fixes for JSONfeed item.content

changelog.d/20240422_124749_bbolli_jsonfeed_content.rst

@@ -0,0 +1,5 @@
+Fixed
+-----
+
+*   JSONfeed item content is now a list that prefers HTML content.
+*   A JSONfeed item without content sets the bozo flag.

feedparser/parsers/json.py

@@ -85,16 +85,23 @@ def parse_entry(self, e):
             if src in e:
                 entry[dst] = e[src]
 
-        if "content_text" in e:
-            entry["content"] = c = FeedParserDict()
-            c["value"] = e["content_text"]
-            c["type"] = "text"
-        elif "content_html" in e:
-            entry["content"] = c = FeedParserDict()
+        entry["content"] = content = []
+        if "content_html" in e:
+            c = FeedParserDict()
             c["value"] = sanitize_html(
                 e["content_html"], self.encoding, "application/json"
             )
             c["type"] = "html"
+            content.append(c)
+        if "content_text" in e:
+            c = FeedParserDict()
+            c["value"] = e["content_text"]
+            c["type"] = "text"
+            content.append(c)
+        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"]

tests/json/html_first.json

@@ -0,0 +1,18 @@
+{
+    "__TEST__": "Description: basic JSON tests Expect: not bozo and items[0].content[0].type == 'html' -->",
+    "version": "https://jsonfeed.org/version/1",
+    "title": "html_preferred",
+    "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" },
+            "content_text": "Hello, world!\n",
+            "content_html": "<p>Hello, world!</p>\n\n<script>alert();</script>",
+            "url": "https://example.org/initial-post"
+        }
+    ]
+}

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"
+        }
+    ]
+}