tests: exercise upstream Org examples

.gitmodules

@@ -0,0 +1,4 @@
+[submodule "external/org-mode"]
+	path = testdata/external/org-mode
+	url = https://github.com/bzg/org-mode.git
+	shallow = true

README.md

@@ -116,9 +116,14 @@ The tests show additional supported features:
 
 ## Development and documentation
 
+Clone with `git clone --recurse-submodules https://github.com/karlicoss/orgparse.git` to include the test corpus.
+For an existing checkout, run `git submodule update --init --recursive`.
+
 Run the tests with `uv tool run --with tox-uv tox -e tests`.
 This also checks the examples in this README.
 
+The normal test suite parses the `.org` files in `testdata/external/org-mode/testing/examples` and checks tree structure and attribute access.
+
 Edit `README.qmd`, then regenerate `README.md` with `uv tool run --with tox-uv tox -e quarto`.
 Quarto computes links to source code and tests from their definitions, so line numbers are refreshed when rendering.
 Commit both files together; CI checks that the generated README is current.

README.qmd

@@ -145,9 +145,14 @@ The tests show additional supported features:
 
 ## Development and documentation
 
+Clone with `git clone --recurse-submodules https://github.com/karlicoss/orgparse.git` to include the test corpus.
+For an existing checkout, run `git submodule update --init --recursive`.
+
 Run the tests with `uv tool run --with tox-uv tox -e tests`.
 This also checks the examples in this README.
 
+The normal test suite parses the `.org` files in `testdata/external/org-mode/testing/examples` and checks tree structure and attribute access.
+
 Edit `README.qmd`, then regenerate `README.md` with `uv tool run --with tox-uv tox -e quarto`.
 Quarto computes links to source code and tests from their definitions, so line numbers are refreshed when rendering.
 Commit both files together; CI checks that the generated README is current.

pyproject.toml

@@ -66,6 +66,10 @@ quarto = [
 requires = ["hatchling", "hatch-vcs"]
 build-backend = "hatchling.build"
 
+[tool.hatch.build]
+# Upstream test corpora are development inputs and retain their own licenses.
+exclude = ["/testdata/external"]
+
 # unfortunately have to duplicate project name here atm, see https://github.com/pypa/hatch/issues/1894
 [tool.hatch.build.targets.wheel]
 packages = ["src/orgparse"]
@@ -111,7 +115,7 @@ enable_error_code = [
 # suppressing rules/changing error level goes here
 
 [tool.ty.src]
-# exclude = [...]
+exclude = ["testdata/external"]
 
 
 [tool.ruff]
@@ -231,13 +235,16 @@ lint.ignore = [
 #       "scratch/**",
 #   ]
 extend-exclude = [
+    "testdata/external",
 ]
 
 [tool.ruff.format]
 quote-style = "preserve"
 
 
 [tool.pytest.ini_options]
+testpaths = ["src/orgparse"]
+
 # discover files that don't follow test_ naming. Useful to keep tests along with the source code
 python_files = ["*.py"]
 

src/orgparse/tests/test_corpus.py

@@ -0,0 +1,70 @@
+"""Exercise the public parser API on Org mode's upstream example files."""
+
+from pathlib import Path
+
+import pytest
+
+from .. import load
+from ..extra import Table
+from ..node import OrgNode
+
+CORPUS = Path(__file__).resolve().parents[3] / 'testdata' / 'external' / 'org-mode' / 'testing' / 'examples'
+
+
+def corpus_files() -> list[Path]:
+    paths = sorted(CORPUS.rglob('*.org'))
+    assert len(paths) > 0, 'Missing Org corpus: run git submodule update --init --recursive'
+    return paths
+
+
+@pytest.mark.parametrize('path', corpus_files(), ids=lambda path: path.relative_to(CORPUS).as_posix())
+def test_corpus(path: Path) -> None:
+    """Check parser robustness and tree consistency across upstream Org examples.
+
+    Exercise heading/body formatting, property lookups, table row/block iteration,
+        and formatting of populated timestamps.
+    Check tag inclusion, parent/child links, a shared root, increasing heading line numbers,
+        and heading levels against the source text.
+    These are smoke and consistency checks; exact parsed values require separate expected-output tests.
+    """
+    root = load(path)
+    lines = path.read_text(encoding='utf-8').splitlines()
+    assert root.is_root()
+    assert root.parent is None
+
+    previous_line = 0
+    for node in root:
+        # Access lazy formatting as well as the eagerly parsed attributes.
+        assert isinstance(node.heading, str)
+        assert isinstance(node.body, str)
+        assert node.shallow_tags <= node.tags
+        for key, value in node.properties.items():
+            assert node.get_property(key) == value
+        for part in node.body_rich:
+            if isinstance(part, Table):
+                list(part.rows)
+                list(part.blocks)
+
+        dates = node.get_timestamps(active=True, inactive=True)
+        if isinstance(node, OrgNode):
+            dates = [*dates, node.scheduled, node.deadline, node.closed, *node.clock, *node.repeated_tasks]
+        for date in dates:
+            if date.start is not None:
+                str(date)
+
+        assert node.root is root
+        for child in node.children:
+            assert child.parent is node
+            assert child.level > node.level
+        if node is root:
+            continue
+
+        assert previous_line < node.linenumber <= len(lines)
+        previous_line = node.linenumber
+        heading_line = lines[node.linenumber - 1]
+        assert heading_line[: node.level] == '*' * node.level
+        assert heading_line[node.level].isspace()
+        parent = node.parent
+        assert parent is not None
+        assert parent.level < node.level
+        assert any(child is node for child in parent.children)