fix(slides): reindent xml-get output with a stdlib-only formatter

[tianyouskrrr]

Summary

Reland of #1987 (reverted in #2013): slides +xml-get again reindents presentation/slide XML on the text output surfaces — with the formatter rebuilt on the Go standard library only. The BSD-licensed github.com/beevik/etree dependency that triggered the revert is gone; go.mod gains nothing.

• --raw and --output reindent the XML so each structural element (presentation/slide/shape/style/...) sits on its own line. Reindentation never enters the schema's mixed-content text-bearing elements (p, span, strong, em, u, del, a, shadow, outline, chartTitle, chartSubTitle) and never touches leaf elements, so document text is never altered.
• The default JSON envelope carries the server's XML verbatim — never parsed, byte-exact, no pretty_printed field (unchanged from the reverted PR).
• If formatting --raw/--output content fails (non-strict XML), the command falls back to the original content, warns on stderr, and reports pretty_printed: false in --output file metadata (unchanged).

How the new formatter works (and why it needs no XML library)

encoding/xml serves purely as a tokenizer: every token is mapped back to its byte range in the original input via Decoder.InputOffset, and the output is assembled exclusively from verbatim slices of the input plus \n+indent runs inserted between the children of structural elements. Nothing is parsed-and-reserialized, so:

• CDATA sections, whitespace character references in any spelling ( ,  , 	, 
, 
, ...), entity lexical forms, attribute quoting, and in-tag whitespace all survive byte-for-byte by construction;
• the character-reference masking machinery the etree implementation needed is deleted, not ported — ~90 lines of code and a whole failure class gone;
• the full document is decoded before anything is emitted, so malformed input still errors into the existing fallback path — including element-free responses (plain text, comment-only), which the tolerant decoder would otherwise pass through while misreporting pretty_printed: true.

Fidelity evidence

• Every contract test from the reverted PR carries over without weakening a single assertion (golden strings, escaped-reference table incl. hex/zero-padded/CR/LF spellings, leaf whitespace, CDATA, the bare  -between-inline-siblings case, idempotency, malformed rejection, all three output-surface behaviors incl. fallbacks).
• A differential probe against the etree implementation over 53 inputs (including the real ~60KB chart-demo deck) was byte-identical except six cases, each of which now preserves the original bytes more faithfully (e.g. non-whitespace character references keep their lexical form, single-quoted attributes keep their quoting, whitespace-only CDATA between structural children is kept). Each divergence is pinned in TestPrettyPrintXMLPreservesLexicalFormsEtreeChanged with the old etree output recorded alongside.
• New engine-level tests: malformed-input table (incl. late-error no-partial-output), structural table (comments/PIs/prolog/DOCTYPE/CRLF/multi-byte UTF-8/namespace-prefixed p), masking-era placeholder regression, and a real-fixture convergence test (format(minified fixture) == format(shipped fixture), idempotent) plus benchmarks.
• Perf on the 60KB chart demo: ~0.74ms minified / ~0.85ms preformatted per format, ~40% fewer allocations than the etree implementation on identical input.

Test plan

• make unit-test
• go vet ./...
• gofmt -l .
• go mod tidy (removes github.com/beevik/etree; nothing else changes)
• make build
• go run github.com/golangci/golangci-lint/v2/cmd/golangci-lint@v2.1.6 run --new-from-rev=origin/main
• Live verification on a real presentation: envelope content byte-identical to api get raw response (both scopes), --raw/--output identical formatted output, pretty_printed/size metadata correct

Summary by CodeRabbit

• New Features

  • XML returned through --raw or saved with --output is now formatted for easier reading.
  • Output metadata indicates whether formatting succeeded.
  • Formatting failures preserve the original XML and display a warning.

• Documentation

  • Updated command-line help and dry-run messages to describe formatted XML output.

• Bug Fixes

  • Preserved verbatim XML envelope responses and content when formatting is not requested.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: 4ebb6e66-4af1-48dd-b0b4-afc662d36c1e

📥 Commits

Reviewing files that changed from the base of the PR and between af8507e and 7f9c6d9.

📒 Files selected for processing (4)

• shortcuts/slides/slides_xml_get.go
• shortcuts/slides/slides_xml_get_test.go
• shortcuts/slides/slides_xml_prettyprint.go
• shortcuts/slides/slides_xml_prettyprint_test.go

---

📝 Walkthrough

Walkthrough

Slides XML fetching now pretty-prints XML for --raw and --output, preserves envelope content verbatim, and reports formatting fallbacks. A new byte-preserving formatter and extensive tests cover malformed input, lexical preservation, idempotency, and output behavior.

Changes

Slides XML formatting

Layer / File(s)	Summary
Byte-preserving XML formatter shortcuts/slides/slides_xml_prettyprint.go	Strictly tokenizes XML, preserves raw lexical content, and reindents structural elements while keeping text-bearing subtrees verbatim.
SlidesXMLGet output integration shortcuts/slides/slides_xml_get.go	Formats --raw and --output content, records pretty_printed for files, and falls back to original XML with a warning on formatting failure.
Formatter validation shortcuts/slides/slides_xml_prettyprint_test.go	Adds golden, malformed-input, lexical-preservation, idempotency, fixture, and benchmark coverage.
SlidesXMLGet validation shortcuts/slides/slides_xml_get_test.go	Verifies formatted output, verbatim envelopes, fallback behavior, metadata, warnings, and updated flag descriptions.

Estimated code review effort: 4 (Complex) | ~60 minutes

Sequence Diagram(s)

sequenceDiagram
  participant SlidesXMLGet
  participant prettyPrintXMLOrOriginal
  participant Output
  SlidesXMLGet->>prettyPrintXMLOrOriginal: format raw or output XML
  prettyPrintXMLOrOriginal->>Output: return formatted or original XML
  SlidesXMLGet->>Output: write XML and metadata

Loading

Possibly related PRs

• larksuite/cli#1585: Introduced the SlidesXMLGet shortcut and shares the same command area.
• larksuite/cli#1887: Refactored SlidesXMLGet output and envelope behavior used by this formatting change.
• larksuite/cli#2013: Directly conflicts on whether SlidesXMLGet preserves server XML or pretty-prints it.

Suggested labels: feature

Suggested reviewers: ethan-zhx, fangshuyu-768

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title is concise and accurately summarizes the main change: reindenting slides xml-get output with a standard-library formatter.
Description check	✅ Passed	The description covers the summary and test plan well, with detailed implementation notes, though it omits the exact Changes and Related Issues sections.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch codex/cli-xml-pretty-print-stdlib

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[github-actions]

🚀 PR Preview Install Guide

🧰 CLI update

npm i -g https://pkg.pr.new/larksuite/cli/@larksuite/cli@7f9c6d9bfb91f057bbe9f171d95340a6d75b2603

🧩 Skill update

npx skills add larksuite/cli#codex/cli-xml-pretty-print-stdlib -y -g

[codecov]

Codecov Report

❌ Patch coverage is 89.83051% with 12 lines in your changes missing coverage. Please review.
✅ Project coverage is 75.08%. Comparing base (af8507e) to head (7f9c6d9).

Files with missing lines	Patch %	Lines
shortcuts/slides/slides_xml_prettyprint.go	91.34%	4 Missing and 5 partials ⚠️
shortcuts/slides/slides_xml_get.go	78.57%	2 Missing and 1 partial ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #2018      +/-   ##
==========================================
+ Coverage   75.06%   75.08%   +0.01%     
==========================================
  Files         902      903       +1     
  Lines       95944    96057     +113     
==========================================
+ Hits        72025    72129     +104     
- Misses      18380    18384       +4     
- Partials     5539     5544       +5     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.