Support v2019-12-12 for blob, file share, file datalake (#37)

* add new version

* upgrade version

* upgrade version

* refine

Co-authored-by: Ubuntu <zunli@zuhvm.etyrgwjlsqfeplvzbzef2qjagg.cbnx.internal.cloudapp.net>

Author: Juliehzl

README.rst

@@ -17,6 +17,15 @@ Handles multi-API versions of Azure Storage Data Plane originally from https://g
 
 Change Log
 ----------
+0.4.1
++++++
+* Add tags support for blob
+* Add new api support for azure-multiapi-storagev2:
+    - filedatalake
+        - v2019-12-12
+    - fileshare
+        - v2019-12-12
+
 0.4.0
 +++++
 * Add v2019-12-12 for azure.multiapi.storagev2.blob

azure/multiapi/storagev2/blob/v2019_12_12/__init__.py

@@ -52,11 +52,12 @@
     CustomerProvidedEncryptionKey,
     ContainerEncryptionScope,
     BlobQueryError,
-    DelimitedJSON,
+    DelimitedJsonDialect,
     DelimitedTextDialect,
     ObjectReplicationPolicy,
     ObjectReplicationRule
 )
+from ._list_blobs_helper import BlobPrefix
 
 __version__ = VERSION
 
@@ -195,6 +196,7 @@ def download_blob_from_url(
     'CorsRule',
     'ContainerProperties',
     'BlobProperties',
+    'BlobPrefix',
     'FilteredBlob',
     'LeaseProperties',
     'ContentSettings',
@@ -215,7 +217,7 @@ def download_blob_from_url(
     'PartialBatchErrorException',
     'ContainerEncryptionScope',
     'BlobQueryError',
-    'DelimitedJSON',
+    'DelimitedJsonDialect',
     'DelimitedTextDialect',
     'BlobQueryReader',
     'ObjectReplicationPolicy',

azure/multiapi/storagev2/blob/v2019_12_12/_blob_client.py

@@ -39,8 +39,8 @@
     from ._shared.models import UserDelegationKey
     from ._lease import BlobLeaseClient
     from ._models import (
-        BlobProperties,
         ContainerProperties,
+        BlobProperties,
         PublicAccess,
         BlobAnalyticsLogging,
         Metrics,

azure/multiapi/storagev2/blob/v2019_12_12/_blob_service_client.py

@@ -40,9 +40,8 @@
 from ._models import ( # pylint: disable=unused-import
     ContainerProperties,
     BlobProperties,
-    BlobPropertiesPaged,
-    BlobType,
-    BlobPrefix)
+    BlobType)
+from ._list_blobs_helper import BlobPrefix, BlobPropertiesPaged
 from ._lease import BlobLeaseClient, get_access_conditions
 from ._blob_client import BlobClient
 
@@ -783,6 +782,12 @@ def upload_blob(
             and act according to the condition specified by the `match_condition` parameter.
         :keyword ~azure.core.MatchConditions match_condition:
             The match condition to use upon the etag.
+        :keyword str if_tags_match_condition
+            Specify a SQL where clause on blob tags to operate only on blob with a matching value.
+            eg. "\"tagname\"='my tag'"
+
+            .. versionadded:: 12.4.0
+
         :keyword int timeout:
             The timeout parameter is expressed in seconds. This method may make
             multiple calls to the Azure service and the timeout will apply to
@@ -893,6 +898,12 @@ def delete_blob(
             and act according to the condition specified by the `match_condition` parameter.
         :keyword ~azure.core.MatchConditions match_condition:
             The match condition to use upon the etag.
+        :keyword str if_tags_match_condition
+            Specify a SQL where clause on blob tags to operate only on blob with a matching value.
+            eg. "\"tagname\"='my tag'"
+
+            .. versionadded:: 12.4.0
+
         :keyword int timeout:
             The timeout parameter is expressed in seconds.
         :rtype: None
@@ -952,6 +963,12 @@ def download_blob(self, blob, offset=None, length=None, **kwargs):
             and act according to the condition specified by the `match_condition` parameter.
         :keyword ~azure.core.MatchConditions match_condition:
             The match condition to use upon the etag.
+        :keyword str if_tags_match_condition
+            Specify a SQL where clause on blob tags to operate only on blob with a matching value.
+            eg. "\"tagname\"='my tag'"
+
+            .. versionadded:: 12.4.0
+
         :keyword ~azure.storage.blob.CustomerProvidedEncryptionKey cpk:
             Encrypts the data on the service-side with the given key.
             Use of customer-provided keys must be done over HTTPS.
@@ -998,6 +1015,9 @@ def _generate_delete_blobs_subrequest_options(
         if_none_match = None
         if modified_access_conditions is not None:
             if_none_match = modified_access_conditions.if_none_match
+        if_tags = None
+        if modified_access_conditions is not None:
+            if_tags = modified_access_conditions.if_tags
 
         # Construct parameters
         timeout = kwargs.pop('timeout', None)
@@ -1027,6 +1047,8 @@ def _generate_delete_blobs_subrequest_options(
         if if_none_match is not None:
             header_parameters['If-None-Match'] = self._client._serialize.header(  # pylint: disable=protected-access
                 "if_none_match", if_none_match, 'str')
+        if if_tags is not None:
+            header_parameters['x-ms-if-tags'] = self._client._serialize.header("if_tags", if_tags, 'str')  # pylint: disable=protected-access
 
         return query_parameters, header_parameters
 
@@ -1039,6 +1061,7 @@ def _generate_delete_blobs_options(self,
         delete_snapshots = kwargs.pop('delete_snapshots', None)
         if_modified_since = kwargs.pop('if_modified_since', None)
         if_unmodified_since = kwargs.pop('if_unmodified_since', None)
+        if_tags_match_condition = kwargs.pop('if_tags_match_condition', None)
         kwargs.update({'raise_on_any_failure': raise_on_any_failure,
                        'sas': self._query_str.replace('?', '&'),
                        'timeout': '&timeout=' + str(timeout) if timeout else ""
@@ -1057,18 +1080,21 @@ def _generate_delete_blobs_options(self,
                     if_modified_since=if_modified_since or blob.get('if_modified_since'),
                     if_unmodified_since=if_unmodified_since or blob.get('if_unmodified_since'),
                     etag=blob.get('etag'),
+                    if_tags_match_condition=if_tags_match_condition or blob.get('if_tags_match_condition'),
                     match_condition=blob.get('match_condition') or MatchConditions.IfNotModified if blob.get('etag')
                     else None,
                     timeout=blob.get('timeout'),
                 )
-                query_parameters, header_parameters = self._generate_delete_blobs_subrequest_options(**options)
             except AttributeError:
-                query_parameters, header_parameters = self._generate_delete_blobs_subrequest_options(
+                options = BlobClient._generic_delete_blob_options(  # pylint: disable=protected-access
                     delete_snapshots=delete_snapshots,
                     if_modified_since=if_modified_since,
-                    if_unmodified_since=if_unmodified_since
+                    if_unmodified_since=if_unmodified_since,
+                    if_tags_match_condition=if_tags_match_condition
                 )
 
+            query_parameters, header_parameters = self._generate_delete_blobs_subrequest_options(**options)
+
             req = HttpRequest(
                 "DELETE",
                 "/{}/{}{}".format(quote(container_name), quote(blob_name, safe='/~'), self._query_str),
@@ -1113,6 +1139,8 @@ def delete_blobs(self, *blobs, **kwargs):
                     key: 'etag', value type: str
                 match the etag or not:
                     key: 'match_condition', value type: MatchConditions
+                tags match condition:
+                    key: 'if_tags_match_condition', value type: str
                 lease:
                     key: 'lease_id', value type: Union[str, LeaseClient]
                 timeout for subrequest:
@@ -1135,6 +1163,12 @@ def delete_blobs(self, *blobs, **kwargs):
             If a date is passed in without timezone info, it is assumed to be UTC.
             Specify this header to perform the operation only if
             the resource has not been modified since the specified date/time.
+        :keyword str if_tags_match_condition
+            Specify a SQL where clause on blob tags to operate only on blob with a matching value.
+            eg. "\"tagname\"='my tag'"
+
+            .. versionadded:: 12.4.0
+
         :keyword bool raise_on_any_failure:
             This is a boolean param which defaults to True. When this is set, an exception
             is raised even if there is a single operation failure.
@@ -1152,19 +1186,25 @@ def delete_blobs(self, *blobs, **kwargs):
                 :dedent: 8
                 :caption: Deleting multiple blobs.
         """
+        if len(blobs) == 0:
+            return iter(list())
+
         reqs, options = self._generate_delete_blobs_options(*blobs, **kwargs)
 
         return self._batch_send(*reqs, **options)
 
     def _generate_set_tiers_subrequest_options(
-        self, tier, rehydrate_priority=None, lease_access_conditions=None, **kwargs
+        self, tier, snapshot=None, version_id=None, rehydrate_priority=None, lease_access_conditions=None, **kwargs
     ):
         """This code is a copy from _generated.
 
         Once Autorest is able to provide request preparation this code should be removed.
         """
         if not tier:
             raise ValueError("A blob tier must be specified")
+        if snapshot and version_id:
+            raise ValueError("Snapshot and version_id cannot be set at the same time")
+        if_tags = kwargs.pop('if_tags', None)
 
         lease_id = None
         if lease_access_conditions is not None:
@@ -1174,6 +1214,10 @@ def _generate_set_tiers_subrequest_options(
         timeout = kwargs.pop('timeout', None)
         # Construct parameters
         query_parameters = {}
+        if snapshot is not None:
+            query_parameters['snapshot'] = self._client._serialize.query("snapshot", snapshot, 'str')  # pylint: disable=protected-access
+        if version_id is not None:
+            query_parameters['versionid'] = self._client._serialize.query("version_id", version_id, 'str')  # pylint: disable=protected-access
         if timeout is not None:
             query_parameters['timeout'] = self._client._serialize.query("timeout", timeout, 'int', minimum=0)  # pylint: disable=protected-access
         query_parameters['comp'] = self._client._serialize.query("comp", comp, 'str')  # pylint: disable=protected-access, specify-parameter-names-in-call
@@ -1186,6 +1230,8 @@ def _generate_set_tiers_subrequest_options(
                 "rehydrate_priority", rehydrate_priority, 'str')
         if lease_id is not None:
             header_parameters['x-ms-lease-id'] = self._client._serialize.header("lease_id", lease_id, 'str')  # pylint: disable=protected-access
+        if if_tags is not None:
+            header_parameters['x-ms-if-tags'] = self._client._serialize.header("if_tags", if_tags, 'str')  # pylint: disable=protected-access
 
         return query_parameters, header_parameters
 
@@ -1197,6 +1243,7 @@ def _generate_set_tiers_options(self,
         timeout = kwargs.pop('timeout', None)
         raise_on_any_failure = kwargs.pop('raise_on_any_failure', True)
         rehydrate_priority = kwargs.pop('rehydrate_priority', None)
+        if_tags = kwargs.pop('if_tags_match_condition', None)
         kwargs.update({'raise_on_any_failure': raise_on_any_failure,
                        'sas': self._query_str.replace('?', '&'),
                        'timeout': '&timeout=' + str(timeout) if timeout else ""
@@ -1211,13 +1258,16 @@ def _generate_set_tiers_options(self,
                 tier = blob_tier or blob.get('blob_tier')
                 query_parameters, header_parameters = self._generate_set_tiers_subrequest_options(
                     tier=tier,
+                    snapshot=blob.get('snapshot'),
+                    version_id=blob.get('version_id'),
                     rehydrate_priority=rehydrate_priority or blob.get('rehydrate_priority'),
                     lease_access_conditions=blob.get('lease_id'),
+                    if_tags=if_tags or blob.get('if_tags_match_condition'),
                     timeout=timeout or blob.get('timeout')
                 )
             except AttributeError:
                 query_parameters, header_parameters = self._generate_set_tiers_subrequest_options(
-                    blob_tier, rehydrate_priority=rehydrate_priority)
+                    blob_tier, rehydrate_priority=rehydrate_priority, if_tags=if_tags)
 
             req = HttpRequest(
                 "PUT",
@@ -1270,12 +1320,24 @@ def set_standard_blob_tier_blobs(
                     key: 'rehydrate_priority', value type: RehydratePriority
                 lease:
                     key: 'lease_id', value type: Union[str, LeaseClient]
+                snapshot:
+                    key: "snapshost", value type: str
+                version id:
+                    key: "version_id", value type: str
+                tags match condition:
+                    key: 'if_tags_match_condition', value type: str
                 timeout for subrequest:
                     key: 'timeout', value type: int
 
         :type blobs: list[str], list[dict], or list[~azure.storage.blob.BlobProperties]
         :keyword ~azure.storage.blob.RehydratePriority rehydrate_priority:
             Indicates the priority with which to rehydrate an archived blob
+        :keyword str if_tags_match_condition
+            Specify a SQL where clause on blob tags to operate only on blob with a matching value.
+            eg. "\"tagname\"='my tag'"
+
+            .. versionadded:: 12.4.0
+
         :keyword int timeout:
             The timeout parameter is expressed in seconds.
         :keyword bool raise_on_any_failure:

azure/multiapi/storagev2/blob/v2019_12_12/_container_client.py

@@ -9,8 +9,11 @@
     TYPE_CHECKING
 )
 
+from ._models import BlobType, CopyProperties, ContentSettings, LeaseProperties, BlobProperties
+from ._shared.models import get_enum_value
+
 from ._shared.response_handlers import deserialize_metadata
-from ._models import BlobProperties, ContainerProperties, BlobAnalyticsLogging, Metrics, CorsRule, RetentionPolicy, \
+from ._models import ContainerProperties, BlobAnalyticsLogging, Metrics, CorsRule, RetentionPolicy, \
     StaticWebsite, ObjectReplicationPolicy, ObjectReplicationRule
 
 if TYPE_CHECKING:
@@ -20,7 +23,7 @@
 def deserialize_blob_properties(response, obj, headers):
     blob_properties = BlobProperties(
         metadata=deserialize_metadata(response, obj, headers),
-        object_replication_source_properties=deserialize_ors_policies(response),
+        object_replication_source_properties=deserialize_ors_policies(response.headers),
         **headers
     )
     if 'Content-Range' in headers:
@@ -31,20 +34,21 @@ def deserialize_blob_properties(response, obj, headers):
     return blob_properties
 
 
-def deserialize_ors_policies(response):
+def deserialize_ors_policies(policy_dictionary):
+
+    if policy_dictionary is None:
+        return None
     # For source blobs (blobs that have policy ids and rule ids applied to them),
     # the header will be formatted as "x-ms-or-<policy_id>_<rule_id>: {Complete, Failed}".
     # The value of this header is the status of the replication.
-    or_policy_status_headers = {key: val for key, val in response.headers.items()
-                                if key.startswith('x-ms-or') and key != 'x-ms-or-policy-id'}
+    or_policy_status_headers = {key: val for key, val in policy_dictionary.items()
+                                if 'or-' in key and key != 'x-ms-or-policy-id'}
 
     parsed_result = {}
 
-    # all the ors headers have the same prefix, so we note down its length here to avoid recalculating it repeatedly
-    header_prefix_length = len('x-ms-or-')
-
     for key, val in or_policy_status_headers.items():
-        policy_and_rule_ids = key[header_prefix_length:].split('_')
+        # list blobs gives or-policy_rule and get blob properties gives x-ms-or-policy_rule
+        policy_and_rule_ids = key.split('or-')[1].split('_')
         policy_id = policy_and_rule_ids[0]
         rule_id = policy_and_rule_ids[1]
 
@@ -106,3 +110,48 @@ def service_properties_deserialize(generated):
         'delete_retention_policy': RetentionPolicy._from_generated(generated.delete_retention_policy),  # pylint: disable=protected-access
         'static_website': StaticWebsite._from_generated(generated.static_website),  # pylint: disable=protected-access
     }
+
+
+def get_blob_properties_from_generated_code(generated):
+    blob = BlobProperties()
+    blob.name = generated.name
+    blob_type = get_enum_value(generated.properties.blob_type)
+    blob.blob_type = BlobType(blob_type) if blob_type else None
+    blob.etag = generated.properties.etag
+    blob.deleted = generated.deleted
+    blob.snapshot = generated.snapshot
+    blob.is_append_blob_sealed = generated.properties.is_sealed
+    blob.metadata = generated.metadata.additional_properties if generated.metadata else {}
+    blob.encrypted_metadata = generated.metadata.encrypted if generated.metadata else None
+    blob.lease = LeaseProperties._from_generated(generated)  # pylint: disable=protected-access
+    blob.copy = CopyProperties._from_generated(generated)  # pylint: disable=protected-access
+    blob.last_modified = generated.properties.last_modified
+    blob.creation_time = generated.properties.creation_time
+    blob.content_settings = ContentSettings._from_generated(generated)  # pylint: disable=protected-access
+    blob.size = generated.properties.content_length
+    blob.page_blob_sequence_number = generated.properties.blob_sequence_number
+    blob.server_encrypted = generated.properties.server_encrypted
+    blob.encryption_scope = generated.properties.encryption_scope
+    blob.deleted_time = generated.properties.deleted_time
+    blob.remaining_retention_days = generated.properties.remaining_retention_days
+    blob.blob_tier = generated.properties.access_tier
+    blob.rehydrate_priority = generated.properties.rehydrate_priority
+    blob.blob_tier_inferred = generated.properties.access_tier_inferred
+    blob.archive_status = generated.properties.archive_status
+    blob.blob_tier_change_time = generated.properties.access_tier_change_time
+    blob.version_id = generated.version_id
+    blob.is_current_version = generated.is_current_version
+    blob.tag_count = generated.properties.tag_count
+    blob.tags = parse_tags(generated.blob_tags)  # pylint: disable=protected-access
+    blob.object_replication_source_properties = deserialize_ors_policies(generated.object_replication_metadata)
+    return blob
+
+
+def parse_tags(generated_tags):
+    # type: (Optional[List[BlobTag]]) -> Union[Dict[str, str], None]
+    """Deserialize a list of BlobTag objects into a dict.
+    """
+    if generated_tags:
+        tag_dict = {t.key: t.value for t in generated_tags.blob_tag_set}
+        return tag_dict
+    return None

azure/multiapi/storagev2/blob/v2019_12_12/_deserialize.py

@@ -258,7 +258,7 @@ class StorageStreamDownloader(object):  # pylint: disable=too-many-instance-attr
         The properties of the blob being downloaded. If only a range of the data is being
         downloaded, this will be reflected in the properties.
     :ivar int size:
-        The size of the total data in the stream. This will be the byte range if speficied,
+        The size of the total data in the stream. This will be the byte range if specified,
         otherwise the total size of the blob.
     """