acryl-datahub 0.15.0rc9__py3-none-any.whl → 0.15.0rc11__py3-none-any.whl

{acryl_datahub-0.15.0rc9.dist-info → acryl_datahub-0.15.0rc11.dist-info}/RECORD

@@ -1,4 +1,4 @@
-datahub/__init__.py,sha256=-XVob0PhZLue6g_4duveVmhOndcfh_8ifCuVB_YBxkk,574
+datahub/__init__.py,sha256=RtQFMiAfUcKAAt_1AITPz1UOKRx2gIW0yLbLYUEChqU,575
 datahub/__main__.py,sha256=pegIvQ9hzK7IhqVeUi1MeADSZ2QlP-D3K0OQdEg55RU,106
 datahub/entrypoints.py,sha256=3-qSfXAx3Z0FEkBV5tlO8fQr4xk4ySeDRMVTpS5Xd6A,7793
 datahub/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -61,7 +61,7 @@ datahub/cli/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 datahub/cli/check_cli.py,sha256=9dXNyzZayHeoFjwFjLkMVyx6DiCZfeESyI-sYtGA6bE,12850
 datahub/cli/cli_utils.py,sha256=xMEK4tmoX2f_5BihxWrApOkayVLwhpPvMUU842x6FsI,13111
 datahub/cli/config_utils.py,sha256=yuXw7RzpRY5x_-MAoqWbv46qUkIeRNAJL4_OeJpYdBE,4879
-datahub/cli/delete_cli.py,sha256=YEbPDFcMtbIHDE0yNB9opfVI5r8f7kFOnfyc0rlyIqU,20906
+datahub/cli/delete_cli.py,sha256=Z7iXNr4ZMmghCnldU8laK4SwTNrhQEEnnUH_TeaBKog,21838
 datahub/cli/docker_check.py,sha256=rED4wHXqxcQ_qNFyIgFEZ85BHT9ZTE5YC-oUKqbRqi0,9432
 datahub/cli/docker_cli.py,sha256=QGoWFp8ZZsXOSMbgu0Q4snMmMmtP3epWAN-fYglUNEc,36491
 datahub/cli/env_utils.py,sha256=RQzjg4JE29hjPt4v7p-RuqoOr99w8E3DBHWiN2Sm7T4,252
@@ -97,7 +97,7 @@ datahub/configuration/git.py,sha256=s55eUHxKqVZgtVsISaDyS-1F4iZBiybbjYsjbp5LU5o,
 datahub/configuration/import_resolver.py,sha256=b4Ie9L7knN1LALEVMxTcNFSklDD6CVE-4Ipy4ZYhNYA,369
 datahub/configuration/json_loader.py,sha256=vIDnjwXWi9yHDO8KW64EupOzOb_sspehGCD7xGHzg84,302
 datahub/configuration/kafka.py,sha256=MlIwpd5FFyOyjdDXW_X9JTLNk7f988sPMgevkcZYVgI,2579
-datahub/configuration/kafka_consumer_config.py,sha256=14UWK6kTAnKTgHM43aeWjg67fddACsSLzhO6wgp6cm4,1175
+datahub/configuration/kafka_consumer_config.py,sha256=LivsObTt9yC3WoGnslJbF_x4ojfNdxMIMEhb8vvJfcA,2133
 datahub/configuration/pattern_utils.py,sha256=Q5IB9RfWOOo5FvRVBU7XkhiwHCxSQ1NTMfUlWtWI9qc,699
 datahub/configuration/pydantic_migration_helpers.py,sha256=4C_COAVZ5iJ8yxcWNgXZNWsY7ULogICNZ368oNF7zWg,1462
 datahub/configuration/source_common.py,sha256=68LZOuB23zSEcfgQJE1wZQnyYQHVVnEZK3Sniv_nEQs,2107
@@ -201,7 +201,7 @@ datahub/ingestion/source/nifi.py,sha256=ttsjZ9aRUvINmewvKFIQD8Rwa4jcl35WFG-F-jPG
 datahub/ingestion/source/openapi.py,sha256=3ea2ORz1cuq4e7L2hSjxG9Cw3__pVoJ5UNYTJS3EnKU,17386
 datahub/ingestion/source/openapi_parser.py,sha256=1_68wHWe_SzWYEyC1YVDw9vxoadKjW1yv8DecvyIhwY,13606
 datahub/ingestion/source/preset.py,sha256=eq7h1qKs8nfSBVot1ofN-YgZhw_rzq8DG4cKOGfDHko,3948
-datahub/ingestion/source/pulsar.py,sha256=bKtBzBTPuKwdTN4nSs7M0FYfA-q6FDc554W2NhnuWEY,19475
+datahub/ingestion/source/pulsar.py,sha256=H8XJC7xIX8Kdkd7006PxllAGVO_Pjza8Xx9VUBOvpPc,19827
 datahub/ingestion/source/redash.py,sha256=E-a14X19zppPun7_-S-pZ2lRiw1-68QiT-jL7bDzG10,32057
 datahub/ingestion/source/salesforce.py,sha256=S6LSM6mzl8-zKbrJPoINhM1SCpYfM244Xb74pbEI-J0,31792
 datahub/ingestion/source/source_registry.py,sha256=a2mLjJPLkSI-gYCTb_7U7Jo4D8jGknNQ_yScPIihXFk,1208
@@ -301,7 +301,7 @@ datahub/ingestion/source/fivetran/fivetran_log_api.py,sha256=EAak3hJpe75WZSgz6wP
 datahub/ingestion/source/fivetran/fivetran_query.py,sha256=vLrTj7e-0NxZ2U4bWTB57pih42WirqPlUvwtIRfStlQ,5275
 datahub/ingestion/source/gc/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 datahub/ingestion/source/gc/datahub_gc.py,sha256=f6Erj3KfD0Hx3ydwL5MUVCZgFzS9c6U2Pkr54JLIUOA,12394
-datahub/ingestion/source/gc/dataprocess_cleanup.py,sha256=rGIwUKPlNu2XR0YT8DuJGg2pPGIr3MM-YDa5Slo2vNY,14470
+datahub/ingestion/source/gc/dataprocess_cleanup.py,sha256=GOdLVYYfmmHV7Xtr-cwCTiQiRSVwq80Nu0EmARyoEeo,15323
 datahub/ingestion/source/gc/execution_request_cleanup.py,sha256=cHJmxz4NmA7VjTX2iGEo3wZ_SDrjC_rCQcnRxKgfUVI,8713
 datahub/ingestion/source/gc/soft_deleted_entity_cleanup.py,sha256=_tms5AqNAJRDRzQmyN_VydzXbdME2lkvTwa5u1La5z8,7353
 datahub/ingestion/source/gcs/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -331,7 +331,7 @@ datahub/ingestion/source/looker/looker_constant.py,sha256=GMKYtNXlpojPxa9azridKf
 datahub/ingestion/source/looker/looker_dataclasses.py,sha256=ULWLFWsV2cKmTuOFavD8QjEBmnXmvjyr8RbUB62DwJQ,12178
 datahub/ingestion/source/looker/looker_file_loader.py,sha256=c1ewDrIb9VJg1o-asbwX9gL83kgL01vIETzzbmZIhmw,4267
 datahub/ingestion/source/looker/looker_lib_wrapper.py,sha256=QTTCW-rPNUoazQG_sTJbCARXJzQ7NKS-XKURp2AAWls,11106
-datahub/ingestion/source/looker/looker_liquid_tag.py,sha256=GxK-NkP0Evzv_se3l5f-z-HZZNQRjYr9py_h0T6ReHY,2902
+datahub/ingestion/source/looker/looker_liquid_tag.py,sha256=mO4G4MNA4YZFvZaDBpdiJ2vP3irC82kY34RdaK4Pbfs,3100
 datahub/ingestion/source/looker/looker_query_model.py,sha256=N0jBbFruiCIIGT6sJn6tNeppeQ78KGTkOwTLirhxFNc,2144
 datahub/ingestion/source/looker/looker_source.py,sha256=AByQxWVfOBqOtZPaR_cw9SB-tFZtfppiKRkFSbcK1GA,65346
 datahub/ingestion/source/looker/looker_template_language.py,sha256=EG4ZfVZ0x53lgaYh2ohzL4ZCy9KsX0TA51XqCmsCd2Q,14328
@@ -485,7 +485,7 @@ datahub/ingestion/source/state_provider/datahub_ingestion_checkpointing_provider
 datahub/ingestion/source/state_provider/file_ingestion_checkpointing_provider.py,sha256=xsH7Ao_05VTjqpkzLkhdf5B1ULMzFoD8vkJJIJU9w-U,4077
 datahub/ingestion/source/state_provider/state_provider_registry.py,sha256=SVq4mIyGNmLXE9OZx1taOiNPqDoQp03-Ot9rYnB5F3k,401
 datahub/ingestion/source/tableau/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-datahub/ingestion/source/tableau/tableau.py,sha256=LgsVZPMRiJiyyB9-ljj53T8WGUyBSiQwByW_1TSuxC4,130417
+datahub/ingestion/source/tableau/tableau.py,sha256=AFlDng8EfvBvZL692hMf_sfzGwpHpUU6FW_ElR4uitQ,131551
 datahub/ingestion/source/tableau/tableau_common.py,sha256=Dy_2pvkPucZJsG_LvQZLlxNEkjh-yOXHlZ4jurq9opM,26069
 datahub/ingestion/source/tableau/tableau_constant.py,sha256=nWElhtDo5kj5mWivZFmtVF_4Ugw0-EatBYWyDVzu5hE,2501
 datahub/ingestion/source/unity/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -974,8 +974,8 @@ datahub_provider/operators/datahub_assertion_operator.py,sha256=uvTQ-jk2F0sbqqxp
 datahub_provider/operators/datahub_assertion_sensor.py,sha256=lCBj_3x1cf5GMNpHdfkpHuyHfVxsm6ff5x2Z5iizcAo,140
 datahub_provider/operators/datahub_operation_operator.py,sha256=aevDp2FzX7FxGlXrR0khoHNbxbhKR2qPEX5e8O2Jyzw,174
 datahub_provider/operators/datahub_operation_sensor.py,sha256=8fcdVBCEPgqy1etTXgLoiHoJrRt_nzFZQMdSzHqSG7M,168
-acryl_datahub-0.15.0rc9.dist-info/METADATA,sha256=lTHIKVVIg5N9xzphcWZ_fhOb0yL1MEx4lRidtr2GvJU,172484
-acryl_datahub-0.15.0rc9.dist-info/WHEEL,sha256=PZUExdf71Ui_so67QXpySuHtCi3-J3wvF4ORK6k_S8U,91
-acryl_datahub-0.15.0rc9.dist-info/entry_points.txt,sha256=Yj0PWB0LQOq4Rj2fyR6ETx4BUGw4TOcNL0ZNoAZ9kQg,9504
-acryl_datahub-0.15.0rc9.dist-info/top_level.txt,sha256=iLjSrLK5ox1YVYcglRUkcvfZPvKlobBWx7CTUXx8_GI,25
-acryl_datahub-0.15.0rc9.dist-info/RECORD,,
+acryl_datahub-0.15.0rc11.dist-info/METADATA,sha256=0GnLLvLM_Fvdw4XNusJLRrDZ0vemeu_B0ftz8XQXmbs,174408
+acryl_datahub-0.15.0rc11.dist-info/WHEEL,sha256=PZUExdf71Ui_so67QXpySuHtCi3-J3wvF4ORK6k_S8U,91
+acryl_datahub-0.15.0rc11.dist-info/entry_points.txt,sha256=Yj0PWB0LQOq4Rj2fyR6ETx4BUGw4TOcNL0ZNoAZ9kQg,9504
+acryl_datahub-0.15.0rc11.dist-info/top_level.txt,sha256=iLjSrLK5ox1YVYcglRUkcvfZPvKlobBWx7CTUXx8_GI,25
+acryl_datahub-0.15.0rc11.dist-info/RECORD,,

datahub/__init__.py

@@ -3,7 +3,7 @@ import warnings
 
 # Published at https://pypi.org/project/acryl-datahub/.
 __package_name__ = "acryl-datahub"
-__version__ = "0.15.0rc9"
+__version__ = "0.15.0rc11"
 
 
 def is_dev_mode() -> bool:

datahub/cli/delete_cli.py

@@ -214,14 +214,47 @@ def references(urn: str, dry_run: bool, force: bool) -> None:
 
 
 @delete.command()
-@click.option("--urn", required=True, type=str, help="the urn of the entity")
-def undo_by_filter(urn: str) -> None:
+@click.option("--urn", required=False, type=str, help="the urn of the entity")
+@click.option(
+    "-p",
+    "--platform",
+    required=False,
+    type=str,
+    help="Platform filter (e.g. snowflake)",
+)
+@click.option(
+    "-b",
+    "--batch-size",
+    required=False,
+    default=3000,
+    type=int,
+    help="Batch size when querying for entities to un-soft delete."
+    "Maximum 10000. Large batch sizes may cause timeouts.",
+)
+def undo_by_filter(
+    urn: Optional[str], platform: Optional[str], batch_size: int
+) -> None:
     """
-    Undo a soft deletion of an entity
+    Undo soft deletion by filters
     """
     graph = get_default_graph()
     logger.info(f"Using {graph}")
-    graph.set_soft_delete_status(urn=urn, delete=False)
+    if urn:
+        graph.set_soft_delete_status(urn=urn, delete=False)
+    else:
+        urns = list(
+            graph.get_urns_by_filter(
+                platform=platform,
+                query="*",
+                status=RemovedStatusFilter.ONLY_SOFT_DELETED,
+                batch_size=batch_size,
+            )
+        )
+        logger.info(f"Going to un-soft delete {len(urns)} urns")
+        urns_iter = progressbar.progressbar(urns, redirect_stdout=True)
+        for urn in urns_iter:
+            assert urn
+            graph.set_soft_delete_status(urn=urn, delete=False)
 
 
 @delete.command(no_args_is_help=True)

datahub/configuration/kafka_consumer_config.py

@@ -1,3 +1,4 @@
+import inspect
 import logging
 from typing import Any, Dict, Optional
 
@@ -34,5 +35,34 @@ class CallableConsumerConfig:
             "oauth_cb must be a string representing python function reference "
             "in the format <python-module>:<function-name>."
         )
+
+        call_back_fn = import_path(call_back)
+        self._validate_call_back_fn_signature(call_back_fn)
+
         # Set the callback
-        self._config[CallableConsumerConfig.CALLBACK_ATTRIBUTE] = import_path(call_back)
+        self._config[CallableConsumerConfig.CALLBACK_ATTRIBUTE] = call_back_fn
+
+    def _validate_call_back_fn_signature(self, call_back_fn: Any) -> None:
+        sig = inspect.signature(call_back_fn)
+
+        num_positional_args = len(
+            [
+                param
+                for param in sig.parameters.values()
+                if param.kind
+                in (
+                    inspect.Parameter.POSITIONAL_ONLY,
+                    inspect.Parameter.POSITIONAL_OR_KEYWORD,
+                )
+                and param.default == inspect.Parameter.empty
+            ]
+        )
+
+        has_variadic_args = any(
+            param.kind == inspect.Parameter.VAR_POSITIONAL
+            for param in sig.parameters.values()
+        )
+
+        assert num_positional_args == 1 or (
+            has_variadic_args and num_positional_args <= 1
+        ), "oauth_cb function must accept single positional argument."

datahub/ingestion/source/gc/dataprocess_cleanup.py

@@ -208,22 +208,28 @@ class DataProcessCleanup:
         dpis = []
         start = 0
         while True:
-            job_query_result = self.ctx.graph.execute_graphql(
-                DATA_PROCESS_INSTANCES_QUERY,
-                {"dataJobUrn": job_urn, "start": start, "count": batch_size},
-            )
-            job_data = job_query_result.get("dataJob")
-            if not job_data:
-                raise ValueError(f"Error getting job {job_urn}")
-
-            runs_data = job_data.get("runs")
-            if not runs_data:
-                raise ValueError(f"Error getting runs for {job_urn}")
-
-            runs = runs_data.get("runs")
-            dpis.extend(runs)
-            start += batch_size
-            if len(runs) < batch_size:
+            try:
+                job_query_result = self.ctx.graph.execute_graphql(
+                    DATA_PROCESS_INSTANCES_QUERY,
+                    {"dataJobUrn": job_urn, "start": start, "count": batch_size},
+                )
+                job_data = job_query_result.get("dataJob")
+                if not job_data:
+                    logger.error(f"Error getting job {job_urn}")
+                    break
+
+                runs_data = job_data.get("runs")
+                if not runs_data:
+                    logger.error(f"Error getting runs for {job_urn}")
+                    break
+
+                runs = runs_data.get("runs")
+                dpis.extend(runs)
+                start += batch_size
+                if len(runs) < batch_size:
+                    break
+            except Exception as e:
+                logger.error(f"Exception while fetching DPIs for job {job_urn}: {e}")
                 break
         return dpis
 
@@ -243,8 +249,12 @@ class DataProcessCleanup:
                 futures[future] = dpi
 
             for future in as_completed(futures):
-                deleted_count_last_n += 1
-                futures[future]["deleted"] = True
+                try:
+                    future.result()
+                    deleted_count_last_n += 1
+                    futures[future]["deleted"] = True
+                except Exception as e:
+                    logger.error(f"Exception while deleting DPI: {e}")
 
             if deleted_count_last_n % self.config.batch_size == 0:
                 logger.info(f"Deleted {deleted_count_last_n} DPIs from {job.urn}")
@@ -279,7 +289,7 @@ class DataProcessCleanup:
         dpis = self.fetch_dpis(job.urn, self.config.batch_size)
         dpis.sort(
             key=lambda x: x["created"]["time"]
-            if x["created"] and x["created"]["time"]
+            if "created" in x and "time" in x["created"]
             else 0,
             reverse=True,
         )
@@ -314,15 +324,23 @@ class DataProcessCleanup:
             if dpi.get("deleted"):
                 continue
 
-            if dpi["created"]["time"] < retention_time * 1000:
+            if (
+                "created" not in dpi
+                or "time" not in dpi["created"]
+                or dpi["created"]["time"] < retention_time * 1000
+            ):
                 future = executor.submit(
                     self.delete_entity, dpi["urn"], "dataprocessInstance"
                 )
                 futures[future] = dpi
 
         for future in as_completed(futures):
-            deleted_count_retention += 1
-            futures[future]["deleted"] = True
+            try:
+                future.result()
+                deleted_count_retention += 1
+                futures[future]["deleted"] = True
+            except Exception as e:
+                logger.error(f"Exception while deleting DPI: {e}")
 
             if deleted_count_retention % self.config.batch_size == 0:
                 logger.info(
@@ -378,8 +396,11 @@ class DataProcessCleanup:
             dataFlows[flow.urn] = flow
 
         scroll_id: Optional[str] = None
+        previous_scroll_id: Optional[str] = None
+
         dataJobs: Dict[str, List[DataJobEntity]] = defaultdict(list)
         deleted_jobs: int = 0
+
         while True:
             result = self.ctx.graph.execute_graphql(
                 DATAJOB_QUERY,
@@ -426,9 +447,11 @@ class DataProcessCleanup:
                 else:
                     dataJobs[datajob_entity.flow_urn].append(datajob_entity)
 
-            if not scroll_id:
+            if not scroll_id or previous_scroll_id == scroll_id:
                 break
 
+            previous_scroll_id = scroll_id
+
         logger.info(f"Deleted {deleted_jobs} DataJobs")
         # Delete empty dataflows if needed
         if self.config.delete_empty_data_flows:
@@ -443,4 +466,5 @@ class DataProcessCleanup:
                     if deleted_jobs % self.config.batch_size == 0:
                         logger.info(f"Deleted {deleted_data_flows} DataFlows")
             logger.info(f"Deleted {deleted_data_flows} DataFlows")
+
         return []

datahub/ingestion/source/looker/looker_liquid_tag.py

@@ -4,6 +4,7 @@ from typing import ClassVar, Optional, TextIO
 from liquid import Environment
 from liquid.ast import Node
 from liquid.context import Context
+from liquid.filter import string_filter
 from liquid.parse import expect, get_parser
 from liquid.stream import TokenStream
 from liquid.tag import Tag
@@ -81,12 +82,18 @@ class ConditionTag(Tag):
 custom_tags = [ConditionTag]
 
 
+@string_filter
+def sql_quote_filter(variable: str) -> str:
+    return f"'{variable}'"
+
+
 @lru_cache(maxsize=1)
 def _create_env() -> Environment:
-    env: Environment = Environment()
+    env: Environment = Environment(strict_filters=False)
     # register tag. One time activity
     for custom_tag in custom_tags:
         env.add_tag(custom_tag)
+    env.add_filter("sql_quote", sql_quote_filter)
     return env
 
 

datahub/ingestion/source/pulsar.py

@@ -78,7 +78,17 @@ class PulsarSchema:
     def __init__(self, schema):
         self.schema_version = schema.get("version")
 
-        avro_schema = json.loads(schema.get("data"))
+        schema_data = schema.get("data")
+        if not schema_data:
+            logger.warning("Schema data is empty or None. Using default empty schema.")
+            schema_data = "{}"
+
+        try:
+            avro_schema = json.loads(schema_data)
+        except json.JSONDecodeError as e:
+            logger.error(f"Invalid JSON schema: {schema_data}. Error: {str(e)}")
+            avro_schema = {}
+
         self.schema_name = avro_schema.get("namespace") + "." + avro_schema.get("name")
         self.schema_description = avro_schema.get("doc")
         self.schema_type = schema.get("type")

datahub/ingestion/source/tableau/tableau.py

@@ -68,6 +68,7 @@ from datahub.ingestion.api.source import (
     CapabilityReport,
     MetadataWorkUnitProcessor,
     Source,
+    StructuredLogLevel,
     TestableSource,
     TestConnectionReport,
 )
@@ -289,16 +290,12 @@ class TableauConnectionConfig(ConfigModel):
             server.auth.sign_in(authentication)
             return server
         except ServerResponseError as e:
+            message = f"Unable to login (invalid/expired credentials or missing permissions): {str(e)}"
             if isinstance(authentication, PersonalAccessTokenAuth):
                 # Docs on token expiry in Tableau:
                 # https://help.tableau.com/current/server/en-us/security_personal_access_tokens.htm#token-expiry
-                logger.info(
-                    "Error authenticating with Tableau. Note that Tableau personal access tokens "
-                    "expire if not used for 15 days or if over 1 year old"
-                )
-            raise ValueError(
-                f"Unable to login (invalid/expired credentials or missing permissions): {str(e)}"
-            ) from e
+                message = f"Error authenticating with Tableau. Note that Tableau personal access tokens expire if not used for 15 days or if over 1 year old: {str(e)}"
+            raise ValueError(message) from e
         except Exception as e:
             raise ValueError(
                 f"Unable to login (check your Tableau connection and credentials): {str(e)}"
@@ -700,6 +697,7 @@ class TableauSource(StatefulIngestionSourceBase, TestableSource):
                         config=self.config,
                         ctx=self.ctx,
                         site=site,
+                        site_id=site.id,
                         report=self.report,
                         server=self.server,
                         platform=self.platform,
@@ -707,11 +705,19 @@ class TableauSource(StatefulIngestionSourceBase, TestableSource):
                     logger.info(f"Ingesting assets of site '{site.content_url}'.")
                     yield from site_source.ingest_tableau_site()
             else:
-                site = self.server.sites.get_by_id(self.server.site_id)
+                site = None
+                with self.report.report_exc(
+                    title="Unable to fetch site details. Site hierarchy may be incomplete and external urls may be missing.",
+                    message="This usually indicates missing permissions. Ensure that you have all necessary permissions.",
+                    level=StructuredLogLevel.WARN,
+                ):
+                    site = self.server.sites.get_by_id(self.server.site_id)
+
                 site_source = TableauSiteSource(
                     config=self.config,
                     ctx=self.ctx,
                     site=site,
+                    site_id=self.server.site_id,
                     report=self.report,
                     server=self.server,
                     platform=self.platform,
@@ -722,6 +728,7 @@ class TableauSource(StatefulIngestionSourceBase, TestableSource):
                 title="Failed to Retrieve Tableau Metadata",
                 message="Unable to retrieve metadata from tableau.",
                 context=str(md_exception),
+                exc=md_exception,
             )
 
     def close(self) -> None:
@@ -743,7 +750,8 @@ class TableauSiteSource:
         self,
         config: TableauConfig,
         ctx: PipelineContext,
-        site: SiteItem,
+        site: Optional[SiteItem],
+        site_id: Optional[str],
         report: TableauSourceReport,
         server: Server,
         platform: str,
@@ -752,9 +760,16 @@ class TableauSiteSource:
         self.report = report
         self.server: Server = server
         self.ctx: PipelineContext = ctx
-        self.site: SiteItem = site
         self.platform = platform
 
+        self.site: Optional[SiteItem] = site
+        if site_id is not None:
+            self.site_id: str = site_id
+        else:
+            assert self.site is not None, "site or site_id is required"
+            assert self.site.id is not None, "site_id is required when site is provided"
+            self.site_id = self.site.id
+
         self.database_tables: Dict[str, DatabaseTable] = {}
         self.tableau_stat_registry: Dict[str, UsageStat] = {}
         self.tableau_project_registry: Dict[str, TableauProject] = {}
@@ -808,7 +823,7 @@ class TableauSiteSource:
     def _re_authenticate(self):
         tableau_auth: Union[
             TableauAuth, PersonalAccessTokenAuth
-        ] = self.config.get_tableau_auth(self.site.content_url)
+        ] = self.config.get_tableau_auth(self.site_id)
         self.server.auth.sign_in(tableau_auth)
 
     @property
@@ -826,6 +841,7 @@ class TableauSiteSource:
             if not view.id:
                 continue
             self.tableau_stat_registry[view.id] = UsageStat(view_count=view.total_views)
+        logger.info(f"Got Tableau stats for {len(self.tableau_stat_registry)} assets")
         logger.debug("Tableau stats %s", self.tableau_stat_registry)
 
     def _populate_database_server_hostname_map(self) -> None:
@@ -876,7 +892,7 @@ class TableauSiteSource:
                 ancestors = [cur_proj.name]
                 while cur_proj.parent_id is not None:
                     if cur_proj.parent_id not in all_project_map:
-                        self.report.report_warning(
+                        self.report.warning(
                             "project-issue",
                             f"Parent project {cur_proj.parent_id} not found. We need Site Administrator Explorer permissions.",
                         )
@@ -974,8 +990,11 @@ class TableauSiteSource:
                 self.datasource_project_map[ds.id] = ds.project_id
         except Exception as e:
             self.report.get_all_datasources_query_failed = True
-            logger.info(f"Get all datasources query failed due to error {e}")
-            logger.debug("Error stack trace", exc_info=True)
+            self.report.warning(
+                title="Unexpected Query Error",
+                message="Get all datasources query failed due to error",
+                exc=e,
+            )
 
     def _init_workbook_registry(self) -> None:
         if self.server is None:
@@ -1141,7 +1160,6 @@ class TableauSiteSource:
                     )
 
                 if node_limit_errors:
-                    logger.debug(f"Node Limit Error. query_data {query_data}")
                     self.report.warning(
                         title="Tableau Data Exceed Predefined Limit",
                         message="The numbers of record in result set exceeds a predefined limit. Increase the tableau "
@@ -1257,9 +1275,10 @@ class TableauSiteSource:
                     wrk_id: Optional[str] = workbook.get(c.ID)
                     prj_name: Optional[str] = workbook.get(c.PROJECT_NAME)
 
-                    logger.debug(
-                        f"Skipping workbook {wrk_name}({wrk_id}) as it is project {prj_name}({project_luid}) not "
-                        f"present in project registry"
+                    self.report.warning(
+                        title="Skipping Missing Workbook",
+                        message="Skipping workbook as its project is not present in project registry",
+                        context=f"workbook={wrk_name}({wrk_id}), project={prj_name}({project_luid})",
                     )
                     continue
 
@@ -1453,7 +1472,7 @@ class TableauSiteSource:
                 c.COLUMNS_CONNECTION
             ].get("totalCount")
             if not is_custom_sql and not num_tbl_cols:
-                logger.debug(
+                logger.warning(
                     f"Skipping upstream table with id {table[c.ID]}, no columns: {table}"
                 )
                 continue
@@ -1469,7 +1488,12 @@ class TableauSiteSource:
                     table, default_schema_map=self.config.default_schema_map
                 )
             except Exception as e:
-                logger.info(f"Failed to generate upstream reference for {table}: {e}")
+                self.report.warning(
+                    title="Potentially Missing Lineage Issue",
+                    message="Failed to generate upstream reference",
+                    exc=e,
+                    context=f"table={table}",
+                )
                 continue
 
             table_urn = ref.make_dataset_urn(
@@ -1917,10 +1941,12 @@ class TableauSiteSource:
                 self.datasource_project_map[ds_result.id] = ds_result.project_id
         except Exception as e:
             self.report.num_get_datasource_query_failures += 1
-            logger.warning(
-                f"Failed to get datasource project_luid for {ds_luid} due to error {e}"
+            self.report.warning(
+                title="Unexpected Query Error",
+                message="Failed to get datasource details",
+                exc=e,
+                context=f"ds_luid={ds_luid}",
             )
-            logger.debug("Error stack trace", exc_info=True)
 
     def _get_workbook_project_luid(self, wb: dict) -> Optional[str]:
         if wb.get(c.LUID) and self.workbook_project_map.get(wb[c.LUID]):
@@ -3181,10 +3207,10 @@ class TableauSiteSource:
             else:
                 # This is a root Tableau project since the parent_project_id is None.
                 # For a root project, either the site is the parent, or the platform is the default parent.
-                if self.config.add_site_container and self.site and self.site.id:
+                if self.config.add_site_container:
                     # The site containers have already been generated by emit_site_container, so we
                     # don't need to emit them again here.
-                    parent_project_key = self.gen_site_key(self.site.id)
+                    parent_project_key = self.gen_site_key(self.site_id)
 
             yield from gen_containers(
                 container_key=project_key,
@@ -3201,12 +3227,12 @@ class TableauSiteSource:
             yield from emit_project_in_topological_order(project)
 
     def emit_site_container(self):
-        if not self.site or not self.site.id:
+        if not self.site:
             logger.warning("Can not ingest site container. No site information found.")
             return
 
         yield from gen_containers(
-            container_key=self.gen_site_key(self.site.id),
+            container_key=self.gen_site_key(self.site_id),
             name=self.site.name or "Default",
             sub_types=[c.SITE],
         )