dagster-datacontract 0.3.1__py3-none-any.whl → 0.4.0__py3-none-any.whl

dagster_datacontract/__init__.py

@@ -10,9 +10,11 @@ from loguru import logger
 from dagster_datacontract.description import get_description
 from dagster_datacontract.metadata import (
     get_column_lineage,
+    get_links,
     get_server_information,
     get_table_column,
 )
+from dagster_datacontract.owners import get_owner
 from dagster_datacontract.tags import get_tags
 from dagster_datacontract.utils import normalize_path
 
@@ -35,7 +37,7 @@ class DataContractLoader:
             self.asset_name,
             self.data_contract_specification,
         )
-        self.owner = self._load_owner()
+        self.owner = get_owner(self.data_contract_specification)
         self.version = self._load_version()
         self.cron_schedule = self._load_cron_schedule()
 
@@ -44,7 +46,7 @@ class DataContractLoader:
     ) -> dict[str, TableColumnLineage | TableSchema | Any] | None:
         metadata = (
             {
-                "data contract path": dg.MetadataValue.url(
+                "datacontract/path": dg.MetadataValue.url(
                     normalize_path(self.data_contract._data_contract_file)
                 ),
             }
@@ -75,12 +77,10 @@ class DataContractLoader:
         )
         metadata.update(server_information)
 
-        return metadata
-
-    def _load_owner(self) -> list[str] | None:
-        owner = self.data_contract_specification.info.owner
+        links = get_links(self.data_contract_specification.links)
+        metadata.update(links)
 
-        return [f"team:{owner}"] if owner else None
+        return metadata
 
     def _load_version(self) -> str | None:
         version = self.data_contract_specification.info.version

dagster_datacontract/metadata/__init__.py

@@ -1,7 +1,13 @@
+from dagster_datacontract.metadata.links import get_links
 from dagster_datacontract.metadata.server_information import get_server_information
 from dagster_datacontract.metadata.table_colums import (
     get_column_lineage,
     get_table_column,
 )
 
-__all__ = ["get_table_column", "get_column_lineage", "get_server_information"]
+__all__ = [
+    "get_column_lineage",
+    "get_links",
+    "get_table_column",
+    "get_server_information",
+]

dagster_datacontract/metadata/links.py

@@ -0,0 +1,17 @@
+import dagster as dg
+
+
+def get_links(links: dict[str, str]) -> dict[str, str]:
+    """Return a dictionary with keys prefixed by 'link/' and values as Dagster URL metadata.
+
+    Args:
+        links (dict[str, str]): A dictionary where each key is a name/label and each
+            value is a URL string.
+
+    Returns:
+        dict[str, str]: A dictionary where each key is prefixed with 'link/' and
+            each value is a `MetadataValue.url`.
+    """
+    links = {f"link/{key}": dg.MetadataValue.url(value) for key, value in links.items()}
+
+    return links

dagster_datacontract/metadata/server_information.py

@@ -15,6 +15,8 @@ def get_server_information(
     the specified server by name and constructs a dictionary with keys such as
     "dagster/uri" and "dagster/table_name" depending on the server type.
 
+    Server information can be obtained from: https://datacontract.com/#server-object
+
     Parameters:
         data_contract_specification (DataContractSpecification):
             The data contract specification containing server configurations.
@@ -36,43 +38,71 @@ def get_server_information(
     match server.type:
         case "azure":
             server_information["dagster/uri"] = server.location
+            server_information["azure/storage_account"] = server.storageAccount
+            server_information["file/format"] = server.format
+            server_information["file/delimiter"] = server.delimiter
+        case "bigquery":
+            server_information["bigquery/project"] = server.project
+            server_information["bigquery/dataset"] = server.dataset
         case "databricks":
             server_information["dagster/uri"] = server.host
             server_information["dagster/table_name"] = (
                 f"{server.catalog}.{server.schema}.{asset_name}"
             )
+        case "glue":
+            server_information = {}
         case "kafka":
             server_information["dagster/uri"] = server.host
+            server_information["kafka/topic"] = server.topic
+            server_information["kafka/format"] = server.format
         case "kinesis":
-            server_information = {}
+            server_information["kinesis/stream"] = server.stream
+            server_information["kinesis/region"] = server.region
+            server_information["kinesis/format"] = server.format
         case "local":
             server_information["dagster/uri"] = normalize_path(server.path)
+            server_information["file/format"] = server.format
         case "oracle":
             server_information["dagster/uri"] = f"{server.host}:{server.port}"
+            server_information["oracle/service_name"] = server.serviceName
         case "postgres":
             server_information["dagster/uri"] = f"{server.host}:{server.port}"
             server_information["dagster/table_name"] = (
                 f"{server.database}.{server.schema}.{asset_name}"
             )
         case "pubsub":
-            server_information = {}
+            server_information["pubsub/project"] = server.project
+            server_information["pubsub/topic"] = server.topic
         case "redshift":
             server_information["dagster/uri"] = server.endpoint
             server_information["dagster/table_name"] = (
                 f"{server.database}.{server.schema}.{asset_name}"
             )
+            server_information["redshift/account"] = server.account
+            server_information["redshift/host"] = server.host
+            server_information["redshift/port"] = server.port
+            server_information["redshift/cluster"] = server.clusterIdentifier
         case "s3":
             server_information["dagster/uri"] = server.location
+            server_information["s3/endpoint"] = server.endpointUrl
+            server_information["file/format"] = server.format
+            server_information["file/delimiter"] = server.delimiter
         case "sftp":
             server_information["dagster/uri"] = server.location
+            server_information["file/format"] = server.format
+            server_information["file/delimiter"] = server.delimiter
         case "snowflake":
             server_information["dagster/table_name"] = (
                 f"{server.database}.{server.schema}.{asset_name}"
             )
+            server_information["snowflake/account"] = server.account
         case "sqlserver":
             server_information["dagster/table_name"] = (
                 f"{server.database}.{server.schema}.{asset_name}"
             )
+            server_information["sqlserver/host"] = server.host
+            server_information["sqlserver/port"] = server.port
+            server_information["sqlserver/driver"] = server.driver
         case "trino":
             server_information["dagster/uri"] = f"{server.host}:{server.port}"
             server_information["dagster/table_name"] = (

dagster_datacontract/owners/__init__.py

@@ -0,0 +1,23 @@
+from datacontract.data_contract import DataContractSpecification
+
+
+def get_owner(
+    data_contract_specification: DataContractSpecification,
+    is_team: bool = True,
+) -> list[str] | None:
+    """Return the owner of a data contract, optionally formatted as a team identifier.
+
+    Args:
+        data_contract_specification (DataContractSpecification): The data contract specification containing ownership metadata.
+        is_team (bool, optional): If True, formats the owner as a team identifier (e.g., 'team:owner').
+                                  If False, returns the raw owner string. Defaults to True.
+
+    Returns:
+        list[str] | None: A list containing the owner string, formatted depending on `is_team`, or None if no owner is found.
+    """
+    owner = data_contract_specification.info.owner
+
+    if is_team:
+        return [f"team:{owner}"]
+
+    return [owner]

{dagster_datacontract-0.3.1.dist-info → dagster_datacontract-0.4.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dagster-datacontract
-Version: 0.3.1
+Version: 0.4.0
 Summary: Load metadata and asset check spesifications from data contracts.
 Author-email: Fredrik Bakken <fredrik@dataheim.io>
 Requires-Python: >=3.10.0

dagster_datacontract-0.4.0.dist-info/RECORD

@@ -0,0 +1,16 @@
+dagster_datacontract/__init__.py,sha256=nlLGDivblIOJqE5CUA7wzSb_MwSVO_3YwhO3dyVjwgo,5711
+dagster_datacontract/description/__init__.py,sha256=ulWqPp5jIPvCzaDFZcjLjcDkljJ5j_FRsE0dXhK8Wlc,104
+dagster_datacontract/description/description.py,sha256=FmjgCYDpJ9UHrvAv0sAthfRohDjdG0lL1XcMKK8QMmI,1646
+dagster_datacontract/metadata/__init__.py,sha256=Gj7Htl3rYRXsE-631yr4LWqL7Tf5bZKFEFjTeglxZek,359
+dagster_datacontract/metadata/links.py,sha256=ckSTZLhj6ZETrJMeXDEKVFIoZz871lhL8ZgRtsEIIvs,562
+dagster_datacontract/metadata/server_information.py,sha256=Wd7hPZ8IfeISBe5VThz0pBGkmiPv4Ia2MGNhBJ-aWik,5397
+dagster_datacontract/metadata/table_colums.py,sha256=Q7ZCiMReWU4-T2YfBvtt5vvoVXEoUgzK5OPMxQEgzpQ,4013
+dagster_datacontract/owners/__init__.py,sha256=c0AhLQRzfw-QPmsF9rPXRyE6VoLmgDRRNgVqG8JUvFs,882
+dagster_datacontract/tags/__init__.py,sha256=2Ph-M0WbBKUjJWIzM_cEBW3SQZh7Nq8oy5MbD5bt_lc,76
+dagster_datacontract/tags/tags.py,sha256=aZ_HTkc-vjJ_rofT32fT_zrLCt9x1ZGn8XoihhOMhfU,1414
+dagster_datacontract/utils/__init__.py,sha256=Zfbuf20Eorf7BD8gSDASiPqgPwWu8Mz03r3aa2zE4NA,1106
+dagster_datacontract-0.4.0.dist-info/licenses/LICENSE,sha256=9ULsEM1ICzCaGoso40plwO-d_SCQ7nsU6ZA4xgfaRq8,11338
+dagster_datacontract-0.4.0.dist-info/METADATA,sha256=RRJoWluvktDqQhXBH7dsQOsctgrUQYIgiuo4QAbxYu4,3029
+dagster_datacontract-0.4.0.dist-info/WHEEL,sha256=CmyFI0kx5cdEMTLiONQRbGQwjIoR1aIYB7eCAQ4KPJ0,91
+dagster_datacontract-0.4.0.dist-info/top_level.txt,sha256=_HUQ6OJ50Q0VZxEkdocTtxk1QkJpztb1QY7A0rcvtCE,21
+dagster_datacontract-0.4.0.dist-info/RECORD,,

dagster_datacontract-0.3.1.dist-info/RECORD

@@ -1,14 +0,0 @@
-dagster_datacontract/__init__.py,sha256=Pzq4AblLnDhW0QBmV8ntV5KLD0a7D0-mA2Lw9hjS_V8,5685
-dagster_datacontract/description/__init__.py,sha256=ulWqPp5jIPvCzaDFZcjLjcDkljJ5j_FRsE0dXhK8Wlc,104
-dagster_datacontract/description/description.py,sha256=FmjgCYDpJ9UHrvAv0sAthfRohDjdG0lL1XcMKK8QMmI,1646
-dagster_datacontract/metadata/__init__.py,sha256=e-xmcWWoAhmKTwosshsxnyrjI1j-UyY6YpdpzA2ggF4,269
-dagster_datacontract/metadata/server_information.py,sha256=jk_H8aI5PdGzIeoYThlhhZMSOtBh-6xc8QAFz0BFesU,3512
-dagster_datacontract/metadata/table_colums.py,sha256=Q7ZCiMReWU4-T2YfBvtt5vvoVXEoUgzK5OPMxQEgzpQ,4013
-dagster_datacontract/tags/__init__.py,sha256=2Ph-M0WbBKUjJWIzM_cEBW3SQZh7Nq8oy5MbD5bt_lc,76
-dagster_datacontract/tags/tags.py,sha256=aZ_HTkc-vjJ_rofT32fT_zrLCt9x1ZGn8XoihhOMhfU,1414
-dagster_datacontract/utils/__init__.py,sha256=Zfbuf20Eorf7BD8gSDASiPqgPwWu8Mz03r3aa2zE4NA,1106
-dagster_datacontract-0.3.1.dist-info/licenses/LICENSE,sha256=9ULsEM1ICzCaGoso40plwO-d_SCQ7nsU6ZA4xgfaRq8,11338
-dagster_datacontract-0.3.1.dist-info/METADATA,sha256=P98oLsqB-GCJz4uanki-xtG3UnZCHD0biDMUYIZMEhc,3029
-dagster_datacontract-0.3.1.dist-info/WHEEL,sha256=CmyFI0kx5cdEMTLiONQRbGQwjIoR1aIYB7eCAQ4KPJ0,91
-dagster_datacontract-0.3.1.dist-info/top_level.txt,sha256=_HUQ6OJ50Q0VZxEkdocTtxk1QkJpztb1QY7A0rcvtCE,21
-dagster_datacontract-0.3.1.dist-info/RECORD,,