feat: default values, general item excludes, images, queueManager to manage multi worker messaging to telegram to prevent too many connections

Signed-off-by: Omar Sánchez Pizarro <omar.sanchez@pistacero.net>

datalayer/general_monitor.py

@@ -0,0 +1,32 @@
+class GeneralMonitor:
+    def __init__(self, title_exclude, description_exclude, title_must_include, description_must_include, title_first_word_exclude):
+        self._title_exclude = title_exclude
+        self._description_exclude = description_exclude
+        self._title_must_include = title_must_include
+        self._description_must_include = description_must_include
+        self._title_first_word_exclude = title_first_word_exclude
+    
+    @classmethod
+    def load_from_json(cls, json_data):
+        return cls(
+            json_data.get('title_exclude', []),
+            json_data.get('description_exclude', []),   
+            json_data.get('title_must_include', []), 
+            json_data.get('description_must_include', []), 
+            json_data.get('title_first_word_exclude', [])
+        )
+    
+    def get_title_exclude(self):
+        return self._title_exclude
+    
+    def get_description_exclude(self):
+        return self._description_exclude
+    
+    def get_title_must_include(self):
+        return self._title_must_include
+    
+    def get_description_must_include(self):
+        return self._description_must_include
+    
+    def get_title_first_word_exclude(self):
+        return self._title_first_word_exclude

datalayer/item_monitor.py

@@ -1,9 +1,10 @@
 
 class ItemMonitor:
-    def __init__(self, search_query, latitude, longitude, max_distance,
+    def __init__(self, name,search_query, latitude, longitude, max_distance,
                  condition, min_price, max_price, title_exclude,
                  description_exclude, title_must_include, description_must_include,
-                 title_first_word_exclude):
+                 title_first_word_exclude, check_every):
+        self._name = name
         self._search_query = search_query
         self._latitude = latitude
         self._longitude = longitude
@@ -16,24 +17,33 @@ class ItemMonitor:
         self._title_must_include = title_must_include
         self._description_must_include = description_must_include
         self._title_first_word_exclude = title_first_word_exclude
-    
+        self._check_every = check_every
     @classmethod
     def load_from_json(cls, json_data):
+        # search_query is mandatory
+        if 'search_query' not in json_data:
+            raise ValueError("Missing mandatory field: search_query")
+
         return cls(
+            json_data['name'],
             json_data['search_query'],
-            json_data['latitude'],
-            json_data['longitude'],
-            json_data['max_distance'],
-            json_data['condition'],
-            json_data['min_price'],
-            json_data['max_price'],
-            json_data['title_exclude'],
-            json_data['description_exclude'],
-            json_data['title_must_include'],
-            json_data['description_must_include'],
-            json_data['title_first_word_exclude']
+            json_data.get('latitude', 0),
+            json_data.get('longitude', 0),
+            json_data.get('max_distance', 0),
+            json_data.get('condition', 'all'),
+            json_data.get('min_price', 0),
+            json_data.get('max_price', 0),
+            json_data.get('title_exclude', []),
+            json_data.get('description_exclude', []),
+            json_data.get('title_must_include', []),
+            json_data.get('description_must_include', []),
+            json_data.get('title_first_word_exclude', []),
+            json_data.get('check_every', 30)
         )
 
+    def get_name(self):
+        return self._name
+
     def get_search_query(self):
         return self._search_query
 
@@ -68,4 +78,7 @@ class ItemMonitor:
         return self._description_must_include
 
     def get_title_first_word_exclude(self):
-        return self._title_first_word_exclude
+        return self._title_first_word_exclude
+
+    def get_check_every(self):
+        return self._check_every

datalayer/wallapop_article.py

@@ -1,6 +1,8 @@
+import datetime
+import pandas as pd
 
 class WallapopArticle:
-    def __init__(self, id, title, description, price, currency, location, allows_shipping, url):
+    def __init__(self, id, title, description, price, currency, location, allows_shipping, url, images, modified_at):
         self._id = id
         self._title = title
         self._description = description
@@ -9,7 +11,8 @@ class WallapopArticle:
         self._location = location
         self._allows_shipping = allows_shipping
         self._url = url
-
+        self._images = images
+        self._modified_at = modified_at
     @classmethod
     def load_from_json(cls, json_data):
         return cls(
@@ -20,7 +23,9 @@ class WallapopArticle:
             json_data['price']['currency'],
             json_data['location']['city'],
             json_data['shipping']['user_allows_shipping'],
-            json_data['web_slug']
+            json_data['web_slug'],
+            json_data['images'],
+            json_data['modified_at']
         )
 
     def get_id(self):
@@ -30,7 +35,8 @@ class WallapopArticle:
         return self._title
 
     def get_description(self):
-        return self._description
+        #return only 500 characters
+        return self._description[:500] + "..." if len(self._description) > 500 else self._description
 
     def get_price(self):
         return self._price
@@ -42,14 +48,23 @@ class WallapopArticle:
         return self._location
 
     def get_allows_shipping(self):
-        return self._allows_shipping
+        return "✅" if self._allows_shipping else "❌"
 
     def get_url(self):
         return self._url
 
+    def get_images(self):
+        return [img['urls']['medium'] for img in self._images[:3]]
+
+    def get_modified_at(self):
+        # Convert timestamp in milliseconds to datetime string "YYYY-MM-DD HH:MM:SS"
+        ts = int(self._modified_at)
+        dt = datetime.date.fromtimestamp(ts / 1000)
+        return dt.strftime("%Y-%m-%d %H:%M:%S")
+
     def __eq__(self, article):
         return self.get_id() == article.get_id()
 
     def __str__(self):
         return f"Article(id={self._id}, title='{self._title}', " \
-               f"price={self._price} {self._currency}, url='{self._url}')"
+               f"price={self._price} {self._currency}, url='{self._url}', modified_at='{self._modified_at}')"