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>
This commit is contained in:
Omar Sánchez Pizarro
2025-10-10 00:03:44 +02:00
parent 08c1577b2a
commit 0245b603b2
9 changed files with 275 additions and 114 deletions

View File

@@ -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}')"