抓取项目详情

Project/index.js

@@ -91,6 +91,83 @@ async function extractDataFromHtml($, origin) {
     return data.filter(item => item !== null);
 }
 
+async function scrapeProjectDetails(project) {
+    if (!project || !project['项目链接']) {
+        return project;
+    }
+
+    try {
+        console.log(`正在抓取详情: ${project['项目名称']}`);
+        const response = await axios.get(project['项目链接'], {
+            headers: {
+                'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36'
+            }
+        });
+        const $ = cheerio.load(response.data);
+
+        // 抓取 housedetail 表格中的所有信息
+        $('.housedetail tr').each((i, row) => {
+            const cells = $(row).find('td');
+            if (cells.length >= 2) {
+                const key = $(cells[0]).text().trim().replace('：', '').replace(':', '');
+                const value = $(cells[1]).text().trim();
+                if (key) project[key] = value;
+            }
+            if (cells.length >= 4) {
+                const key = $(cells[2]).text().trim().replace('：', '').replace(':', '');
+                const value = $(cells[3]).text().trim();
+                if (key) project[key] = value;
+            }
+        });
+
+        // 预售许可证 special handling, since it's more complex
+        const licenses = [];
+        const presellInfo = $('#presellInfo').val();
+        if (presellInfo) {
+            const licenseItems = presellInfo.split(';;');
+            for (const item of licenseItems) {
+                const parts = item.split(',,');
+                if (parts.length === 2 && parts[1]) {
+                    const licenseData = {
+                        '许可证号': parts[1].trim(),
+                        '楼幢': []
+                    };
+                    
+                    const licenseDiv = $(`.three:contains('${parts[1]}')`);
+                    const buildingSpans = licenseDiv.nextUntil('.three', 'span');
+                    
+                    buildingSpans.each((i, span) => {
+                        const buildingText = $(span).text().trim();
+                        const building = {};
+                        const match = buildingText.match(/(.*)\(成交均价：(.*)\)/);
+                        if(match){
+                            building['楼幢名称'] = match[1].trim();
+                            building['成交均价'] = match[2].trim().replace(')','');
+                        } else {
+                            building['楼幢名称'] = buildingText;
+                        }
+
+                        // Find the preceding input and get the 'bid'
+                        const bid = $(span).prev('input[name="radiobuild"]').attr('bid');
+                        if (bid) {
+                            building['bid'] = bid.trim();
+                        }
+
+                        licenseData['楼幢'].push(building);
+                    });
+                    licenses.push(licenseData);
+                }
+            }
+        }
+        project['预售许可证'] = licenses;
+
+    } catch (error) {
+        console.error(`抓取详情失败: ${project['项目名称']}`, error.message);
+    }
+    return project;
+}
+
+
 /**
  * 抓取网站所有数据，处理分页
  * @returns {Promise<Array<Object>>} - 所有抓取的数据记录数组
@@ -173,7 +250,13 @@ async function scrapeWebsite() {
     }
     
     console.log(`\n抓取全站数据完毕！共 ${allData.length} 条原始记录。`);
-    return allData;
+    
+    console.log('\n开始抓取项目详情...');
+    const detailedDataPromises = allData.map(project => scrapeProjectDetails(project));
+    const detailedData = await Promise.all(detailedDataPromises);
+    console.log('所有项目详情抓取完毕。');
+
+    return detailedData;
 }
 
 /**