For example, if I write "测试" in txt file, and then pull this txt file into zotero, searching "测试" will not find it.

Zotero 6 isn't properly detecting the format of text files with Chinese characters.

This will be fixed in Zotero 7.

For now, you can select the affected attachments and run this from Tools → Developer → Run JavaScript to reindex them properly:

var items = ZoteroPane.getSelectedItems();
for (let item of items) {
if (!item.isAttachment()) continue;
if (item.attachmentCharset != 'utf-8') {
item.attachmentCharset = 'utf-8'
await item.saveTx({ skipDateModifiedUpdate: true})
await Zotero.FullText.indexItems([item.id])
}
}

By the way, it can be found in search using the method provided, but it cannot be found by filtering the attachment content in advanced search.