focus-product-extractor2
Version:
Extract product information from chat/order data
41 lines (37 loc) • 1.64 kB
JavaScript
;
var _fs = _interopRequireDefault(require("fs"));
function _interopRequireDefault(e) { return e && e.__esModule ? e : { default: e }; }
// 读取converted_data_empty.json文件
const emptyData = JSON.parse(_fs.default.readFileSync('./converted_data_empty.json', 'utf-8'));
console.log(`共有 ${emptyData.length} 条空结果数据`);
// 提取所需字段
const extractedFields = emptyData.map((item, index) => {
const message = item.input.rawData.messages[0].message;
return {
index: item.index,
type: message.type,
sub_type: message.sub_type,
sku_id: message.sku_id || "",
spu_id: message.spu_id || "",
order_id: message.order_id || ""
};
});
// 统计各字段的分布情况
const typeStats = {};
const subTypeStats = {};
const hasSkuId = extractedFields.filter(item => item.sku_id).length;
const hasSpuId = extractedFields.filter(item => item.spu_id).length;
const hasOrderId = extractedFields.filter(item => item.order_id).length;
extractedFields.forEach(item => {
typeStats[item.type] = (typeStats[item.type] || 0) + 1;
subTypeStats[item.sub_type] = (subTypeStats[item.sub_type] || 0) + 1;
});
console.log('\n=== 统计信息 ===');
console.log('Type分布:', typeStats);
console.log('Sub_type分布:', subTypeStats);
console.log(`有sku_id的数量: ${hasSkuId}`);
console.log(`有spu_id的数量: ${hasSpuId}`);
console.log(`有order_id的数量: ${hasOrderId}`);
// 保存提取的字段到文件
_fs.default.writeFileSync('./empty_data_fields.json', JSON.stringify(extractedFields, null, 2), 'utf-8');
console.log('\n提取完成,结果已保存到 empty_data_fields.json');