Refactor project structure to organize APIs (#24)
* Refactor project structure "api" * fix bug conversione delle valute fiat in stablecoin in BinanceWrapper * Refactor: WrapperHandler for managing API wrappers with retry logic; update related modules and tests * Refactor: Update ProductInfo and Price classes to include aggregation methods; remove standalone aggregation functions * fix docs
This commit was merged in pull request #24.
This commit is contained in:
committed by
GitHub
parent
517842c834
commit
093a7f5a48
78
src/app/api/news/__init__.py
Normal file
78
src/app/api/news/__init__.py
Normal file
@@ -0,0 +1,78 @@
|
||||
from agno.tools import Toolkit
|
||||
from app.api.wrapper_handler import WrapperHandler
|
||||
from app.api.base.news import NewsWrapper, Article
|
||||
from app.api.news.news_api import NewsApiWrapper
|
||||
from app.api.news.googlenews import GoogleNewsWrapper
|
||||
from app.api.news.cryptopanic_api import CryptoPanicWrapper
|
||||
from app.api.news.duckduckgo import DuckDuckGoWrapper
|
||||
|
||||
__all__ = ["NewsAPIsTool", "NewsApiWrapper", "GoogleNewsWrapper", "CryptoPanicWrapper", "DuckDuckGoWrapper", "Article"]
|
||||
|
||||
|
||||
class NewsAPIsTool(NewsWrapper, Toolkit):
|
||||
"""
|
||||
Aggregates multiple news API wrappers and manages them using WrapperHandler.
|
||||
This class supports retrieving top headlines and latest news articles by querying multiple sources:
|
||||
- GoogleNewsWrapper
|
||||
- DuckDuckGoWrapper
|
||||
- NewsApiWrapper
|
||||
- CryptoPanicWrapper
|
||||
|
||||
By default, it returns results from the first successful wrapper.
|
||||
Optionally, it can be configured to collect articles from all wrappers.
|
||||
If no wrapper succeeds, an exception is raised.
|
||||
"""
|
||||
|
||||
def __init__(self):
|
||||
"""
|
||||
Initialize the NewsAPIsTool with multiple news API wrappers.
|
||||
The tool uses WrapperHandler to manage and invoke the different news API wrappers.
|
||||
The following wrappers are included in this order:
|
||||
- GoogleNewsWrapper.
|
||||
- DuckDuckGoWrapper.
|
||||
- NewsApiWrapper.
|
||||
- CryptoPanicWrapper.
|
||||
"""
|
||||
wrappers: list[type[NewsWrapper]] = [GoogleNewsWrapper, DuckDuckGoWrapper, NewsApiWrapper, CryptoPanicWrapper]
|
||||
self.handler = WrapperHandler.build_wrappers(wrappers)
|
||||
|
||||
Toolkit.__init__( # type: ignore
|
||||
self,
|
||||
name="News APIs Toolkit",
|
||||
tools=[
|
||||
self.get_top_headlines,
|
||||
self.get_latest_news,
|
||||
self.get_top_headlines_aggregated,
|
||||
self.get_latest_news_aggregated,
|
||||
],
|
||||
)
|
||||
|
||||
def get_top_headlines(self, limit: int = 100) -> list[Article]:
|
||||
return self.handler.try_call(lambda w: w.get_top_headlines(limit))
|
||||
def get_latest_news(self, query: str, limit: int = 100) -> list[Article]:
|
||||
return self.handler.try_call(lambda w: w.get_latest_news(query, limit))
|
||||
|
||||
def get_top_headlines_aggregated(self, limit: int = 100) -> dict[str, list[Article]]:
|
||||
"""
|
||||
Calls get_top_headlines on all wrappers/providers and returns a dictionary mapping their names to their articles.
|
||||
Args:
|
||||
limit (int): Maximum number of articles to retrieve from each provider.
|
||||
Returns:
|
||||
dict[str, list[Article]]: A dictionary mapping providers names to their list of Articles
|
||||
Raises:
|
||||
Exception: If all wrappers fail to provide results.
|
||||
"""
|
||||
return self.handler.try_call_all(lambda w: w.get_top_headlines(limit))
|
||||
|
||||
def get_latest_news_aggregated(self, query: str, limit: int = 100) -> dict[str, list[Article]]:
|
||||
"""
|
||||
Calls get_latest_news on all wrappers/providers and returns a dictionary mapping their names to their articles.
|
||||
Args:
|
||||
query (str): The search query to find relevant news articles.
|
||||
limit (int): Maximum number of articles to retrieve from each provider.
|
||||
Returns:
|
||||
dict[str, list[Article]]: A dictionary mapping providers names to their list of Articles
|
||||
Raises:
|
||||
Exception: If all wrappers fail to provide results.
|
||||
"""
|
||||
return self.handler.try_call_all(lambda w: w.get_latest_news(query, limit))
|
||||
79
src/app/api/news/cryptopanic_api.py
Normal file
79
src/app/api/news/cryptopanic_api.py
Normal file
@@ -0,0 +1,79 @@
|
||||
import os
|
||||
from typing import Any
|
||||
import requests
|
||||
from enum import Enum
|
||||
from app.api.base.news import NewsWrapper, Article
|
||||
|
||||
|
||||
class CryptoPanicFilter(Enum):
|
||||
RISING = "rising"
|
||||
HOT = "hot"
|
||||
BULLISH = "bullish"
|
||||
BEARISH = "bearish"
|
||||
IMPORTANT = "important"
|
||||
SAVED = "saved"
|
||||
LOL = "lol"
|
||||
ANY = ""
|
||||
|
||||
class CryptoPanicKind(Enum):
|
||||
NEWS = "news"
|
||||
MEDIA = "media"
|
||||
ALL = "all"
|
||||
|
||||
def extract_articles(response: dict[str, Any]) -> list[Article]:
|
||||
articles: list[Article] = []
|
||||
if 'results' in response:
|
||||
for item in response['results']:
|
||||
article = Article()
|
||||
article.source = item.get('source', {}).get('title', '')
|
||||
article.time = item.get('published_at', '')
|
||||
article.title = item.get('title', '')
|
||||
article.description = item.get('description', '')
|
||||
articles.append(article)
|
||||
return articles
|
||||
|
||||
class CryptoPanicWrapper(NewsWrapper):
|
||||
"""
|
||||
A wrapper for the CryptoPanic API (Documentation: https://cryptopanic.com/developers/api/)
|
||||
Requires an API key set in the environment variable CRYPTOPANIC_API_KEY.
|
||||
It is free to use, but has rate limits and restrictions based on the plan type (the free plan is 'developer' with 100 req/month).
|
||||
Supports different plan types via the CRYPTOPANIC_API_PLAN environment variable (developer, growth, enterprise).
|
||||
"""
|
||||
|
||||
def __init__(self):
|
||||
self.api_key = os.getenv("CRYPTOPANIC_API_KEY", "")
|
||||
assert self.api_key, "CRYPTOPANIC_API_KEY environment variable not set"
|
||||
|
||||
# Set here for the future, but currently not needed
|
||||
plan_type = os.getenv("CRYPTOPANIC_API_PLAN", "developer").lower()
|
||||
assert plan_type in ["developer", "growth", "enterprise"], "Invalid CRYPTOPANIC_API_PLAN value"
|
||||
|
||||
self.base_url = f"https://cryptopanic.com/api/{plan_type}/v2"
|
||||
self.filter = CryptoPanicFilter.ANY
|
||||
self.kind = CryptoPanicKind.NEWS
|
||||
|
||||
def get_base_params(self) -> dict[str, str]:
|
||||
params: dict[str, str] = {}
|
||||
params['public'] = 'true' # recommended for app and bots
|
||||
params['auth_token'] = self.api_key
|
||||
params['kind'] = self.kind.value
|
||||
if self.filter != CryptoPanicFilter.ANY:
|
||||
params['filter'] = self.filter.value
|
||||
return params
|
||||
|
||||
def set_filter(self, filter: CryptoPanicFilter):
|
||||
self.filter = filter
|
||||
|
||||
def get_top_headlines(self, limit: int = 100) -> list[Article]:
|
||||
return self.get_latest_news("", limit) # same endpoint so just call the other method
|
||||
|
||||
def get_latest_news(self, query: str, limit: int = 100) -> list[Article]:
|
||||
params = self.get_base_params()
|
||||
params['currencies'] = query
|
||||
|
||||
response = requests.get(f"{self.base_url}/posts/", params=params)
|
||||
assert response.status_code == 200, f"Error fetching data: {response}"
|
||||
|
||||
json_response = response.json()
|
||||
articles = extract_articles(json_response)
|
||||
return articles[:limit]
|
||||
34
src/app/api/news/duckduckgo.py
Normal file
34
src/app/api/news/duckduckgo.py
Normal file
@@ -0,0 +1,34 @@
|
||||
import json
|
||||
from typing import Any
|
||||
from agno.tools.duckduckgo import DuckDuckGoTools
|
||||
from app.api.base.news import Article, NewsWrapper
|
||||
|
||||
|
||||
def extract_article(result: dict[str, Any]) -> Article:
|
||||
article = Article()
|
||||
article.source = result.get("source", "")
|
||||
article.time = result.get("date", "")
|
||||
article.title = result.get("title", "")
|
||||
article.description = result.get("body", "")
|
||||
return article
|
||||
|
||||
class DuckDuckGoWrapper(NewsWrapper):
|
||||
"""
|
||||
A wrapper for DuckDuckGo News search using the Tool from agno.tools.duckduckgo.
|
||||
It can be rewritten to use direct API calls if needed in the future, but currently is easy to write and use.
|
||||
"""
|
||||
|
||||
def __init__(self):
|
||||
self.tool = DuckDuckGoTools()
|
||||
self.query = "crypto"
|
||||
|
||||
def get_top_headlines(self, limit: int = 100) -> list[Article]:
|
||||
results = self.tool.duckduckgo_news(self.query, max_results=limit)
|
||||
json_results = json.loads(results)
|
||||
return [extract_article(result) for result in json_results]
|
||||
|
||||
def get_latest_news(self, query: str, limit: int = 100) -> list[Article]:
|
||||
results = self.tool.duckduckgo_news(query or self.query, max_results=limit)
|
||||
json_results = json.loads(results)
|
||||
return [extract_article(result) for result in json_results]
|
||||
|
||||
38
src/app/api/news/googlenews.py
Normal file
38
src/app/api/news/googlenews.py
Normal file
@@ -0,0 +1,38 @@
|
||||
from typing import Any
|
||||
from gnews import GNews # type: ignore
|
||||
from app.api.base.news import Article, NewsWrapper
|
||||
|
||||
|
||||
def extract_article(result: dict[str, Any]) -> Article:
|
||||
article = Article()
|
||||
article.source = result.get("source", "")
|
||||
article.time = result.get("publishedAt", "")
|
||||
article.title = result.get("title", "")
|
||||
article.description = result.get("description", "")
|
||||
return article
|
||||
|
||||
class GoogleNewsWrapper(NewsWrapper):
|
||||
"""
|
||||
A wrapper for the Google News RSS Feed (Documentation: https://github.com/ranahaani/GNews/?tab=readme-ov-file#about-gnews)
|
||||
It does not require an API key and is free to use.
|
||||
"""
|
||||
|
||||
def get_top_headlines(self, limit: int = 100) -> list[Article]:
|
||||
gnews = GNews(language='en', max_results=limit, period='7d')
|
||||
results: list[dict[str, Any]] = gnews.get_top_news() # type: ignore
|
||||
|
||||
articles: list[Article] = []
|
||||
for result in results:
|
||||
article = extract_article(result)
|
||||
articles.append(article)
|
||||
return articles
|
||||
|
||||
def get_latest_news(self, query: str, limit: int = 100) -> list[Article]:
|
||||
gnews = GNews(language='en', max_results=limit, period='7d')
|
||||
results: list[dict[str, Any]] = gnews.get_news(query) # type: ignore
|
||||
|
||||
articles: list[Article] = []
|
||||
for result in results:
|
||||
article = extract_article(result)
|
||||
articles.append(article)
|
||||
return articles
|
||||
54
src/app/api/news/news_api.py
Normal file
54
src/app/api/news/news_api.py
Normal file
@@ -0,0 +1,54 @@
|
||||
import os
|
||||
from typing import Any
|
||||
import newsapi # type: ignore
|
||||
from app.api.base.news import Article, NewsWrapper
|
||||
|
||||
|
||||
def extract_article(result: dict[str, Any]) -> Article:
|
||||
article = Article()
|
||||
article.source = result.get("source", {}).get("name", "")
|
||||
article.time = result.get("publishedAt", "")
|
||||
article.title = result.get("title", "")
|
||||
article.description = result.get("description", "")
|
||||
return article
|
||||
|
||||
class NewsApiWrapper(NewsWrapper):
|
||||
"""
|
||||
A wrapper for the NewsAPI (Documentation: https://newsapi.org/docs/get-started)
|
||||
Requires an API key set in the environment variable NEWS_API_KEY.
|
||||
It is free to use, but has rate limits and restrictions based on the plan type (the free plan is 'developer' with 100 req/day).
|
||||
"""
|
||||
|
||||
def __init__(self):
|
||||
api_key = os.getenv("NEWS_API_KEY")
|
||||
assert api_key, "NEWS_API_KEY environment variable not set"
|
||||
|
||||
self.client = newsapi.NewsApiClient(api_key=api_key)
|
||||
self.category = "business" # Cryptocurrency is under business
|
||||
self.language = "en"
|
||||
self.max_page_size = 100
|
||||
|
||||
def __calc_pages(self, limit: int, page_size: int) -> tuple[int, int]:
|
||||
page_size = min(self.max_page_size, limit)
|
||||
pages = (limit // page_size) + (1 if limit % page_size > 0 else 0)
|
||||
return pages, page_size
|
||||
|
||||
def get_top_headlines(self, limit: int = 100) -> list[Article]:
|
||||
pages, page_size = self.__calc_pages(limit, self.max_page_size)
|
||||
articles: list[Article] = []
|
||||
|
||||
for page in range(1, pages + 1):
|
||||
headlines: dict[str, Any] = self.client.get_top_headlines(q="", category=self.category, language=self.language, page_size=page_size, page=page) # type: ignore
|
||||
results = [extract_article(article) for article in headlines.get("articles", [])] # type: ignore
|
||||
articles.extend(results)
|
||||
return articles
|
||||
|
||||
def get_latest_news(self, query: str, limit: int = 100) -> list[Article]:
|
||||
pages, page_size = self.__calc_pages(limit, self.max_page_size)
|
||||
articles: list[Article] = []
|
||||
|
||||
for page in range(1, pages + 1):
|
||||
everything: dict[str, Any] = self.client.get_everything(q=query, language=self.language, sort_by="publishedAt", page_size=page_size, page=page) # type: ignore
|
||||
results = [extract_article(article) for article in everything.get("articles", [])] # type: ignore
|
||||
articles.extend(results)
|
||||
return articles
|
||||
Reference in New Issue
Block a user