feat(injections): enable language-specific code block syntax highlighting

# Add support for parsing [source,language] attributes in code blocks

- Enhance block_attributes grammar to parse source block language specifications
- Add source_block_attributes rule with type and language fields
- Add language_identifier rule to capture programming language names
- Update injections.scm with language-specific injection patterns for:
  - Python, JavaScript, Rust, Go, HTML, CSS, JSON, YAML, Bash
- Add comprehensive test files for various programming languages
- Resolve conflicts between attribute parsing rules

This enables proper syntax highlighting within AsciiDoc code blocks in editors
that support Tree Sitter language injections (like Zed, Neovim, etc.).

BREAKING CHANGE: block_attributes now has a more structured AST with
source_block_attributes as a distinct node type instead of generic content.

Author: louiss0

grammar.js

@@ -42,7 +42,8 @@ module.exports = grammar({
     [$.ordered_list_item], 
     [$.description_item],
     [$.callout_item],
-    [$.inline_element, $.explicit_link]
+    [$.inline_element, $.explicit_link],
+    [$.attribute_content, $.role_list]
   ],
 
   rules: {
@@ -301,11 +302,23 @@ module.exports = grammar({
 
     block_attributes: $ => prec(3, seq(
       '[',
-      /[^\]\r\n]+/,  // attribute content
+      choice(
+        $.source_block_attributes,
+        field('content', $.attribute_content)
+      ),
       ']',
       $._line_ending
     )),
 
+    source_block_attributes: $ => prec(100, seq(
+      field('type', token('source')),
+      token(','),
+      field('language', $.language_identifier)
+    )),
+    
+    language_identifier: $ => /[a-zA-Z][a-zA-Z0-9_+-]*/,
+    attribute_content: $ => /[^\]\r\n]+/,
+    
     id_and_roles: $ => seq(
       '[',
       /#[^\]\r\n]+/,  // content starting with # for ID

queries/injections.scm

@@ -1,14 +1,92 @@
 ; AsciiDoc language injections for embedded syntax highlighting
 
-; Code blocks with language specified in metadata
-; TODO: This would require parsing block attributes to detect language
-; For now, we'll use simpler patterns
+; Language-specific code blocks - Now supported!
+; Extract language from [source,language] attributes
+(listing_block
+  (metadata
+    (block_attributes
+      (source_block_attributes
+        language: (language_identifier) @injection.language)))
+  content: (block_content) @injection.content)
+
+; Fallback: Generic source blocks with language detection from content
+((listing_block
+  (metadata 
+    (block_attributes
+      content: (attribute_content) @_lang
+      (#match? @_lang "source,python")))
+  content: (block_content) @injection.content)
+ (#set! injection.language "python"))
+
+((listing_block
+  (metadata 
+    (block_attributes
+      content: (attribute_content) @_lang
+      (#match? @_lang "source,javascript")))
+  content: (block_content) @injection.content)
+ (#set! injection.language "javascript"))
+
+((listing_block
+  (metadata 
+    (block_attributes
+      content: (attribute_content) @_lang
+      (#match? @_lang "source,rust")))
+  content: (block_content) @injection.content)
+ (#set! injection.language "rust"))
+
+((listing_block
+  (metadata 
+    (block_attributes
+      content: (attribute_content) @_lang
+      (#match? @_lang "source,json")))
+  content: (block_content) @injection.content)
+ (#set! injection.language "json"))
+
+((listing_block
+  (metadata 
+    (block_attributes
+      content: (attribute_content) @_lang
+      (#match? @_lang "source,yaml")))
+  content: (block_content) @injection.content)
+ (#set! injection.language "yaml"))
+
+((listing_block
+  (metadata 
+    (block_attributes
+      content: (attribute_content) @_lang
+      (#match? @_lang "source,html")))
+  content: (block_content) @injection.content)
+ (#set! injection.language "html"))
+
+((listing_block
+  (metadata 
+    (block_attributes
+      content: (attribute_content) @_lang
+      (#match? @_lang "source,css")))
+  content: (block_content) @injection.content)
+ (#set! injection.language "css"))
+
+((listing_block
+  (metadata 
+    (block_attributes
+      content: (attribute_content) @_lang
+      (#match? @_lang "source,go")))
+  content: (block_content) @injection.content)
+ (#set! injection.language "go"))
+
+((listing_block
+  (metadata 
+    (block_attributes
+      content: (attribute_content) @_lang
+      (#match? @_lang "source,bash")))
+  content: (block_content) @injection.content)
+ (#set! injection.language "bash"))
 
 ; Math content injection (math_macro is a simple token)
 (math_macro) @injection.content
 (#set! injection.language "latex")
 
-; Generic code blocks (listing blocks often contain code)
+; Generic code blocks without language specification
 (listing_block
   (block_content) @injection.content
   (#set! injection.language "text"))