In recent years, eXtensible Business Reporting Language (XBRL) already has become a research focus in the recent years because of its economic significance. In actual engineering practice, parsing technologies for XBRL taxonomy are under the spotlight because they are the key factors that affect the performance of business software systems. Thus, in this paper, the principle of typical parsing models available for XBRL taxonomy are analyzed at first, and then quantitative analysis for respective business scenarios is provided by parsing the real XBRL taxonomy using those parsing models under the same condition (the same machine, the same programming language and the same compiling environment). Through the quantitative analysis, we found that those parsing models are hard to satisfy the needs of the actual XBRL project. Therefore, at last of this paper, a new parsing model for XBRL taxonomy is proposed, and the performance data are provided. Those analysis and comparison results have certain practical meaning and reference value for analogous projects.