210 lines
6.7 KiB
JavaScript
210 lines
6.7 KiB
JavaScript
'use strict';
|
||
|
||
const fs = require('fs');
|
||
const path = require('path');
|
||
|
||
const fileUtils = require('./utils/fileUtils');
|
||
const requestUtils = require('./utils/requestUtils');
|
||
|
||
const DATA_FOLDER = path.join(path.dirname(__dirname), process.env.DATA_FOLDER ?? 'data', 'weibo');
|
||
console.log("DATA_FOLDER", DATA_FOLDER);
|
||
fileUtils.createFolder(DATA_FOLDER); // 程序运行就保证 data 目录存在
|
||
|
||
async function main() {
|
||
let requestTimestamp = Date.now();
|
||
let now = new Date(requestTimestamp + 8 * 3600 * 1000).toISOString();
|
||
|
||
let result = await requestUtils.getApiResult("https://weibo.com/ajax/statuses/hot_band");
|
||
if (result.ok != 1) {
|
||
console.log(new Date(Date.now() + 8 * 60 * 60 * 1000).toISOString(), "weibo", "请求成功,但服务器处理失败,正在重试。");
|
||
result = await requestUtils.getApiResult("https://weibo.com/ajax/statuses/hot_band");
|
||
if (result.ok != 1) {
|
||
console.log(new Date(Date.now() + 8 * 60 * 60 * 1000).toISOString(), "weibo", "请求成功,但服务器处理失败,保存失败信息。");
|
||
// ok 不为 1,那么久直接保存便于后续分析,不进行后续处理
|
||
fileUtils.saveJSON({
|
||
saveFolder: DATA_FOLDER,
|
||
now: now,
|
||
fileNameSuffix: `origin-error`,
|
||
object: result,
|
||
compress: true,
|
||
uncompress: false
|
||
});
|
||
return;
|
||
}
|
||
}
|
||
|
||
console.log(new Date(Date.now() + 8 * 60 * 60 * 1000).toISOString(), "weibo", "请求成功");
|
||
// console.log("result", result);
|
||
|
||
/**
|
||
* 保存原始数据
|
||
*/
|
||
fileUtils.saveJSON({
|
||
saveFolder: DATA_FOLDER,
|
||
now: now,
|
||
fileNameSuffix: `origin`,
|
||
object: result,
|
||
compress: true,
|
||
uncompress: false
|
||
});
|
||
|
||
let data = JSON.parse(JSON.stringify(result.data));
|
||
|
||
|
||
/**
|
||
* 过滤掉不需要的数据
|
||
*/
|
||
// hotgov
|
||
delete data.hotgov["mblog"];
|
||
// 重复字段只保留一个
|
||
delete data.hotgov["note"]; // note word
|
||
delete data.hotgov["small_icon_desc"]; // icon_desc small_icon_desc
|
||
delete data.hotgov["small_icon_desc_color"]; // icon_desc_color small_icon_desc_color
|
||
|
||
// band_list
|
||
for (let i = 0; i < data.band_list.length; i++) {
|
||
const item = data.band_list[i];
|
||
|
||
// 过滤广告
|
||
if (item.is_ad) {
|
||
data.band_list.splice(i, 1);
|
||
i--;
|
||
}
|
||
|
||
// 过滤空字段
|
||
delete item["ad_info"];
|
||
|
||
// 重复字段只保留一个
|
||
delete item["note"]; // note word
|
||
delete item["icon_desc"]; delete item["small_icon_desc"]; // label_name icon_desc small_icon_desc
|
||
delete item["small_icon_desc_color"]; // icon_desc_color small_icon_desc_color
|
||
delete item["flag_desc"]; // flag_desc subject_label 这两个有值的时候相同,没有值的时候,前一个为 undefined,后一个为 ""
|
||
}
|
||
|
||
|
||
/**
|
||
* 获取需要的数据,进行转换
|
||
*/
|
||
let convert = [];
|
||
data.band_list.forEach(item => {
|
||
let detail = "";
|
||
let pic_ids = [];
|
||
if (item.mblog) { // 有些热搜没有 mblog
|
||
var regex = /(<([^>]+)>)/ig
|
||
detail = item.mblog.text.replace(regex, "");
|
||
if (item.mblog.pics) {
|
||
pic_ids = item.mblog.pics.map(pic => `${pic}`);
|
||
}
|
||
}
|
||
convert.push({
|
||
// 热搜排行顺序
|
||
rank: item.rank,
|
||
realpos: item.realpos,
|
||
|
||
// 热搜信息
|
||
word: item.word, // 热搜标题
|
||
word_scheme: item.word_scheme, // 热搜话题 "#热搜标题#"
|
||
emoticon: item.emoticon, // 热搜小表情,如 "[泪]"
|
||
label_name: item.label_name, // 热搜标签,如 "爆" "热" "新" ""
|
||
onboard_time: item.onboard_time, // 热搜上线时间,秒级时间戳,如 1658565575
|
||
|
||
/**
|
||
* 热搜数据
|
||
*
|
||
* 大部分的 num 和 raw_hot 是相同的,页面上显示的是 num,可能是人工调控的热搜
|
||
*
|
||
* 两者差值通过观测似乎最大是 1250000
|
||
* 例如 【爆】唐山打架事件8名违法嫌疑人已到案 这条热搜一开始 delta 首先不断增大,最大达到 1250000
|
||
* 然后热搜数量增加到 12600000 左右的时候,delta 逐渐减小到 1040000 左右
|
||
*/
|
||
num: item.num,
|
||
raw_hot: item.raw_hot,
|
||
detla: item.num - item.raw_hot, // 计算值
|
||
|
||
url: `https://s.weibo.com/weibo?q=${encodeURIComponent(item.word_scheme)}`, // 热搜话题链接
|
||
|
||
// 分类
|
||
category: item.category ? item.category.split(',') : "",
|
||
subject_label: item.subject_label,
|
||
|
||
// 其他
|
||
more: {
|
||
is_new: item.is_new,
|
||
subject_querys: item.subject_querys,
|
||
mid: item.mid,
|
||
icon_desc_color: item.icon_desc_color,
|
||
detail: detail,
|
||
},
|
||
});
|
||
});
|
||
fileUtils.saveJSON({
|
||
saveFolder: DATA_FOLDER,
|
||
now: now,
|
||
fileNameSuffix: `final`,
|
||
object: convert,
|
||
compress: true,
|
||
// uncompress: true,
|
||
uncompress: false,
|
||
});
|
||
|
||
|
||
/**
|
||
* 只统计微博调控信息
|
||
*/
|
||
let convert2 = [];
|
||
let total = 0;
|
||
data.band_list.forEach(item => {
|
||
total += item.num;
|
||
total -= item.raw_hot;
|
||
if (item.num - item.raw_hot == 0) return;
|
||
convert2.push([
|
||
`[${item.realpos}] ${item.word}【${item.label_name}】`,
|
||
`原始:${item.raw_hot} 显示:${item.num} 调控: ${item.num - item.raw_hot}`
|
||
]);
|
||
});
|
||
fileUtils.saveJSON({
|
||
saveFolder: DATA_FOLDER,
|
||
now: now,
|
||
fileNameSuffix: `regulation`,
|
||
object: {
|
||
total_delta: total, // 所有调控值之和
|
||
data: convert2
|
||
},
|
||
compress: false,
|
||
uncompress: true
|
||
});
|
||
|
||
|
||
/**
|
||
* 保存预处理后数据
|
||
*/
|
||
// 过滤掉不需要的数据
|
||
// band_list
|
||
data.band_list.forEach(function (item) {
|
||
delete item["mblog"];
|
||
});
|
||
fileUtils.saveJSON({
|
||
saveFolder: DATA_FOLDER,
|
||
now: now,
|
||
fileNameSuffix: `simplify`,
|
||
object: data,
|
||
compress: true,
|
||
// uncompress: true,
|
||
// compress: false,
|
||
uncompress: false,
|
||
});
|
||
|
||
|
||
/**
|
||
* 更新最新的
|
||
*/
|
||
fs.writeFileSync(`${DATA_FOLDER}/latest.json`, JSON.stringify({
|
||
update_time: requestTimestamp,
|
||
update_time_friendly: now.substring(0, 19).replace(/T/g, " "),
|
||
regulation: convert2,
|
||
data: convert
|
||
}));
|
||
}
|
||
|
||
exports.main = main;
|