Fixup: remove overwrite of schema.org data with 'mainEntity' contents…

… during schema.org processing (#666)
hhursev · Oct 21, 2022 · 557cfd9 · 557cfd9
1 parent d839b76
commit 557cfd9
Show file tree

Hide file tree

Showing 2 changed files with 5 additions and 7 deletions.
diff --git a/recipe_scrapers/_schemaorg.py b/recipe_scrapers/_schemaorg.py
@@ -52,21 +52,15 @@ def __init__(self, page_data, raw=False):
                 if in_context and item_type.lower() in low_schema:
                     self.format = syntax
                     self.data = item
-                    if item_type.lower() == "webpage":
-                        self.data = self.data.get("mainEntity")
                     return
                 elif in_context and "@graph" in item:
                     for graph_item in item.get("@graph", ""):
                         graph_item_type = graph_item.get("@type", "")
                         if not isinstance(graph_item_type, str):
                             continue
                         if graph_item_type.lower() in low_schema:
-                            in_graph = SCHEMA_ORG_HOST in graph_item.get("@context", "")
                             self.format = syntax
-                            if graph_item_type.lower() == "webpage" and in_graph:
-                                self.data = self.data.get("mainEntity")
-                                return
-                            elif graph_item_type.lower() == "recipe":
+                            if graph_item_type.lower() == "recipe":
                                 self.data = graph_item
                                 return
 

diff --git a/recipe_scrapers/sallysblog.py b/recipe_scrapers/sallysblog.py
@@ -41,3 +41,7 @@ def instructions(self):
                 for instruction in instructions
             ]
         )
+
+    def image(self):
+        image = self.soup.find("meta", {"property", "og:image"})
+        return image["content"] if image else None