fix(http-client-python): synthesize filename in multipart Content-Disposition for bare file inputs#10843
Open
iscai-msft wants to merge 8 commits into
Open
fix(http-client-python): synthesize filename in multipart Content-Disposition for bare file inputs#10843iscai-msft wants to merge 8 commits into
iscai-msft wants to merge 8 commits into
Conversation
iscai-msft
requested review from
ChenxiJiang333,
catalinaperalta,
lmazuel,
msyyc,
tadelesh and
timotheeguerin
as code owners
microsoft-github-policy-service
Bot
added
the
emitter:client:python
commit: |
Contributor
|
All changed packages have been documented.
Show changes
|
|
You can try these changes here
|
…ition When callers pass bare bytes/str/IO instead of a (filename, content) tuple for multipart file fields, the prepare_multipart_form_data helper now wraps them with a synthesized filename derived from IO.name or the field name. This ensures the Content-Disposition header includes filename= so servers that require it (e.g. for .zip detection) don't reject the upload. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
iscai-msft
force-pushed
the
fix/python-multipart-filename
branch
from
196355c to
4ad096a
Compare
…esis Add sync and async test variants that pass bare open() IO (without explicit filename tuples) to the uploadFileRequiredFilename, uploadFileSpecificContentType, and uploadFileArray Spector scenarios. These exercise _normalize_multipart_file_entry and verify the server receives filename= in Content-Disposition. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
msyyc
reviewed
Jun 1, 2026
| # (non-tuple) values - a tuple bypasses that default and falls back to the | ||
| # HTTP "text/plain" default instead. Setting it explicitly preserves the | ||
| # pre-existing behavior for bare bytes/IO across all transports. | ||
| return (filename, entry, "application/octet-stream") |
Contributor
There was a problem hiding this comment.
I add default content-type "application/octet-stream" to keep consistent with sdk core behavior https://github.com/Azure/azure-sdk-for-python/blob/c8abdc18a0d51368f3ff7fc9dea71dbf131550e4/sdk/core/azure-core/azure/core/utils/_pipeline_transport_rest_shared.py#L374 so that SDK keep compatible in runtime. cc @iscai-msft / @tadelesh
msyyc
approved these changes
Jun 1, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Problem
When callers pass bare
bytes/str/IO(theFileContentvariant ofFileType) for multipart file fields, the generatedprepare_multipart_form_datahelper creates(field_name, bare_content). The HTTP library interprets this asContent-Disposition: form-data; name="field_name"with nofilename=attribute.Many servers require
filename=in theContent-Dispositionheader to recognize file uploads (e.g., servers that use the file extension to detect package type will reject with "At least one file must be uploaded").The tuple variants
(filename, content)and(filename, content, content_type)already work correctly since they producefilename=in the header.Fix
Added
_normalize_multipart_file_entryhelper inutils.py.jinja2that wraps bare content into a(filename, content)tuple:.name: derives filename viaos.path.basename()(e.g.,open('path/to/image.jpg')→filename="image.jpg")"profileImage") or"field_0","field_1"for list entriesAlso changed
elif multipart_entry:toelif multipart_entry is not None:to allow empty bytes (b"") to be uploaded.Files Changed
generator/pygen/codegen/templates/utils.py.jinja2: Added_normalize_multipart_file_entryhelper, updatedprepare_multipart_form_datagenerator/pygen/codegen/serializers/general_serializer.py: Addedimport osto generated utils imports