Add sitemap generation

omicron/ssg/output/__init__.py

@@ -7,6 +7,7 @@ from omicron.ssg.output.memory import Memory
 from omicron.ssg.output.directory import Directory
 from omicron.ssg.output.symlink import Symlink
 from omicron.ssg.output.tags import Tags, discover_tags
+from omicron.ssg.output.sitemap import Sitemap
 
 __all__ = [
     "Output",
@@ -23,4 +24,5 @@ __all__ = [
     "Symlink",
     "Tags",
     "discover_tags",
+    "Sitemap",
 ]

omicron/ssg/output/content.py

@@ -44,6 +44,7 @@ class Content(Output):
         self.tags: set[str] = tags
         self.summary: str | None = summary
         self.draft: bool = draft
+        self.in_sitemap = True
 
     @staticmethod
     def build_path(section: str, slug: str) -> Path:

omicron/ssg/output/index.py

@@ -30,6 +30,7 @@ class Index(Output):
         self.page_num = page_num
         self.total_pages = total_pages
         self.label = label
+        self.in_sitemap = kind != "tag" and page_num == 1
 
     def pagination_url(self, n: int) -> str:
         path = Index.build_path(self.kind, n, self.label)

omicron/ssg/output/output.py

@@ -31,6 +31,8 @@ class Output(ABC):
             url = url.parent
         self.url = url.as_posix()
 
+        self.in_sitemap = False
+
     @property
     def site(self) -> Site:
         obj = self._site_ref()

omicron/ssg/output/sitemap.py

@@ -0,0 +1,38 @@
+import logging
+from pathlib import Path
+from typing import TYPE_CHECKING
+from xml.etree.ElementTree import Element, ElementTree, SubElement
+from omicron.ssg.output.output import Output
+from omicron.ssg.output.content import Content
+
+log = logging.getLogger(__name__)
+
+if TYPE_CHECKING:
+    from omicron.ssg.site import Site
+else:
+    Site = "omicron.ssg.site.Site"
+
+
+class Sitemap(Output):
+    PATH = Path("sitemap.xml")
+    NAMESPACE = "http://www.sitemaps.org/schemas/sitemap/0.9"
+
+    def __init__(self, site: Site):
+        super().__init__(site, Sitemap.PATH)
+
+    def write(self) -> None:
+        urlset = Element("urlset", xmlns=Sitemap.NAMESPACE)
+        for entry in self.site.by_path.values():
+            if not entry.in_sitemap:
+                continue
+
+            url = SubElement(urlset, "url")
+            SubElement(url, "loc").text = f"{self.site.origin}{entry.url}"
+            if isinstance(entry, Content):
+                lastmod = entry.updated or entry.created
+                SubElement(url, "lastmod").text = lastmod.isoformat()
+
+        dest = self.site.output_path / self.destination
+        dest.parent.mkdir(parents=True, exist_ok=True)
+        ElementTree(urlset).write(dest, encoding="utf-8", xml_declaration=True)
+        log.debug("Wrote %s", dest)

omicron/ssg/output/tags.py

@@ -17,6 +17,7 @@ class Tags(Output):
     def __init__(self, site: Site, tags: list[tuple[str, int]]):
         super().__init__(site, Tags.PATH)
         self.tags = tags
+        self.in_sitemap = True
 
     def write(self) -> None:
         template = self.site.jinja_env.get_template("tags.html")

omicron/ssg/site.py

@@ -15,6 +15,7 @@ from omicron.ssg.output import (
     Index,
     Tags,
     Memory,
+    Sitemap,
     discover_index,
     discover_tags,
 )
@@ -194,6 +195,12 @@ class Site:
         tags_page = discover_tags(self)
         self.add_by_path(tags_page)
         self.other_outputs.append(tags_page)
+        if self.origin:
+            sitemap = Sitemap(self)
+            self.add_by_path(sitemap)
+            self.other_outputs.append(sitemap)
+        else:
+            log.debug("Skipping sitemap, base_url has no domain name")
         log.info("Discovered %d content items", len(self.content))
 
     def clean(self) -> None: