Wrote a php page which uses the curl library in php to send a query to google, store the results in a variable and then process those for display.

I configured the search to look in the uvic.ca domain for only pdf files which contained the string provided by the user in a text box.

The results coming back from google are not well-formed xml snippets and are not the easiest to deal with (they may or may not contain various elements depending on whether google thinks the files are not in English etc.), so there will be a fair bit of tedious text processing to get the output I want.

My test file grabs what Google identifies as the title line (which is a link to the PDF file), the links to Google's html version and text version of the pdf, and the KeyWordInContext provided by Google. Because some of the hits coming back are structured other than my code expects it's possible to break my test page depending on what search string you use.

Assuming LC is OK in principle with using the Google-based approach, I'll create an account, upload the PDFs that Nadine is making and wait for them to be indexed. Then I can fine-tune the php page.