Contract framework with markdown-native contracts utilizing fenced YAML blocks

src/markitect_tool/core/parser.py

@@ -29,7 +29,7 @@ def parse_markdown(markdown: str, source_path: str | None = None) -> Document:
 
     frontmatter, body, body_line_offset = _split_frontmatter(markdown)
     tokens = _parse_tokens(body)
-    blocks, headings = _blocks_and_headings(tokens, body_line_offset)
+    blocks, headings = _blocks_and_headings(tokens, body_line_offset, body)
     sections = _sections_from_blocks(blocks, headings)
     return Document(
         source_path=source_path,
@@ -97,7 +97,7 @@ def _token_to_dict(token: Token) -> dict[str, Any]:
 
 
 def _blocks_and_headings(
-    tokens: list[dict[str, Any]], line_offset: int
+    tokens: list[dict[str, Any]], line_offset: int, markdown: str
 ) -> tuple[list[ContentBlock], list[Heading]]:
     blocks: list[ContentBlock] = []
     headings: list[Heading] = []
@@ -126,6 +126,8 @@ def _blocks_and_headings(
             if not text and token_type.endswith("_open"):
                 inline = _next_inline(tokens, index)
                 text = inline.get("content", "") if inline else ""
+            if not text:
+                text = _source_text(token, line_offset, markdown)
             blocks.append(
                 ContentBlock(
                     type=_block_type(token_type),
@@ -151,6 +153,16 @@ def _line_range(token: dict[str, Any], line_offset: int) -> tuple[int | None, in
     return line_map[0] + line_offset + 1, line_map[1] + line_offset
 
 
+def _source_text(token: dict[str, Any], line_offset: int, markdown: str) -> str:
+    line_start, line_end = _line_range(token, line_offset)
+    if line_start is None or line_end is None:
+        return ""
+    lines = markdown.splitlines()
+    start_index = max(line_start - line_offset - 1, 0)
+    end_index = max(line_end - line_offset, start_index)
+    return "\n".join(lines[start_index:end_index]).strip()
+
+
 def _block_type(token_type: str) -> str:
     return {
         "paragraph_open": "paragraph",