I wish it can support pipelined data in parsing.

We do usually use Http parsers in this scene. Tcp protocol will seperate a large data part into some small ones; also , if you get data from a file whitch is too large to load into memory , handling like a std::stream is better;