unit_tests/freshclam_test.py: fix unit test

[arjendekorte]

Somehow the tests returns a different result than expected. Not sure if the expected result should be changed (this patch) or that the message emitted by freshclam should be changed.

[arjendekorte]

String to match is generated in libfreshclam/libfreshclam_internal.c line 2144

[opoplawski]

This certainly does resolve the test failure. Still to be determined I suppose if this is the correct fix.

[arjendekorte]

Indeed. Changing line 2144 in libfreshclam/libfreshclam_internal.c will have the same result, but I have no idea either whether or not this is intentionally different.

[val-ms]

After thinking about this for some time, I suspect you are carrying a patch that makes Debug-level messages go to stderr. This test is checking for a Debug-level message in stdout. Can you confirm?

@opoplawski's proposed change switches to look for an Info-level message in stdout that is roughly equivalent. I'm also happy with this change.

[val-ms]

Before I merge, if you can confirm the debug/stderr suspicion that may allow us to fix up the commit message.

[opoplawski]

I don't believe I have any patches that are changing where any messages might be going.

[val-ms]

I don't believe I have any patches that are changing where any messages might be going.

Thanks for checking. I'm very confused how the test is failing for you without this change. I have never seen that test fail, and just tried reproducing it locally on fedora:latest. Perhaps it is something in your build environment, then? I don't know.

In the interests of understanding why it is failing for you, I am very curious if this change would cause the test to pass for you as well, instead of the change in this PR:

diff --git a/libfreshclam/libfreshclam.c b/libfreshclam/libfreshclam.c
index eda780374..33a9c0e45 100644
--- a/libfreshclam/libfreshclam.c
+++ b/libfreshclam/libfreshclam.c
@@ -640,7 +640,7 @@ fc_error_t fc_update_database(
                     if (*bUpdated) {
                         logg(LOGG_DEBUG, "fc_update_database: %s updated.\n", dbFilename);
                     } else {
-                        logg(LOGG_DEBUG, "fc_update_database: %s already up-to-date.\n", dbFilename);
+                        logg(LOGG_INFO, "fc_update_database: %s already up-to-date.\n", dbFilename);
                     }
                     goto success;
                 }

Can you please test it? It's not a good change, but would prove or disprove my thoughts about the LOG_DEBUG messages going to stderr instead of stdout.

[opoplawski]

That patch fixes the test for me as well. FWIW - my ctest line is:

+ /usr/bin/ctest --test-dir redhat-linux-build --output-on-failure --force-new-ctest-process -j4 -- -E valgrind

[val-ms]

I merged the commit and updated the commit message to explain the reason for the change and the outstanding issue. I am still curious to solve why debug-level messages aren't bring printed for you.