From ed163cc842ced03e4babad050de42c97754fdfff Mon Sep 17 00:00:00 2001 From: felinae98 <731499577@qq.com> Date: Sat, 27 Feb 2021 13:49:09 +0800 Subject: [PATCH] add support for wechat --- .../nonebot_hk_reporter/platform/__init__.py | 1 + .../nonebot_hk_reporter/platform/utils.py | 4 +- .../nonebot_hk_reporter/platform/wechat.py | 73 +++++++++++++++++++ src/plugins/nonebot_hk_reporter/utils.py | 11 ++- 4 files changed, 85 insertions(+), 4 deletions(-) create mode 100644 src/plugins/nonebot_hk_reporter/platform/wechat.py diff --git a/src/plugins/nonebot_hk_reporter/platform/__init__.py b/src/plugins/nonebot_hk_reporter/platform/__init__.py index ea37e30..4c806b4 100644 --- a/src/plugins/nonebot_hk_reporter/platform/__init__.py +++ b/src/plugins/nonebot_hk_reporter/platform/__init__.py @@ -1,6 +1,7 @@ from .bilibili import Bilibili from .rss import Rss from .weibo import Weibo +from .wechat import Wechat from .utils import check_sub_target from .platform import PlatformNoTarget from .utils import platform_manager diff --git a/src/plugins/nonebot_hk_reporter/platform/utils.py b/src/plugins/nonebot_hk_reporter/platform/utils.py index 53449d5..27fa172 100644 --- a/src/plugins/nonebot_hk_reporter/platform/utils.py +++ b/src/plugins/nonebot_hk_reporter/platform/utils.py @@ -5,6 +5,7 @@ from typing import Type from .weibo import Weibo from .bilibili import Bilibili from .rss import Rss +from .wechat import Wechat from .platform import PlatformProto from ..config import Config from ..post import Post @@ -16,7 +17,8 @@ async def check_sub_target(target_type, target): platform_manager: dict[str, PlatformProto] = { 'bilibili': Bilibili(), 'weibo': Weibo(), - 'rss': Rss() + 'rss': Rss(), + 'wechat': Wechat(), } async def fetch_and_send(target_type: str): diff --git a/src/plugins/nonebot_hk_reporter/platform/wechat.py b/src/plugins/nonebot_hk_reporter/platform/wechat.py new file mode 100644 index 0000000..a3d3d6c --- /dev/null +++ b/src/plugins/nonebot_hk_reporter/platform/wechat.py @@ -0,0 +1,73 @@ +from datetime import datetime +import hashlib +import json +import re +from typing import Any, Optional + +from bs4 import BeautifulSoup as bs +import httpx + +from ..post import Post +from ..types import * +from .platform import Platform + + +class Wechat(Platform): + + categories = {} + enable_tag = False + platform_name = 'wechat' + + @classmethod + def _get_query_url(cls, target: Target): + return 'https://weixin.sogou.com/weixin?type=1&s_from=input&query={}&ie=utf8&_sug_=n&_sug_type_='.format(target) + + @classmethod + async def _get_target_soup(cls, target: Target) -> Optional[bs]: + target_url = cls._get_query_url(target) + async with httpx.AsyncClient() as client: + res = await client.get(target_url) + soup = bs(res.text, 'html.parser') + blocks = soup.find(class_='news-list2').find_all('li',recursive=False) + for block in blocks: + if block.find(string=[target]): + return block + + @classmethod + async def get_account_name(cls, target: Target) -> Optional[str]: + if not (block := await cls._get_target_soup(target)): + return None + return block.find('p', class_='tit').find('a').text + + async def get_sub_list(self, target: Target) -> list[RawPost]: + block = await self._get_target_soup(target) + if (last_post_dt := block.find('dt', string='最近文章：')): + post = { + 'title': last_post_dt.find_parent().find('a').text, + 'target': target, + 'page_url': self._get_query_url(target) + } + return [post] + else: + return [] + + def get_id(self, post: RawPost) -> Any: + return post['title'] + + def get_date(self, post: RawPost): + return None + + def get_tags(self, post: RawPost): + return None + + def get_category(self, post: RawPost): + return None + + async def parse(self, raw_post: RawPost) -> Post: + # TODO get content of post + return Post(target_type='wechat', + text='{}\n详细内容请自行查看公众号'.format(raw_post['title']), + pics=[], + url='' + ) + diff --git a/src/plugins/nonebot_hk_reporter/utils.py b/src/plugins/nonebot_hk_reporter/utils.py index 4fbcfd9..51bf93e 100644 --- a/src/plugins/nonebot_hk_reporter/utils.py +++ b/src/plugins/nonebot_hk_reporter/utils.py @@ -1,10 +1,11 @@ import os import asyncio -from typing import Optional +from typing import Awaitable, Callable, Optional import nonebot from nonebot import logger import base64 from pyppeteer import launch +from pyppeteer.page import Page from pyppeteer.chromium_downloader import check_chromium, download_chromium from html import escape from hashlib import sha256 @@ -30,14 +31,18 @@ class Render(metaclass=Singleton): def __init__(self): self.lock = asyncio.Lock() - async def render(self, url: str, viewport: Optional[dict] = None, target: Optional[str] = None) -> str: + async def render(self, url: str, viewport: Optional[dict] = None, target: Optional[str] = None, + operation: Optional[Callable[[Page], Awaitable[None]]] = None) -> str: async with self.lock: if plugin_config.hk_reporter_use_local: browser = await launch(executablePath='/usr/bin/chromium', args=['--no-sandbox']) else: browser = await launch(args=['--no-sandbox']) page = await browser.newPage() - await page.goto(url) + if operation: + await operation(page) + else: + await page.goto(url) if viewport: await page.setViewport(viewport) if target: