add a markdown source field to recipe

comfyrecipes/parsing.py

@@ -6,6 +6,37 @@ import comfyrecipes.settings as settings
 import comfyrecipes.issues as issues
 
 import jsonschema
+import lxml.html.clean
+import mistune
+
+
+class SafeHTML:
+    def __init__(self, clean_html: str) -> None:
+        self.html = clean_html
+
+    @classmethod
+    def from_html(cls, dirty_html: str) -> Self:
+        cleaner = lxml.html.clean.Cleaner(
+            allow_tags=["a", "b", "em", "i", "strong"],
+            safe_attrs_only=True,
+            safe_attrs=["href"],
+        )
+
+        html = cleaner.clean_html(dirty_html).strip()
+        assert isinstance(html, str)
+
+        assert html.startswith("<div>")
+        assert html.endswith("</div>")
+        html = html[5:-6]
+
+        return cls(html)
+
+    @classmethod
+    def from_markdown(cls, markdowntxt: str) -> Self:
+        dirty_html = mistune.html(markdowntxt)
+        assert isinstance(dirty_html, str)
+        return cls.from_html(dirty_html)
+
 
 class Context:
     def __init__(self) -> None:
@@ -219,6 +250,7 @@ class Recipe(Element):
         self,
         ctx: Context,
         title: str,
+        source: Optional[SafeHTML],
         ingredients: List[IngredientInstance],
         subrecipes: List["Recipe"],
         price: Optional["PriceDB"],
@@ -232,6 +264,7 @@ class Recipe(Element):
         self.subrecipes = subrecipes
         self.price = price
         self.stepsections = stepsections
+        self.source = source
 
     @classmethod
     def from_dict(cls, ctx: Context, dct: Dict[str, Any]) -> Self:
@@ -287,9 +320,14 @@ class Recipe(Element):
                     section = StepSection.from_dict(ctx, step)
                     stepsections.append(section)
 
+        source = None
+        if "source" in dct:
+            source = SafeHTML.from_markdown(dct["source"])
+
         return cls(
             ctx=ctx,
             title=dct["title"],
+            source=source,
             ingredients=ingredients,
             subrecipes=subrecipes,
             price=price,