fix: group title on first ingredient + multi-site parser registry

- Fix ingredient groups creating empty entries in Mealie: set title
  field on the first ingredient after the group marker instead
- Refactor scraper with @_register decorator for URL-based site dispatch
- Update README with structured ingredients, groups, MEALIE_INTERNAL_URL

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

app/mealie.py

@@ -145,27 +145,26 @@ class MealieClient:
 
     def _build_payload(self, recipe: dict) -> dict:
         ingredients = []
+        pending_group = ""
         for item in recipe.get("ingredients", []):
             if isinstance(item, dict):
-                # Group header marker
+                # Group header marker — apply title to the next real ingredient
                 if "group" in item and "food" not in item:
-                    ingredients.append({
-                        "referenceId": str(uuid.uuid4()),
-                        "title": item["group"],
-                        "note": "",
-                        "isFood": False,
-                        "disableAmount": True,
-                    })
-                else:
-                    ingredients.append(self._build_ingredient(item))
+                    pending_group = item["group"]
+                    continue
+                ing = self._build_ingredient(item)
             else:
                 # Legacy: plain string
-                ingredients.append({
+                ing = {
                     "referenceId": str(uuid.uuid4()),
                     "note": str(item),
                     "isFood": False,
                     "disableAmount": True,
-                })
+                }
+            if pending_group:
+                ing["title"] = pending_group
+                pending_group = ""
+            ingredients.append(ing)
 
         instructions = []
         for text in recipe.get("instructions", []):

app/scraper.py

@@ -1,6 +1,7 @@
 """Recipe scraper — parses Hungarian recipe sites into a structured dict.
 
-Currently supported: mindmegette.hu
+Each supported site has a parser registered via _PARSERS.
+Unsupported sites fall back to generic schema.org / og-tag extraction.
 """
 
 import re
@@ -12,6 +13,19 @@ _HEADERS = {
     "Accept-Language": "hu-HU,hu;q=0.9,en;q=0.5",
 }
 
+# Maps a substring of the hostname to a parser function.
+# Order matters: first match wins.
+_PARSERS: list[tuple[str, "callable"]] = []
+
+
+def _register(host_substring: str):
+    """Decorator: register a parser for URLs whose hostname contains *host_substring*."""
+    def decorator(fn):
+        _PARSERS.append((host_substring, fn))
+        return fn
+    return decorator
+
+
 # ---------------------------------------------------------------------------
 # Public API
 # ---------------------------------------------------------------------------
@@ -39,11 +53,17 @@ def scrape(url: str) -> dict:
     soup = BeautifulSoup(resp.text, "lxml")
 
     host = _host(url)
-    if "mindmegette" in host:
-        return _parse_mindmegette(soup, url)
-    else:
-        # Fallback: try generic schema.org / og-tag extraction
-        return _parse_generic(soup, url)
+    for substring, parser in _PARSERS:
+        if substring in host:
+            return parser(soup, url)
+
+    # Fallback: try generic schema.org / og-tag extraction
+    return _parse_generic(soup, url)
+
+
+def supported_sites() -> list[str]:
+    """Return list of supported site hostname substrings."""
+    return [s for s, _ in _PARSERS]
 
 
 # ---------------------------------------------------------------------------
@@ -51,6 +71,7 @@ def scrape(url: str) -> dict:
 # ---------------------------------------------------------------------------
 
 
+@_register("mindmegette")
 def _parse_mindmegette(soup: BeautifulSoup, url: str) -> dict:
     title = _og(soup, "og:title") or _text(soup.find("title"))
     # Strip " | Mindmegette.hu" suffix