pybiolib 1.1.2155__py3-none-any.whl → 1.1.2169__py3-none-any.whl

biolib/_data_record/data_record.py

@@ -2,12 +2,14 @@ import os
 from collections import namedtuple
 from datetime import datetime
 from fnmatch import fnmatch
+from pathlib import Path
 from struct import Struct
 from typing import Callable, Dict, List, Union, cast
 
 from biolib import api, utils
 from biolib._internal import types
 from biolib._internal.data_record import get_data_record_state_from_uri
+from biolib._internal.data_record.data_record import validate_sqlite_v1
 from biolib._internal.data_record.remote_storage_endpoint import DataRecordRemoteStorageEndpoint
 from biolib._internal.file_utils import get_files_and_size_of_directory, get_iterable_zip_stream
 from biolib._internal.http_client import HttpClient
@@ -106,6 +108,13 @@ class DataRecord:
             logger.info(f"Validating data record of type {data_record_type['name']}")
             for rule in data_record_type['validation_rules']:
                 logger.info(f"Validating rule {rule['type']} for {rule['path']}...")
+                if rule['type'] == "sqlite-v1":
+                    try:
+                        validate_sqlite_v1(schema=rule['rule'], sqlite_file=Path(rule['path']))
+                    except Exception as error:
+                        raise Exception("Data Record Validation failed") from error
+                else:
+                    raise Exception(f"Error processing data record validation: unknown rule type {rule['type']}")
 
         min_chunk_size_bytes = 10_000_000
         chunk_size_in_bytes: int

biolib/_internal/data_record/__init__.py

@@ -1 +1 @@
-from .data_record import get_data_record_state_from_uri
+from .data_record import get_data_record_state_from_uri, validate_sqlite_v1

biolib/_internal/data_record/data_record.py

@@ -1,8 +1,86 @@
+import sqlite3
+from pathlib import Path
+
+from biolib._internal.types.data_record import SqliteV1DatabaseSchema
 from biolib.api import client as api_client
 from biolib.biolib_api_client import AppGetResponse
 from biolib.biolib_api_client.lfs_types import DataRecordVersionInfo
 
 
+def get_actual_schema(db_path):
+    if not db_path.exists():
+        raise Exception(f'File {db_path} not found.')
+    conn = sqlite3.connect(db_path)
+    cursor = conn.cursor()
+
+    cursor.execute("SELECT name FROM sqlite_master WHERE type='table';")
+    tables = cursor.fetchall()
+
+    actual_schema: SqliteV1DatabaseSchema = {'tables': []}
+    for table in tables:
+        table_name = table[0]
+        cursor.execute(f'PRAGMA table_info({table_name});')
+        columns = cursor.fetchall()
+        actual_schema['tables'][table_name] = {'columns': {}}
+        for column in columns:
+            actual_schema['tables'][table_name]['columns'][column[1]] = {
+                'type': column[2],
+                'nullable': not bool(column[3]),
+            }
+
+        cursor.execute(f'PRAGMA foreign_key_list({table_name});')
+        foreign_keys = cursor.fetchall()
+        for fk in foreign_keys:
+            actual_schema['tables'][table_name]['columns'][fk[3]]['foreign_key'] = {'table': fk[2], 'column': fk[4]}
+
+    conn.close()
+    return actual_schema
+
+
+def verify_schema(specification: SqliteV1DatabaseSchema, actual_schema: SqliteV1DatabaseSchema):
+    for table_name, table_spec in specification['tables'].items():
+        if table_name not in actual_schema['tables']:
+            raise Exception(f"Error: Table '{table_name}' is missing.")
+
+        for column_name, column_spec in table_spec['columns'].items():
+            if column_name not in actual_schema['tables'][table_name]['columns']:
+                raise Exception(f"Error: Column '{column_name}' in table '{table_name}' is missing.")
+
+            actual_column = actual_schema['tables'][table_name]['columns'][column_name]
+            if actual_column['type'] != column_spec['type']:
+                raise Exception(
+                    f"Error: Column '{column_name}' in table '{table_name}' "
+                    "has type '{actual_column['type']}' but expected '{column_spec['type']}'."
+                )
+
+            if not actual_column['nullable'] and column_spec.get('nullable', True):
+                raise Exception(
+                    f"Error: Column '{column_name}' in table '{table_name}' is "
+                    'not nullable but should be nullable according to the specification.'
+                )
+
+        for column_name, column_spec in table_spec['columns'].items():
+            if column_spec.get('foreign_key'):
+                foreign_key_spec = column_spec['foreign_key']
+                if actual_schema['tables'][table_name]['columns'][column_name].get('foreign_key'):
+                    fk = actual_schema['tables'][table_name]['columns'][column_name]['foreign_key']
+                    if (
+                        fk
+                        and foreign_key_spec
+                        and fk['table'] == foreign_key_spec['table']
+                        and fk['column'] == foreign_key_spec['column']
+                    ):
+                        raise Exception(
+                            f"Error: Column '{column_name}' in table '{table_name}' does "
+                            'not have the correct foreign key constraint.'
+                        )
+                else:
+                    raise Exception(
+                        f"Error: Column '{column_name}' in table '{table_name}' does "
+                        'not have a foreign key constraint.'
+                    )
+
+
 def get_data_record_state_from_uri(uri) -> 'DataRecordVersionInfo':
     app_response: AppGetResponse = api_client.get(path='/app/', params={'uri': uri}).json()
     return DataRecordVersionInfo(
@@ -10,3 +88,10 @@ def get_data_record_state_from_uri(uri) -> 'DataRecordVersionInfo':
         resource_uuid=app_response['app']['public_id'],
         resource_version_uuid=app_response['app_version']['public_id'],
     )
+
+
+def validate_sqlite_v1(schema: SqliteV1DatabaseSchema, sqlite_file: Path):
+    actual_schema = get_actual_schema(sqlite_file)
+    print(schema)
+    print(sqlite_file)
+    verify_schema(specification=schema, actual_schema=actual_schema)

biolib/_internal/types/data_record.py

@@ -1,15 +1,35 @@
-from .typing import Dict, List, Optional, TypedDict
+from .typing import Dict, List, Literal, Optional, TypedDict, Union
+
+
+class SqliteV1ForeignKey(TypedDict):
+    table: str
+    column: str
+
+
+class SqliteV1Column(TypedDict):
+    type: Literal['INTEGER', 'REAL', 'TEXT', 'JSON']  # noqa:F821
+    nullable: Optional[bool]
+    foreign_key: Optional[SqliteV1ForeignKey]
+    json_schema: Optional[Dict]
+
+
+class SqliteV1Table(TypedDict):
+    columns: Dict[str, SqliteV1Column]
+
+
+class SqliteV1DatabaseSchema(TypedDict):
+    tables: Dict[str, SqliteV1Table]
 
 
 class DataRecordValidationRuleDict(TypedDict):
     path: str
     type: str
-    rule: Dict
+    rule: Union[SqliteV1DatabaseSchema]
 
 
 class DataRecordTypeDict(TypedDict):
     name: str
-    validation_rules: List[Dict]
+    validation_rules: List[DataRecordValidationRuleDict]
 
 
 class DataRecordSlimDict(TypedDict):

{pybiolib-1.1.2155.dist-info → pybiolib-1.1.2169.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: pybiolib
-Version: 1.1.2155
+Version: 1.1.2169
 Summary: BioLib Python Client
 Home-page: https://github.com/biolib
 License: MIT

{pybiolib-1.1.2155.dist-info → pybiolib-1.1.2169.dist-info}/RECORD

@@ -1,10 +1,10 @@
 LICENSE,sha256=F2h7gf8i0agDIeWoBPXDMYScvQOz02pAWkKhTGOHaaw,1067
 README.md,sha256=_IH7pxFiqy2bIAmaVeA-iVTyUwWRjMIlfgtUbYTtmls,368
 biolib/__init__.py,sha256=_tThyzISH81yS9KXP_X3qEiKXmsIp5XOBcJIODfLVnc,4338
-biolib/_data_record/data_record.py,sha256=Sud8yXz7yR6YW4V6OqE7nO6I4a0TdqijmMTZwwU59j8,12152
+biolib/_data_record/data_record.py,sha256=XC3BsxnmA20odM7r4dsRHNnwYqhYp054f4BNkA5J5dA,12685
 biolib/_internal/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-biolib/_internal/data_record/__init__.py,sha256=wLOy3Pb7dWYik5eQtQM00DH2AWC-M5RbTjwh9InPiqo,56
-biolib/_internal/data_record/data_record.py,sha256=If4SQj-XwKSPzCpaWA01LEGKalZ6DEjD5PJZRtl3Mao,556
+biolib/_internal/data_record/__init__.py,sha256=fGdME6JGRU_2VxpJbYpGXYndjN-feUkmKY4fuMyq3cg,76
+biolib/_internal/data_record/data_record.py,sha256=iUwLGQEQtXt09iLUMWrE_-Gj3ZkQfXxB7Zcp1WcwsF0,4297
 biolib/_internal/data_record/remote_storage_endpoint.py,sha256=eCptuZ4DMAPnaNCVDvpWXwXGI6Jac9U1N5dqU8Cj95Q,1732
 biolib/_internal/file_utils.py,sha256=4jT6j7bB21c0JNn5BfnyWQib_zt0CVtJ_TiOFOStRcE,2604
 biolib/_internal/fuse_mount/__init__.py,sha256=B_tM6RM2dBw-vbpoHJC4X3tOAaN1H2RDvqYJOw3xFwg,55
@@ -18,7 +18,7 @@ biolib/_internal/push_application.py,sha256=8P7eXvySn7CRp5XBDkO3xjTGixS8g7-jD-_i
 biolib/_internal/runtime.py,sha256=9pZ3s3L7LGxdqOgnHh1KK3Jjyn_9MjhQmKHI-6hMT3U,448
 biolib/_internal/types/__init__.py,sha256=11ZucS8jKeLGAAswXyKI7FH2KLHd6T9Sh8ZK2Ar3jlk,152
 biolib/_internal/types/app.py,sha256=Mz2QGD_jESX-K9JYnLWPo4YA__Q_1FQQTk9pvidCohU,118
-biolib/_internal/types/data_record.py,sha256=U1e9mqEAOOcjR2QzL8eK_xdttkN44yhxvslTyQk1QOo,369
+biolib/_internal/types/data_record.py,sha256=AHoIiwVqeHj0HozQxFRAyxk-d3XJgLWno4ic1z9eTrQ,865
 biolib/_internal/types/experiment.py,sha256=D94iBdn2nS92lRW-TOs1a2WKXJD5ZtmzL4ypggKX2ys,176
 biolib/_internal/types/resource.py,sha256=G-vPkZoe4Um6FPxsQZtRzAlbSW5sDW4NFkbjn21I3V4,372
 biolib/_internal/types/typing.py,sha256=D4EKKEe7kDx0K6lJi-H_XLtk-8w6nu2fdqn9bvzI-Xo,288
@@ -116,8 +116,8 @@ biolib/utils/cache_state.py,sha256=u256F37QSRIVwqKlbnCyzAX4EMI-kl6Dwu6qwj-Qmag,3
 biolib/utils/multipart_uploader.py,sha256=XvGP1I8tQuKhAH-QugPRoEsCi9qvbRk-DVBs5PNwwJo,8452
 biolib/utils/seq_util.py,sha256=jC5WhH63FTD7SLFJbxQGA2hOt9NTwq9zHl_BEec1Z0c,4907
 biolib/utils/zip/remote_zip.py,sha256=0wErYlxir5921agfFeV1xVjf29l9VNgGQvNlWOlj2Yc,23232
-pybiolib-1.1.2155.dist-info/LICENSE,sha256=F2h7gf8i0agDIeWoBPXDMYScvQOz02pAWkKhTGOHaaw,1067
-pybiolib-1.1.2155.dist-info/METADATA,sha256=tu73TYRGUjbJ3MfO15gYq2UO81SYCy508-wwmoHN3WQ,1508
-pybiolib-1.1.2155.dist-info/WHEEL,sha256=sP946D7jFCHeNz5Iq4fL4Lu-PrWrFsgfLXbbkciIZwg,88
-pybiolib-1.1.2155.dist-info/entry_points.txt,sha256=p6DyaP_2kctxegTX23WBznnrDi4mz6gx04O5uKtRDXg,42
-pybiolib-1.1.2155.dist-info/RECORD,,
+pybiolib-1.1.2169.dist-info/LICENSE,sha256=F2h7gf8i0agDIeWoBPXDMYScvQOz02pAWkKhTGOHaaw,1067
+pybiolib-1.1.2169.dist-info/METADATA,sha256=Ah2VUsAv-zV4lYA_0OTbPHmZuRHNBbNUE8hyRY5icYs,1508
+pybiolib-1.1.2169.dist-info/WHEEL,sha256=sP946D7jFCHeNz5Iq4fL4Lu-PrWrFsgfLXbbkciIZwg,88
+pybiolib-1.1.2169.dist-info/entry_points.txt,sha256=p6DyaP_2kctxegTX23WBznnrDi4mz6gx04O5uKtRDXg,42
+pybiolib-1.1.2169.dist-info/RECORD,,