Merge pull request #441 from Backblaze/header-args-v3

Header args v3

Author: mpnowacki-reef

CHANGELOG.md

@@ -11,6 +11,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 ### Added
 * Add `*_PART_SIZE`, `BUCKET_NAME_*`, `STDOUT_FILEPATH` constants
 * Add `points_to_fifo`, `points_to_stdout` functions
+* Add `expires`, `content_disposition`, `content_encoding`, `content_language` arguments to various `Bucket` methods
 
 ### Changed
 * Mark `TempDir` as deprecated in favor of `tempfile.TemporaryDirectory`

b2sdk/_v3/__init__.py

@@ -84,6 +84,7 @@
 from b2sdk.file_version import FileVersionFactory
 from b2sdk.large_file.part import Part
 from b2sdk.large_file.unfinished_large_file import UnfinishedLargeFile
+from b2sdk.large_file.services import LargeFileServices
 from b2sdk.utils.range_ import Range
 
 # file lock

b2sdk/api.py

@@ -71,6 +71,7 @@ class Services:
     UPLOAD_MANAGER_CLASS = staticmethod(UploadManager)
     COPY_MANAGER_CLASS = staticmethod(CopyManager)
     DOWNLOAD_MANAGER_CLASS = staticmethod(DownloadManager)
+    LARGE_FILE_SERVICES_CLASS = staticmethod(LargeFileServices)
 
     def __init__(
         self,
@@ -95,7 +96,7 @@ def __init__(
         """
         self.api = api
         self.session = api.session
-        self.large_file = LargeFileServices(self)
+        self.large_file = self.LARGE_FILE_SERVICES_CLASS(self)
         self.upload_manager = self.UPLOAD_MANAGER_CLASS(
             services=self, max_workers=max_upload_workers
         )

b2sdk/bucket.py

@@ -9,6 +9,7 @@
 ######################################################################
 from __future__ import annotations
 
+import datetime as dt
 import re
 from copy import deepcopy
 from typing import TYPE_CHECKING, Any
@@ -19,6 +20,7 @@
 from .progress import AbstractProgressListener
 from .replication.types import ReplicationStatus
 from .utils import Sha1HexDigest, b2_url_decode
+from .utils.http_date import parse_http_date
 from .utils.range_ import Range
 
 if TYPE_CHECKING:
@@ -49,7 +51,6 @@ class BaseFileVersion:
         'file_retention',
         'mod_time_millis',
         'replication_status',
-        'cache_control',
     ]
     _TYPE_MATCHER = re.compile('[a-z0-9]+_[a-z0-9]+_f([0-9]).*')
     _FILE_TYPE = {
@@ -73,7 +74,6 @@ def __init__(
         file_retention: FileRetentionSetting = NO_RETENTION_FILE_SETTING,
         legal_hold: LegalHold = LegalHold.UNSET,
         replication_status: ReplicationStatus | None = None,
-        cache_control: str | None = None,
     ):
         self.api = api
         self.id_ = id_
@@ -87,7 +87,6 @@ def __init__(
         self.file_retention = file_retention
         self.legal_hold = legal_hold
         self.replication_status = replication_status
-        self.cache_control = cache_control
 
         if SRC_LAST_MODIFIED_MILLIS in self.file_info:
             self.mod_time_millis = int(self.file_info[SRC_LAST_MODIFIED_MILLIS])
@@ -128,7 +127,6 @@ def _get_args_for_clone(self):
             'file_retention': self.file_retention,
             'legal_hold': self.legal_hold,
             'replication_status': self.replication_status,
-            'cache_control': self.cache_control,
         }  # yapf: disable
 
     def as_dict(self):
@@ -140,7 +138,6 @@ def as_dict(self):
             'serverSideEncryption': self.server_side_encryption.as_dict(),
             'legalHold': self.legal_hold.value,
             'fileRetention': self.file_retention.as_dict(),
-            'cacheControl': self.cache_control,
         }
 
         if self.size is not None:
@@ -259,7 +256,6 @@ def __init__(
         file_retention: FileRetentionSetting = NO_RETENTION_FILE_SETTING,
         legal_hold: LegalHold = LegalHold.UNSET,
         replication_status: ReplicationStatus | None = None,
-        cache_control: str | None = None,
     ):
         self.account_id = account_id
         self.bucket_id = bucket_id
@@ -279,9 +275,36 @@ def __init__(
             file_retention=file_retention,
             legal_hold=legal_hold,
             replication_status=replication_status,
-            cache_control=cache_control,
         )
 
+    @property
+    def cache_control(self) -> str | None:
+        return self.file_info.get('b2-cache-control')
+
+    @property
+    def expires(self) -> str | None:
+        return self.file_info.get('b2-expires')
+
+    def expires_parsed(self) -> dt.datetime | None:
+        """Return the expiration date as a datetime object, or None if there is no expiration date.
+        Raise ValueError if `expires` property is not a valid HTTP-date."""
+
+        if self.expires is None:
+            return None
+        return parse_http_date(self.expires)
+
+    @property
+    def content_disposition(self) -> str | None:
+        return self.file_info.get('b2-content-disposition')
+
+    @property
+    def content_encoding(self) -> str | None:
+        return self.file_info.get('b2-content-encoding')
+
+    @property
+    def content_language(self) -> str | None:
+        return self.file_info.get('b2-content-language')
+
     def _get_args_for_clone(self):
         args = super()._get_args_for_clone()
         args.update(
@@ -352,7 +375,6 @@ def _get_upload_headers(self) -> bytes:
             server_side_encryption=sse,
             file_retention=self.file_retention,
             legal_hold=self.legal_hold,
-            cache_control=self.cache_control,
         )
 
         headers_str = ''.join(
@@ -381,8 +403,8 @@ class DownloadVersion(BaseFileVersion):
         'content_disposition',
         'content_length',
         'content_language',
-        '_expires',
-        '_cache_control',
+        'expires',
+        'cache_control',
         'content_encoding',
     ]
 
@@ -412,8 +434,8 @@ def __init__(
         self.content_disposition = content_disposition
         self.content_length = content_length
         self.content_language = content_language
-        self._expires = expires  # TODO: parse the string representation of this timestamp to datetime in DownloadVersionFactory
-        self._cache_control = cache_control  # TODO: parse the string representation of this mapping to dict in DownloadVersionFactory
+        self.expires = expires
+        self.cache_control = cache_control
         self.content_encoding = content_encoding
 
         super().__init__(
@@ -429,9 +451,30 @@ def __init__(
             file_retention=file_retention,
             legal_hold=legal_hold,
             replication_status=replication_status,
-            cache_control=cache_control,
         )
 
+    def expires_parsed(self) -> dt.datetime | None:
+        """Return the expiration date as a datetime object, or None if there is no expiration date.
+        Raise ValueError if `expires` property is not a valid HTTP-date."""
+
+        if self.expires is None:
+            return None
+        return parse_http_date(self.expires)
+
+    def as_dict(self) -> dict:
+        result = super().as_dict()
+        if self.cache_control is not None:
+            result['cacheControl'] = self.cache_control
+        if self.expires is not None:
+            result['expires'] = self.expires
+        if self.content_disposition is not None:
+            result['contentDisposition'] = self.content_disposition
+        if self.content_encoding is not None:
+            result['contentEncoding'] = self.content_encoding
+        if self.content_language is not None:
+            result['contentLanguage'] = self.content_language
+        return result
+
     def _get_args_for_clone(self):
         args = super()._get_args_for_clone()
         args.update(
@@ -440,8 +483,8 @@ def _get_args_for_clone(self):
                 'content_disposition': self.content_disposition,
                 'content_length': self.content_length,
                 'content_language': self.content_language,
-                'expires': self._expires,
-                'cache_control': self._cache_control,
+                'expires': self.expires,
+                'cache_control': self.cache_control,
                 'content_encoding': self.content_encoding,
             }
         )
@@ -516,7 +559,6 @@ def from_api_response(self, file_version_dict, force_action=None):
         replication_status_value = file_version_dict.get('replicationStatus')
         replication_status = replication_status_value and ReplicationStatus[
             replication_status_value.upper()]
-        cache_control = file_version_dict.get('cacheControl')
 
         return self.FILE_VERSION_CLASS(
             self.api,
@@ -535,7 +577,6 @@ def from_api_response(self, file_version_dict, force_action=None):
             file_retention,
             legal_hold,
             replication_status,
-            cache_control,
         )
 
 
@@ -575,11 +616,6 @@ def from_response_headers(self, headers):
             size = content_length = int(headers['Content-Length'])
             range_ = Range(0, max(size - 1, 0))
 
-        if 'Cache-Control' in headers:
-            cache_control = b2_url_decode(headers['Cache-Control'])
-        else:
-            cache_control = None
-
         return DownloadVersion(
             api=self.api,
             id_=headers['x-bz-file-id'],
@@ -595,7 +631,7 @@ def from_response_headers(self, headers):
             content_length=content_length,
             content_language=headers.get('Content-Language'),
             expires=headers.get('Expires'),
-            cache_control=cache_control,
+            cache_control=headers.get('Cache-Control'),
             content_encoding=headers.get('Content-Encoding'),
             file_retention=FileRetentionSetting.from_response_headers(headers),
             legal_hold=LegalHold.from_response_headers(headers),

b2sdk/file_version.py

@@ -17,6 +17,9 @@
 
 
 class LargeFileServices:
+
+    UNFINISHED_LARGE_FILE_CLASS = staticmethod(UnfinishedLargeFile)
+
     def __init__(self, services):
         self.services = services
 
@@ -57,7 +60,7 @@ def list_unfinished_large_files(
                 bucket_id, start_file_id, batch_size, prefix
             )
             for file_dict in batch['files']:
-                yield UnfinishedLargeFile(file_dict)
+                yield self.UNFINISHED_LARGE_FILE_CLASS(file_dict)
             start_file_id = batch.get('nextFileId')
             if start_file_id is None:
                 break
@@ -86,7 +89,6 @@ def start_large_file(
         encryption: EncryptionSetting | None = None,
         file_retention: FileRetentionSetting | None = None,
         legal_hold: LegalHold | None = None,
-        cache_control: str | None = None,
     ):
         """
         Start a large file transfer.
@@ -97,9 +99,8 @@ def start_large_file(
         :param b2sdk.v2.EncryptionSetting encryption: encryption settings (``None`` if unknown)
         :param b2sdk.v2.FileRetentionSetting file_retention: file retention setting
         :param b2sdk.v2.LegalHold legal_hold: legal hold setting
-        :param str,None cache_control: an optional cache control setting. Syntax based on the section 14.9 of RFC 2616. Example string value: 'public, max-age=86400, s-maxage=3600, no-transform'.
         """
-        return UnfinishedLargeFile(
+        return self.UNFINISHED_LARGE_FILE_CLASS(
             self.services.session.start_large_file(
                 bucket_id,
                 file_name,
@@ -108,7 +109,6 @@ def start_large_file(
                 server_side_encryption=encryption,
                 file_retention=file_retention,
                 legal_hold=legal_hold,
-                cache_control=cache_control,
             )
         )
 

b2sdk/large_file/services.py

@@ -9,8 +9,11 @@
 ######################################################################
 from __future__ import annotations
 
+import datetime as dt
+
 from b2sdk.encryption.setting import EncryptionSettingFactory
 from b2sdk.file_lock import FileRetentionSetting, LegalHold
+from b2sdk.utils.http_date import parse_http_date
 
 
 class UnfinishedLargeFile:
@@ -38,7 +41,34 @@ def __init__(self, file_dict):
         self.encryption = EncryptionSettingFactory.from_file_version_dict(file_dict)
         self.file_retention = FileRetentionSetting.from_file_version_dict(file_dict)
         self.legal_hold = LegalHold.from_file_version_dict(file_dict)
-        self.cache_control = file_dict.get('cacheControl')
+
+    @property
+    def cache_control(self) -> str | None:
+        return (self.file_info or {}).get('b2-cache-control')
+
+    @property
+    def expires(self) -> str | None:
+        return (self.file_info or {}).get('b2-expires')
+
+    def expires_parsed(self) -> dt.datetime | None:
+        """Return the expiration date as a datetime object, or None if there is no expiration date.
+        Raise ValueError if `expires` property is not a valid HTTP-date."""
+
+        if self.expires is None:
+            return None
+        return parse_http_date(self.expires)
+
+    @property
+    def content_disposition(self) -> str | None:
+        return (self.file_info or {}).get('b2-content-disposition')
+
+    @property
+    def content_encoding(self) -> str | None:
+        return (self.file_info or {}).get('b2-content-encoding')
+
+    @property
+    def content_language(self) -> str | None:
+        return (self.file_info or {}).get('b2-content-language')
 
     def __repr__(self):
         return f'<{self.__class__.__name__} {self.bucket_id} {self.file_name}>'