Fix UTF-8 corruption in HTML serialization - #784
Merged
Merged
Conversation
mrobinson
approved these changes
Sep 13, 2026
mrobinson
enabled auto-merge
September 13, 2026 15:30
Goldziher
added a commit
to xberg-io/html-to-markdown
that referenced
this pull request
Sep 14, 2026
Minor rather than patch: the regen carries source-breaking enum renames in several bindings. Wire values are unchanged throughout, so no serialized payload moves -- only the names consumers spell in source. Java enum constants become SCREAMING_SNAKE (LinkStyle.Inline -> INLINE) across every generated enum, which is the broad one. The RDFa structured-data variant previously generated seven different spellings from one Rust variant; each binding now uses the correct acronym segmentation for its own convention, and Dart keeps rdFa because flutter_rust_bridge declares that name, not alef. Three of these are bug fixes rather than cosmetics. Python's .pyi declared StructuredDataType.RD_FA while the extension declares RDFA, so a type checker accepted the name that raises AttributeError and rejected the one that works. Elixir's VisitResult typedoc named a payload key the NIF struct never had. Ruby's from_hash read `_0` for an adjacently tagged enum whose content key is `output`, silently discarding the payload in both directions. html5ever stays at 0.39.0. 0.40.0's serializer drops the C2 lead byte for every codepoint in U+0080-U+00BF except NBSP, emitting invalid UTF-8; here that makes repair_with_html5ever fail its from_utf8 check and silently skip the repair pass on any document containing a section sign, copyright sign or degree sign. Fixed upstream in servo/html5ever#784, merged but unpublished. Caught by the corpus_fixpoint_sweep ratchet, which went from 8 unexplained documents to 10.
OGKevin
pushed a commit
to OGKevin/html5ever
that referenced
this pull request
Sep 15, 2026
Prepare html5ever patch release for servo#784 Signed-off-by: Simon Sapin <simon@igalia.com>
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
2 participants
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.
Serializing ordinary HTML such as
<p>©</p>produces invalid UTF-8 in 0.40.0:©(C2 A9) becomes a loneA9byte. Attribute values and other characters in U+0080–U+00BF, except NBSP, are also affected. Consumers that validate UTF-8 can therefore reject otherwise valid HTML.The HTML specification's escaping algorithm replaces
&, NBSP,<,>, and double quotes in attribute mode. Other characters must be preserved.The byte scanner introduced in #740 looks for
0xC2to recognize NBSP (C2 A0). When the following byte is notA0, it advances pastC2without writing it. This patch writes that byte before continuing; the next iteration writes the continuation byte. NBSP escaping and the separate raw-text path remain unchanged.Adds text and attribute regressions mixing ©, ®, « and » with NBSP and ampersand escaping. Both tests fail before the fix.
Verification:
html2htmlexample matches independently calculated escaping rules for 198 cases covering all 64 code points U+0080–U+00BF, adjacent escapes, repeated characters, long spans, and wider Unicode; script/style raw text is preserved.Local checks used an untracked lockfile with
jiff 0.2.35: the published0.2.36package fails to compile because included documentation files are missing. No dependency changes are included.