[DanielNg0729]

Summary

This is a follow-up to #4773 (issue #4772).

python-pptx has no plot class for some plot types: 3-D bar, 3-D line, 3-D pie, stock, surface and bar-of-pie. #4773 classifies such a plot as other_chart and skips its series with a warning. A chart that has only such a plot therefore has no data table.

python-pptx has a plot class for a 3-D area plot, but no series class for it. When the backend reads its series, python-pptx raises NotImplementedError. The backend does not catch this error, so one 3-D area chart stops the conversion of the whole presentation (RuntimeError: Pipeline SimplePipeline failed).

This PR changes the PowerPoint backend only.

• The backend reads the series of every plot. All c:ser elements have the
  same structure, so one python-pptx series class can read the series of any
  plot type. XySeries reads scatter and bubble plots, which keep their
  values in c:yVal. BarSeries reads all other plots, which keep their
  values in c:val.
• The backend reads the category labels from the first series. python-pptx's
  plot.categories does the same. A multi-level axis still gives the labels of
  its leaf level.
• The backend classifies each plot by its XML tag. The mapping is the same as
  in the Word and Excel backends. bar3DChart is now bar_chart,
  line3DChart is line_chart, and pie3DChart is pie_chart. Stock,
  surface, 3-D area and bar-of-pie plots stay other_chart, as they are in
  Word and Excel.
• The warning about skipped plots is removed, because the backend does not
  skip a plot now.

The output does not change for the plot types that python-pptx could already read. I compared column, line, pie, doughnut, area, radar, scatter, bubble and multi-level category charts before and after this change.

Change in the reference data

Slide 2 of tests/data/pptx/sources/pptx_chart.pptx has a 3-D clustered column chart (c:bar3DChart). This PR changes its output in pptx_chart.pptx.json and pptx_chart.pptx.md:

• Before: "Other chart", with no data table.

• After: "Bar chart", with this data table:

  	2026
  Q1	10
  Q2	20
  Q3	30

These values are the values in the data cache of the chart (ppt/charts/chart2.xml). The .itxt file does not change. The other
reference files do not change. The second commit contains only this reference-data change.

Tests

• test_plot_without_a_python_pptx_class_keeps_its_data makes one chart for
  each plot type: 3-D bar, 3-D line, 3-D pie, 3-D area, bar-of-pie, stock,
  surface and 3-D surface. Each chart must keep its label and its data. All 8
  cases fail on main. The 3-D area case fails with the pipeline error.
• test_stock_plot_keeps_its_series_next_to_the_other_plots replaces
  test_unreadable_plot_keeps_the_data_of_the_other_plots. The stock series
  now come after the column series in the data table.
• test_chart_parsed_as_classified_picture_with_data now checks the
  classification and the data of slide 2.

Related PR

PR #4380 by @chrikrah fixes the same 3-D area error. It catches NotImplementedError, so the chart is kept but has no data. With this PR, the chart keeps its data too. #4380 is based on the code before #4773. I can help to coordinate the two PRs if the maintainers prefer a different order.

Checklist:

• Documentation has been updated, if necessary.
• Examples have been added, if necessary.
• Tests have been added, if necessary.

[github-actions]

✅ DCO Check Passed

Thanks @DanielNg0729, all your commits are properly signed off. 🎉

[mergify]

Merge Protections

🟢 All 2 merge protections satisfied — ready to merge.

Show 2 satisfied protections

🟢 Enforce conventional commit

Make sure that we follow https://www.conventionalcommits.org/en/v1.0.0/

• title ~= ^(fix|feat|docs|style|refactor|perf|test|build|ci|chore|revert)(?:\(.+\))?(!)?:

🟢 Require two reviewer for test updates

When test data is updated, we require two reviewers

• #approved-reviews-by >= 2

[github-actions]

AI triage (advisory)

Checked commit d572724bd8. The analysis read the diff only. It did not run code from this pull request.

The PowerPoint backend now reads the series of every chart plot through the XML tag. 3-D bar, line and pie plots keep their type and data, and a 3-D area chart no longer stops the conversion of the whole presentation.

CI safety: no concern found. A maintainer must still approve CI.

4 related pull request(s) in the same code

PR	State	Signal	Assessment
#4773	merged	referenced as #4773; also references issue #4772	related: This merged PR added the multi-plot chart code. This PR is a follow-up that changes how that code reads plots and series. It does not repeat the same change.
#4443	open	changes the same code	related: It changes run-level text formatting and hyperlinks in the same backend file. It does not touch chart code.
#4812	merged	also references issue #4772	related: This merged PR moves the chart table builder into a shared module and applies the same plot-series idea to Excel. It does not make the same change as this PR.
#4380	closed	referenced as #4380	related: This closed, unmerged PR also targets the NotImplementedError from a 3-D area chart. It drops the chart data. This PR reads the data instead.

Reference data changes

Change	Files
tables	1
text	1

• tables tests/data/pptx/groundtruth/pptx_chart.pptx.md: +7 / -1 lines; Markdown table rows changed
• text tests/data/pptx/groundtruth/pptx_chart.pptx.json: item attributes changed

Assessment: expected. The PR now reads the data of the c:bar3DChart plot on slide 2 of pptx_chart.pptx. A new data table in the Markdown and JSON output matches that change. The updated test asserts the same grid.

_{The ai:* labels are advisory. They do not approve CI and they do not add tests:full. A new push runs this check again.}

[codecov]

Codecov Report

❌ Patch coverage is 90.62500% with 3 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
docling/backend/mspowerpoint_backend.py	90.62%	1 Missing and 2 partials ⚠️

📢 Thoughts on this report? Let us know!

[dolfim-ibm]

lgtm

[dolfim-ibm]

One blocker, plus a follow-up suggestion.

Blocker: a malformed series in a 3-D, stock, surface or bar-of-pie plot now fails the whole presentation.

_plot_series and _category_labels now read c:order and c:pt/c:v for plot types that main skipped. When one of these elements is missing, python-pptx raises InvalidXmlError, but the call site around _chart_to_table_data catches only (AttributeError, ValueError, KeyError).

Sample	main	this PR
stockChart, c:ser without c:order	OK (other_chart, no data)	RuntimeError: Pipeline SimplePipeline failed
stockChart, c:pt without c:v	OK	RuntimeError: Pipeline SimplePipeline failed
barChart, same two defects	fails	fails (shared with main)

The fix is to add InvalidXmlError (already imported) to that except tuple. That also fixes the barChart case that is shared with main. Please add a one-case regression test, for example a stock plot whose c:ser has no c:order. The purpose of this PR is that one bad chart must not stop the conversion, so this gap belongs here.

Suggestion (not blocking): _CHART_TAGNAME_TO_CLASSIFICATION is now identical in the Word, Excel and PowerPoint backends. A shared module would keep the three in sync.

Classifying by tag aligns the PowerPoint backend with Word and Excel, which is good. Using BarSeries for series that are not bar series relies on all c:ser elements having the same structure. That is acceptable here.

[DanielNg0729]

hi Dr @dolfim-ibm, thank you very much, i have added some commit to resolve your point for this PR as well. This PR is ready for your review again when you have a moment. Thanks again!

[dolfim-ibm]

#4812 is now merged, and this PR conflicts with main. Please update the branch with the latest main (merge or rebase), and regenerate the affected ground truth if needed. After CI is green, this is ready for the final review.

This PR also covers the c:area3DChart crash that @chrikrah reported and fixed first in #4380. We will continue with this PR and close #4380, with credit to @chrikrah for finding the issue.

[DanielNg0729]

Thank you very much @dolfim-ibm, I merged the latest main. The chart table code now uses the shared build_chart_table from #4812. The PowerPoint reference data did not need another update. When I regenerate it, the only change is still slide 2 of pptx_chart.pptx, as before.
Thank you @chrikrah for finding the c:area3DChart crash in #4380.

[dolfim-ibm]

lgtm

[chrikrah]

@DanielNg0729 closing my #4380 for this was the right call: the c:area3DChart deck I attached there now converts with its data table, where mine kept only a picture. The InvalidXmlError path @dolfim-ibm flagged degrades to a chart without a table, and I'd merge this at d572724.

$ PYTHONPATH=$PWD python -m pytest tests/test_backend_pptx.py tests/test_backend_pptx_charts.py -q
72 passed in 12.54s

# same tests, docling/backend/mspowerpoint_backend.py taken from merge base 0b62ddd2
12 failed, 60 passed in 10.57s
# area3DChart at base: NotImplementedError "series class ... not yet implemented",
# then RuntimeError: Pipeline SimplePipeline failed
# ruff 0.15.12 check and format: clean; python-pptx 1.0.2, pytest 9.0.3
# not run: suites beyond the pptx backend

non-blocking: slide 2 of pptx_chart.pptx changes from "Other chart" with no table to "Bar chart" with one. Anyone filtering on OTHER_CHART for 3-D bars will see that move, so it deserves a line in the release notes.

@PeterStaar-IBM, Mergify wants a second approval for the ground-truth update, the same rule you cleared on #4812. @ceberam, slide 2 comes from your #3972, so the move to "Bar chart" may want your view. Could one of you review it this week?

[PeterStaar-IBM]

lgtm