Design And Implementation Of A Server-side Mashup Framework

With the Web2.0development for nearly10years, mashup applications can seeeverywhere on the Internet. But based on the RIA browser technology client-sidemashup applications still have some deficiencies: the browser security sandboxmechanism and origin policy, making the client-side mashup applications in datafusion and security limitations; a large number of complex, multi-level JavaScriptcode makes the browser download, parsing, rendering, relatively slow, resulting in agreater impact on performance. The way of the currently mashup service providersprovide resources mainly is OpenAPI.OpenAPI request protocol and OpenAPI interactive data format is a wide variety,and some resources need to be authorized to get, which gave developers to mashupapplication adds great complexity. Few server-side mashup development tools,mostly client-side mashup development tools. Some minority support for server-sidemashup development tools only support some open platform, and someenterprise-class heavyweights are not suitable for small and medium-sizedapplication development.The paper designed and implemented a lightweight server-side mashup framework.The framework contains a code generator automatically generate lightweightpackage each open platform OpenAPI Data Access Components and componentlibrary access OpenAPI data mechanism. Through analysising of the principles andfeatures of the server-side mashup applications, this paper established server-sidemashup of the OpenAPI data access framework. It includes the Session componentprovides the context and encapsulates the entire access process, the proxy agentcomponent of the remote resource server, the data parser component to bind the data,cache components, authorized components, access parameters class, resource dataclass. the automatic code generator is based on template technology in thisframework: input the open platform for information and OpenAPI description, afterrequest parameter class identification and data class identification process, generatemetadata that belong to parameter classes, data classes and data accesscomponents facade class metadata, finally FreeMarker embed metadata to generateclass files in the template file. In this framework, the external resources data bindingis different from the traditional data binding, but for a group of related XML databinding. This kind of data binding proposed in this paper is: first extracted the XMLmode tree diagram for this set of XML data files. Then each mode tree diagram nodeis represented as a vector in the vector pace. Then use their similarity and distance toidentify the mode node corresponded to a predefined mode node types.Then get the data classes according to the mode node type to class mapping rules.Finally based onthe mapping rules XML data bound to an instance of the class object. The methodused to implement the automatic code generator of the data class identificationprocess and the data access of binding data parser component.Compared with other mashup tool, this mashup framework is used for server-sidemashup, avoiding the limitations of client-side mashups. The framework usesautomatic code generation technology, can automatically generate access componentclass to help develop applications based on user input, suits for the majority openplatform. The data binding used in this framework avoids the tranditional databinding ubiquitous redundant classes and data class system large scale. Thecomponent library for accessing OpenAPI data mechanism is convenient theprogramer to get external resources. The framework adds the cache function,avoiding short period of time repeated access to the same resources to accelerate dataresponse speed to improve operational performance. The realization of theframework is based on the interface, combined with design patterns, which has goodscalability...