Not a data source, but a data tool. Sometimes the only format available is PDF, and it looks like you'll have to type it all in by hand. Tabula to the rescue! (Note: Downloadable software which, once installed, opens a new web browser window. Data extraction happens in the browser.) Read more about it.
Repository for all file formats and types of data--searchable, browsable, and taggable. Figshare allows users to upload any file format to be previewed in the browser so that any research output, from posters and presentations to datasets and code, can be disseminated in a way that the current scholarly publishing model does not allow.
articles, books, conference papers, technical reports, theses, and dissertations from CalTech's faculty and students. There are over a dozen collections and more coming. CODA is particularly strong in engineering, but covers all subjects.
The CIA World FactBook and the Occupational Outlook Handbook are just a few of the many government documents available online. Other agencies have collections of historical material, such as the National Parks Service Parks History site. USA.gov will search the whole federal government. Science.gov searches just the science related agencies (NASA, EPA, FWS, NOAA, etc.)