manahl / PyBloqs

89%

master: 89%

LAST BUILD ON BRANCH fix/plot-format-restore

branch: fix/plot-format-restore

CHANGE BRANCH

x

Reset

• fix/plot-format-restore
• 1.1.9
• Add-Collapsible-Block
• DataFrameMapCompat
• add_typing
• blog-post-photos
• bugfix_cfg
• bugfix_foo
• bugfix_server_tabs
• bugfix_utf8
• cherry-pick-1.1.9-changelog-and-version-bump
• code_block
• decompress
• docs_folders
• feature/common_table_formatter
• feature/increase-frontend-coverage
• feature/initial-release
• fix/pandas3-table-formatters
• fix_docs
• htmx_fixes
• increase_circleci_coverage
• master
• no_async
• p_140
• p_141
• p_142
• patch-1
• permit_no_compression
• pull/133
• pybloqs_server
• python-312-compat
• readme_touchup
• release
• render_resources
• replace_inner_html
• rmorotti-pandas-applymap
• rspencer01-docs-patch
• rspencer01-patch-1
• rspencer01-patch-2
• rspencer01-v-1-4-3
• set_server_id
• unique_ids
• update-contributors
• use_old_compression

Build # 7a4aae92-f0aa-4775-91aa-8745174ad27d

Build Type

Pull #176

circleci

Committed by WatchTree-19

Commit Message

fix: plot_format leaks its settings when the wrapped block raises

plot_format mutates module-level state and restores it after the yield with no
try/finally:

    old = get_plot_format()
    set_plot_format(plot_format, dpi)
    yield
    set_plot_format(*old)

If the wrapped code raises, which rendering readily does, the restore never
runs and _PLOT_FORMAT, _PLOT_MIME_TYPE and _PLOT_DPI stay overridden for the
rest of the process. Nothing warns; every later block silently renders at the
wrong format and DPI. Against released 1.4.3, a block that raises inside
plot_format("svg", 300) leaves the globals at ('svg', 300).

Wrap the yield in try/finally.

Separately, set_plot_format assigns _PLOT_FORMAT before looking up the mime
type, so an unsupported format raises KeyError with _PLOT_FORMAT already
changed and _PLOT_MIME_TYPE still holding the old value. On 1.4.3,
set_plot_format("jpeg") leaves format 'jpeg' paired with mime 'png'. Look the
mime type up first so a bad format changes nothing.

Adds two tests alongside the existing test_plot_format_ctx_manager, which only
covers the success path.

Signed-Off By Sandeep Singh Rai sandeeprai_dsp@hotmail.com

Pull Request Pull Request #176: fix: plot_format leaks its settings when the wrapped block raises

Coverage Stats

5 of 5 new or added lines in 1 file covered. (100.0%)

2512 of 2835 relevant lines covered (88.61%)

1.77 hits per line