@willbooster/tree-sitter-markdown 1.0.5 → 1.0.7

package/README.md

@@ -19,6 +19,7 @@ digits of an ordered list marker with `'0'` and `'9'` instead of calling `isdigi
 own ABI number (`0x57420001`) and load only with that fork's runtime, version 2.0.0 or later: upstream's tree-sitter
 runtime rejects them. They parse as the parsers in upstream's 0.5.3 packages do, with these exceptions:
 
+- HTML declarations end on the first line containing `>`, leaving subsequent Markdown outside the HTML block.
 - Short HTML comments such as `<!-->` and `<!--->` end on their opening line. Malformed `<!-` prefixes remain ordinary text.
 - Block containers nest to a maximum depth of 255. Additional quote, list, code, or HTML block openers remain literal text inside the deepest container.
 - Final headings, thematic breaks, empty list items, and opening code fences are recognized without a trailing newline.
@@ -137,6 +138,7 @@ parser.set_language(&tree_sitter_md::LANGUAGE.into())?;
 
 The crate keeps the API of upstream's `tree-sitter-md` crate: `tree_sitter_md::INLINE_LANGUAGE` is the inline grammar,
 and the `parser` feature adds `MarkdownParser`, which parses a document with both grammars into a `MarkdownTree`.
+The `parser` feature requires runtime version 2.0.18 or later, which supports UTF-8 characters split across short input chunks.
 `MarkdownCursor::field_name` returns a name that borrows from the tree (`Option<&'tree str>`), since the
 willbooster-tree-sitter runtime returns field names with that lifetime, where upstream's returns `Option<&'static str>`.
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@willbooster/tree-sitter-markdown",
-  "version": "1.0.5",
+  "version": "1.0.7",
   "description": "Markdown grammar for tree-sitter",
   "keywords": [
     "incremental",

package/tree-sitter-markdown/grammar.js

@@ -221,7 +221,7 @@ module.exports = grammar({
         ),
         _html_block_2: $ => build_html_block($, $._html_block_2_start, '-->', true),
         _html_block_3: $ => build_html_block($, $._html_block_3_start, '?>', true),
-        _html_block_4: $ => build_html_block($, $._html_block_4_start, '>', true),
+        _html_block_4: $ => build_html_block($, $._html_block_4_start, '>', true, $._declaration_line),
         _html_block_5: $ => build_html_block($, $._html_block_5_start, ']]>', true),
         _html_block_6: $ => build_html_block(
             $,
@@ -390,6 +390,7 @@ module.exports = grammar({
             optional($.block_continuation)
         ),
         // Some symbols get parsed as single tokens so that html blocks get detected properly
+        _declaration_line: $ => prec.right(repeat1(choice($._word, $._whitespace, common.punctuation_without($, ['>'])))),
         _line: $ => prec.right(repeat1(choice($._word, $._whitespace, common.punctuation_without($, [])))),
         _word: $ => choice(
             new RegExp('[^' + PUNCTUATION_CHARACTERS_REGEX + ' \\t\\n\\r]+'),
@@ -599,11 +600,11 @@ module.exports = grammar({
 // General purpose structure for html blocks. The different kinds mostly work the same but have
 // different openling and closing conditions. Some html blocks may not interrupt a paragraph and
 // have to be marked as such.
-function build_html_block($, open, close, interrupt_paragraph) {
+function build_html_block($, open, close, interrupt_paragraph, line = $._line) {
     return seq(
         open,
         repeat(choice(
-            $._line,
+            line,
             $._newline,
             seq(close, $._close_block),
         )),

package/tree-sitter.json

@@ -30,7 +30,7 @@
     }
   ],
   "metadata": {
-    "version": "1.0.5",
+    "version": "1.0.7",
     "license": "MIT",
     "description": "Markdown grammar for tree-sitter",
     "authors": [