Outline, node-get, write outlines, the render summary and replace-text's
opaque-node match name no library ID unless the library group is visible
and the user has divi_library; read-only still names them.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
One PostIdParser for {{link:}} and select_post; post_link_url_<type> needs a
living, readable, non-attachment target of exactly that type.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Own fixture pod lit_fixture_person (ruling Q1), template and body layout
built through Divi's Theme Builder and Visual Builder UI, brief render with
the template limited to one person (Q5), findings 10-12 in the README.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
- G1: every decoded string value of a Divi 5 opener is scanned for shortcodes;
outside the openers only the remaining content is.
- G2: attributes are parsed like shortcode_parse_atts() and global_module is
cast with (int), as Divi does.
- G3: a non-empty global_module other than "0" that casts to 0 makes Divi
render the newest layout, so it counts for every entry (fail closed).
- The prefilter LIKEs use LOWER(post_content), so they are case-insensitive
under any collation.
- The guide describes the fail-closed rule without an absolute claim.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Candidates come from one ID/status/type query: content with globalModule,
global_module or a JSON-escaped letter. Their content is read in chunks of
50 and parsed by UsageMatcher: Divi 5 openers are json-decoded (escaped keys
and values, leading zeros, any depth), and Divi 4 global_module attributes
are read in any quoting. An undecodable escaped opener counts as a possible
reference. Beyond 5000 candidates the rest count unread. The guide names
the formats and the wp_block parity. Also fixes the grammar of the
blocked-hidden count.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
synthetic-divi-filters.html now stores its paragraph as Divi does (unicode
escapes, bytes built with chr(92) and checked with od). A data-provider test
inserts a section as a user without unfiltered_html into every Divi 5 fixture
page read from disk and asserts that every other byte stays.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Per-run editor login and role, deleted only when this run created them (and
refused when they exist); the allowed-roles option is deleted again when it
was absent; the run-token fallback covers pages too; readOnlyHint must be
false, not just missing.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
The header says "mit OAuth für KI-Assistenten" like the readme; the spec names
the client neutrally and keeps claude.ai only where the connector is meant.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
search uses search_columns post_title; users without edit_others_posts on
et_pb_layout query their own entries only (PostGuard::ownPostsOnly); the
cursor counts the entries shown, never the rows left out.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
StoredNodeGuard accepts a zero-padded library id (Divi intval()s it) but
refuses instances whose opener holds more than one or an escaped
globalModule, so the entry Divi renders is the one checked. GlobalUsages
matches "globalModule":"0*<id>" with REGEXP behind a cheap LIKE prefilter.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Divi renders any referenced et_pb_layout, whatever its scope term. A normal
entry with usages now gets the dry run, confirm, P8a, backup and the
site-wide GlobalItem scope; lit-divi-delete's notice and the library-list
detail name its pages too. The usage scan is the same single query.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
et_header_layout, et_body_layout, et_footer_layout and et_template showing a
global entry now block a confirmed edit unless et_pb_is_allowed('theme_builder'),
like Divi's own check for TB layouts.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
GlobalUsages runs one query per request (ID, status, type – no content),
checks read_post/edit_post in chunks of 200 with a primed and released post
cache, lists at most 50 readable posts and refuses a confirmed global edit
above 2000 live references (unsupported, edit in the Divi builder). Posts the
caller may not read appear only as hidden_count / blocked_hidden_count;
blocked_ids names readable posts only.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
With the library group disabled the entries stay unreadable and unwritable, but
page outline, node-get and render summary still name global module IDs; full
hiding follows in 3b.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
A library entry as source now requires LibraryAccess::canInsert() (add_library,
writable library group) and follows lit-library-insert's Tools-tab state
(disabled -> unsupported). DiviPageLoader::library() gets its caller.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Reuses NewDraft for insert and rollback (P6), requires edit_global_library
for scope global (P7), refuses opaque and global roots before wrapping (P11)
and checks the node with StoredNodeGuard. The source page is never written.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
lottie.innerContent (src → data-lottie-src) makes the frontend load an
animation whose expressions lottie-web runs with eval; it joins the S7 gate
in ValueCleaner, which covers removals and StoredNodeGuard.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
A link without scheme and without whitespace, quotes, <, >, backslashes or
control characters passes although esc_url_raw() would prefix http://; every
value with a scheme, and protocol-relative //host, must survive url() unchanged.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
URLs must be sanitizer-stable, keys inside values follow ValueCleaner::VALUE_KEY,
instances need a digits-only globalModule; state keys reuse AttrWriter::STATES.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Divi's frontend applies the interaction effect "Attribute" with setAttribute
(e.g. an onclick handler) and sets cookies; *.decoration.interactions joins
the S7 gate in ValueCleaner, which also covers removals and StoredNodeGuard.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Refuses, for users without unfiltered_html, non-Divi blocks, raw text parts,
unreadable opaque blocks, code modules, builder-only paths (also below the
state value), strings that are not sanitizer-stable and stored $variable
references that VariableRef::canonical() refuses or would change (P2, P3).
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Created through Divi's own save paths (Visual Builder add-to-library dialog,
divi/v1/divi-library/create-item, divi/v1/sync-to-server, Theme Builder
admin-ajax create_layout/save) and exported by divi-export-stage3-fixtures.php.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Control characters removed by kses (wp_kses_no_null: C0, \\0 sequences)
or ControlChars (DEL) could hide a divi/* delimiter or an et_pb shortcode
from the raw-input check. The shared DiviMarkup helper checks the raw
and the normalized form and now guards every raw-content path:
content-create/-update content and media-update description (attachment
post_content). MediaImporter logs import failures with scheme, host,
port and path only; uninstall.php lists the removed user meta.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01At3V6SjwbMe6eMzVGW3Hs7
- m-1: the no-change Divi result carries preview_url/edit_url too
- m-2: overview names replace-text's per-page tree_rev
- m-3: divi-shortform documents validation_failed vs not_found for unknown IDs
- m-4: DiviWriter/DiviPageLoader final (no mocks left), neutral NewDraft
log detail, snapshots are Etappe 4
- m-5: Tested up to 7.1, changelog completed
- m-6: content-create warns when the extras changed the stored layout
- m-7: deleting a whole section outlines nothing instead of the next one
- m-8: uninstall removes the user meta lit_mcp_log_per_page; readme lists it
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01At3V6SjwbMe6eMzVGW3Hs7
Every file under src/ exits outside WordPress (defined('ABSPATH') ||
exit; after the use block, so a direct request no longer prints a fatal
error with the server path); divi-convert-fixture.php gets the WP_CLI
guard the other integration scripts have. A unit test keeps both true;
PSR1 side-effects warning excluded for the guard.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01At3V6SjwbMe6eMzVGW3Hs7
mcp-oauth has no filter for this, so AuthorizeGate runs on
template_redirect before its router and stops the browser endpoints
(authorize, authorize-callback, consent) for logged-in users without
lit_mcp_use with "Dein Benutzer hat keinen Zugriff auf LIT MCP." (403).
Login redirect, token and revoke stay mcp-oauth's.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01At3V6SjwbMe6eMzVGW3Hs7
PodsFieldSource maps hidden, read_only, admin_only and
restrict_role/-capability like Pods' own view: hidden -> new status
HIDDEN, read_only -> READONLY, restricted and not permitted -> READONLY
with read_only_restricted, else HIDDEN. HIDDEN fields are not described,
read out, queried or written (they look unknown) but stay in the schema
for the media usage search; write() re-checks before Pods' save().
Required-but-restricted fields no longer block a create. Pods is not
active on DDEV: covered by unit tests with a field-option fixture.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01At3V6SjwbMe6eMzVGW3Hs7
PublicAddress blocks every private/reserved IPv4 and IPv6 range
(169.254/16, 100.64/10, fc00::/7, fe80::/10, IPv4-mapped/NAT64/6to4 by
their embedded address, ...) independent of the core version; checked
before the request and on every redirect (Requests before_redirect).
Network failures return only "Die Datei konnte nicht geladen werden.";
the detail goes to the error log. DNS-rebinding pinning stays deferred.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01At3V6SjwbMe6eMzVGW3Hs7
A contact-field item whose type is no allowed child now gets a hint that
type is the node type, with the attrs path (fieldItem.advanced.type) and
the lit-divi-update set alternative, instead of "unbekanntes Modul
divi/email". The metadata fixture gains contact-form/contact-field
(regenerated from Divi 5.2.1, additions only).
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01At3V6SjwbMe6eMzVGW3Hs7
contact-form, cta, toggle, accordion-item, slide, number-counter and
team-member (name) take level like heading/blurb; paths from Divi's
conversion outlines, checked live with divi-shortform-check.php. A test
keeps the guide's field table in sync with ShortformMap.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01At3V6SjwbMe6eMzVGW3Hs7
DiviPage::builderUrl() is shared by DiviPage::editUrl() and the content
composer, which now returns edit_url; content-create without divi keeps
the wp-admin link. divi-write-smoke checks the builder URL and, new in
section 9, the M1/M2 gates live (raw Divi markup refused; Divi read-only
stops content-create divi and execute-ability lit/divi-insert).
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01At3V6SjwbMe6eMzVGW3Hs7
The divi input of lit/content-create is refused with unsupported when
lit/divi-insert is disabled or the Divi group is disabled or read-only
(ExposurePolicy::allows as a cross-group gate, asked at execute time).
content-create without divi keeps working. Adds a registrar test that
read-only write tools are never registered, so execute-ability can't
reach them either.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01At3V6SjwbMe6eMzVGW3Hs7
lit/content-create and lit/content-update refuse content containing the
divi/* block delimiters WordPress' block parser accepts or Divi 4
shortcodes, so the builder-only gates (S7/S26) can't be bypassed. Other
block markup stays allowed (normal kses).
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
The change summary is the wide primary column (key "title", so it cannot
be hidden) with the tool's human label below and row actions "Revision
ansehen" / "Objekt öffnen". Objects get human names ("Seite: Kontakt",
"Menü: …", "Einstellungen: Titel der Website"; deleted/trashed marked),
primed per page with one query each for posts, menus and users. Times are
relative within 24 h. Non-ok results show a badge and a marked row instead
of an Ergebnis column. Page size and columns via Screen Options; the tool
filter shows labels grouped by integration, dates have visible labels,
and an active filter offers "Filter zurücksetzen" (also in the empty state).
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01At3V6SjwbMe6eMzVGW3Hs7
MenuWriter::fail guards wp_delete_post for created items: the touched items
are still restored and the client gets partial_write (with previous_items)
instead of internal_error.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01At3V6SjwbMe6eMzVGW3Hs7
New rows no longer accept 4_5 (the builder does not offer it). On update,
the flexType of every non-desktop breakpoint (tablet, phone, …) is
format-checked against the builder list like desktop. Guide updated.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01At3V6SjwbMe6eMzVGW3Hs7
PostChangeResult is PostSaveFollowUp (the after-save pipeline, not a
result); DiviChange::changed() is record() (moved nodes are recorded but
not touched). cleanValue/withoutControls were renamed in wave 2.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01At3V6SjwbMe6eMzVGW3Hs7
Composer::autoLabel is private (tested through compose()); the
DiviComposition getters used only by tests are marked @internal;
Limits::snapshotCount is documented as reserved; the ShortformWriter test
seam is named $modulesForTests.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01At3V6SjwbMe6eMzVGW3Hs7
Integrations name their own admin group (Integration::label, registry
labels()); ToolsTab no longer hard-codes ids and explains why it re-applies
withGroups. AccessTab and LogTab share AdminPostAction (guard, redirect,
terminate seam). ExposurePolicy parses option keys in one place, has one
isRequired() check and a clear $groups docblock (null vs []).
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01At3V6SjwbMe6eMzVGW3Hs7
The block/ID/file-URL matching moves out of MediaUsageFinder into a static
class; the finder keeps queries, limits and result building. Behaviour
unchanged; direct tests for the matcher added.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01At3V6SjwbMe6eMzVGW3Hs7
One test class per Divi write tool on a shared DiviWriteAbilityTestCase;
the 30 tests move verbatim (per-test assertion counts identical). Adds the
ledger T4 test for a TreeEditor::move() StructureRules refusal.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01At3V6SjwbMe6eMzVGW3Hs7
The SQL prefilter and the per-page hit computation move to DiviTextSearch;
DiviPageLoader::loadWritable serves DiviWriter::page and the dry run;
PostStatuses::NOT_CONTENT replaces the duplicated status list (same set,
order in the SQL list only). The private log() wrapper is inlined.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01At3V6SjwbMe6eMzVGW3Hs7
Block::isPlaceholder replaces the divi/placeholder literals (ledger T4);
NodeIndex::parentBlock/sectionLabel replace the two parentBlock copies and
DiviWriter::sectionLabel; DiviChange::idOf and Outline::summary replace the
duplicated id resolution and outline text; DiviSchema holds the shared
input properties. Behaviour unchanged.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01At3V6SjwbMe6eMzVGW3Hs7
The five Divi write tools extend DiviWriteAbility (final permission and
execute, change() per tool, shared output schema). DiviErrors holds the
unknown-node, position and compose-error mapping; DiviContentComposer uses
the same mapping (ledger T9). ToolNamingTest skips abstract classes.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01At3V6SjwbMe6eMzVGW3Hs7
Composer::row uses ColumnWidths::parse; rowWidthsError and its error-state
save/restore are gone. The update width check moves out of
DiviUpdateAbility into ColumnWidths::updateErrors. Behaviour unchanged.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01At3V6SjwbMe6eMzVGW3Hs7
writeContent($postId, $content, ?$before) holds the kses window plus
wp_update_post(wp_slash([ID, post_content]), true); save() passes the restore
point capture as $before, so the window still holds exactly RevisionPoint::
capture() and wp_update_post() (S37), saveNew() and the restore() rollback use it
without. refusal() turns WP_Error / 0 into the message ("returned 0" as before)
for save() and saveNew(); restore() keeps its own RuntimeException text. The
class docblock (the M-5 window description) is unchanged.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01At3V6SjwbMe6eMzVGW3Hs7
- HtmlTokens (static, with the S45 docblock): SPLIT, RAW_OPEN, tokenize(),
structure(), closes(), unterminated() moved unchanged; new RAW_ELEMENTS list.
- StoredForms: of() (was TextReplacer::forms), mentions(), unicodeEscaped() –
the DB prefilter forms; TextReplacer::mentions() delegates, the ability's
candidate query uses StoredForms::of().
- TextReplacer keeps problems(), apply() and the splice/safe() verification;
safe() still runs the tokenizer and HtmlContext as two independent scans, in
the same order.
HtmlTokensTest asserts RAW_ELEMENTS === HtmlContext::RAW and that RAW_OPEN opens
exactly those elements. TextReplacerTest and HtmlContextTest unchanged.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01At3V6SjwbMe6eMzVGW3Hs7
enum ValueKind: string { Html, Code, Url, Text } with strictness() and
tightens(self $routed) (the S33a rule formerly AttrWriter/ValueCleaner::tightens
over STRICTNESS). ValueCleaner::clean() takes ?ValueKind, kindOf()/kind()/
unknownKind() return it, the sanitizer match is exhaustive over the four cases;
the "Feldart" message prints ->value, so its text is unchanged. ShortformWriter
drops the identity map CLEAN_KIND for ValueKind::tryFrom($spec['kind']) ??
ValueKind::Text; Composer and ShortformWriter pass ValueKind::Text/Url.
The installed PHPCompatibility sniffs do not know enums: two sniffs are disabled
in ValueKind.php only, with the reason.
Tests: ValueCleanerTest passes/asserts ValueKind cases instead of strings;
+ValueKindTest (full tightens matrix, case values).
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01At3V6SjwbMe6eMzVGW3Hs7
The choke point moves out of AttrWriter into ValueCleaner (catalog, the
composition's VariableRef, sanitizer): clean() (was AttrWriter::cleanValue),
kindOf(), guardPath() and the private cleanNode/kind/unknownKind/tightens with
VALUE_KEY, URL_KEY, URL_SEGMENT, STRICTNESS, BUILDER_ONLY (its docblock now sits
on it; STRUCTURE stays in AttrWriter). Code of the pipeline moved unchanged.
- ValueCleaner's docblock is the one canonical description: steps 0-8 with their
rulings in code order, invariants I1-I3; the step comments in cleanNode carry
the same numbers. AttrWriter/ShortformWriter/Composer/TextReplacer docblocks
point there and keep only their own rules.
- ValueCleaner::textReplacementFor($module) holds the D10 rule that was a
closure in DiviReplaceTextAbility::replace (text/html only, clean() by kind).
- AttrWriter, ShortformWriter and Composer take the ValueCleaner; DiviComposition
builds one per composition (cleaner()).
Tests: the eight AttrWriterTest methods that only exercised cleanValue/kindOf
moved verbatim to ValueCleanerTest (clean/kindOf, ComposeFixture::cleaner());
mixed tests call the cleaner in place; +1 test for textReplacementFor. Every
provider row and per-test assertion count unchanged.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01At3V6SjwbMe6eMzVGW3Hs7
- DiviComposition::references() builds the one VariableRef with the entry's
live map (S22/S31) and passes it to AttrWriter; Placeholders only needs the
design system now, Placeholders::references() is gone (no reach-through).
- DesignSystem::colorIdFor()/variableFor() replace the dead colorId()/variable()
and carry Placeholders' lookup incl. the S13a ambiguity check (byLabel moved,
messages unchanged); isColorVariable() decides color vs content for {{var:}}.
- DesignSystem::colorReferenceIds() (VariableRef color names) and
contentVariableIds() (DynamicContentRules design names) replace the two
inline computations; same lists, same order.
Tests: fixtures/tests get the VariableRef from ComposeFixture::references() or
build it with the map; DesignSystemTest uses the new lookups (+2 tests).
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01At3V6SjwbMe6eMzVGW3Hs7
- ControlChars::strip()/stripDeep() (was VariableRef::withoutControls/Deep) with
the S32/S32a/S35 docblock; callers AttrWriter, ShortformWriter, TextReplacer and
the setting cleaner use it.
- DynamicContentRules (design system + dynamic content map): fieldsFor() (was
contentFields: meta prefix -> map -> loop_ -> design names -> unknown),
cleanSetting() (was setting(): raw strip -> raw marker -> field type ->
final-string check -> meta key check, same order), one private method per
field family (textSetting, selectPostSetting, scalarSetting), isMetaKey(),
checkMap(), and the constants DEFAULT_DYNAMIC_CONTENT, META_SETTINGS,
DESIGN_SETTINGS, META_KEY.
- VariableRef keeps the syntax: const END = ')$' replaces the six literals,
payloadOf() is shared by typeOf() and canonical(); unknownName() is public for
the rules. Error messages unchanged.
Tests unchanged apart from a class name in DynamicContentMapTest.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01At3V6SjwbMe6eMzVGW3Hs7
Without /D, "$" also matches before a trailing "\n", so "custom_meta_telefon\n",
a color setting "5\n" (stored as sent) or a whole reference followed by "\n"
passed VariableRef. Added /D to every ^...$ pattern that validates client or
admin input:
- VariableRef META_KEY, NUMBER, WHOLE (canonical/typeOf; Placeholders::field
shares META_KEY)
- Cursor::decode, GuideRepository topic names (MCP input and filter keys)
- ExposurePolicy::sanitize ability keys and group ids, LogTab list filters
Not changed: Placeholders::PATTERN (runs on trim()), Warnings hex checks (hints
only), NodeSummary/AttrPathIndex/LogListTable (stored or Divi data, no client
input). Each fixed pattern has a test with a trailing "\n" that is rejected.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01At3V6SjwbMe6eMzVGW3Hs7
tests/Support/AssertsToolErrors (used by TestCase): assertToolError($result, code)
checks the WP_Error code lit_<code> and the code of the client JSON and returns the
payload; clientError() decodes the payload; errorFields() returns the fields and
asserts they exist. Replaces the private clientError() decoders of
DiviWriteAbilitiesTest/DiviReplaceTextAbilityTest, 216 single-line
assertSame('lit_…', …->get_error_code()) assertions, 24 get_error_data()['fields']
digs and two json_decode(message) reads. ErrorsTest keeps json_decode: it tests
the format itself. Tests 1404 -> 1409 (AssertsToolErrorsTest); no test has fewer
assertions.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
- tests/Support/FakeChangeLog extends ChangeLog: records record()/recordPartial()
calls, increasing ids, optional failure; new FakeChangeLog(id) / partialsOnly() /
any(id) / forbidden() mirror what the replaced mocks allowed. assertRecorded() /
assertPartials() check count and the named arguments. Calls a fake does not allow
throw and are also verified in TestCase::tearDown (one assertion per test that
uses a fake), so swallowed change-log calls still fail like a never() expectation.
- All ChangeLog mocks of ability/integration tests replaced (the admin tests that
stub query/purge/clear keep their mocks); expectException tests that checked the
partial log now catch and assert.
- SchemaFixture::managed(?divi) replaces the 21 isManaged-only ContentSchema mocks;
DiviFixtures::schema() uses it.
Tests 1397 -> 1404 (FakeChangeLogTest); no test has fewer assertions.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
- DiviFixtures: markup(), givenPost(), schema(), designSystem(), catalog(),
dynamicContentRegistry(), maliciousRefs() - moved verbatim out of
DiviPageLoaderTest, DesignSystemTest, ModuleCatalogTest, DynamicContentMapTest
and VariableRefTest (its malicious() provider now returns maliciousRefs()).
- ComposeFixture moves to tests/Support.
- DiviWriterFixture: modules(), designReader(), composition(), a recording saver(),
log() and writer(&$saved, &$failures, unfiltered, loader) for the five tests that
wired the Divi write stack by hand.
No test cases or assertions changed (1397 / 5834).
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
NewDraft owns insert (partial write on a throwing hook), refuse and discard (via
Rollback::attempt and WriteFailed::*) and the partial logging; ContentCreateAbility
keeps checkDivi/prepare/combined (Errors::fieldsOf) and the result. Both discard
callers still pass REASON_EXTRAS. Behaviour unchanged; NewDraftTest adds the
throwing-delete branch.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
- PostRow::update(id, columns, failure) / ::save(): the slashed wp_update_post of
internal row writes, refusals as RuntimeException '<failure>: <why>'. Used by
ContentUpdate::restoreRow, PostCloner::updateRow, MediaMetaWriter apply/restore.
- PostGuard::assertNoMissedSchedule(postId, message) replaces the re-read + check in
ContentUpdate::restoreRow, PostCloner and DiviSaver::restore (DiviSaver keeps its
own kses window; M4).
- PostCloner::restore uses Rollback::attempt.
Log messages unchanged.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
