assert subcrate.get("conformsTo") == "https://w3id.org/ro/crate"
 
     # check that at this point, we have not yet loaded the subcrate
     assert subcrate._crate is None

From ad51c860c37134f1cadf312ddc15e04f2180df36 Mon Sep 17 00:00:00 2001
From: simleo <simleo@crs4.it>
Date: Fri, 12 Dec 2025 09:52:59 +0100
Subject: [PATCH 24/25] reindent metadata files to reduce diffs

---
 test/test-data/crate_with_subcrate/ro-crate-metadata.json | 8 ++++++--
 .../crate_with_subcrate/subcrate/ro-crate-metadata.json   | 8 ++++++--
 .../subcrate/subsubcrate/ro-crate-metadata.json           | 4 +++-
 3 files changed, 15 insertions(+), 5 deletions(-)

diff --git a/test/test-data/crate_with_subcrate/ro-crate-metadata.json b/test/test-data/crate_with_subcrate/ro-crate-metadata.json
index ae5a7eb..89e7413 100644
--- a/test/test-data/crate_with_subcrate/ro-crate-metadata.json
+++ b/test/test-data/crate_with_subcrate/ro-crate-metadata.json
@@ -9,8 +9,12 @@
             "license": "https://spdx.org/licenses/MIT.html",
             "datePublished": "2025-12-02T08:39:54+00:00",
             "hasPart": [
-                    {"@id": "file.txt"},
-                    {"@id":"subcrate/"}
+                {
+                    "@id": "file.txt"
+                },
+                {
+                    "@id": "subcrate/"
+                }
             ]
         },
         {
diff --git a/test/test-data/crate_with_subcrate/subcrate/ro-crate-metadata.json b/test/test-data/crate_with_subcrate/subcrate/ro-crate-metadata.json
index 86273d1..eca9026 100644
--- a/test/test-data/crate_with_subcrate/subcrate/ro-crate-metadata.json
+++ b/test/test-data/crate_with_subcrate/subcrate/ro-crate-metadata.json
@@ -6,8 +6,12 @@
             "@type": "Dataset",
             "datePublished": "2025-12-02T08:39:54+00:00",
             "hasPart": [
-                    {"@id": "subfile.txt"},
-                    {"@id": "subsubcrate/"}
+                {
+                    "@id": "subfile.txt"
+                },
+                {
+                    "@id": "subsubcrate/"
+                }
             ]
         },
         {
diff --git a/test/test-data/crate_with_subcrate/subcrate/subsubcrate/ro-crate-metadata.json b/test/test-data/crate_with_subcrate/subcrate/subsubcrate/ro-crate-metadata.json
index 1b5390f..b552c79 100644
--- a/test/test-data/crate_with_subcrate/subcrate/subsubcrate/ro-crate-metadata.json
+++ b/test/test-data/crate_with_subcrate/subcrate/subsubcrate/ro-crate-metadata.json
@@ -6,7 +6,9 @@
             "@type": "Dataset",
             "datePublished": "2025-12-02T08:39:54+00:00",
             "hasPart": [
-                    {"@id": "deepfile.txt"}
+                {
+                    "@id": "deepfile.txt"
+                }
             ]
         },
         {

From d2649a218960a87ecb3cb37fe385dfef98380b78 Mon Sep 17 00:00:00 2001
From: simleo <simleo@crs4.it>
Date: Fri, 12 Dec 2025 12:27:28 +0100
Subject: [PATCH 25/25] don't use dereference to check for unlisted files

---
 rocrate/rocrate.py | 12 ++++++------
 test/test_write.py | 24 ++++++++++++++++--------
 2 files changed, 22 insertions(+), 14 deletions(-)

diff --git a/rocrate/rocrate.py b/rocrate/rocrate.py
index 97ce542..c51710f 100644
--- a/rocrate/rocrate.py
+++ b/rocrate/rocrate.py
@@ -417,6 +417,10 @@ def get_entities(self):
     def _get_root_jsonld(self):
         self.root_dataset.properties()
 
+    def __contains__(self, entity_id):
+        canonical_id = self.resolve_id(entity_id)
+        return canonical_id in self.__entity_map
+
     def dereference(self, entity_id, default=None):
         canonical_id = self.resolve_id(entity_id)
 
@@ -563,7 +567,7 @@ def _copy_unlisted(self, top, base_path):
             for name in files:
                 source = root / name
                 rel = source.relative_to(top)
-                if not self.dereference(str(rel)):
+                if str(rel) not in self:
                     dest = base_path / rel
                     if not dest.exists() or not dest.samefile(source):
                         shutil.copyfile(source, dest)
@@ -621,7 +625,7 @@ def _stream_zip(self, chunk_size=8192, out_path=None):
                             continue
 
                         rel = source.relative_to(self.source)
-                        if not self.dereference(str(rel)) and not str(rel) in listed_files:
+                        if str(rel) not in self and not str(rel) in listed_files:
                             with archive.open(str(rel), mode='w') as out_file, open(source, 'rb') as in_file:
                                 while chunk := in_file.read(chunk_size):
                                     out_file.write(chunk)
@@ -891,10 +895,6 @@ def _load_subcrate(self):
             # parse_subcrate=True to load further nested RO-Crate (on-demand / lazily too)
             self._crate = ROCrate(self.source, parse_subcrate=True)
 
-    def write(self, base_path):
-        self.get_crate().write(base_path / self.id)
-        # TODO check with URL
-
 
 def make_workflow_rocrate(workflow_path, wf_type, include_files=[],
                           fetch_remote=False, cwl=None, diagram=None):
diff --git a/test/test_write.py b/test/test_write.py
index d82b2c2..3b55897 100644
--- a/test/test_write.py
+++ b/test/test_write.py
@@ -627,20 +627,28 @@ def test_write_zip_nested_dest(tmpdir, helpers):
     assert (unpack_path / "subdir" / "a b" / "j k" / "l m.txt").is_file()
 
 
-def test_write_subcrate(test_data_dir, tmpdir):
+@pytest.mark.parametrize("to_zip", [False, True])
+def test_write_subcrate(test_data_dir, tmpdir, to_zip):
     """Read the test crate with subcrate and write it to a new location.
     Check that the subcrate contents are correctly written."""
     crate = ROCrate(test_data_dir / "crate_with_subcrate", parse_subcrate=True)
-    crate.write(tmpdir / "ro_crate_out")
+    out_path = tmpdir / "ro_crate_out"
+    if to_zip:
+        zip_path = tmpdir / 'ro_crate_out.crate.zip'
+        crate.write_zip(zip_path)
+        with zipfile.ZipFile(zip_path, "r") as zf:
+            zf.extractall(out_path)
+    else:
+        crate.write(out_path)
 
-    assert (tmpdir / "ro_crate_out" / "file.txt").is_file()
-    assert (tmpdir / "ro_crate_out" / "ro-crate-metadata.json").is_file()
+    assert (out_path / "file.txt").is_file()
+    assert (out_path / "ro-crate-metadata.json").is_file()
 
-    assert (tmpdir / "ro_crate_out" / "subcrate" / "ro-crate-metadata.json").is_file()
-    assert (tmpdir / "ro_crate_out" / "subcrate" / "subfile.txt").is_file()
+    assert (out_path / "subcrate" / "ro-crate-metadata.json").is_file()
+    assert (out_path / "subcrate" / "subfile.txt").is_file()
 
-    assert (tmpdir / "ro_crate_out" / "subcrate" / "subsubcrate" / "deepfile.txt").is_file()
-    assert (tmpdir / "ro_crate_out" / "subcrate" / "subsubcrate" / "ro-crate-metadata.json").is_file()
+    assert (out_path / "subcrate" / "subsubcrate" / "deepfile.txt").is_file()
+    assert (out_path / "subcrate" / "subsubcrate" / "ro-crate-metadata.json").is_file()
 
 
 @pytest.mark.parametrize("version", ["1.0", "1.1", "1.2"])