Align html5ever rcdom dependencies (#278)

Cargo.toml

@@ -22,8 +22,10 @@ clap = { version = "4", features = ["derive"] }
 regex = "1"
 once_cell = "1"
 rayon = "1.11"
-html5ever = "0.27"
-markup5ever_rcdom = "0.3"
+html5ever = "0.39.0"
+# `markup5ever_rcdom` must stay on the same parser stack as `html5ever`
+# because `RcDom` implements `TreeSink` from that shared `markup5ever` line.
+markup5ever_rcdom = "0.39.0"
 textwrap = "0.16.2"
 unicode-width = "0.2"
 
@@ -35,6 +37,7 @@ insta = "1.47"
 tempfile = "3"
 libc = "0.2.174"
 predicates = "3"
+trybuild = "1"
 
 [lints.clippy]
 pedantic = "warn"

docs/architecture.md

@@ -257,7 +257,10 @@ immediately follows an H2 heading when these conditions are met.
 
 `mdtablefix` can format simple HTML `<table>` elements embedded in Markdown.
 These HTML tables are transformed into Markdown before the main table reflow
-logic runs. That preprocessing is handled by the `convert_html_tables` function.
+logic runs. That preprocessing is handled by the `convert_html_tables`
+function. The parser path uses `html5ever` to build a temporary `RcDom` tree
+through `markup5ever_rcdom`, then walks that tree to extract table rows and
+cells.
 
 Only straightforward tables with `<tr>`, `<th>` and `<td>` tags are detected.
 Attributes and tag casing are ignored, and complex nested or styled tables are

docs/developers-guide.md

@@ -82,6 +82,18 @@ restores the separator row with widths derived from the final table body.
 - `format_separator_cells`: Expands separator cells to the target widths while
   preserving Markdown alignment markers.
 
+## HTML parser dependency coupling
+
+HTML table conversion uses `html5ever` for parsing and `markup5ever_rcdom` for
+the temporary DOM sink. These crates must stay on the same `markup5ever` parser
+stack because `RcDom` implements the `TreeSink` trait from that shared
+dependency line. If `html5ever` is upgraded, update `markup5ever_rcdom` in the
+same change and run the compile-time parser integration test before merging.
+
+The manifest uses caret requirements rather than exact pins, so compatible
+patch updates remain available. The lockfile records the concrete crate release
+selected for the branch.
+
 ## Fence normalization module
 
 `src/fences.rs` exposes the preprocessing helpers used by the `--fences` option.

tests/breaks.rs

@@ -4,11 +4,12 @@
 
 use std::borrow::Cow;
 
+use assert_cmd::Command;
 use mdtablefix::{THEMATIC_BREAK_LEN, format_breaks};
 
 #[macro_use]
-mod prelude;
-use prelude::*;
+#[path = "common/mod.rs"]
+mod common;
 
 #[test]
 fn test_format_breaks_basic() {

tests/cli.rs

@@ -11,12 +11,14 @@ use std::{
     io::Write,
 };
 
+use assert_cmd::Command;
 use rstest::rstest;
 use tempfile::tempdir;
 
 #[macro_use]
-mod prelude;
-use prelude::*;
+#[path = "common/mod.rs"]
+mod common;
+use common::broken_table;
 
 /// Verifies that the CLI fails when the `--in-place` flag is used without specifying a file.
 ///

tests/cli_fences.rs

@@ -1,8 +1,9 @@
 //! CLI regression tests for fence normalization edge cases.
 
 #[macro_use]
-mod prelude;
-use prelude::*;
+#[path = "common/mod.rs"]
+mod common;
+use common::run_cli_with_stdin;
 
 #[test]
 fn test_cli_fences_preserves_nested_backtick_block() {

tests/code_emphasis.rs

@@ -2,10 +2,11 @@
 //!
 //! Verifies that emphasis markers adjacent to inline code are normalised.
 
-mod prelude;
 use std::fs;
 
-use prelude::{run_cli_with_args, run_cli_with_stdin};
+#[path = "common/mod.rs"]
+mod common;
+use common::{run_cli_with_args, run_cli_with_stdin};
 use tempfile::tempdir;
 
 #[test]

tests/compile.rs

@@ -0,0 +1,7 @@
+//! Compile-time regression tests for dependency integration.
+
+#[test]
+fn html5ever_rcdom_parser_stack_compiles() {
+    let cases = trybuild::TestCases::new();
+    cases.pass("tests/ui/html5ever_rcdom_pass.rs");
+}

tests/fences.rs

@@ -1,7 +1,8 @@
 //! Tests for fence normalisation functionality.
 
 #[macro_use]
-mod prelude;
+#[path = "common/mod.rs"]
+mod common;
 use mdtablefix::{attach_orphan_specifiers, compress_fences};
 use rstest::rstest;
 

tests/footnotes.rs

@@ -2,20 +2,20 @@
 //!
 //! Each test processes a complete Markdown document using
 //! `convert_footnotes`. Inputs are loaded from fixture files through the
-//! `include_lines!` and `lines_vec!` macros re-exported by `tests::prelude`.
+//! `include_lines!` and `lines_vec!` macros from the shared test utilities.
 //! The cases mix headings, code blocks and ordinary text to confirm that
 //! inline references become footnote links; eligible trailing numeric lists are
 //! rewritten as definition-style footnotes when at least one footnote reference exists;
 //! footnotes are renumbered sequentially with definitions reordered to match.
 //!
-//! A simple check ensures these macros are available so the prelude exports
-//! are correctly wired for all integration tests.
+//! A simple check ensures these macros are available for integration tests.
 
 use mdtablefix::{convert_footnotes, process_stream};
 use rstest::rstest;
 
 #[macro_use]
-mod prelude;
+#[path = "common/mod.rs"]
+mod common;
 
 #[test]
 fn macros_available() {

tests/lists.rs

@@ -1,11 +1,12 @@
 //! Integration tests for list renumbering.
 
+use assert_cmd::Command;
 use mdtablefix::renumber_lists;
 use rstest::rstest;
 
 #[macro_use]
-mod prelude;
-use prelude::*;
+#[path = "common/mod.rs"]
+mod common;
 
 #[test]
 fn restart_after_equal_indent_paragraph() {

tests/markdownlint.rs

@@ -5,7 +5,8 @@
 //! after processing. Regular comments should still be wrapped normally.
 
 #[macro_use]
-mod prelude;
+#[path = "common/mod.rs"]
+mod common;
 use mdtablefix::process_stream;
 use rstest::rstest;
 

tests/parallel.rs

@@ -2,12 +2,14 @@
 
 use std::{fs::File, io::Write};
 
+use assert_cmd::Command;
 use rstest::rstest;
 use tempfile::tempdir;
 
 #[macro_use]
-mod prelude;
-use prelude::*;
+#[path = "common/mod.rs"]
+mod common;
+use common::{broken_table, run_cli_with_args};
 
 #[rstest]
 fn test_cli_parallel_empty_file_list() { run_cli_with_args(&[]).success().stdout("\n"); }

tests/table/mod.rs

@@ -17,10 +17,11 @@
 //! across different scenarios whilst avoiding duplication.
 
 use mdtablefix::{convert_html_tables, process_stream, reflow_table};
+use rstest::fixture;
 
 #[macro_use]
-mod prelude;
-use prelude::*;
+#[path = "../common/mod.rs"]
+mod common;
 
 #[fixture]
 fn malformed_table() -> Vec<String> {

tests/ui/html5ever_rcdom_pass.rs

@@ -0,0 +1,12 @@
+//! Ensures `markup5ever_rcdom::RcDom` implements the `TreeSink` expected by
+//! the active `html5ever` parser stack.
+
+use html5ever::{driver::ParseOpts, parse_document, tendril::TendrilSink};
+use markup5ever_rcdom::RcDom;
+
+fn main() {
+    let opts = ParseOpts::default();
+    let dom: RcDom = parse_document(RcDom::default(), opts).one("<table></table>");
+
+    let _document = dom.document;
+}

tests/wrap/mod.rs

@@ -6,8 +6,9 @@
 use mdtablefix::process_stream;
 
 #[macro_use]
-mod prelude;
-use prelude::*;
+#[path = "../common/mod.rs"]
+mod common;
+use common::{assert_wrapped_blockquote, assert_wrapped_list_item, run_cli_with_stdin};
 
 mod paragraphs;
 mod lists;

tests/wrap_renumber.rs

@@ -3,7 +3,8 @@
 use mdtablefix::{process_stream, renumber_lists};
 
 #[macro_use]
-mod prelude;
+#[path = "common/mod.rs"]
+mod common;
 
 #[test]
 fn wrap_then_renumber_preserves_order() {