Read report/label templates in binary mode

[impuls42]

Fix

file_from_template reads default report and label templates in text mode, so the resulting ContentFile reports its size as a character count. For a template with multi-byte UTF-8 characters that count is smaller than the encoded byte length, and django-storages sends it as the Content-Length. Backblaze B2 and other strict S3 backends reject the upload with IncompleteBody. AWS S3 tolerates the mismatch, so the bug is not visible there.

inventree_transfer_order_report.html is the only default template with a non-ASCII character (1458 chars, 1460 bytes), so its upload fails during create_default_reports, the transaction rolls back, and it is retried on every startup.

Reading the file in binary mode makes ContentFile.size match the uploaded bytes.

Fixes #12431

Testing

Reproduced on a live instance backed by Backblaze B2. Before the change, create_default_reports fails for the transfer-order template on every startup with IncompleteBody. After the change the template is created and the error is gone. Confirmed the mechanism in a shell for the same file: ContentFile(open(path, 'r').read()) fails, ContentFile(open(path, 'rb').read()) succeeds.

The mismatch only shows up against an S3 backend that enforces Content-Length, so it is not reproducible against the local storage used in CI.

[netlify]

✅ Deploy Preview for inventree-web-pui-preview ready!

Name	Link
🔨 Latest commit	44b845b
🔍 Latest deploy log	https://app.netlify.com/projects/inventree-web-pui-preview/deploys/6a5f1ddc2b44b000088a6c07
😎 Deploy Preview	https://deploy-preview-12432--inventree-web-pui-preview.netlify.app
📱 Preview on mobile	Toggle QR Code... Use your smartphone camera to open QR code link.
Lighthouse	1 paths audited Performance: 98 (🟢 up 1 from production) Accessibility: 81 (no change from production) Best Practices: 100 (no change from production) SEO: 78 (no change from production) PWA: - View the detailed breakdown and full score reports
🤖 Make changes	Run an agent on this branch

---

To edit notification comments on pull requests, go to your Netlify project configuration.

[impuls42]

Updated after a self-review:

• Switched to Path.read_bytes() instead of open('rb').read(), so the file handle is closed rather than relying on refcounting.
• Added a regression test in report/tests.py asserting that file_from_template sizes the ContentFile by the file's byte length. It fails on the old text-mode read (character count) and passes with this change, and it needs no S3 backend.

[SchrodingersGat]

@impuls42 thanks for the fix, this seems clean

[SchrodingersGat]

This is the same issue as #12410

[SchrodingersGat]

@impuls42 can you please add some testing against different text encodings as per the comments in #12410

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 86.88%. Comparing base (b723f5d) to head (44b845b).

Additional details and impacted files

@@           Coverage Diff           @@
##           master   #12432   +/-   ##
=======================================
  Coverage   86.87%   86.88%           
=======================================
  Files        1442     1442           
  Lines       94942    94971   +29     
  Branches    11101    11101           
=======================================
+ Hits        82482    82511   +29     
  Misses      12395    12395           
  Partials       65       65           

Flag	Coverage Δ
backend	91.07% <100.00%> (+<0.01%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Components	Coverage Δ
Backend Apps	92.57% <100.00%> (+<0.01%)	⬆️
Backend General	93.49% <ø> (ø)
Frontend	79.66% <ø> (ø)

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[impuls42]

Added a regression test (test_file_from_template_is_encoding_agnostic) that runs file_from_template over content in utf-8, utf-16 and latin-1 and asserts the ContentFile is byte-identical to the file on disk, with the size equal to the byte length. The utf cases include multi-byte characters where the byte length differs from the character count, which is the mismatch the old text-mode read produced. Reading the raw bytes never decodes, so non-English content can't be mis-sized or corrupted regardless of the file's encoding.

[SchrodingersGat]

@impuls42 thanks for the contribution!

[github-actions]

💚 All backports created successfully

Status	Branch	Result
✅	1.4.x

Questions ?

Please refer to the Backport tool documentation and see the Github Action logs for details