There is a resource designed to ease the process of sampling repositories or developers in Git. The resource is designed to allow users to perform basic restrictions then return a list of developers/repositories that fit the selected criteria. The resource uses the metadata collections described in described in this paper http://arxiv.org/abs/2008.03439. This API is in development and will have all the options listed in the above paper for researchers to perform sampling.
The API allows users to perform sampling using the active timeframe of an author/project. If the given timeframe is between the start and end date of an author/project it is counted. The timeframe of the Git object is based on the first and last commit time.
The API allows users to perform sampling based on the languages used by the author/project. If the given language is one of many languages the author/project used it is counted. The language information is based off the file extensions associated with each file related to the given Git object.
The API allows users to perform sampling based on the total number of files related to the author/project. It is possible to sample based on the total number of any language file contained within the resource. If the user wants to sample projects that have 50+ python files it is possible. It is also possible to restrict based on the total number of files assosicated with the author/project.
The data is stored within a MongoDB server, thus queries may take time to retrieve all the associated data. When the sampling is finished data is returned in JSON format.
This sampling resource has some restrictions to what is possible. For more information on restrictions to the timeframe/language/file information see the above paper.
