Updated database and wrote a few scripts to test/demo it 🗃️

color_analysis/download_dataset.py

@@ -0,0 +1,11 @@
+import os
+import pandas as pd
+
+# df = pd.read_csv(f'grailed-dataset/styles_grailed.csv')
+# print(df.head())
+# print(len(df.loc[df["Department"]=="womenswear"]))
+num_hits = 40
+page_range = 500
+for department in ["menswear", "womenswear"]:
+    for page_num in range(page_range):
+        os.system(f"python3 get_grailed_items.py -g {department} -n {num_hits} -p {page_num}") 

color_analysis/get_grailed_items.py

@@ -26,17 +26,23 @@ def download_image(save_path, image_url):
                 handle.write(block)
 
 def get_products(client, product_specifications):
-
+    department = product_specifications["department"]
+    product_specifications.pop("department")
+    if department=="menswear":
+        department = Departments.MENSWEAR
+    else:
+        department = Departments.WOMENSWEAR
     products = client.find_products(
+        sold=True,
+        on_sale=True,
         conditions=[Conditions.IS_GENTLY_USED, Conditions.IS_NEW],
         markets=[Markets.BASIC, Markets.GRAILED],
         locations=[Locations.US, Locations.ASIA, Locations.EUROPE],
-        # department=Departments.MENSWEAR,
-        department=Departments.WOMENSWEAR,
+        department=department,
         **product_specifications
     )
 
-    print(f"Fetched {len(products)} products...")
+    # print(f"Fetched {len(products)} products...")
 
     return products
 
@@ -49,22 +55,23 @@ def download_and_save_products(products):
         image_url = product["cover_photo"]["image_url"]
         save_path = f"{data_path}/images/{product_id}.jpg"
         # print(product["category"].split('.'))
-        data.append(
-            {
-                "Id": product_id,
-                "Department": product["department"],
-                "MasterCategory": product["category_path_size"].split('.')[0],
-                "SubCategory": product["category_path_size"].split('.')[1],
-                "Size": product["category_path_size"].split('.')[2],
-                "Color": product["color"],
-                "Designers": product["designers"],
-                "Hashtags": product["hashtags"],
-                "ProductDisplayName": product["description"],
-                "ItemUrl": product["cover_photo"]["url"]
-            }
-        )
         # check if photo already downloaded
         if not os.path.exists(save_path):
+
+            data.append(
+                {
+                    "Id": product_id,
+                    "Department": product["department"],
+                    "MasterCategory": product["category_path_size"].split('.')[0],
+                    "SubCategory": product["category_path_size"].split('.')[1],
+                    "Size": product["category_path_size"].split('.')[2],
+                    "Color": product["color"],
+                    "Designers": product["designers"],
+                    "Hashtags": product["hashtags"],
+                    "ProductDisplayName": product["description"],
+                    "ItemUrl": product["cover_photo"]["url"]
+                }
+            )
             download_image(save_path, image_url)
             num_downloaded+=1
         else:
@@ -73,15 +80,22 @@ def download_and_save_products(products):
     print(f"Downloaded {num_downloaded} images; already downloaded {already_downloaded} images in results")
     # create empty dataframe
     df = pd.DataFrame(data=data, columns=["Id", "Department", "MasterCategory", "SubCategory", "Size", "Color", "Designers", "Hashtags", "ProductDisplayName", "ItemUrl"])
-    print(df.head())
+    # print(df.head())
+    # print(len(df))
+
+    if os.path.exists(f'{data_path}/styles_grailed.csv'):
+        df_old = pd.read_csv(f'{data_path}/styles_grailed.csv')
+        df = pd.concat([df_old, df])
 
+    # print(len(df))
     df.to_csv(f'{data_path}/styles_grailed.csv', index=False)
 
 if __name__ == "__main__":
     parser = argparse.ArgumentParser(description="Fetching and downloading items from Grailed")
-    # parser.add_argument("--gender", "-g", help="Gender, mens or womens", dest="department")
+    parser.add_argument("--gender", "-g", help="Gender, mens or womens", dest="department")
     parser.add_argument("--query_search", "-q", help="Query to look up", dest="query_search")
     parser.add_argument("--num_hits", "-n", help="Number of hits to pull up", dest="hits_per_page", default=1)
+    parser.add_argument("--page", "-p", help="Page number", dest="page", default=1)
     # parser.add_argument("--categories", "-c", helper="Iterable of categories to include")
     parser.add_argument("--price_from", "-l", help="Min price", dest="price_from", default=0)
     parser.add_argument("--price_to", "-m", help="Max price", dest="price_to", default=100)

main.py

@@ -14,7 +14,7 @@ class AuraRequest(BaseModel):
     Department: Literal['menswear', 'womenswear']
     ColorSeason: Literal['autumn', 'winter', 'spring', 'summer']
     # MasterCategory: Literal['Tops', 'Bottoms', 'Outerwear', 'Footwear', 'Tailoring', 'Accessories']
-    num_outfits: int = 1
+    n: int = 1
 
 
 class Items(SQLModel, table=True):
@@ -132,28 +132,28 @@ async def generate_outfit(
     aura_request: AuraRequest
 ):
     # Get tops
-    tops = session.exec(select(Items).where(Items.ColorSeason == aura_request.ColorSeason).where(Items.Department == aura_request.Department).where(or_(Items.MasterCategory == "tops", Items.MasterCategory == "womens_tops")).order_by(func.random()).limit(aura_request.num_outfits)).all()
+    tops = session.exec(select(Items).where(Items.ColorSeason == aura_request.ColorSeason).where(Items.Department == aura_request.Department).where(or_(Items.MasterCategory == "tops", Items.MasterCategory == "womens_tops")).order_by(func.random()).limit(aura_request.n)).all()
     if not tops:
         raise HTTPException(status_code=404, detail="No matching tops found")
 
     # Get bottoms
-    bottoms = session.exec(select(Items).where(Items.ColorSeason == aura_request.ColorSeason).where(Items.Department == aura_request.Department).where(or_(Items.MasterCategory == "bottoms", Items.MasterCategory == "womens_bottoms")).order_by(func.random()).limit(aura_request.num_outfits)).all()
+    bottoms = session.exec(select(Items).where(Items.ColorSeason == aura_request.ColorSeason).where(Items.Department == aura_request.Department).where(or_(Items.MasterCategory == "bottoms", Items.MasterCategory == "womens_bottoms")).order_by(func.random()).limit(aura_request.n)).all()
     if not bottoms:
         raise HTTPException(status_code=404, detail="No matching bottoms found")
 
     # Get shoes
-    shoes = session.exec(select(Items).where(Items.ColorSeason == aura_request.ColorSeason).where(Items.Department == aura_request.Department).where(or_(Items.MasterCategory == "footwear", Items.MasterCategory == "womens_footwear")).order_by(func.random()).limit(aura_request.num_outfits)).all()
+    shoes = session.exec(select(Items).where(Items.ColorSeason == aura_request.ColorSeason).where(Items.Department == aura_request.Department).where(or_(Items.MasterCategory == "footwear", Items.MasterCategory == "womens_footwear")).order_by(func.random()).limit(aura_request.n)).all()
     if not bottoms:
         raise HTTPException(status_code=404, detail="No matching shoes found")
 
     # Get outerwear
-    outerwear = session.exec(select(Items).where(Items.ColorSeason == aura_request.ColorSeason).where(Items.Department == aura_request.Department).where(or_(Items.MasterCategory == "outerwear", Items.MasterCategory == "outerwear")).order_by(func.random()).limit(aura_request.num_outfits)).all()
+    outerwear = session.exec(select(Items).where(Items.ColorSeason == aura_request.ColorSeason).where(Items.Department == aura_request.Department).where(or_(Items.MasterCategory == "outerwear", Items.MasterCategory == "womens_outerwear")).order_by(func.random()).limit(aura_request.n)).all()
     if not outerwear:
         print("No matching outerwear found")
         # raise HTTPException(status_code=404, detail="No matching outerwear found")
 
     # Get accessories
-    accessories = session.exec(select(Items).where(Items.ColorSeason == aura_request.ColorSeason).where(Items.Department == aura_request.Department).where(or_(Items.MasterCategory == "accessories", Items.MasterCategory == "womens_accessories")).order_by(func.random()).limit(aura_request.num_outfits)).all()
+    accessories = session.exec(select(Items).where(Items.ColorSeason == aura_request.ColorSeason).where(Items.Department == aura_request.Department).where(or_(Items.MasterCategory == "accessories", Items.MasterCategory == "womens_accessories")).order_by(func.random()).limit(aura_request.n)).all()
     if not bottoms:
         print("No matching accessories found")
         # raise HTTPException(status_code=404, detail="No matching outerwear found")

tests/database_contents.sql

@@ -0,0 +1,49 @@
+SELECT COUNT(*) FROM items WHERE ColorSeason="spring" AND MasterCategory="tops"; --217
+SELECT COUNT(*) FROM items WHERE ColorSeason="summer" AND MasterCategory="tops"; --132
+SELECT COUNT(*) FROM items WHERE ColorSeason="autumn" AND MasterCategory="tops"; --220
+SELECT COUNT(*) FROM items WHERE ColorSeason="winter" AND MasterCategory="tops"; --160
+
+SELECT COUNT(*) FROM items WHERE ColorSeason="spring" AND MasterCategory="bottoms"; --84
+SELECT COUNT(*) FROM items WHERE ColorSeason="summer" AND MasterCategory="bottoms"; --54
+SELECT COUNT(*) FROM items WHERE ColorSeason="autumn" AND MasterCategory="bottoms"; --72
+SELECT COUNT(*) FROM items WHERE ColorSeason="winter" AND MasterCategory="bottoms"; --54
+
+SELECT COUNT(*) FROM items WHERE ColorSeason="spring" AND MasterCategory="footwear"; --13
+SELECT COUNT(*) FROM items WHERE ColorSeason="summer" AND MasterCategory="footwear"; --7
+SELECT COUNT(*) FROM items WHERE ColorSeason="autumn" AND MasterCategory="footwear"; --22
+SELECT COUNT(*) FROM items WHERE ColorSeason="winter" AND MasterCategory="footwear"; --11
+
+SELECT COUNT(*) FROM items WHERE ColorSeason="spring" AND MasterCategory="outerwear"; --37
+SELECT COUNT(*) FROM items WHERE ColorSeason="summer" AND MasterCategory="outerwear"; --30
+SELECT COUNT(*) FROM items WHERE ColorSeason="autumn" AND MasterCategory="outerwear"; --51
+SELECT COUNT(*) FROM items WHERE ColorSeason="winter" AND MasterCategory="outerwear"; --35
+
+SELECT COUNT(*) FROM items WHERE ColorSeason="spring" AND MasterCategory="accessories"; --160
+SELECT COUNT(*) FROM items WHERE ColorSeason="summer" AND MasterCategory="accessories"; --87
+SELECT COUNT(*) FROM items WHERE ColorSeason="autumn" AND MasterCategory="accessories"; --137
+SELECT COUNT(*) FROM items WHERE ColorSeason="winter" AND MasterCategory="accessories"; --99
+
+SELECT COUNT(*) FROM items WHERE ColorSeason="spring" AND MasterCategory="womens_tops"; --231
+SELECT COUNT(*) FROM items WHERE ColorSeason="summer" AND MasterCategory="womens_tops"; --125
+SELECT COUNT(*) FROM items WHERE ColorSeason="autumn" AND MasterCategory="womens_tops"; --163
+SELECT COUNT(*) FROM items WHERE ColorSeason="winter" AND MasterCategory="womens_tops"; --133
+
+SELECT COUNT(*) FROM items WHERE ColorSeason="spring" AND MasterCategory="womens_bottoms"; --116 
+SELECT COUNT(*) FROM items WHERE ColorSeason="summer" AND MasterCategory="womens_bottoms"; --69
+SELECT COUNT(*) FROM items WHERE ColorSeason="autumn" AND MasterCategory="womens_bottoms"; --85
+SELECT COUNT(*) FROM items WHERE ColorSeason="winter" AND MasterCategory="womens_bottoms"; --57
+
+SELECT COUNT(*) FROM items WHERE ColorSeason="spring" AND MasterCategory="womens_footwear"; --56
+SELECT COUNT(*) FROM items WHERE ColorSeason="summer" AND MasterCategory="womens_footwear"; --66
+SELECT COUNT(*) FROM items WHERE ColorSeason="autumn" AND MasterCategory="womens_footwear"; --43
+SELECT COUNT(*) FROM items WHERE ColorSeason="winter" AND MasterCategory="womens_footwear"; --39
+
+SELECT COUNT(*) FROM items WHERE ColorSeason="spring" AND MasterCategory="womens_outerwear"; --81
+SELECT COUNT(*) FROM items WHERE ColorSeason="summer" AND MasterCategory="womens_outerwear"; --63
+SELECT COUNT(*) FROM items WHERE ColorSeason="autumn" AND MasterCategory="womens_outerwear"; --56
+SELECT COUNT(*) FROM items WHERE ColorSeason="winter" AND MasterCategory="womens_outerwear"; --41
+
+SELECT COUNT(*) FROM items WHERE ColorSeason="spring" AND MasterCategory="womens_accessories"; --81
+SELECT COUNT(*) FROM items WHERE ColorSeason="summer" AND MasterCategory="womens_accessories"; --66
+SELECT COUNT(*) FROM items WHERE ColorSeason="autumn" AND MasterCategory="womens_accessories"; --67
+SELECT COUNT(*) FROM items WHERE ColorSeason="winter" AND MasterCategory="womens_accessories"; --38