Fix picnicapi search (again)

python_picnic_api/client.py

@@ -1,6 +1,6 @@
 from hashlib import md5
 
-from .helper import _tree_generator, _url_generator, _get_category_name
+from .helper import _tree_generator, _url_generator, _get_category_name, _extract_search_results
 from .session import PicnicAPISession, PicnicAuthError
 
 DEFAULT_URL = "https://storefront-prod.{}.picnicinternational.com/api/{}"
@@ -23,7 +23,7 @@ def __init__(
         # Login if not authenticated
         if not self.session.authenticated and username and password:
             self.login(username, password)
-        
+
         self.high_level_categories = None
 
     def initialize_high_level_categories(self):
@@ -36,8 +36,8 @@ def _get(self, path: str, add_picnic_headers=False):
 
         # Make the request, add special picnic headers if needed
         headers = {
-            "x-picnic-agent": "30100;1.15.183-14941;",
-            "x-picnic-did": "00DE6414C744E7CB"
+            "x-picnic-agent": "30100;1.15.232-15154;",
+            "x-picnic-did": "3C417201548B2E3B"
         } if add_picnic_headers else None
         response = self.session.get(url, headers=headers).json()
 
@@ -77,8 +77,9 @@ def get_user(self):
         return self._get("/user")
 
     def search(self, term: str):
-        path = "/search?search_term=" + term
-        return self._get(path)
+        path = f"/pages/search-page-results?search_term={term}"
+        raw_results = self._get(path, add_picnic_headers=True)
+        return _extract_search_results(raw_results)
 
     def get_lists(self, list_id: str = None):
         if list_id:
@@ -101,7 +102,7 @@ def get_sublist(self, list_id: str, sublist_id: str) -> list:
 
     def get_cart(self):
         return self._get("/cart")
-        
+
     def get_article(self, article_id: str, add_category_name=False):
         path = "/articles/" + article_id
         article = self._get(path)
@@ -111,7 +112,7 @@ def get_article(self, article_id: str, add_category_name=False):
                 category_name=_get_category_name(article['category_link'], self.high_level_categories)
             )
         return article
-        
+
     def get_article_category(self, article_id: str):
         path = "/articles/" + article_id + "/category"
         return self._get(path)

python_picnic_api/helper.py

@@ -1,4 +1,6 @@
+import json
 import re
+from typing import List, Dict, Any, Optional
 
 # prefix components:
 space = "    "
@@ -10,6 +12,9 @@
 IMAGE_SIZES = ["small", "medium", "regular", "large", "extra-large"]
 IMAGE_BASE_URL = "https://storefront-prod.nl.picnicinternational.com/static/images"
 
+SOLE_ARTICLE_ID_PATTERN = re.compile(r'"sole_article_id":"(\w+)"')
+
+
 def _tree_generator(response: list, prefix: str = ""):
     """A recursive tree generator,
     will yield a visual tree structure line by line
@@ -37,42 +42,70 @@ def _url_generator(url: str, country_code: str, api_version: str):
     return url.format(country_code.lower(), api_version)
 
 
-def _get_category_id_from_link(category_link: str) -> str:
-    pattern = r'categories/(\d+)'
+def _get_category_id_from_link(category_link: str) -> Optional[str]:
+    pattern = r"categories/(\d+)"
     first_number = re.search(pattern, category_link)
     if first_number:
         result = str(first_number.group(1))
         return result
     else:
         return None
-    
-    
-def _get_category_name(category_link: str, categories: list) -> str:
+
+
+def _get_category_name(category_link: str, categories: list) -> Optional[str]:
     category_id = _get_category_id_from_link(category_link)
     if category_id:
-        category = next((item for item in categories if item["id"] == category_id), None)
+        category = next(
+            (item for item in categories if item["id"] == category_id), None
+        )
         if category:
             return category["name"]
         else:
             return None
     else:
         return None
 
+
 def get_recipe_image(id: str, size="regular"):
     sizes = IMAGE_SIZES + ["1250x1250"]
     assert size in sizes, "size must be one of: " + ", ".join(sizes)
     return f"{IMAGE_BASE_URL}/recipes/{id}/{size}.png"
 
 
 def get_image(id: str, size="regular", suffix="webp"):
-    assert "tile" in size if suffix == "webp" else True, (
-        "webp format only supports tile sizes"
-    )
+    assert (
+        "tile" in size if suffix == "webp" else True
+    ), "webp format only supports tile sizes"
     assert suffix in ["webp", "png"], "suffix must be webp or png"
     sizes = IMAGE_SIZES + [f"tile-{size}" for size in IMAGE_SIZES]
 
-    assert size in sizes, (
-        "size must be one of: " + ", ".join(sizes)
-    )
+    assert size in sizes, "size must be one of: " + ", ".join(sizes)
     return f"{IMAGE_BASE_URL}/{id}/{size}.{suffix}"
 
+def _extract_search_results(raw_results, max_items: int = 10):
+    """Extract search results from the nested dictionary structure returned by Picnic search.
+    Number of max items can be defined to reduce excessive nested search"""
+    search_results = []
+
+    def find_articles(node):
+        if len(search_results) >= max_items:
+            return
+
+        content = node.get("content", {})
+        if content.get("type") == "SELLING_UNIT_TILE" and "sellingUnit" in content:
+            selling_unit = content["sellingUnit"]
+            sole_article_ids = SOLE_ARTICLE_ID_PATTERN.findall(json.dumps(node))
+            sole_article_id = sole_article_ids[0] if sole_article_ids else None
+            result_entry = {
+                **selling_unit,
+                "sole_article_id": sole_article_id,
+            }
+            search_results.append(result_entry)
+
+        for child in node.get("children", []):
+            find_articles(child)
+
+    body = raw_results.get("body", {})
+    find_articles(body.get("child", {}))
+
+    return [{"items": search_results}]

tests/test_client.py

@@ -6,7 +6,7 @@
 from python_picnic_api.session import PicnicAuthError
 
 PICNIC_HEADERS = {
-    "x-picnic-agent": "30100;1.15.77-10293",
+    "x-picnic-agent": "30100;1.15.232-15154",
     "x-picnic-did": "3C417201548B2E3B",
 }
 
@@ -34,7 +34,7 @@ def test_login_credentials(self):
         PicnicAPI(username='[email protected]', password='test')
         self.session_mock().post.assert_called_with(
             self.expected_base_url + '/user/login',
-            json={'key': '[email protected]', 'secret': '098f6bcd4621d373cade4e832627b4f6', "client_id": 1}
+            json={'key': '[email protected]', 'secret': '098f6bcd4621d373cade4e832627b4f6', "client_id": 30100}
         )
 
     def test_login_auth_token(self):
@@ -83,7 +83,7 @@ def test_get_user(self):
     def test_search(self):
         self.client.search("test-product")
         self.session_mock().get.assert_called_with(
-            self.expected_base_url + "/search?search_term=test-product", headers=None
+            self.expected_base_url + "/pages/search-page-results?search_term=test-product", headers=PICNIC_HEADERS
         )
 
     def test_get_lists(self):