feat: improve KitchenOwl import parsing and tagging

home/doris-kitchen/app/services/recipe_importer.py

@@ -57,6 +57,91 @@ class ScriptExtractor(HTMLParser):
             self.title = (self.title or "") + data
 
 
+
+
+class RecipeDomExtractor(HTMLParser):
+    def __init__(self) -> None:
+        super().__init__()
+        self._stack: list[str] = []
+        self._ingredient_depth: int | None = None
+        self._direction_depth: int | None = None
+        self._current_li: list[str] | None = None
+        self._current_heading: list[str] | None = None
+        self.ingredients: list[str] = []
+        self.sections: list[dict[str, Any]] = []
+        self._current_section: dict[str, Any] | None = None
+
+    def handle_starttag(self, tag: str, attrs: list[tuple[str, str | None]]) -> None:
+        attrs_d = dict(attrs)
+        classes = set((attrs_d.get("class") or "").split())
+        self._stack.append(tag)
+
+        if "ingredients-list" in classes:
+            self._ingredient_depth = len(self._stack)
+        if "directions-list" in classes:
+            self._direction_depth = len(self._stack)
+
+        if self.in_ingredients and tag == "li":
+            self._current_li = []
+        elif self.in_directions and tag == "li":
+            self._current_li = []
+
+        if self.in_directions and tag == "p":
+            self._current_heading = []
+
+    def handle_endtag(self, tag: str) -> None:
+        if self.in_ingredients and tag == "li" and self._current_li is not None:
+            text = normalize_whitespace("".join(self._current_li))
+            if text:
+                self.ingredients.append(text)
+            self._current_li = None
+        elif self.in_directions and tag == "li" and self._current_li is not None:
+            text = normalize_whitespace("".join(self._current_li))
+            if text:
+                self._ensure_section().setdefault("itemListElement", []).append({"@type": "HowToStep", "text": text})
+            self._current_li = None
+
+        if self.in_directions and tag == "p" and self._current_heading is not None:
+            heading = normalize_whitespace("".join(self._current_heading)).rstrip(":*").strip()
+            if heading:
+                self._current_section = {"@type": "HowToSection", "name": heading, "itemListElement": []}
+                self.sections.append(self._current_section)
+            self._current_heading = None
+
+        if self._stack:
+            self._stack.pop()
+
+        if self._ingredient_depth is not None and len(self._stack) < self._ingredient_depth:
+            self._ingredient_depth = None
+        if self._direction_depth is not None and len(self._stack) < self._direction_depth:
+            self._direction_depth = None
+
+    def handle_data(self, data: str) -> None:
+        if self.in_ingredients and self._current_li is not None:
+            self._current_li.append(data)
+            return
+        if not self.in_directions:
+            return
+        if self._current_li is not None:
+            self._current_li.append(data)
+        elif self._current_heading is not None:
+            self._current_heading.append(data)
+
+    @property
+    def in_ingredients(self) -> bool:
+        return self._ingredient_depth is not None
+
+    @property
+    def in_directions(self) -> bool:
+        return self._direction_depth is not None
+
+    def _ensure_section(self) -> dict[str, Any]:
+        if self._current_section is None:
+            self._current_section = {"@type": "HowToSection", "name": "Instructions", "itemListElement": []}
+            self.sections.append(self._current_section)
+        return self._current_section
+
+
 def fetch_url(url: str, timeout: int = 30) -> str:
     req = request.Request(
         url,
@@ -134,6 +219,17 @@ def parse_json_ld_recipe(html: str) -> tuple[dict[str, Any] | None, str | None]:
         nodes.extend(flatten_graph(data))
     return choose_recipe_node(nodes), (parser.title.strip() if parser.title else None)
 
+def enrich_recipe_node_from_dom(recipe_node: dict[str, Any], html: str) -> dict[str, Any]:
+    extractor = RecipeDomExtractor()
+    extractor.feed(html)
+    enriched = dict(recipe_node)
+    if extractor.ingredients and not (enriched.get("recipeIngredient") or []):
+        enriched["recipeIngredient"] = extractor.ingredients
+    if extractor.sections and not extract_instruction_steps(enriched.get("recipeInstructions")):
+        enriched["recipeInstructions"] = extractor.sections
+    return enriched
+
+
 
 def coerce_text(value: Any) -> str:
     if value is None:
@@ -223,12 +319,18 @@ def build_description(recipe_node: dict[str, Any]) -> str:
     if steps:
         rendered_steps = []
         step_number = 1
-        for idx, step in enumerate(steps, start=1):
+        in_notes = False
+        for step in steps:
             if step.endswith(":"):
                 rendered_steps.append(step)
+                in_notes = step[:-1].strip().lower() == "notes"
+                step_number = 1
             else:
-                rendered_steps.append(f"{step_number}. {step}")
-                step_number += 1
+                if in_notes:
+                    rendered_steps.append(f"- {step}")
+                else:
+                    rendered_steps.append(f"{step_number}. {step}")
+                    step_number += 1
         parts.append("\n".join(rendered_steps))
     elif intro_parts[1:]:
         parts.append("\n".join(intro_parts[1:]))
@@ -376,4 +478,5 @@ def normalize_recipe_from_url(url: str, timeout: int = 30) -> dict[str, Any]:
     recipe_node, title = parse_json_ld_recipe(html)
     if not recipe_node:
         raise RuntimeError(f"No schema.org Recipe JSON-LD found for {url}")
+    recipe_node = enrich_recipe_node_from_dom(recipe_node, html)
     return normalize_recipe(url, recipe_node, title)