FIRST

(In reply to comment #0)

During the student testing, we encountered an exam that did not have any
question numbers, and an exam that was a scanned document. We can probably
detect these cases, and should warn the user about them, as the parser is never
going to do well in these situations.

Reference: PDFAssessment2_4.java

In the event of question numbers being absent, questionNumbers.size() is zero at the end of separateQuestions() in parse()! This can be used to pop-up a warning to the user. Is there a more appropriate check or will the above test suffice?

I think it would be good to test each page and see if it contains anything other than an image to detect the case where the document is a scanned PDF.

But, for now, checking that the question list size is > 0 and warning the user if not, telling them that one possible reason would be a scanned document, would be good.

Since scanned documents are detected and handled in AssessmentFileSelection, this bug only corresponds to cases wherein the question numbers are not detected in the input document as they are actually absent.

Now, the question numbers are not actually processed until separateQuestions() function under "PDFAssessment*.java". Hence, this bug will be resolved in that function and not in AssessmentFileSelection!

When separateQuestions() returns with # of questions equal to zero, the current bug has occurred. Prompt a warning message to the user about potential faulty recognition if continued and provide an option for termination of the parser.

Original Bugzilla ID was 3659