A similar need in recent work is: There are about 300 million data dictionaries in a csv file A as a data source. For any word M entered by the user, it is necessary to quickly match whether the M word exists in A. (A file is about 3G in size, the total number of lines is 300 million). Get this demand. What is your first thought?