diff --git a/b2sdk/_internal/bucket.py b/b2sdk/_internal/bucket.py index 6c8c8d24..72591115 100644 --- a/b2sdk/_internal/bucket.py +++ b/b2sdk/_internal/bucket.py @@ -40,7 +40,11 @@ ) from .file_version import DownloadVersion, FileIdAndName, FileVersion from .filter import Filter, FilterMatcher -from .http_constants import LIST_FILE_NAMES_MAX_LIMIT +from .http_constants import ( + LARGE_FILE_SHA1, + LIST_FILE_NAMES_MAX_LIMIT, + SSE_C_KEY_ID_FILE_INFO_KEY_NAME, +) from .progress import AbstractProgressListener, DoNothingProgressListener from .raw_api import LifecycleRule, NotificationRule, NotificationRuleResponse from .replication.setting import ReplicationConfiguration, ReplicationConfigurationFactory @@ -1441,7 +1445,39 @@ def copy( Example string value: 'gzip'. :param str,None content_language: an optional content language setting. Syntax based on the section 14.12 of RFC 2616. Example string value: 'mi, en_US'. - """ + + .. note:: + When header overrides are provided without ``content_type``, source metadata is used + to preserve the content type and, if ``file_info`` is omitted, the source file info. + If the required source metadata is not supplied, fetching it requires the ``readFiles`` + capability. Supplying ``source_content_type`` and either ``source_file_info`` or + destination ``file_info`` avoids that lookup. + For partial copies and other copies with a specified range, the inherited + ``large_file_sha1`` is omitted because it may not describe the copied bytes. + """ + + if content_type is None and any( + value is not None + for value in ( + cache_control, + expires, + content_disposition, + content_encoding, + content_language, + ) + ): + if source_content_type is None or (file_info is None and source_file_info is None): + source_metadata = self.api.session.get_file_info_by_id(file_id) + if source_content_type is None: + source_content_type = source_metadata['contentType'] + if source_file_info is None: + source_file_info = source_metadata['fileInfo'] + content_type = source_content_type + if file_info is None: + file_info = source_file_info.copy() + file_info.pop(SSE_C_KEY_ID_FILE_INFO_KEY_NAME, None) + if offset or length: + file_info.pop(LARGE_FILE_SHA1, None) copy_source = CopySource( file_id, diff --git a/changelog.d/+copy-header-content-type.fixed.md b/changelog.d/+copy-header-content-type.fixed.md new file mode 100644 index 00000000..b4a65939 --- /dev/null +++ b/changelog.d/+copy-header-content-type.fixed.md @@ -0,0 +1 @@ +Preserve the source content type when copying with header overrides and avoid inheriting an incorrect checksum for ranged copies. diff --git a/test/unit/bucket/test_copy_header_content_type.py b/test/unit/bucket/test_copy_header_content_type.py new file mode 100644 index 00000000..252d8f73 --- /dev/null +++ b/test/unit/bucket/test_copy_header_content_type.py @@ -0,0 +1,191 @@ +###################################################################### +# +# File: test/unit/bucket/test_copy_header_content_type.py +# +# Copyright 2026 Backblaze Inc. All Rights Reserved. +# +# License https://www.backblaze.com/using_b2_code.html +# +###################################################################### +from hashlib import sha1 +from unittest import mock + +import pytest + +from b2sdk._internal.exception import AccessDenied +from b2sdk._internal.http_constants import LARGE_FILE_SHA1, SSE_C_KEY_ID_FILE_INFO_KEY_NAME + +from .test_bucket import SSE_C_AES, SSE_C_AES_2 + + +@pytest.mark.apiver(from_ver=2) +@pytest.mark.parametrize('length', [None, 5]) +@pytest.mark.parametrize('provide_source_metadata', [False, True]) +def test_copy_headers_without_content_type_preserves_source_metadata( + bucket, length, provide_source_metadata, monkeypatch +): + source = bucket.upload_bytes( + b'hello', 'source.txt', content_type='application/x-example', file_info={'owner': 'alice'} + ) + source_metadata = ( + {'source_content_type': source.content_type, 'source_file_info': source.file_info} + if provide_source_metadata + else {} + ) + if provide_source_metadata: + + def fail_lookup(*args, **kwargs): + raise AssertionError('source metadata lookup was unnecessary') + + monkeypatch.setattr(bucket.api.session, 'get_file_info_by_id', fail_lookup) + copied = bucket.copy( + source.id_, + 'destination.txt', + length=length, + expires='Thu, 01 Dec 2050 16:00:00 GMT', + content_language='en', + **source_metadata, + ) + assert copied.content_type == 'application/x-example' + assert copied.file_info['owner'] == 'alice' + assert copied.file_info['b2-expires'] == 'Thu, 01 Dec 2050 16:00:00 GMT' + assert copied.file_info['b2-content-language'] == 'en' + + +@pytest.mark.apiver(from_ver=2) +def test_copy_headers_uses_explicit_destination_metadata(bucket, monkeypatch): + source = bucket.upload_bytes( + b'hello', 'source.txt', content_type='application/x-example', file_info={'old': 'value'} + ) + + def fail_lookup(*args, **kwargs): + raise AssertionError('source metadata lookup was unnecessary') + + monkeypatch.setattr(bucket.api.session, 'get_file_info_by_id', fail_lookup) + copied = bucket.copy( + source.id_, + 'destination.txt', + source_content_type='application/x-example', + file_info={'new': 'value'}, + content_disposition='attachment', + ) + assert copied.content_type == 'application/x-example' + assert copied.file_info == {'new': 'value', 'b2-content-disposition': 'attachment'} + + +@pytest.mark.apiver(from_ver=2) +@pytest.mark.parametrize('length', [None, 5]) +@pytest.mark.parametrize('destination_encryption', [None, SSE_C_AES_2]) +def test_copy_headers_replaces_source_encryption_key_id(bucket, length, destination_encryption): + source = bucket.upload_bytes( + b'hello', + 'source.txt', + content_type='application/x-example', + file_info={'owner': 'alice'}, + encryption=SSE_C_AES, + ) + copied = bucket.copy( + source.id_, + 'destination.txt', + length=length, + source_encryption=SSE_C_AES, + destination_encryption=destination_encryption, + source_file_info=source.file_info, + source_content_type=source.content_type, + content_language='en', + ) + assert copied.file_info['owner'] == 'alice' + assert copied.file_info['b2-content-language'] == 'en' + assert copied.file_info.get(SSE_C_KEY_ID_FILE_INFO_KEY_NAME) == ( + destination_encryption.key.key_id if destination_encryption else None + ) + assert source.file_info[SSE_C_KEY_ID_FILE_INFO_KEY_NAME] == SSE_C_AES.key.key_id + + +@pytest.mark.apiver(from_ver=2) +@pytest.mark.parametrize('provide_source_metadata', [False, True]) +def test_partial_multipart_copy_does_not_inherit_source_checksum(bucket, provide_source_metadata): + data = b'a' * 300 + b'b' * 300 + source_checksum = sha1(data).hexdigest() + source = bucket.upload_bytes( + data, + 'source.txt', + content_type='application/x-example', + file_info={'owner': 'alice'}, + large_file_sha1=source_checksum, + ) + source_metadata = ( + {'source_content_type': source.content_type, 'source_file_info': source.file_info} + if provide_source_metadata + else {} + ) + with mock.patch.object( + bucket.api.session, 'copy_part', wraps=bucket.api.session.copy_part + ) as copy_part: + copied = bucket.copy( + source.id_, + 'partial.txt', + offset=100, + length=400, + max_part_size=200, + content_language='en', + **source_metadata, + ) + assert copy_part.call_count == 2 + assert copied.size == 400 + assert LARGE_FILE_SHA1 not in copied.file_info + assert copied.get_content_sha1() is None + assert copied.file_info['owner'] == 'alice' + assert copied.file_info['b2-content-language'] == 'en' + assert source.file_info[LARGE_FILE_SHA1] == source_checksum + + +@pytest.mark.apiver(from_ver=2) +def test_whole_file_copy_keeps_source_checksum(bucket): + data = b'hello' + checksum = sha1(data).hexdigest() + source = bucket.upload_bytes(data, 'source.txt', file_info={LARGE_FILE_SHA1: checksum}) + copied = bucket.copy(source.id_, 'destination.txt', content_language='en') + assert copied.file_info[LARGE_FILE_SHA1] == checksum + + +@pytest.mark.apiver(from_ver=2) +def test_partial_copy_keeps_explicit_destination_checksum(bucket): + data = b'a' * 300 + b'b' * 300 + source = bucket.upload_bytes(data, 'source.txt') + checksum = sha1(data[100:500]).hexdigest() + copied = bucket.copy( + source.id_, + 'partial.txt', + offset=100, + length=400, + max_part_size=200, + source_content_type=source.content_type, + file_info={LARGE_FILE_SHA1: checksum}, + content_language='en', + ) + assert copied.file_info[LARGE_FILE_SHA1] == checksum + assert copied.get_content_sha1() == checksum + + +@pytest.mark.apiver(from_ver=2) +def test_copy_headers_propagates_source_info_access_denied(bucket, monkeypatch): + source = bucket.upload_bytes(b'hello', 'source.txt') + + def deny_lookup(*args, **kwargs): + raise AccessDenied() + + monkeypatch.setattr(bucket.api.session, 'get_file_info_by_id', deny_lookup) + with mock.patch.object( + bucket.api.session, 'copy_file', wraps=bucket.api.session.copy_file + ) as copy_file: + with pytest.raises(AccessDenied): + bucket.copy(source.id_, 'destination.txt', content_language='en') + copy_file.assert_not_called() + + +@pytest.mark.apiver(from_ver=2) +def test_copy_documents_omission_of_source_checksum_for_partial_copies(bucket): + doc = bucket.copy.__doc__ + assert 'partial cop' in doc.lower() + assert 'large_file_sha1' in doc diff --git a/test/unit/bucket/test_regression_bucket.py b/test/unit/bucket/test_regression_bucket.py index 4a676ea5..4b009845 100644 --- a/test/unit/bucket/test_regression_bucket.py +++ b/test/unit/bucket/test_regression_bucket.py @@ -43,7 +43,7 @@ UploadSourceBytes, WriteIntent, ) -from apiver_deps_exception import CopyArgumentsMismatch, NonExistentBucket +from apiver_deps_exception import NonExistentBucket from test.known_defect import DefectStillPresent, defect_raises, expect, known_defect @@ -53,22 +53,16 @@ class TestRegressionCopyAndLs(TestCaseWithBucket): - @known_defect( - 'C1', - 'copy() with expires/content_disposition/cache_control but no content_type is refused', - raises=DefectStillPresent, - ) def test_c1_copy_with_expiry_but_no_content_type(self): """C1: copy() with expires/headers but no content_type must work (2.13.3 notes).""" file_id = self.bucket.upload_bytes(b'hello world', 'src.txt').id_ - with defect_raises(CopyArgumentsMismatch, match='content type'): - copied = self.bucket.copy( - file_id, - 'copied.txt', - expires='Thu, 01 Dec 2050 16:00:00 GMT', - content_disposition='attachment', - cache_control='max-age=5', - ) + copied = self.bucket.copy( + file_id, + 'copied.txt', + expires='Thu, 01 Dec 2050 16:00:00 GMT', + content_disposition='attachment', + cache_control='max-age=5', + ) assert copied.file_info['b2-expires'] == 'Thu, 01 Dec 2050 16:00:00 GMT', copied.file_info assert copied.file_info['b2-content-disposition'] == 'attachment', copied.file_info