diff --git a/sites/arbitrary.py b/sites/arbitrary.py
index 668f92a..b40a3da 100644
--- a/sites/arbitrary.py
+++ b/sites/arbitrary.py
@@ -1,5 +1,6 @@
 #!/usr/bin/python
 
+import attr
 import datetime
 import json
 import os.path
@@ -24,8 +25,20 @@ difference would be whether a `--toc` arg was given.)
 """
 
 
+@attr.s
+class SiteDefinition:
+    url = attr.ib()
+    title = attr.ib()
+    author = attr.ib()
+    content_selector = attr.ib()
+    # If this is present, it looks for chapters linked from `url`. If not, it assumes `url` points to a chapter.
+    chapter_selector = attr.ib(default=False)
+    # If this is present, it's used to filter out content that matches the selector
+    filter_selector = attr.ib(default=False)
+
+
 @register
-class ArbitraryIndex(Site):
+class Arbitrary(Site):
     """A way to describe an arbitrary side for a one-off fetch
     """
     @staticmethod
@@ -36,20 +49,27 @@ class ArbitraryIndex(Site):
 
     def extract(self, url):
         with open(url) as definition_file:
-            definition = json.load(definition_file)
-
-        soup = self._soup(definition['url'])
+            definition = SiteDefinition(**json.load(definition_file))
 
         story = Section(
-            title=definition['title'],
-            author=definition['author']
+            title=definition.title,
+            author=definition.author
         )
 
-        for chapter in soup.select(definition['chapter_selector']):
-            chapter_url = str(chapter.get('href'))
+        if definition.chapter_selector:
+            soup = self._soup(definition.url)
+            for chapter in soup.select(definition.chapter_selector):
+                chapter_url = str(chapter.get('href'))
+                story.add(Chapter(
+                    title=chapter.string,
+                    contents=self._chapter(chapter_url, definition),
+                    # TODO: better date detection
+                    date=datetime.datetime.now()
+                ))
+        else:
             story.add(Chapter(
-                title=chapter.string,
-                contents=self._chapter(chapter_url, definition),
+                title=definition.title,
+                contents=self._chapter(definition.url, definition),
                 # TODO: better date detection
                 date=datetime.datetime.now()
             ))
@@ -57,12 +77,14 @@ class ArbitraryIndex(Site):
         return story
 
     def _chapter(self, url, definition):
+        # TODO: refactor so this can meaningfully handle multiple matches on content_selector.
+        # Probably by changing it so that this returns a Chapter / Section.
         print("Extracting chapter from", url)
         soup = self._soup(url)
-        content = soup.select(definition['content_selector'])[0]
+        content = soup.select(definition.content_selector)[0]
 
-        if 'filter_selector' in definition:
-            for filtered in content.select(definition['filter_selector']):
+        if definition.filter_selector:
+            for filtered in content.select(definition.filter_selector):
                 filtered.decompose()
 
         return content.prettify()