Dataset collected by JSIsolate

This dataset contains: 1) the object access logs, 2) script isolation policies and 3) script write conflicts collected by JSIsolate on Alexa top 1K websites. We analyze the access logs to generate the conflict summary files and script isolation policies that assign static scripts to an execution context. We split the whole dataset of object access logs into 10 subsets, i.e., access-0.zip ~ access-9.zip. The isolation policies are released in url-level-policies.zip and domain-level-policies.zip. The object accesses (i.e., reads and writes) are saved in [rank].[main/sub].[frame_cnt].access (e.g., 1.main.0.access) files. The URLs of frames (i.e., main frames and iframes) are saved in [rank].[main/sub].[frame_cnt].frame (e.g., 1.main.0.frame) files. The maps from script IDs to script URLs are saved in [rank].[main/sub].[frame_cnt].id2url (e.g., 1.main.0.id2url) files. The maps from script IDs to their parent script (script that includes it, <null> for static scripts) IDs are saved in [rank].[main/sub].[frame_cnt].id2parentid (e.g., 1.main.0.id2parentid) files. The source code of scripts are saved in [rank].[main/sub].[frame_cnt].[script_ID].script (e.g., 1.main.0.17.script) files. Note that we perform monkey testing during the data collection, which may cause the page to navigate to a different URL. Therefore, there could be multiple main frame files. The conflicts are dumped to [rank].conflicts (e.g., 1.conflicts) files. The isolation policies are dumped to [rank].configs (e.g., 1.configs) and [rank].configs-simple (e.g., 1.configs-simple) files. Note that the *.configs files also include the read/write operations that cause JSIsolate to assign a script from third-party domain to the first-party context.