integrated notion database

config.yaml

@@ -1,12 +1,14 @@
 PROMPT_ENGINEERING_DICT:
   SYSTEM_INSTRUCTION: |
-    YOU ARE A SEARCH ENGINE HAVING FULL ACCESS TO WEB PAGES, YOU GIVE EXTREMELY DETAILED AND ACCURATE INFORMATION ACCORDING TO USER PROMPTS.
+    YOU ARE A SEARCH ENGINE AND AN INVENTORY MANAGER HAVING FULL ACCESS TO WEB PAGES AND NOTION DATABASE IN JSON, 
+    YOU GIVE EXTREMELY DETAILED AND ACCURATE INFORMATION ACCORDING TO USER PROMPTS.
+
   SYSTEM_RESPONSE: |
     Certainly! I'm here to help. What information are you looking for? 
     Please provide me with a specific topic or question, and I'll do my 
     best to provide you with detailed and accurate information.
 
-  PRE_CONTEXT: NOW YOU ARE SEARCHING THE WEB, AND HERE ARE THE CHUNKS RETRIEVED FROM THE WEB.
+  PRE_CONTEXT: NOW YOU ARE SEARCHING THE WEB, AND HERE ARE THE CHUNKS RETRIEVED FROM THE WEB, YOU ALSO HAVE ACCESS TO INVENTORY DATABASE IN JSON FORMAT.
   POST_CONTEXT: ""
   PRE_PROMPT: NOW ACCORDING TO THE CONTEXT RETRIEVED FROM THE GENERATE THE CONTENT FOR THE FOLLOWING SUBJECT
   POST_PROMPT: PRIORITIZE DATA, FACTS AND STATISTICS OVER PERSONAL EXPERIENCES AND OPINIONS, FOCUS MORE ON STATISTICS AND DATA.

middlewares/utils.py

@@ -1,4 +1,7 @@
 from sentence_transformers import CrossEncoder
+from pprint import pprint, pformat
+from notion_client import Client
+import json
 import math
 import numpy as np
 from middlewares.search_client import SearchClient
@@ -20,6 +23,74 @@ googleSearchClient = SearchClient(
 )
 bingSearchClient = SearchClient("bing", api_key=BING_SEARCH_API_KEY, engine_id=None)
 
+def safe_get(data, dot_chained_keys):
+    '''
+        {'a': {'b': [{'c': 1}]}}
+        safe_get(data, 'a.b.0.c') -> 1
+    '''
+    keys = dot_chained_keys.split('.')
+    for key in keys:
+        try:
+            if isinstance(data, list):
+                data = data[int(key)]
+            else:
+                data = data[key]
+        except (KeyError, TypeError, IndexError):
+            return None
+    return data
+
+def get_notion_data() :
+    integration_token = "secret_lTOe0q9dqqKQLRRb2KJwi7QFSl0vqoztroRFHW6MeQE"
+    notion_database_id = "6c0d877b823a4e3699016fa7083f3006"
+
+    client = Client(auth=integration_token)
+
+    first_db_rows = client.databases.query(notion_database_id)
+    rows = []
+
+
+    for row in first_db_rows['results']:
+            price = safe_get(row, 'properties.($) Per Unit.number')
+            store_link = safe_get(row,  'properties.Store Link.url')
+            supplier_email = safe_get(row,  'properties.Supplier Email.email')
+            exp_del = safe_get(row,  'properties.Expected Delivery.date')
+            
+            collections = safe_get(row,  'properties.Collection.multi_select')
+            collection_names = []
+            for collection in collections :
+            collection_names.append(collection['name'])
+
+            status = safe_get(row,  'properties.Status.select.name')
+            sup_phone = safe_get(row, 'properties.Supplier Phone.phone_number')
+            stock_alert = safe_get(row, 'properties.Stock Alert.status.name') 
+            prod_name = safe_get(row, 'properties.Product .title.0.text.content')
+            sku = safe_get(row, 'properties.SKU.number')
+            shipped_date = safe_get(row, 'properties.Shipped On.date')
+            on_order = safe_get(row, 'properties.On Order.number')
+            on_hand = safe_get(row, 'properties.On Hand.number')
+            size_names = []
+            sizes = safe_get(row, 'properties.Size.multi_select')
+            for size in sizes :
+            size_names.append(size['name'])
+            
+            rows.append({
+            'Price Per unit': price,
+            'Store Link' : store_link,
+            'Supplier Email' : supplier_email,
+            'Expected Delivery' : exp_del,
+            'Collection' : collection_names,
+            'Status' : status,
+            'Supplier Phone' : sup_phone,
+            'Stock Alert' : stock_alert,
+            'Product Name' : prod_name,
+            'SKU' : sku,
+            'Sizes' : size_names,
+            })
+
+    notion_data_string = pformat(rows)
+    return notion_data_string
+
+
 
 def rerank(query, top_k, search_results, chunk_size=512):
     chunks = []
@@ -67,6 +138,7 @@ def gen_augmented_prompt_via_websearch(
     chunk_size=512,
 ):
    
+    notion_data= get_notion_data()
 
     search_results = []
     reranked_results = []
@@ -91,12 +163,16 @@ def gen_augmented_prompt_via_websearch(
     prev_output = prev_output if pass_prev else ""
 
     augmented_prompt = f"""
-    {pre_context}
+    
+{pre_context}
 
     {context}
 
+    {notion_data}
     {post_context}
     
+
+
     {pre_prompt} 
     
     {prompt} 

requirements.txt

@@ -1,3 +1,4 @@
+notion-client==2.2.1
 altair==5.1.2
 asttokens==2.2.1
 attrs==23.1.0