JSONfeed: validate the presence of content_html or content_text

The spec says that at least one element is mandatory. Verify this and raise an exception if both are missing. Signed-off-by: Beat Bolli <[email protected]>
kurtmckee · Apr 22, 2024 · 6a68b8d · 6a68b8d
1 parent 97c648c
commit 6a68b8d
Show file tree

Hide file tree

Showing 2 changed files with 22 additions and 3 deletions.
diff --git a/feedparser/parsers/json.py b/feedparser/parsers/json.py
@@ -85,7 +85,7 @@ def parse_entry(self, e):
             if src in e:
                 entry[dst] = e[src]
 
-        content = []
+        entry["content"] = content = []
         if "content_html" in e:
             c = FeedParserDict()
             c["value"] = sanitize_html(
@@ -98,8 +98,10 @@ def parse_entry(self, e):
             c["value"] = e["content_text"]
             c["type"] = "text"
             content.append(c)
-        if content:
-            entry["content"] = content
+        if not content:
+            raise ValueError(
+                f"item {entry['id']=} has neither 'content_text' nor 'content_html'"
+            )
 
         if "date_published" in e:
             entry["published"] = e["date_published"]

diff --git a/tests/json/no_content.json b/tests/json/no_content.json
@@ -0,0 +1,17 @@
+{
+    "__TEST__": "Description: basic JSON tests Expect: bozo and 'neither' in str(bozo_exception) -->",
+    "version": "https://jsonfeed.org/version/1",
+    "title": "no content",
+    "home_page_url": "https://example.org/",
+    "feed_url": "https://example.org/feed.json",
+    "icon": "https://example.org/feed.png",
+    "author": { "name": "me" },
+    "items": [
+        {
+            "id": "1",
+            "author": { "name": "you", "url": "http://example.net/~you" },
+            "summary": "Hello, world!\n",
+            "url": "https://example.org/initial-post"
+        }
+    ]
+}