Parse Markdown files into a structured Python object

tests/test_parse_contract.py

@@ -0,0 +1,89 @@
+from pathlib import Path
+
+import pytest
+from click.testing import CliRunner
+
+from markitect_tool import MarkdownParseError, parse_markdown, parse_markdown_file
+from markitect_tool.cli import main
+
+
+def test_parse_markdown_preserves_headings_and_paragraphs():
+    document = parse_markdown("# Heading\n\nThis is a paragraph.")
+
+    assert document.frontmatter == {}
+    assert document.headings[0].level == 1
+    assert document.headings[0].text == "Heading"
+    assert [block.type for block in document.blocks] == ["heading", "paragraph"]
+    assert document.sections[0].heading.text == "Heading"
+    assert document.sections[0].blocks[0].text == "This is a paragraph."
+    assert document.tokens[0]["type"] == "heading_open"
+
+
+def test_parse_markdown_extracts_yaml_frontmatter():
+    markdown = """---
+title: YAML Frontmatter Test Document
+tags:
+  - yaml
+  - frontmatter
+published: true
+nested:
+  priority: high
+---
+
+# YAML Frontmatter Test Document
+
+Body text.
+"""
+
+    document = parse_markdown(markdown)
+
+    assert document.frontmatter["title"] == "YAML Frontmatter Test Document"
+    assert document.frontmatter["tags"] == ["yaml", "frontmatter"]
+    assert document.frontmatter["published"] is True
+    assert document.frontmatter["nested"]["priority"] == "high"
+    assert document.headings[0].line == 11
+    assert document.body.lstrip().startswith("# YAML Frontmatter Test Document")
+
+
+def test_parse_markdown_without_frontmatter_is_graceful():
+    document = parse_markdown("# Document Without Frontmatter\n\nText.")
+
+    assert document.frontmatter == {}
+    assert document.headings[0].text == "Document Without Frontmatter"
+
+
+def test_parse_markdown_rejects_non_mapping_frontmatter():
+    with pytest.raises(MarkdownParseError, match="Frontmatter must be a mapping"):
+        parse_markdown("---\n- nope\n---\n\n# Bad")
+
+
+def test_parse_markdown_file_records_source_path(tmp_path: Path):
+    source = tmp_path / "doc.md"
+    source.write_text("# Test Document\n\nBody", encoding="utf-8")
+
+    document = parse_markdown_file(source)
+
+    assert document.source_path == str(source)
+    assert document.headings[0].text == "Test Document"
+
+
+def test_mkt_parse_outputs_json(tmp_path: Path):
+    source = tmp_path / "doc.md"
+    source.write_text("# Test Document\n\nBody", encoding="utf-8")
+
+    result = CliRunner().invoke(main, ["parse", str(source)])
+
+    assert result.exit_code == 0
+    assert '"headings"' in result.output
+    assert "Test Document" in result.output
+
+
+def test_mkt_parse_outputs_tree(tmp_path: Path):
+    source = tmp_path / "doc.md"
+    source.write_text("# One\n\n## Two\n", encoding="utf-8")
+
+    result = CliRunner().invoke(main, ["parse", str(source), "--format", "tree"])
+
+    assert result.exit_code == 0
+    assert "# One" in result.output
+    assert "## Two" in result.output