Rspamd: adjust learning of ham (#3334)

* adjust learning of ham

See #3333

When moving a mail from the Junk folder to the Trash folder, the mail
previously classified as ham due to the wildcard match of `*`. Because
the syntax does not allow for negation, we can only change the behavior
in a way that mails are learned as ham when they are moved into `INBOX`
from `Junk`. This is reasonable though.

* adjust tests accordingly

* adjust docs accordingly

docs/content/config/environment.md

@@ -353,7 +353,9 @@ Controls whether the [Rspamd Greylisting module][rspamd-greylisting-module] is e
 When enabled,
 
 1. the "[autolearning][rspamd-autolearn]" feature is turned on;
-2. the Bayes classifier will be trained when moving mails from or to the Junk folder (with the help of Sieve scripts).
+2. the Bayes classifier will be trained (with the help of Sieve scripts) when moving mails
+    1. from anywhere to the `Junk` folder (learning this email as spam);
+    2. from the `Junk` folder into the `INBOX` (learning this email as ham).
 
 !!! warning "Attention"
 

target/scripts/startup/setup.d/security/rspamd.sh

@@ -222,13 +222,13 @@ function __rspamd__setup_learning
     sedfile -i -E '/^}/d' /etc/dovecot/conf.d/90-sieve.conf
     cat >>/etc/dovecot/conf.d/90-sieve.conf << EOF
 
-  # From elsewhere to Junk folder
+  # From anyhwere to Junk
   imapsieve_mailbox1_name = Junk
   imapsieve_mailbox1_causes = COPY
   imapsieve_mailbox1_before = file:${SIEVE_PIPE_BIN_DIR}/learn-spam.sieve
 
-  # From Junk folder to elsewhere
-  imapsieve_mailbox2_name = *
+  # From Junk to Inbox
+  imapsieve_mailbox2_name = INBOX
   imapsieve_mailbox2_from = Junk
   imapsieve_mailbox2_causes = COPY
   imapsieve_mailbox2_before = file:${SIEVE_PIPE_BIN_DIR}/learn-ham.sieve

test/test-files/nc_templates/rspamd_imap_move_to_inbox.txt

@@ -1,4 +1,4 @@
 A LOGIN user1@localhost.localdomain 123
 B SELECT Junk
-A UID MOVE 1:1 INBOX
-A4 LOGOUT
+A MOVE 1:1 INBOX
+A LOGOUT

test/test-files/nc_templates/rspamd_imap_move_to_junk.txt

@@ -1,4 +1,4 @@
 A LOGIN user1@localhost.localdomain 123
 B SELECT INBOX
-A UID MOVE 1:1 Junk
-A4 LOGOUT
+A MOVE 1:1 Junk
+A LOGOUT

test/tests/parallel/set1/spam_virus/rspamd_full.bats

@@ -237,14 +237,8 @@ function teardown_file() { _default_teardown ; }
   _run_in_container grep -F 'sieve_pipe_bin_dir = /usr/lib/dovecot/sieve-pipe' "${SIEVE_CONFIG_FILE}"
   assert_success
 
-  # Move an email to the "Junk" folder from "INBOX"; the first email we
-  # sent should pass fine, hence we can now move it
-  _send_email 'nc_templates/rspamd_imap_move_to_junk' '0.0.0.0 143'
-  sleep 1 # wait for the transaction to finish
-
-  local MOVE_TO_JUNK_LINES=(
+  local LEARN_SPAM_LINES=(
     'imapsieve: mailbox Junk: MOVE event'
-    'imapsieve: Matched static mailbox rule [1]'
     "sieve: file storage: script: Opened script \`learn-spam'"
     'sieve: file storage: Using Sieve script path: /usr/lib/dovecot/sieve-pipe/learn-spam.sieve'
     "sieve: Executing script from \`/usr/lib/dovecot/sieve-pipe/learn-spam.svbin'"
@@ -254,20 +248,7 @@ function teardown_file() { _default_teardown ; }
     "left message in mailbox 'Junk'"
   )
 
-  _run_in_container cat /var/log/mail/mail.log
-  assert_success
-  for LINE in "${MOVE_TO_JUNK_LINES[@]}"
-  do
-    assert_output --partial "${LINE}"
-  done
-
-  # Move an email to the "INBOX" folder from "Junk"; there should be two mails
-  # in the "Junk" folder
-  _send_email 'nc_templates/rspamd_imap_move_to_inbox' '0.0.0.0 143'
-  sleep 1 # wait for the transaction to finish
-
-  local MOVE_TO_JUNK_LINES=(
-    'imapsieve: Matched static mailbox rule [2]'
+  local LEARN_HAM_LINES=(
     "sieve: file storage: script: Opened script \`learn-ham'"
     'sieve: file storage: Using Sieve script path: /usr/lib/dovecot/sieve-pipe/learn-ham.sieve'
     "sieve: Executing script from \`/usr/lib/dovecot/sieve-pipe/learn-ham.svbin'"
@@ -275,9 +256,30 @@ function teardown_file() { _default_teardown ; }
     "left message in mailbox 'INBOX'"
   )
 
+  # Move an email to the "Junk" folder from "INBOX"; the first email we
+  # sent should pass fine, hence we can now move it.
+  _send_email 'nc_templates/rspamd_imap_move_to_junk' '0.0.0.0 143'
+  sleep 1 # wait for the transaction to finish
+
   _run_in_container cat /var/log/mail/mail.log
   assert_success
-  for LINE in "${MOVE_TO_JUNK_LINES[@]}"
+  assert_output --partial 'imapsieve: Matched static mailbox rule [1]'
+  refute_output --partial 'imapsieve: Matched static mailbox rule [2]'
+  for LINE in "${LEARN_SPAM_LINES[@]}"
+  do
+    assert_output --partial "${LINE}"
+  done
+
+  # Move an email to the "INBOX" folder from "Junk"; there should be two mails
+  # in the "Junk" folder, since the second email we sent during setup should
+  # have landed in the Junk folder already.
+  _send_email 'nc_templates/rspamd_imap_move_to_inbox' '0.0.0.0 143'
+  sleep 1 # wait for the transaction to finish
+
+  _run_in_container cat /var/log/mail/mail.log
+  assert_success
+  assert_output --partial 'imapsieve: Matched static mailbox rule [2]'
+  for LINE in "${LEARN_HAM_LINES[@]}"
   do
     assert_output --partial "${LINE}"
   done