• Why was this made?
  • Why not?

    Also this project gave us a fun use for Elasticsearch, a technology we were interested in. During the process we had a chance to work with many technologies we were unfamiliar with; including Javascript, Node, Vue, Bootstrap, JRuby, Ruby (inluding a fascinating detour into the limitations of parallel processing in the MRI), and CSS.

    Finally of course, it seems like a tool that might be of some use to the wider community.
  • What sources are searchable?
  • The Strategic Review, the Dragon Annuals, Dragon Magazine, the Best of Dragon issues, & Oerth Journal (Dragon issues 147-149 aren't included because there are no decent sources for OCR). See changelog for other additions.
  • Can other sources be added?
  • Yes. We're open to requests.
  • Why aren't the results in order?
  • It's complicated, but may be corrected in the future.
  • I searched for a phrase I'm certain was in some issue, but nothing was returned!
  • The schema used for the database was admittedly naive, and is unable to search across multiple pages. It's good enough for our purposes, i but if enough interest is shown, we can look into implementing a new schema.

    In addition the sources we had to OCR from were terrible, and even after many hours of cleaning them, there are many flaws. Characters are mis-recognized, words broken up, even some ads are accidentally included.

    If you're deeply curious (or especially masochistic) about what we mean when we say "terrible", you can have a look here.
  • You're definitely missing a mention of {something}
  • The sources are deeply flawed, and this is compounded by the weirdness of D&D's prose. OCR software tries to resolve ambiguities by using a dictionary, and a large number of the words in these sources will not be found in any English dictionary.
  • If I send you a report of a problem in your source, will you fix it?
  • Probably not. We're not paid for this. Fixing bad PDFs (that we've already spent many hundreds of hours cleaning up to the best of our ability) and then re-indexing them into the database is a a tedious process.

    On the other hand if you have a cleaner PDF source, we'd be happy to take a look and see if it could help. Please note that the issues of Dragon from the CD Archive were the start of this project, and are horrible sources.
  • How is this better than the DragonDex?
  • We don't say it is, the DragonDex is an amazing resource. That said, we're able to index many more sources, and already have all issues of Dragon (and its related magazines) up to issue 400. We also have Oerth Journal and the LGG index at this time. Beyond that, you can search for anything.. a phrase you partially remember, every instance of 'word', and so on.

    Our opinion is that both are imperfect tools, made stronger by the other.
  • The pages you report don't match the pages in the issue I'm looking at.
  • There are two major challenges that result in some pain here: Firstly Dragon magazine had inconsistent page numbering practices throughout its history, especially the PDF-native issues did some very strange things.
    Secondly, the sources are terribly encoded PDFs; the only page data we were able to extract is the PDF page number.
  • Can I help somehow?
  • We're looking for clean sources, in particular Dragon issues 147, 148, and 149 - which we've so far been unable to index at all. Beyond that, feel free to contact us.

    Also see our Support Us page.
  • I have a question or a request.
  • Contact us at admin (at) wait-wtf.com