The gsearch tool provides an easy access over either a terminal application or a web front end to the GENESI-DR opensearch interface. Both front ends take a set of different search parameters, generates and transmit the search request to a central opensearch site, customizable by the user. The received results will be processed and displayed by gsearch. It also offers the possibilities to download a selection, or all of the found datasets to the local computer or respectively it can create a download script containing all necessary information to load the dataset directly on a Grid node. This generated script can be included inside an JDL start script.
Concerning the terminal application, the development of a generic library for accessing the opensearch interface is ongoing. In the current state it provides most of the necessary functionality to send search requests and receive results. Based upon the libxml2 library the received search results are parsed and displayed by the application. The user has two methods to define search parameters and presentation of the incoming results.
1. For simple usability a pure console application list the available dataset. With command line parameters the user is able to choose the datasets, which then can be downloaded. In addition, it creates a file containing the file name and their location of the downloaded datasets. This file can now be submitted with the user application to the job scheduler.
2. A Text User Interface (TUI), based on ncurses provides with a fully interactive application a more detailed selection and representation of datasets. There are also more features, like a comfortable search form and a detailed information screen about a selected dataset.
In addition to the user interface, there was also the development of a background application started for directly downloading the dataset into the grid storage. This download tool, called gdownload will be able to use the following protocols: http, https, ftp, gritftp and dap. At the moment for testing the tool chain only the http protocol is implemented. The other protocols will be added soon.
In order to test the different parts of the tool, an example was created. As a symbolic search dataset the XBT_XXX containing temperature profiles of the Mediterranean Sea was used. The user selects the datasets and the framework creates a JDL file with the given test application processing the data directly loaded by gdownload into the grid system. The tests have proved the efficiency of the framework and that it saves the user a considerably long time of finding the right dataset without gsearch and manually uploading the data into the grid system.
- More functionality of the ncurses front end.
- Better handling of non conform results
- Implementation of more protocols: for instance opendap, hdf...
- Better test data cases and application