Question

We test an application developed in house using a python test suite which accomplishes web navigations/interactions through Selenium WebDriver. A tricky part of our web testing is in dealing with a series of pdf reports in the app. We are testing a planned upgrade of Firefox from v3.6 to v16.0.1, and it turns out that the way we captured reports before no longer works, because of changes in the directory structure of firefox’s temp folder. I didn’t write the original pdf capturing code, but I will refactor it for whatever we end up using with v16.0.1, so I was wondering if there’ s a better way to save a pdf using Python’s selenium webdriver bindings than what we’re currently doing.

Previously, for Firefox v3.6, after clicking a link that generates a report, we would scan the “C:\Documents and Settings\\Local Settings\Temp\plugtmp” directory for a pdf file (with a specific name convention) to be generated. To be clear, we’re not saving the report from the webpage itself, we’re just using the one generated in firefox’s Temp folder.

In Firefox 16.0.1, after clicking a link that generates a report, the file is generated in “C:\Documents and Settings\ \Local Settings\Temp\tmp*\cache*”, with a random file name, not ending in “.pdf”. This makes capturing this file somewhat more difficult, if using a technique similar to our previous one – each browser has a different tmp*** folder, which has a cache full of folders, inside of which the report is generated with a random file name.

The easiest solution I can see would be to directly save the pdf, but I haven’t found a way to do that yet.

To use the same approach as we used in FF3.6 (finding the pdf in the Temp folder directory), I’m thinking we’ll need to do the following:

1. Figure out which tmp*** folder belongs to this particular browser instance (which we can do be inspecting the tmp*** folders that exist before and after the browser is instantiated)
2. Look inside that browser’s cache for a file generated immedaitely after the pdf report was generated (which we can by comparing timestamps)
3. In cases where multiple files are generated in the cache, we could possibly sort based on size, and take the largest file, since the pdf will almost certainly be the largest temp file (although this seems flaky and will need to be tested in practice).

I’m not feeling great about this approach, and was wondering if there’s a better way to capture pdf files. Can anyone suggest a better approach?

Note: the actual scraping of the PDF file is still working fine.

Answer 1

We ultimately accomplished this by clearing firefox’s temporary internet files before the test, then looking for the most recently created file after the report was generated.