Since error suppression was removed with the call to unlink() to make
sure that there are no previous test image files before testing for the
gs path, make sure that it exists first before deleting to prevent it
from throwing any errors.
Basically we should produce a combined PDF from all the files that it was possible to convert.
If any one submission input file fails conversion, it should just be omitted in the result.
This commit reviews all continue uses in core happening within a
loop / switch / case hierarchy. This does not cover:
- Changes to libraries. Will be handled in another issue / commit.
- Uses out from loops, will be reviewed by other commit.
The policy followed has been:
- When possible, take rid of the continue.
- When clearly the intention was to jump to next element in loop
change to continue 2
- When it was not clear, keep old behavior switching to break, no
matter how weird the behavior may be.
Identical submissions do not require individual conversions from the document converter. Text passing through
format_text cannot be trusted and is likely to be unique each time, even for the same text.
Without this, its possible to get a document in the editpdf conversion queue
that keeps failing - and in some cases completely crashing the PHP script
despite exception handling - which ends up blocking the processing of the queue
entirely.
This change allows for a configurable number of attempts the conversion
task will perform before removing the item from the queue.
This patch adds a check to make sure there is a converter available that
will be able to handle html->pdf conversion before adding onlinetext
submissions to the list of files to combine.
If the user is removed from a course after a submission has been made,
the task will crash and never continue. This removes submissions from
the cron task if the user isn't enrolled in the course.
This lets us convert between common office formats. E.g. docx -> pdf
html -> pdf, html -> ods.
This commit also updates assignment editpdf plugin to use this converter
on all compatible submission files.
Added 'draft = 1' criteria for the delete query in
page_editor::set_annotations in order to prevent the deletion of non-draft
annotations while the PDF is being edited.
There was a big proliferation and reuse of $pdf variable
that was making the code hard to read/review. This commit
does change nothing but makes it more readable.
Also adds a couple of Close() calls, not strictly needed
because they don't have opened files. But think it makes
really clearer the scope of every variable. And, for sure
it frees some resources. That cannot be bad.
When a pdf object is instantiated and any file is loaded
with set_pdf() or load_pdf(), the files remain open until
the pdf is saved with pdf_save() or outputted with Output()..
In that cases is needed to perform an explicit Close() in
order to free resources, parsers and, ultimately, fclose()
the files.
Note that only the uses detected in the editpdf unit tests have
been fixed. I'd recommend to analyse every instance of the tcpdf
libs.
Also, there was one incorrecly reused pdf instance in then
generate_combined_pdf_for_attempt() method. Apparently it was not
leading to problems, but better use a separate instance (#246).
The size of the drawing canvas is now set from the size of the generated page image, with automatic
scrolling if it is too wide to display
Absolutely-positioned nodes (comments + stamps) are repositioned as the drawing canvas is scrolled
During a team submission the stamps were attached to the user that
was graded and not to the other users. This could lead to errors when
generating the PDF for each user.
