✨ use tandoor json, directly from the api

2_to_markdown.py

@@ -13,28 +13,44 @@ from config import *
 include_title = False
 
 
-def valid_duration(duration):
-    if not duration:
-        return False
-    if duration == "0":
-        return False
-    return True
-
 def valid_description(description):
     if not description:
         return False
     return True
 
-def parse_duration(duration):
-    if not duration:
-        return ""
-    match = re.match(r"PT(?:(\d+)H)?(?:(\d+)M)?", duration)
-    if not match:
-        return duration
-    hours, minutes = match.groups()
-    hours = f"{int(hours)}h " if hours else ""
-    minutes = f"{int(minutes)}m" if minutes else ""
-    return f"{hours}{minutes}".strip()
+
+def normalize_amount(a):
+    if a == round(a):
+        return round(a)
+    else:
+        return a
+
+def md_for_step_ingredients(step):
+    md = []
+    for ingredient in step['ingredients']:
+        i_amount = ingredient["amount"]
+        i_amount = normalize_amount(i_amount)
+
+        if not ingredient.get("unit"):
+            i_unit = ""
+        else:
+            i_unit = ingredient["unit"]["name"]
+        # TODO → 1_clean_json.py
+        if i_unit == "g / Gramm":
+            i_unit = "g"
+        elif i_unit == "kg / Kilogramm":
+            i_unit = "kg"
+
+        if not ingredient.get("food"):
+            # TODO → 1_clean_json.py
+            continue
+        i_name = ingredient["food"]["name"]
+
+        if i_amount:
+            md.append(f"- {i_amount} {i_unit} {i_name}")
+        else:
+            md.append(f"- {i_name}")
+    return md
 
 def format_recipe_to_markdown(recipe):
     md = []
@@ -49,68 +65,50 @@ def format_recipe_to_markdown(recipe):
 
     # Details
     details_parts = []
-    if valid_duration(recipe.get('prepTime')):
-        details_parts.append(f"Prep time: {parse_duration(recipe['prepTime'])}")
-    if valid_duration(recipe.get('cookTime')):
-        details_parts.append(f"Cook time: {parse_duration(recipe['cookTime'])}")
-    if valid_duration(recipe.get('totalTime')):
-        details_parts.append(f"Total time: {parse_duration(recipe['totalTime'])}")
-    if 'recipeYield' in recipe:
-        details_parts.append(f"Portionen: {recipe['recipeYield']}")
+    prep_time = recipe.get("working_time") or 0
+    total_time = (recipe.get("working_time") or 0) + (recipe.get("waiting_time") or 0)
+    if prep_time:
+         details_parts.append(f"Prep time: {prep_time}")
+    if total_time:
+         details_parts.append(f"Total time: {total_time}")
+    if 'servings' in recipe:
+        details_parts.append(f"Portionen: {recipe['servings']}")
+        # TODO servings_text
     if details_parts:
         for p in details_parts:
             md.append(f"* {p}\n")
 
     # Ingredients
-    if 'recipeIngredient' in recipe:
-        md.append("\n## Zutaten")
-        for ingredient in recipe['recipeIngredient']:
-            md.append(f"- {ingredient}")
+    md.append("\n## Zutaten (gesamt)")
+    for step in recipe["steps"]:
+        md += md_for_step_ingredients(step)
 
     # Instructions
-    if 'recipeInstructions' in recipe:
+    if 'steps' in recipe:
         md.append("\n## Zubereitung")
-        instructions = recipe['recipeInstructions']
-        if isinstance(instructions, list):
-            for i, step in enumerate(instructions, 1):
-                assert isinstance(step, dict)
-                text = step.get('text', '').strip()
+        steps = recipe['steps']
+        for i, step in enumerate(steps, 1):
+            step_name = step["name"] or "<span></span>"
+            md.append(f"{i}. {step_name}")
+            md += md_for_step_ingredients(step)
+            md.append(f"{step["instructions_markdown"]}")
 
-                # "ingredient-only" steps shouldn't exist, but they do
-                if not text:
-                    continue
-
-                md.append(f"{i}.\n{text}")
-        else:
-            for i, line in enumerate(instructions.strip().split('\n'), 1):
-                if line.strip():
-                    md.append(f"{i}. {line.strip()}")
-
-    # Nutrition
-    if 'nutrition' in recipe:
-        md.append("\n## Nutrition")
-        for key, value in recipe['nutrition'].items():
-            if key != "@type":
-                md.append(f"- **{key.replace('_', ' ').capitalize()}**: {value}")
+    # Nutrition TODO
+    # if 'nutrition' in recipe and recipe['nutrition']:
+    #     md.append("\n## Nutrition")
+    #     for key, value in recipe['nutrition'].items():
+    #         if key != "@type":
+    #             md.append(f"- **{key.replace('_', ' ').capitalize()}**: {value}")
 
     return '\n'.join(md) + '\n'
 
-def main(json_file):
-    with open(json_file, 'r', encoding='utf-8') as f:
-        data = json.load(f)
+def main():
 
-    # If the JSON-LD is embedded in @graph or a list
-    # if isinstance(data, list):
-    #     recipe = next((item for item in data if item.get('@type') == 'Recipe'), data[0])
-    # elif '@graph' in data:
-    #     recipe = next((item for item in data['@graph'] if item.get('@type') == 'Recipe'), data['@graph'][0])
-    if isinstance(data, list) and data[0].get('@type') == 'Recipe':
-        recipes = data
-    elif data.get('@type') == 'Recipe':
-        recipes = [data]
-    else:
-        print("No Recipe object found in JSON.")
-        return
+    recipes = []
+    for json_file in os.listdir(OUTDIR_JSON):
+        with open(os.path.join(OUTDIR_JSON, json_file), 'r', encoding='utf-8') as f:
+            data = json.load(f)
+        recipes.append(data)
 
     for recipe in tqdm(recipes):
         markdown_fn = f"{recipe.get('name', 'Untitled Recipe')}.md"
@@ -121,8 +119,5 @@ def main(json_file):
             f.write(markdown)
 
 if __name__ == "__main__":
-    if len(sys.argv) != 2:
-        print("Usage: python recipe_to_md.py <path_to_recipe_json>")
-    else:
-        main(sys.argv[1])
+    main()