Abstract: XML is the de facto standard for data representation and exchanging on web. The query processing technique of XML streaming data is a hotspot in current research. Focused on the characteristics of processing semi-structure XML streaming data such as the stream arriving continuously, requiring to be read sequentially and only once into memory, the querying must be processed on the fly, a method of processing XPath query based on forest transducer is proposed. Firstly, conversion rules of forest transducer are defined for XPath query. And then the transducer is driven by input streaming data nodes. Stack and abstract syntax tree are applied to implement match and state transformation in running procedure. The relationships between state functions and intermediate results are kept by the abstract syntax tree, and the query results are output in reducing process. Finally, the experimental results show that our approach is effective and efficient on this problem, and outperforms about 30 percent of the state-of-the-art algorithms especially for large processed data. At the same time, memory usage is nearly constant. This method resolves the balance between time and space complexity, and it is a useful reference for other methods.