From 73215df2e8c415c1b19c64cd99b277254db92b80 Mon Sep 17 00:00:00 2001
From: Daniel Fichtinger <daniel@ficd.ca>
Date: Sat, 21 Jun 2025 16:53:24 -0400
Subject: [PATCH] add better metadata handling

---
 src/zona/builder.py   | 38 +++++++++++++++++++++++++++---------
 src/zona/models.py    |  2 +-
 tests/test_builder.py | 45 +++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 75 insertions(+), 10 deletions(-)

diff --git a/src/zona/builder.py b/src/zona/builder.py
index c60599b..01b4785 100644
--- a/src/zona/builder.py
+++ b/src/zona/builder.py
@@ -3,18 +3,38 @@ from zona import markdown as zmd
 from zona import util
 from pathlib import Path
 import frontmatter
-from dacite import from_dict
+from dacite import Config, from_dict, DaciteError
+from datetime import date
+from yaml import YAMLError
 
 
 def split_metadata(path: Path) -> tuple[Metadata, str]:
-    post = frontmatter.load(str(path))
-    # TODO: handle missing metadata
-    meta = post.metadata
-    # TODO: handle malformed metadata
-    metadata = from_dict(
-        data_class=Metadata,
-        data=meta,
-    )
+    """
+    Parses a file and returns parsed Metadata and its content. Defaults
+    are applied for missing fields. If there is no metadata, a Metadata
+    with default values is returned.
+
+    Raises:
+        ValueError: If the metadata block is malformed in any way.
+    """
+    try:
+        post = frontmatter.load(str(path))
+    except YAMLError as e:
+        raise ValueError(f"YAML frontmatter error in {path}: {e}")
+    raw_meta = post.metadata or {}
+    defaults = {
+        "title": util.filename_to_title(path),
+        "date": date.fromtimestamp(path.stat().st_mtime),
+    }
+    meta = {**defaults, **raw_meta}
+    try:
+        metadata = from_dict(
+            data_class=Metadata,
+            data=meta,
+            config=Config(check_types=True, strict=True),
+        )
+    except DaciteError as e:
+        raise ValueError(f"Malformed metadata in {path}: {e}")
     return metadata, post.content
 
 
diff --git a/src/zona/models.py b/src/zona/models.py
index 2351abd..29dafee 100644
--- a/src/zona/models.py
+++ b/src/zona/models.py
@@ -9,7 +9,7 @@ from typing import Literal
 class Metadata:
     title: str
     date: date
-    description: str
+    description: str | None
     # add more options later...
 
 
diff --git a/tests/test_builder.py b/tests/test_builder.py
index 199d36f..f61f7ee 100644
--- a/tests/test_builder.py
+++ b/tests/test_builder.py
@@ -1,3 +1,4 @@
+import pytest
 from datetime import date
 from zona.models import Metadata
 from zona.builder import split_metadata, discover, build
@@ -24,6 +25,50 @@ description: This is a test.
     assert content == "# Hello World"
 
 
+def test_no_metadata(tmp_path: Path):
+    content = "# Hello World"
+    test_file = tmp_path / "hello-world.md"
+    test_file.write_text(content)
+    meta, content = split_metadata(test_file)
+
+    assert isinstance(meta, Metadata)
+    assert meta.title == "Hello World"
+    assert meta.description is None
+    assert meta.date == date.today()
+    assert content == "# Hello World"
+
+
+def test_malformed_metadata(tmp_path: Path):
+    with pytest.raises(ValueError):
+        tests = {
+            """
+---
+title: Test Post
+date: not a date!!!
+description: This is a test.
+---
+        """,
+            """
+---
+title: Test Post
+foobar:
+    something: what???
+description: This is a test.
+---
+        """,
+            """
+---
+title Test Post
+description: This is a test.
+---
+        """,
+        }
+        for i, content in enumerate(tests):
+            test_file = tmp_path / str(i)
+            test_file.write_text(content)
+            split_metadata(test_file)
+
+
 def test_discover(tmp_path: Path):
     contentd = tmp_path / "content"
     staticd = contentd / "static"