feat(router): add 杭州市人民政府

lib/routes/gov/hangzhou/crawler.ts

@@ -0,0 +1,54 @@
+import logger from '@/utils/logger';
+export async function zjzwfwCrawler(item: any, browser: any): Promise<string> {
+    try {
+        const page = await browser.newPage();
+        let response = '';
+        try {
+            let navigationSuccess = false;
+            const navigationAttempt = async (attempt) => {
+                if (attempt >= 3) {
+                    return false;
+                }
+                try {
+                    await page.goto(item.link, {
+                        waitUntil: 'networkidle2',
+                        timeout: 60000,
+                    });
+                    return true;
+                } catch {
+                    if (attempt < 3) {
+                        await new Promise((resolve) => setTimeout(resolve, 5000));
+                        if (page.isClosed()) {
+                            throw new Error('Navigation frame was detached');
+                        }
+                        return navigationAttempt(attempt + 1);
+                    }
+                    return false;
+                }
+            };
+
+            navigationSuccess = await navigationAttempt(0);
+
+            if (!navigationSuccess) {
+                throw new Error('Navigation failed after retries');
+            }
+
+            if (page.isClosed()) {
+                throw new Error('Page was closed unexpectedly');
+            }
+            await page.locator('.item-left .item .title .button').click();
+
+            response = await page.content();
+        } catch (error) {
+            logger.error('Page Error when visiting /gov/hangzhou/zwfw:', error);
+        } finally {
+            if (!page.isClosed()) {
+                await page.close();
+            }
+        }
+        return response || '';
+    } catch (error) {
+        logger.error('Error when visiting /gov/hangzhou/zwfw:', error);
+    }
+    return '';
+}

lib/routes/gov/hangzhou/namespace.ts

@@ -0,0 +1,9 @@
+import type { Namespace } from '@/types';
+
+export const namespace: Namespace = {
+    name: "Hangzhou People's Government",
+    url: 'hangzhou.gov.cn',
+    zh: {
+        name: '杭州市人民政府',
+    },
+};

lib/routes/gov/hangzhou/zwfw.ts

@@ -0,0 +1,106 @@
+import { Route } from '@/types';
+import { load } from 'cheerio';
+import puppeteer from '@/utils/puppeteer';
+import ofetch from '@/utils/ofetch';
+import cache from '@/utils/cache';
+import got from '@/utils/got';
+import { parseDate } from '@/utils/parse-date';
+import { zjzwfwCrawler } from './crawler';
+import timezone from '@/utils/timezone';
+
+export const route: Route = {
+    path: '/hangzhou/zwfw',
+    categories: ['government'],
+    example: '/gov/hangzhou/zwfw',
+    features: {
+        requireConfig: false,
+        requirePuppeteer: true,
+        antiCrawler: true,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    radar: [
+        {
+            source: ['hangzhou.gov.cn/col/col1256349/index.html'],
+        },
+    ],
+    name: '政务服务公开',
+    maintainers: ['Flynn Cao'],
+    handler,
+    url: 'hangzhou.gov.cn/col/col1256349/index.html',
+};
+
+async function handler() {
+    const host = 'https://www.hangzhou.gov.cn/col/col1256349/index.html';
+    const response = await ofetch(host);
+
+    const browser = await puppeteer({ stealth: true });
+    const link = host;
+    const formatted = response
+        .replace('<script type="text/xml">', '')
+        .replace('</script>', '')
+        .replaceAll('<recordset>', '')
+        .replaceAll('</recordset>', '')
+        .replaceAll('<record>', '')
+        .replaceAll('</record>', '')
+        .replaceAll('<![CDATA[', '')
+        .replaceAll(']]>', '');
+    const $ = load(formatted);
+
+    const list = $('li.clearfix')
+        .toArray()
+        .map((item) => {
+            item = $(item);
+            const title = item.find('a').first().text();
+            const time = timezone(parseDate(item.find('span').first().text(), 'YYYY-MM-DD'), 8);
+            const a = item.find('a').first().attr('href');
+            const fullUrl = new URL(a, host).href;
+
+            return {
+                title,
+                link: fullUrl,
+                pubDate: time,
+            };
+        });
+    const items = await Promise.all(
+        list.map((item) =>
+            cache.tryGet(item.link, async () => {
+                const host = new URL(item.link).hostname;
+                if (host === 'www.zjzwfw.gov.cn') {
+                    // 来源为浙江政务服务网
+                    const content = await zjzwfwCrawler(item, browser);
+                    const $ = load(content);
+                    item.description = $('.content .item').html();
+                    item.author = '浙江政务服务网';
+                    item.category = $('meta[name="ColumnType"]').attr('content');
+                } else {
+                    // 其他正常抓取
+                    const response = await got(item.link);
+                    const $ = load(response.data);
+                    if (host === 'police.hangzhou.gov.cn') {
+                        // 来源为杭州市公安局
+                        item.description = $('.art-content .wz_con_content').html();
+                        item.author = $('meta[name="ContentSource"]').attr('content');
+                        item.category = $('meta[name="ColumnType"]').attr('content');
+                    } else {
+                        // 缺省：来源为杭州市政府网
+                        item.description = $('.article').html();
+                        item.author = $('meta[name="ContentSource"]').attr('content');
+                        item.category = $('meta[name="ColumnType"]').attr('content');
+                    }
+                }
+                item.pubDate = $('meta[name="PubDate"]').length ? timezone(parseDate($('meta[name="PubDate"]').attr('content') as string, 'YYYY-MM-DD HH:mm'), 8) : item.pubDate;
+                return item;
+            })
+        )
+    );
+
+    await browser.close();
+    return {
+        allowEmpty: true,
+        title: '杭州市人民政府-政务服务公开',
+        link,
+        item: items,
+    };
+}