feat: nested date support in sitemap (#73)

Adds sitemap support for nested dates, like a review.datePublished field of a Product

blurry/sitemap.py

@@ -1,5 +1,7 @@
 from pathlib import Path
 
+import dpath
+
 from blurry.settings import get_build_directory
 from blurry.types import MarkdownFileData
 
@@ -20,10 +22,12 @@
 def generate_sitemap_for_file_data_list(file_data_list: list[MarkdownFileData]) -> str:
     sitemap_url_data = []
     for file_data in file_data_list:
-        lastmod = file_data.front_matter.get(
-            "dateModified"
-        ) or file_data.front_matter.get("datePublished")
-        lastmod_tag = f"<lastmod>{lastmod}</lastmod>" if lastmod else ""
+        lastmod = (
+            dpath.values(file_data.front_matter, "**/dateModified")
+            or dpath.values(file_data.front_matter, "**/datePublished")
+            or dpath.values(file_data.front_matter, "**/dateCreated")
+        )
+        lastmod_tag = f"<lastmod>{lastmod[0]}</lastmod>" if len(lastmod) else ""
         url = file_data.front_matter.get("url")
         sitemap_url_data.append({"lastmod_tag": lastmod_tag, "url": url})
 

pyproject.toml

@@ -36,6 +36,7 @@ toml = "^0.10.2"
 typer = "^0.6.1"
 htmlmin2 = "^0.1.13"
 pydantic2-schemaorg = "^0.1.1"
+dpath = "^2.1.6"
 
 [tool.poetry.scripts]
 blurry = 'blurry:main'

tests/test_sitemap.py

@@ -9,26 +9,29 @@
     MarkdownFileData(
         front_matter=dict(datePublished=date(2021, 1, 1), url="/blog/a-post-1/"),
         body="",
-        path=blog_path / "a-post-1",
+        path=blog_path / "a-post-1.md",
     ),
     MarkdownFileData(
         front_matter=dict(datePublished=date(2021, 3, 1), url="/blog/b-post-3/"),
         body="",
-        path=blog_path / "b-post-3",
+        path=blog_path / "b-post-3.md",
     ),
     MarkdownFileData(
         front_matter=dict(dateCreated=date(2021, 2, 1), url="/blog/c-post-2/"),
         body="",
-        path=blog_path / "c-post-2",
+        path=blog_path / "c-post-2.md",
     ),
+    # Nested dateCreated and dateModified
     MarkdownFileData(
         front_matter=dict(
-            dateCreated=date(2022, 1, 10),
-            dateModified=date(2022, 1, 13),
+            review=dict(
+                dateCreated=date(2022, 1, 10),
+                dateModified=date(2022, 1, 13),
+            ),
             url="/blog/c-post-4/",
         ),
         body="",
-        path=blog_path / "c-post-4",
+        path=blog_path / "c-post-4.md",
     ),
 ]
 
@@ -39,6 +42,14 @@ def test_generate_sitemap_for_file_data_list():
     )
     assert sitemap_content.startswith('<?xml version="1.0" encoding="UTF-8"?>')
     assert sitemap_content.endswith("</urlset>")
+    assert (
+        "<url><loc>/blog/a-post-1/</loc><lastmod>2021-01-01</lastmod></url>"
+        in sitemap_content
+    )
+    assert (
+        "<url><loc>/blog/c-post-2/</loc><lastmod>2021-02-01</lastmod></url>"
+        in sitemap_content
+    )
     assert (
         "<url><loc>/blog/c-post-4/</loc><lastmod>2022-01-13</lastmod></url>"
         in sitemap_content