From 247fd37560e0b04ab839ff41b839444008fc813d Mon Sep 17 00:00:00 2001
From: Emi Vasilek <emi.vasilek@gmail.com>
Date: Tue, 21 Nov 2023 03:12:41 +0100
Subject: [PATCH] add a markdown source field to recipe

---
 comfyrecipes/parsing.py            | 38 ++++++++++++++++++++++++++++++
 comfyrecipes/schemas/recipe.json   |  1 +
 comfyrecipes/templates/recipe.html |  1 +
 pyproject.toml                     |  2 +-
 4 files changed, 41 insertions(+), 1 deletion(-)

diff --git a/comfyrecipes/parsing.py b/comfyrecipes/parsing.py
index d702d27..5fb5833 100644
--- a/comfyrecipes/parsing.py
+++ b/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,
diff --git a/comfyrecipes/schemas/recipe.json b/comfyrecipes/schemas/recipe.json
index a8f9876..e2220b0 100644
--- a/comfyrecipes/schemas/recipe.json
+++ b/comfyrecipes/schemas/recipe.json
@@ -8,6 +8,7 @@
     "additionalProperties": false,
     "properties": {
         "title": { "type": "string" },
+        "source": { "type": "string" },
         "ingredients": {
             "type": "array",
             "items": {
diff --git a/comfyrecipes/templates/recipe.html b/comfyrecipes/templates/recipe.html
index 552920d..0b1bde7 100644
--- a/comfyrecipes/templates/recipe.html
+++ b/comfyrecipes/templates/recipe.html
@@ -17,6 +17,7 @@
 {{getrecipe(subrecipe)}}
 {% endfor %}
 <h1>{{rec.title}}</h1>
+{% if rec.source != None %}source: {{rec.source.html|safe}}{%endif%}
 {% if rec.ingredients|length != 0 %}
 <h3>Ingredients</h3>
 {% for ing in rec.ingredients %}
diff --git a/pyproject.toml b/pyproject.toml
index 9cac692..12f1a3f 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -9,7 +9,7 @@ description = "Smart recipe static site generator"
 authors = [{ name = "Emi Vasilek", email = "emi.vasilek@gmail.com" }]
 keywords = ["recipes", "recipe", "static site generator"]
 requires-python = ">= 3.8"
-dependencies = ["jsonschema", "jinja2", "PyYAML"]
+dependencies = ["jsonschema", "jinja2", "PyYAML", "mistune", "lxml"]
 classifiers = [
   "Development Status :: 3 - Alpha",
   "Programming Language :: Python",