Fix Weissman recipe time parsing

automation/bin/kitchenowl_recipe_import.py

@@ -295,6 +295,10 @@ def enrich_recipe_node_from_dom(recipe_node: dict[str, Any], html: str) -> dict[
 
 
 DURATION_RE = re.compile(r'^P(?:T(?:(?P<hours>\d+)H)?(?:(?P<minutes>\d+)M)?(?:(?P<seconds>\d+)S)?)?$')
+TEXT_DURATION_PART_RE = re.compile(
+    r'(?P<value>\d+(?:\.\d+)?)\s*(?P<unit>hours?|hrs?|hr|minutes?|mins?|min|seconds?|secs?|sec|s)\b',
+    re.IGNORECASE,
+)
 
 
 def parse_duration_minutes(value: Any) -> int | None:
@@ -314,6 +318,20 @@ def parse_duration_minutes(value: Any) -> int | None:
         seconds = int(m.group('seconds') or 0)
         total = hours * 60 + minutes + (1 if seconds >= 30 else 0)
         return total or None
+    textual_matches = list(TEXT_DURATION_PART_RE.finditer(value))
+    if textual_matches:
+        total_minutes = 0.0
+        for part in textual_matches:
+            amount = float(part.group('value'))
+            unit = part.group('unit').lower()
+            if unit.startswith(('hour', 'hr')):
+                total_minutes += amount * 60
+            elif unit.startswith(('minute', 'min')):
+                total_minutes += amount
+            else:
+                total_minutes += amount / 60
+        rounded = int(total_minutes + 0.5)
+        return rounded or None
     num = re.search(r'(\d+)', value)
     return int(num.group(1)) if num else None
 

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

@@ -20,6 +20,10 @@ UNIT_WORDS = {
 STOPWORDS = {"fresh", "large", "small", "medium", "optional", "to", "taste", "for", "the", "a", "an"}
 TAG_KEYS = ("recipeCategory", "recipeCuisine", "keywords")
 DURATION_RE = re.compile(r"^P(?:T(?:(?P<hours>\d+)H)?(?:(?P<minutes>\d+)M)?(?:(?P<seconds>\d+)S)?)?$")
+TEXT_DURATION_PART_RE = re.compile(
+    r"(?P<value>\d+(?:\.\d+)?)\s*(?P<unit>hours?|hrs?|hr|minutes?|mins?|min|seconds?|secs?|sec|s)\b",
+    re.IGNORECASE,
+)
 
 
 class ScriptExtractor(HTMLParser):
@@ -410,6 +414,20 @@ def parse_duration_minutes(value: Any) -> int | None:
         seconds = int(match.group("seconds") or 0)
         total = hours * 60 + minutes + (1 if seconds >= 30 else 0)
         return total or None
+    textual_matches = list(TEXT_DURATION_PART_RE.finditer(text))
+    if textual_matches:
+        total_minutes = 0.0
+        for part in textual_matches:
+            amount = float(part.group("value"))
+            unit = part.group("unit").lower()
+            if unit.startswith(("hour", "hr")):
+                total_minutes += amount * 60
+            elif unit.startswith(("minute", "min")):
+                total_minutes += amount
+            else:
+                total_minutes += amount / 60
+        rounded = int(total_minutes + 0.5)
+        return rounded or None
     number = re.search(r"(\d+)", text)
     return int(number.group(1)) if number else None
 

home/doris-kitchen/tests/test_tagging.py

@@ -6,10 +6,16 @@ ROOT = Path(__file__).resolve().parents[1]
 sys.path.insert(0, str(ROOT))
 
 from app.services.tagging import curate_recipe_tags
-from app.services.recipe_importer import normalize_recipe_from_url
+from app.services.recipe_importer import normalize_recipe_from_url, parse_duration_minutes
 
 
 class TaggingTests(unittest.TestCase):
+    def test_textual_durations_cover_weissman_style_time_strings(self):
+        self.assertEqual(parse_duration_minutes('1 hr, 15 min'), 75)
+        self.assertEqual(parse_duration_minutes('45 min + 12 hrs'), 765)
+        self.assertEqual(parse_duration_minutes('30 minutes'), 30)
+        self.assertEqual(parse_duration_minutes('90 sec'), 2)
+
     def test_tater_tots_drop_boilerplate_and_gain_meaningful_tags(self):
         payload = normalize_recipe_from_url(
             'https://www.joshuaweissman.com/recipes/the-best-tater-tots-ever-recipe'
@@ -21,6 +27,9 @@ class TaggingTests(unittest.TestCase):
         self.assertIn('tater-tots', payload['tags'])
         self.assertIn('side-dish', payload['tags'])
         self.assertNotIn('beef', payload['tags'])
+        self.assertEqual(payload['prep_time'], 30)
+        self.assertEqual(payload['cook_time'], 45)
+        self.assertEqual(payload['time'], 75)
 
     def test_halal_chicken_rice_drops_boilerplate_and_infers_cuisine(self):
         payload = normalize_recipe_from_url(
@@ -33,6 +42,9 @@ class TaggingTests(unittest.TestCase):
         self.assertIn('chicken', payload['tags'])
         self.assertIn('rice', payload['tags'])
         self.assertNotIn('skillet', payload['tags'])
+        self.assertEqual(payload['prep_time'], 30)
+        self.assertEqual(payload['cook_time'], 45)
+        self.assertEqual(payload['time'], 75)
 
     def test_existing_machine_tags_can_be_replaced_when_recipe_content_disagrees(self):
         tags = curate_recipe_tags(