lamindb_setup 0.75.0__py2.py3-none-any.whl → 0.76.0__py2.py3-none-any.whl

lamindb_setup/__init__.py

@@ -34,7 +34,7 @@ Modules & settings:
 
 """
 
-__version__ = "0.75.0"  # denote a release candidate for 0.1.0 with 0.1rc1
+__version__ = "0.76.0"  # denote a release candidate for 0.1.0 with 0.1rc1
 
 import sys
 from os import name as _os_name

lamindb_setup/_init_instance.py

@@ -15,6 +15,7 @@ from ._close import close as close_instance
 from ._silence_loggers import silence_loggers
 from .core import InstanceSettings
 from .core._settings import settings
+from .core._settings_instance import is_local_db_url
 from .core._settings_storage import StorageSettings, init_storage
 from .core.upath import UPath
 
@@ -245,10 +246,16 @@ def init(
         if instance_state == "connected":
             settings.auto_connect = True  # we can also debate this switch here
             return None
-        # cannot past instance_id here because instance does not yet exist!
-        # the instance_id field of the storage table is populated at the end of
-        # init_instance_hub
-        ssettings = init_storage(storage)
+        # the conditions here match `isettings.is_remote`, but I currently don't
+        # see a way of making this more elegant; should become possible if we
+        # remove the instance.storage_id FK on the hub
+        prevent_register_hub = is_local_db_url(db) if db is not None else False
+        ssettings = init_storage(
+            storage,
+            instance_id=instance_id,
+            init_instance=True,
+            prevent_register_hub=prevent_register_hub,
+        )
         isettings = InstanceSettings(
             id=instance_id,  # type: ignore
             owner=settings.user.handle,

lamindb_setup/core/__init__.py

@@ -11,6 +11,7 @@ Settings:
    StorageSettings
 
 """
+
 from . import django, types, upath
 from ._deprecated import deprecated
 from ._docs import doc_args

lamindb_setup/core/_hub_core.py

@@ -160,7 +160,9 @@ def _init_storage(
         )
         ssettings._instance_id = settings.instance._id
     instance_id_hex = (
-        None if ssettings._instance_id is None else ssettings._instance_id.hex
+        ssettings._instance_id.hex
+        if (ssettings._instance_id is not None and auto_populate_instance)
+        else None
     )
     fields = {
         "id": id.hex,

lamindb_setup/core/_settings_instance.py

@@ -29,6 +29,16 @@ def sanitize_git_repo_url(repo_url: str) -> str:
     return repo_url.replace(".git", "")
 
 
+def is_local_db_url(db_url: str) -> bool:
+    if "@localhost:" in db_url:
+        return True
+    if "@0.0.0.0:" in db_url:
+        return True
+    if "@127.0.0.1" in db_url:
+        return True
+    return False
+
+
 class InstanceSettings:
     """Instance settings."""
 
@@ -370,17 +380,8 @@ class InstanceSettings:
         if not self.storage.type_is_cloud:
             return False
 
-        def is_local_uri(uri: str):
-            if "@localhost:" in uri:
-                return True
-            if "@0.0.0.0:" in uri:
-                return True
-            if "@127.0.0.1" in uri:
-                return True
-            return False
-
         if self.dialect == "postgresql":
-            if is_local_uri(self.db):
+            if is_local_db_url(self.db):
                 return False
         # returns True for cloud SQLite
         # and remote postgres

lamindb_setup/core/_settings_storage.py

@@ -78,7 +78,11 @@ def mark_storage_root(root: UPathStr, uid: str):
 
 
 def init_storage(
-    root: UPathStr, instance_id: UUID | None = None, register_hub: bool | None = None
+    root: UPathStr,
+    instance_id: UUID | None = None,
+    register_hub: bool | None = None,
+    prevent_register_hub: bool = False,
+    init_instance: bool = False,
 ) -> StorageSettings:
     if root is None:
         raise ValueError("`storage` argument can't be `None`")
@@ -114,11 +118,12 @@ def init_storage(
         instance_id=instance_id,
     )
     # the below might update the uid with one that's already taken on the hub
-    if ssettings.type_is_cloud or register_hub:
-        from ._hub_core import delete_storage_record
-        from ._hub_core import init_storage as init_storage_hub
+    if not prevent_register_hub:
+        if ssettings.type_is_cloud or register_hub:
+            from ._hub_core import delete_storage_record
+            from ._hub_core import init_storage as init_storage_hub
 
-        init_storage_hub(ssettings)
+            init_storage_hub(ssettings, auto_populate_instance=not init_instance)
     # below comes last only if everything else was successful
     try:
         # (federated) credentials for AWS access are provisioned under-the-hood

lamindb_setup/core/_setup_bionty_sources.py

@@ -7,6 +7,9 @@ from django.db.utils import OperationalError, ProgrammingError
 if TYPE_CHECKING:
     from ._settings_instance import InstanceSettings
 
+# bionty.Source -> bionty.base
+RENAME = {"name": "source", "description": "source_name"}
+
 
 def write_bionty_sources(isettings: InstanceSettings) -> None:
     """Write bionty sources to Source table."""
@@ -14,10 +17,14 @@ def write_bionty_sources(isettings: InstanceSettings) -> None:
         return None
     import shutil
 
+    import bionty
     import bionty.base as bionty_base
+    from bionty._bionty import list_biorecord_models
     from bionty.base.dev._handle_sources import parse_sources_yaml
     from bionty.models import Source
 
+    bionty_models = list_biorecord_models(bionty)
+
     shutil.copy(
         bionty_base.settings.current_sources, bionty_base.settings.lamindb_sources
     )
@@ -46,7 +53,10 @@ def write_bionty_sources(isettings: InstanceSettings) -> None:
         # won't need this once lamindb is released with the new pin
         kwargs["run"] = None  # can't yet access tracking information
         kwargs["in_db"] = False
-        # kwargs["name"] = kwargs.pop("source")
+        for db_field, base_col in RENAME.items():
+            kwargs[db_field] = kwargs.pop(base_col)
+        if kwargs["entity"] in bionty_models:
+            kwargs["entity"] = f"bionty.{kwargs['entity']}"
         record = Source(**kwargs)
         all_records.append(record)
 
@@ -66,6 +76,12 @@ def load_bionty_sources(isettings: InstanceSettings):
     try:
         # need try except because of integer primary key migration
         active_records = Source.objects.filter(currently_used=True).all().values()
+        for kwargs in active_records:
+            for db_field, base_col in RENAME.items():
+                kwargs[base_col] = kwargs.pop(db_field)
+                # TODO: non-bionty schema?
+                if db_field == "entity":
+                    kwargs["entity"] = kwargs["entity"].split(".")[1]
         write_yaml(
             parse_currently_used_sources(active_records),
             bionty_base.settings.lamindb_sources,

lamindb_setup/core/hashing.py

@@ -17,6 +17,8 @@ from typing import TYPE_CHECKING, Iterable
 
 import psutil
 
+HASH_LENGTH = 22
+
 if TYPE_CHECKING:
     from .types import Path, UPathStr
 
@@ -39,8 +41,8 @@ def b16_to_b64(s: str):
 # a lot to read about this: lamin-notes/2022/hashing
 def hash_set(s: set[str]) -> str:
     bstr = ":".join(sorted(s)).encode("utf-8")
-    # as we're truncating at 20 b64, we choose md5 over sha512
-    return to_b64_str(hashlib.md5(bstr).digest())[:20]
+    # as we're truncating at 22 b64, we choose md5 over sha512
+    return to_b64_str(hashlib.md5(bstr).digest())[:HASH_LENGTH]
 
 
 def hash_md5s_from_dir(hashes: Iterable[str]) -> tuple[str, str]:
@@ -50,7 +52,7 @@ def hash_md5s_from_dir(hashes: Iterable[str]) -> tuple[str, str]:
         hashlib.md5(hash.encode("utf-8")).digest() for hash in sorted(hashes)
     )
     digest = hashlib.md5(digests).digest()
-    return to_b64_str(digest)[:22], "md5-d"
+    return to_b64_str(digest)[:HASH_LENGTH], "md5-d"
 
 
 def hash_code(file_path: UPathStr):
@@ -85,7 +87,7 @@ def hash_file(
                 hashlib.sha1(first_chunk).digest() + hashlib.sha1(last_chunk).digest()
             ).digest()
             hash_type = "sha1-fl"
-    return to_b64_str(digest)[:22], hash_type
+    return to_b64_str(digest)[:HASH_LENGTH], hash_type
 
 
 def hash_dir(path: Path):

lamindb_setup/core/upath.py

@@ -772,7 +772,7 @@ def check_storage_is_empty(
     hint += " ignored"
     message = (
         f"Storage {directory_string} contains {n_objects - n_offset_objects} objects "
-        f"({hint}) - {ask_for_deletion}\n{objects}"
+        f"({hint}) - {ask_for_deletion}"
     )
     if n_diff > 0:
         if raise_error:

{lamindb_setup-0.75.0.dist-info → lamindb_setup-0.76.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: lamindb_setup
-Version: 0.75.0
+Version: 0.76.0
 Summary: Setup & configure LaminDB.
 Author-email: Lamin Labs <open-source@lamin.ai>
 Description-Content-Type: text/markdown

{lamindb_setup-0.75.0.dist-info → lamindb_setup-0.76.0.dist-info}/RECORD

@@ -1,4 +1,4 @@
-lamindb_setup/__init__.py,sha256=h484xhWUuDVBkjDLj8nu5IKkNELvscWtoABYLZIU7dU,1542
+lamindb_setup/__init__.py,sha256=mCahlCfmgFyDdpkDRVZFQnJDPKLOtPIcgiNfMAUqAkw,1542
 lamindb_setup/_cache.py,sha256=wA7mbysANwe8hPNbjDo9bOmXJ0xIyaS5iyxIpxSWji4,846
 lamindb_setup/_check.py,sha256=28PcG8Kp6OpjSLSi1r2boL2Ryeh6xkaCL87HFbjs6GA,129
 lamindb_setup/_check_setup.py,sha256=cNEL9Q4yPpmEkGKHH8JgullWl1VUZwALJ4RHn9wZypY,2613
@@ -8,7 +8,7 @@ lamindb_setup/_delete.py,sha256=Y8KSFYgY0UHAvjd7cCL6hZ_XiLeJwx50BguVATcj_Xo,5524
 lamindb_setup/_django.py,sha256=EoyWvFzH0i9wxjy4JZhcoXCTckztP_Mrl6FbYQnMmLE,1534
 lamindb_setup/_exportdb.py,sha256=43g77-tH-vAlTn8ig1mMD9-KXLKvxUeDLaq0gVu3l-c,2114
 lamindb_setup/_importdb.py,sha256=yYYShzUajTsR-cTW4CZ-UNDWZY2uE5PAgNbp-wn8Ogc,1874
-lamindb_setup/_init_instance.py,sha256=jcKtXLOwkoq7bC-0j0xnNU4_up9nPl0iKQPNoJbBts8,12078
+lamindb_setup/_init_instance.py,sha256=pR_pK38okij7CfnQDecJLKR5lhhTeE4oZ14Lb9bMMQg,12388
 lamindb_setup/_migrate.py,sha256=P4n3x0SYzO9szjF2-JMa7z4mQadtWjHv5ow4HbCDZLI,8864
 lamindb_setup/_register_instance.py,sha256=alQuYp2f8Ct8xvRC1gt8p_HZ0tqCd3gZD3kiPBLPpsI,1269
 lamindb_setup/_schema.py,sha256=b3uzhhWpV5mQtDwhMINc2MabGCnGLESy51ito3yl6Wc,679
@@ -16,31 +16,31 @@ lamindb_setup/_schema_metadata.py,sha256=L0QjUsRmL63txvqptj4t_wh4ZamQ15mKSYsSMrH
 lamindb_setup/_set_managed_storage.py,sha256=mNZrANn-9rwZ0oGWxxg0wS0T0VOQCWyo4nSSyNAE15Q,1419
 lamindb_setup/_setup_user.py,sha256=6Oc7Rke-yRQSZbuntdUAz8QbJ6UuPzYHI9FnYlf_q-A,3670
 lamindb_setup/_silence_loggers.py,sha256=AKF_YcHvX32eGXdsYK8MJlxEaZ-Uo2f6QDRzjKFCtws,1568
-lamindb_setup/core/__init__.py,sha256=dV9S-rQpNK9JcBn4hiEmiLnmNqfpPFJD9pqagMCaIew,416
+lamindb_setup/core/__init__.py,sha256=BxIVMX5HQq8oZ1OuY_saUEJz5Tdd7gaCPngxVu5iou4,417
 lamindb_setup/core/_aws_credentials.py,sha256=uKMQO9q42Hnepz8aj3RxwLKDWUJx8pNOYrFnnNh5X40,5325
 lamindb_setup/core/_aws_storage.py,sha256=nEjeUv4xUVpoV0Lx-zjjmyb9w804bDyaeiM-OqbfwM0,1799
 lamindb_setup/core/_deprecated.py,sha256=3qxUI1dnDlSeR0BYrv7ucjqRBEojbqotPgpShXs4KF8,2520
 lamindb_setup/core/_docs.py,sha256=3k-YY-oVaJd_9UIY-LfBg_u8raKOCNfkZQPA73KsUhs,276
 lamindb_setup/core/_hub_client.py,sha256=Gpt3TmxWQMVenKugCu1agUb-xGN9YFsQOKmrHuFiiDs,5605
-lamindb_setup/core/_hub_core.py,sha256=L9rQXM5eVIFOR6oP5uSqHpkYy4y3Bx0w_6acy-_hEVk,16787
+lamindb_setup/core/_hub_core.py,sha256=XqXZatJOIDXukfdwL3Buq3WXZK-VVGg5OFRm2PrzO1o,16836
 lamindb_setup/core/_hub_crud.py,sha256=b1XF7AJpM9Q-ttm9nPG-r3OTRWHQaGzAGIyvmb83NTo,4859
 lamindb_setup/core/_hub_utils.py,sha256=w5IRtrxZcvxmGSJslzuZF89ewkzXV4cCUmZUVrqmAfo,3026
 lamindb_setup/core/_private_django_api.py,sha256=KIn43HOhiRjkbTbddyJqv-WNTTa1bAizbM1tWXoXPBg,2869
 lamindb_setup/core/_settings.py,sha256=46axQ5HPvI0X9YuotgdpuSOfSo7qYU1DudIx3vxpFk0,4471
-lamindb_setup/core/_settings_instance.py,sha256=OMN6ZROF7iV9Juh0bgE9fOmedE7EJI_YniztwJ5-acI,17623
+lamindb_setup/core/_settings_instance.py,sha256=rPTL_Iwv-ZVMRzu_EV_E7APFkEXLoD0TgUrz5293-vg,17586
 lamindb_setup/core/_settings_load.py,sha256=63PTIVLk3oH1XUmznswygcOk_8VADrJLMB6O9AOppYg,3907
 lamindb_setup/core/_settings_save.py,sha256=n_tYfb9EBSxwm4LHyPRHJptE5uB8lmHhcRkz1JkAmhg,2781
-lamindb_setup/core/_settings_storage.py,sha256=65aobewYX6VfOeYZjZQOOI7ZD_3b4QA9TDmrduU0m4c,13262
+lamindb_setup/core/_settings_storage.py,sha256=bvEOm2aH4M6xBTfTARcWkx6b-jtKYl_AIUbH5yQC9GI,13435
 lamindb_setup/core/_settings_store.py,sha256=VEE8Ff2A7y4j8ksOaa7g48jNaOqe1PBnBjb1PKKGUKU,2115
 lamindb_setup/core/_settings_user.py,sha256=P2lC4WDRAFfT-Xq3MlXJ-wMKIHCoGNhMTQfRGIAyUNQ,1344
-lamindb_setup/core/_setup_bionty_sources.py,sha256=Y6_DiIj1RS8HsUewWtfjIDchYkL1GE98bf75hHN-j5A,2759
+lamindb_setup/core/_setup_bionty_sources.py,sha256=o_MH8Ja_00nA9fQjZh7zlmLHvBeQlTrlNfO0GVivjew,3433
 lamindb_setup/core/cloud_sqlite_locker.py,sha256=reu02M4aE2BT_A5AFqwhv48l91mOMyQ4zTd-hh-wtuU,6922
 lamindb_setup/core/django.py,sha256=QUQm3zt5QIiD8uv6o9vbSm_bshqiSWzKSkgD3z2eJCg,3542
 lamindb_setup/core/exceptions.py,sha256=eoI7AXgATgDVzgArtN7CUvpaMUC067vsBg5LHCsWzDM,305
-lamindb_setup/core/hashing.py,sha256=_JliYeCcjT_foOUJ5ck1rvcCo9q7r4b4SaSclQ_w4Qo,3071
+lamindb_setup/core/hashing.py,sha256=Y2cvEaqtm3KwpHqj5ZG2f_sLaXhsQT4oDrmJdHbOQeo,3116
 lamindb_setup/core/types.py,sha256=bcYnZ0uM_2NXKJCl94Mmc-uYrQlRUUVKG3sK2N-F-N4,532
-lamindb_setup/core/upath.py,sha256=dwudkTVsXuyjS-2xR16WomcWtXJAEfRZ0ZzFq8_EDhE,27157
-lamindb_setup-0.75.0.dist-info/LICENSE,sha256=UOZ1F5fFDe3XXvG4oNnkL1-Ecun7zpHzRxjp-XsMeAo,11324
-lamindb_setup-0.75.0.dist-info/WHEEL,sha256=Sgu64hAMa6g5FdzHxXv9Xdse9yxpGGMeagVtPMWpJQY,99
-lamindb_setup-0.75.0.dist-info/METADATA,sha256=tpmMZNwLSmNudPk6PE-n6mRXDk7DSjAvfEMbtGeU9uA,1638
-lamindb_setup-0.75.0.dist-info/RECORD,,
+lamindb_setup/core/upath.py,sha256=2N2IPRGwTBaamU9E_2CIEbKObwYRU_7yLfi0ATaBwWI,27146
+lamindb_setup-0.76.0.dist-info/LICENSE,sha256=UOZ1F5fFDe3XXvG4oNnkL1-Ecun7zpHzRxjp-XsMeAo,11324
+lamindb_setup-0.76.0.dist-info/WHEEL,sha256=Sgu64hAMa6g5FdzHxXv9Xdse9yxpGGMeagVtPMWpJQY,99
+lamindb_setup-0.76.0.dist-info/METADATA,sha256=6muSAoC177Ogn0kEBEGO7W-kfZrBHLamrTVJW4BuGQE,1638
+lamindb_setup-0.76.0.dist-info/RECORD,,