Repository navigation
fix: add a server-side path from WP block markup to rendered HTML - #689
Merged
ViewFromTheBox merged 2 commits intoAug 2, 2026
Merged
ViewFromTheBox merged 2 commits into
ViewFromTheBox merged 2 commits into
Conversation
Block themes ship their templates as WP block markup, but nothing on
the server bridged cms-framework's parse_blocks() shape
({blockName, attrs, innerHTML}) to the shape BlockRenderer consumes
({name, attributes}). A key rename alone was not enough: Gutenberg
persists most block text in the SAVED HTML rather than the delimiter
JSON — zero of the artisanpack-ui theme's 130 paragraph/heading blocks
serialize a `content` attribute — so renaming keys produced a
structurally correct but completely textless page.
Adds BlockMarkupHydrator, which parses the markup AND replays each
registered block type's block.json attribute definitions back over its
saved HTML to recover content. Recovery is driven entirely by the
registry rather than a hand-maintained per-block table, so a block that
ships a new sourced attribute is picked up with no change here — which
is why this lives beside the block manifests and partials rather than in
cms-framework or a host app.
- BlockMarkupHydrator: hydrate(markup) and hydrateTree(parsedTree),
with a depth cap and a documented trust boundary
- BlockAttributeSourceResolver: server-side port of Gutenberg's
attribute/html/rich-text/text/tag/query/raw matchers over DOM + XPath
- CssSelectorToXPath: hand-rolled translator for the CSS subset the
block manifests use, avoiding a new Composer dependency
- BlockRenderer::renderMarkup(): markup in, HTML out, with
template-part and pattern inlining so a standalone theme template
renders its parts as real content instead of empty wrappers
- TemplatePartInliner: hydrate a part's raw markup when it resolves
with empty blocks, which is the .html-sourced theme-file case
Also drops the leading backslash from TemplatePartInliner's
RESOLVER_CLASS: with it, app() treated the string as a distinct
container key and silently ignored any host- or test-supplied override.
Closes #688
|
Warning Review limit reached
Next review available in: 48 minutes Enable usage-based reviews in Billing to review now. Otherwise, wait until the next included review is available. How can I continue?After more reviews become available, a review can be triggered using the To avoid repeated limits, reduce automatic review volume by pausing incremental auto-reviews earlier, using label-based review opt-in, excluding WIP or generated PR titles, or requesting reviews manually when the PR is ready. If your team needs uninterrupted high-volume reviews, an organization admin can enable usage-based reviews. How do review limits work?CodeRabbit enforces per-developer PR review limits for each organization. Most developers receive the normal plan review availability. For paid Pro and Pro+ PR reviews, CodeRabbit uses adaptive limits for sustained high-volume activity. When a developer's recent PR review activity reaches the 95th percentile or higher among CodeRabbit users, additional reviews become available more gradually as earlier reviews age out of the rolling window. Please refer docs for additional details. Review details⚙️ Run configurationConfiguration used: Path: .coderabbit.yaml Review profile: CHILL Plan: Pro Plus Run ID: 📒 Files selected for processing (3)
📝 WalkthroughWalkthroughThis change adds a server-side path from WordPress block markup to rendered HTML. It recovers block attributes from saved HTML, hydrates nested trees, inlines raw template parts and patterns, and exposes the flow through ChangesMarkup rendering
Estimated code review effort: 4 (Complex) | ~60 minutes Sequence Diagram(s)sequenceDiagram
participant Caller
participant BlockRenderer
participant BlockMarkupHydrator
participant TemplatePartInliner
participant BladeRenderer
Caller->>BlockRenderer: renderMarkup(markup)
BlockRenderer->>BlockMarkupHydrator: hydrate(markup)
BlockMarkupHydrator-->>BlockRenderer: hydrated block tree
BlockRenderer->>TemplatePartInliner: inline template parts
TemplatePartInliner-->>BlockRenderer: resolved block tree
BlockRenderer->>BladeRenderer: render(tree)
BladeRenderer-->>Caller: rendered HTML
Possibly related PRs
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
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. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 2
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@src/Support/BlockMarkupHydrator.php`:
- Around line 259-299: Update hydrateBlock() to normalize core/* block names to
their registered artisanpack/* alias after attribute recovery, reusing
aliasFor() and preserving names already registered under artisanpack/*. Ensure
the normalized name is used for partial resolution and block rendering so
recovered attributes and content-only core blocks match the expected theme
partials.
In `@tests/Unit/VisualEditor/Resources/TemplatePartInlinerRawContentTest.php`:
- Around line 92-100: Rename the test case around TemplatePartInliner so its
description reflects the existing assertion that blank raw markup produces empty
innerBlocks, rather than claiming the part is marked unresolved. Keep the test
setup and assertions unchanged.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro Plus
Run ID: 4bf3ed73-5a9d-437b-8095-3966c16a088e
📒 Files selected for processing (11)
packages/visual-editor-renderer-blade/src/BlockRenderer.phppackages/visual-editor-renderer-blade/tests/Feature/RenderMarkupTest.phpsrc/Resources/TemplatePartInliner.phpsrc/Support/BlockAttributeSourceResolver.phpsrc/Support/BlockMarkupHydrator.phpsrc/Support/CssSelectorToXPath.phpsrc/Support/UnsupportedSelectorException.phpsrc/VisualEditorServiceProvider.phptests/Unit/VisualEditor/Resources/TemplatePartInlinerRawContentTest.phptests/Unit/VisualEditor/Support/BlockMarkupHydratorTest.phptests/Unit/VisualEditor/Support/CssSelectorToXPathTest.php
Comment on lines
+259
to
+299
| protected function recoverAttributes( string $name, string $innerHtml ): array | ||
| { | ||
| if ( '' === trim( $innerHtml ) ) { | ||
| return []; | ||
| } | ||
|
|
||
| $definition = $this->blockTypes->get( $name ); | ||
|
|
||
| if ( null === $definition ) { | ||
| $alias = $this->aliasFor( $name ); | ||
|
|
||
| $definition = null !== $alias ? $this->blockTypes->get( $alias ) : null; | ||
| } | ||
|
|
||
| $attributes = $definition['attributes'] ?? null; | ||
|
|
||
| if ( ! is_array( $attributes ) || [] === $attributes ) { | ||
| return []; | ||
| } | ||
|
|
||
| return $this->sources->recover( $attributes, $innerHtml ); | ||
| } | ||
|
|
||
| /** | ||
| * Map `core/x` ↔ `artisanpack/x`, or null when the name is in | ||
| * neither namespace. | ||
| * | ||
| * @since 1.5.5 | ||
| */ | ||
| protected function aliasFor( string $name ): ?string | ||
| { | ||
| if ( str_starts_with( $name, 'core/' ) ) { | ||
| return 'artisanpack/' . substr( $name, strlen( 'core/' ) ); | ||
| } | ||
|
|
||
| if ( str_starts_with( $name, 'artisanpack/' ) ) { | ||
| return 'core/' . substr( $name, strlen( 'artisanpack/' ) ); | ||
| } | ||
|
|
||
| return null; | ||
| } |
There was a problem hiding this comment.
🗄️ Data Integrity & Integration | 🟠 Major | ⚡ Quick win
🧩 Analysis chain
🏁 Script executed:
#!/bin/bash
# Compare available block partials for the core and artisanpack namespaces.
set -euo pipefail
fd -t d 'blocks' packages/visual-editor-renderer-blade/resources/views
fd -t f -e blade.php . packages/visual-editor-renderer-blade/resources/views | sort
rg -n --glob '*.blade.php' -l 'wp-block-paragraph' packages/visual-editor-renderer-blade/resources/viewsRepository: ArtisanPack-UI/visual-editor
Length of output: 14842
🏁 Script executed:
#!/bin/bash
set -euo pipefail
echo "== Block partial namespace counts =="
find packages/visual-editor-renderer-blade/resources/views/blocks -maxdepth 2 -type f -name '*.blade.php' | sort | awk -F/ '{n=$6; c[n]++} END {for (x in c) print x ": " c[x]}' | sort
echo
echo "== Block type declaration references in block-types / definition files =="
git ls-files | rg '(^|/)(block-types|.*BlockTypes.*|.*Blocks.*\.(php|json|js|ts)|.*yaml$|.*yml$)' | head -200
rg -n "artisanpack/|core/|paragraph|audio|avatar|navigation" --glob '*.{php,json,js,ts,yaml,yml}' .
echo
echo "== Partial missing-core vs artisanpack comparison =="
python3 - <<'PY'
from pathlib import Path
import re
root = Path('packages/visual-editor-renderer-blade/resources/views/blocks')
namespaces = sorted({p/'' for ns,p in {part.parents[1]: ns for ns,part}})
for root, _dirs, files in Path('packages/visual-editor-renderer-blade/resources/views/blocks').walk():
pass
PY
python3 - <<'PY'
from pathlib import Path
p = Path('packages/visual-editor-renderer-blade/resources/views/blocks')
namespaces = sorted({d.name for d in p.iterdir() if d.is_dir()})
print(namespaces)
for ns in namespaces:
files = sorted(f.name.split('.')[0] for f in (p/ns).glob('*.blade.php'))
print(ns, len(files), files[:10], '...', file=__import__('sys').stderr)
PYRepository: ArtisanPack-UI/visual-editor
Length of output: 50385
🏁 Script executed:
#!/bin/bash
set -euo pipefail
echo "== Block partial namespace counts =="
find packages/visual-editor-renderer-blade/resources/views/blocks/*/blocks -maxdepth 1 -type f -name '*.blade.php' 2>/dev/null || true
find packages/visual-editor-renderer-blade/resources/views/blocks -maxdepth 1 -type f -name '*.blade.php' | sort | awk -F/ '{ns=$6; c[ns]++} END {for (x in c) print x ": " c[x]}' | sort
echo
echo "== Files parsed by PHP that include a core block partial =="
python3 - <<'PY'
import ast, pathlib
root = pathlib.Path('packages/visual-editor-renderer-blade/resources/views/blocks')
patterns = []
for f in root.rglob('*.php'):
text = f.read_text(errors='ignore')
if '`@include`(' in text:
patterns.append((f, text))
count = 0
for f, text in patterns:
if 'blocks.core.' in text or '`@include`' in text:
count += 1
print(f)
print("includes_with_core_partial_or_include=", count)
for f, text in patterns[:5]:
print("---", f)
for i, line in enumerate(text.splitlines(), 1):
if '`@include`(' in line or 'Delegates' in line:
print(i, line)
PY
echo
echo "== Hydrator method outline and hydrateBlock section =="
ast-grep outline src/Support/BlockMarkupHydrator.php --view compact || true
sed -n '1,330p' src/Support/BlockMarkupHydrator.phpRepository: ArtisanPack-UI/visual-editor
Length of output: 16423
Normalize the hydrated block name back to the registered alias.
recoverAttributes() loads artisanpack/* attribute definitions for core/* nodes, but hydrateBlock() keeps the original core/* name. Block renders still depend on artisanpack/* partials and core content-only blocks would not match a blocks.core.* partial, so core-namespaced theme markup can render into an unknown-block wrapper and lose recovered attributes/content.
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@src/Support/BlockMarkupHydrator.php` around lines 259 - 299, Update
hydrateBlock() to normalize core/* block names to their registered artisanpack/*
alias after attribute recovery, reusing aliasFor() and preserving names already
registered under artisanpack/*. Ensure the normalized name is used for partial
resolution and block rendering so recovered attributes and content-only core
blocks match the expected theme partials.
Comment on lines
+92
to
+100
| it( 'marks the part unresolved when the raw markup hydrates to nothing', function () { | ||
| bindPartResolver( themeFilePart( ' ' ) ); | ||
|
|
||
| $tree = ( new TemplatePartInliner() )->inline( [ | ||
| [ 'name' => 'core/template-part', 'attributes' => [ 'slug' => 'header' ], 'innerBlocks' => [] ], | ||
| ] ); | ||
|
|
||
| expect( $tree[0]['innerBlocks'] )->toBe( [] ); | ||
| } ); |
There was a problem hiding this comment.
📐 Maintainability & Code Quality | 🟡 Minor | ⚡ Quick win
Rename the test to match what it asserts.
The test name says the part is marked unresolved. The assertion only checks that innerBlocks is empty. findPartBlocks() returns the empty $blocks array on line 242 of src/Resources/TemplatePartInliner.php, not null, so resolvePart() never stamps _resolutionError for blank raw markup. The current name documents a guarantee the code does not provide.
Rename the test, or add the _resolutionError assertion and change findPartBlocks() to return null when the raw markup hydrates to nothing.
💚 Proposed rename
-it( 'marks the part unresolved when the raw markup hydrates to nothing', function () {
+it( 'inlines no children when the raw markup hydrates to nothing', function () {
bindPartResolver( themeFilePart( ' ' ) );📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
Suggested change
| it( 'marks the part unresolved when the raw markup hydrates to nothing', function () { | |
| bindPartResolver( themeFilePart( ' ' ) ); | |
| $tree = ( new TemplatePartInliner() )->inline( [ | |
| [ 'name' => 'core/template-part', 'attributes' => [ 'slug' => 'header' ], 'innerBlocks' => [] ], | |
| ] ); | |
| expect( $tree[0]['innerBlocks'] )->toBe( [] ); | |
| } ); | |
| it( 'inlines no children when the raw markup hydrates to nothing', function () { | |
| bindPartResolver( themeFilePart( ' ' ) ); | |
| $tree = ( new TemplatePartInliner() )->inline( [ | |
| [ 'name' => 'core/template-part', 'attributes' => [ 'slug' => 'header' ], 'innerBlocks' => [] ], | |
| ] ); | |
| expect( $tree[0]['innerBlocks'] )->toBe( [] ); | |
| } ); |
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@tests/Unit/VisualEditor/Resources/TemplatePartInlinerRawContentTest.php`
around lines 92 - 100, Rename the test case around TemplatePartInliner so its
description reflects the existing assertion that blank raw markup produces empty
innerBlocks, rather than claiming the part is marked unresolved. Keep the test
setup and assertions unchanged.
The PHP 8.2 CI leg installs cms-framework from Packagist, and 2.5+ — which introduced BlockMarkupParser — requires PHP 8.3+. So on 8.2 the parser is absent, hydrate() correctly degrades to an empty tree by design, and the 23 tests that assert real recovery failed. Gates exactly those 23 behind BlockMarkupHydrator::canParseMarkup(), matching the existing `templatePartParserAvailable()` convention in TemplateAdapterTest. Tests that exercise hydrateTree() directly, the CSS->XPath translator, and the blank-input paths stay unskipped, since none of them need the parser and all of them work on 8.2.
ViewFromTheBox
marked this pull request as ready for review
August 2, 2026 22:34
github-project-automation
Bot
moved this from In Review
to Done
in ArtisanPack UI Overview
ViewFromTheBox
deleted the
bugfix/688-server-side-wp-block-markup-to-html
branch
August 2, 2026 22:36
This was referenced Aug 2, 2026
EntitySearchController resolves container bindings with a leading backslash, ignoring overrides
#692
Closed
Merged
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
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.
Description
There was no server-side path from WP-serialized block markup to rendered HTML, which made a block theme's
templates/*.htmlandparts/*.htmlunrenderable outside the editor.The gap was not just a shape mismatch. Gutenberg persists most block text in the saved HTML, not in the delimiter JSON — 0 of the
artisanpack-uitheme's 130 paragraph/heading blocks serialize acontentattribute. So a key-rename conversion produces a structurally correct, completely textless page.This adds
BlockMarkupHydrator, which parses the markup and replays each registered block type'sblock.jsonattribute definitions back over its saved HTML to recover the content. Recovery is driven entirely by the block-type registry rather than a hand-maintained per-block table, so a block that ships a new sourced attribute is picked up with no change here — which is exactly why this lives beside the block manifests and partials rather than in cms-framework or a host app.Closes: #688
Type of Change
Related Issue
Issue: #688
Companion (not blocking, see below): ArtisanPack-UI/cms-framework#274
Motivation and Context
Reproductions from the issue, all of which previously produced empty or no output, now render:
Changes Made
BlockMarkupHydrator— the public entry point.hydrate( string $markup )for raw markup;hydrateTree( array $parsed )for callers that already haveparse_blocks()output (accepts the editor shape too, so it is safe over a mixed tree). Carries a depth cap and a documented trust boundary.BlockAttributeSourceResolver— server-side port of Gutenberg's save-shape matchers (attribute,htmlincl.multiline,rich-text,text,tag,query,raw) overDOMDocument+ XPath.CssSelectorToXPath— hand-rolled translator for the CSS subset the manifests actually use, so the package gains no new Composer dependency for a feature this narrow. Unsupported selectors throw and are reported once per process rather than silently mis-matching.BlockRenderer::renderMarkup( $markup, $defaultTheme )— markup in, HTML out. Runs template-part and pattern inlining before the walk so a standalone theme template renders its parts as real content rather than empty wrappers.TemplatePartInliner::findPartBlocks()— falls back to hydrating the part's raw markup when it resolves with emptyblocks, which is the.html-sourced theme-file case. This closes the second impact bullet in No server-side path from WP block markup to rendered HTML (blocks block-theme rendering) #688 without waiting on cms-framework#274.TemplatePartInliner::RESOLVER_CLASS: with it,app()treated the string as a distinct container key and silently ignored any host- or test-supplied override.Deliberate design decisions
block.jsondefault, so the dangerous shape (a delimiter carrying"content":""over populated HTML) is not reachable from editor-produced markup.innerHTMLis preserved on the hydrated block. A block whose content lives only ininnerHTMLand which declares no sourced attributes (core/html) would otherwise hydrate to an empty node, silently losing content.renderMarkup()cannot resolve entity-dependent blocks (post-*,core/query, comments, breadcrumbs) because astringinput carries no post. For a full-fidelity render, hydrate to a tree and hand it to<x-ve-blocks :tree :post />, which runs the complete pipeline. Documented on the method.How Has This Been Tested?
Testing Environment:
release/1.5.5Tests Performed:
src/alt/caption/href, multiline list values, quote value + citation, boolean attribute presence semantics,querysources (table cells), multibyte round-trip, inner-block recursion, and the depth cap.listwhose items arelist-iteminner blocks must leavevaluesempty, or the partial would double-render every item.block.jsonmanifests, so a future manifest introducing an unsupported selector fails CI instead of degrading silently at render time..html-sourced part that resolves with emptyblocks.Accessibility Tests Run
Details: Not applicable — this change is entirely server-side block-tree construction and adds no UI, markup structure, or interactive surface. The HTML emitted is produced by the pre-existing block partials, unchanged by this PR; the change only means those partials now receive the text they were always meant to render. If anything it is an a11y improvement: headings and link text that previously rendered empty now carry their real content.
Tests Added
Test details: 43 new tests across
tests/Unit/VisualEditor/Support/BlockMarkupHydratorTest.php,tests/Unit/VisualEditor/Support/CssSelectorToXPathTest.php,tests/Unit/VisualEditor/Resources/TemplatePartInlinerRawContentTest.php, andpackages/visual-editor-renderer-blade/tests/Feature/RenderMarkupTest.php.Documentation
Documentation details: Each new class carries a full docblock covering the two block shapes, why recovery is registry-driven, the supported CSS subset, and an explicit trust boundary note: recovered rich-text is emitted unescaped by the partials, so the hydrator is only safe over markup already trusted to render (theme files, patterns, editor-authored content) and is not a sanitizer. Wiki/API docs are worth a follow-up once a host wires
renderMarkup()to a route.Screenshots
n/a — server-side rendering change.
Pre-Submission Checklist
Reviewer notes
Security review: run over the branch — no findings at confidence ≥ 8. XXE was cleared (
loadHTMLuses libxml's HTML parser, which does not resolve external entities;LIBXML_NONETset,LIBXML_NOENTcorrectly not set). XPath injection cleared (selectors come only fromblock.jsonon disk; all values quoted, all names charset-constrained). The XSS trust boundary was traced and does not widen: the only wired path reads theme files on disk, already at PHP-execution trust.Known gaps, deliberately out of scope:
core/htmlhas no manifest or partial, so it hydrates with itsinnerHTMLpreserved but still renders via the unknown-block fallback. Worth a follow-up issue.artisanpack/marqueedeclares sourced attributes but is absent from the service provider's registration lists, so they are never recovered — pre-existing, unrelated to this change.EntitySearchController:93-100has the same leading-backslash container-key bug fixed here inTemplatePartInliner; left alone to keep this diff scoped.Pre-existing CI note:
resources/js/visual-editor/site-editor/__tests__/site-editor-app.test.tsxhas 18 failing tests. Verified these fail identically on a cleanrelease/1.5.5with this work stashed; this branch touches zero.tsxfiles.Summary by CodeRabbit
New Features
Bug Fixes