From 481e29145e2e7ede4e3bebe5e9d6e35718693fec Mon Sep 17 00:00:00 2001
From: Sergio Durigan Junior <sergiodj@sergiodj.net>
Date: Sun, 29 Dec 2024 20:32:21 -0500
Subject: [PATCH] Several improvements

---
 spam-ham-learn | 77 +++++++++++++++++++++++++++++++++++++++++++++-----
 1 file changed, 70 insertions(+), 7 deletions(-)

diff --git a/spam-ham-learn b/spam-ham-learn
index 9ac6c9a..d89a170 100644
--- a/spam-ham-learn
+++ b/spam-ham-learn
@@ -3,36 +3,99 @@
 # Useful directories.
 readonly MAILDIR="${HOME}/Maildir"
 readonly SPAMDIR="${MAILDIR}/Spam"
-readonly HAMDIR="${MAILDIR}/Ham"
+#readonly HAMDIR="${MAILDIR}/Ham"
 
 # The spam/ham found are saved for future reference.
-readonly SAVE_SPAM_DIR="${HOME}/spamassassin/spam"
-readonly SAVE_HAM_DIR="${HOME}/spamassassin/ham"
+readonly SAVE_SPAM_DIR="${HOME}/spamassassin/spam/"
+readonly SAVE_HAM_DIR="${HOME}/spamassassin/ham/"
 
 # The minimum age (in days) of spam to be processed.
-readonly SPAM_MIN_AGE=3
+readonly SPAM_MIN_AGE="5"
 
 # "--force" flag.
 FORCE="false"
 
+# The temporary file where a list of spam filenames will be stored.
+SPAM_LIST=$(mktemp)
+readonly SPAM_LIST
+HAM_LIST=$(mktemp)
+readonly HAM_LIST
+
+trap 'rm -f "${SPAM_LIST}" "${HAM_LIST}"' EXIT
+
+readonly LOGFILE="${HOME}/spam-ham-learn.log"
+
+log()
+{
+    printf "I: %s" "$*" >> "${LOGFILE}"
+}
+
 learn_spam()
 {
-    local tmpfile=$(mktemp)
     local findargs=""
 
+    log "Entering learn_spam"
+
     if [ "${FORCE}" = "false" ]; then
+        log "--force detected; not using -ctime on find"
         findargs=" -ctime +${SPAM_MIN_AGE} "
     fi
 
+    # Collecting the spam.
+    log "Collecting spam"
     find "${SPAMDIR}/cur" "${SPAMDIR}/new" -type f "${findargs}" | while read -r msg; do
-        printf "%s\n" "${msg}" >> "${tmpfile}"
+        log "Found ${msg}, adding it to the spam list (${SPAM_LIST})"
+        printf "%s\n" "${msg}" >> "${SPAM_LIST}"
     done
 
-    /usr/bin/sa-learn -f "${tmpfile}"
+    # Invoking sa-learn.
+    log "Invoking sa-learn"
+    /usr/bin/sa-learn -f "${SPAM_LIST}"
+
+    # Saving the spam into a different location.
+    log "Saving the spam into a different location"
+    while read -r msg; do
+        log "Moving ${msg} into ${SAVE_SPAM_DIR}"
+        mv "${msg}" "${SAVE_SPAM_DIR}"
+    done < "${SPAM_LIST}"
 }
 
 learn_ham()
 {
+    local nspam
+
+    log "Entering learn_ham"
+
+    nspam=$(wc -l "${SPAM_LIST}" | cut -d' ' -f1)
+    log "Number of spam handled: ${nspam}"
+
+    # We only learn ham if any spam has been processed.
+    if [ "${nspam}" -le 0 ]; then
+        log "The number of spam is 0, doing nothing"
+        return
+    fi
+
+    # Collecting the ham.
+    log "Collecting the ham"
+    find "${MAILDIR}/cur" -type f -printf "%T@ %p\n" \
+        | sort -n \
+        | cut -d ' ' -f2 \
+        | tail -n"${nspam}" \
+        | while read -r msg; do
+        log "Found ham ${msg}, adding it to the ham list (${HAM_LIST})"
+        printf "%s\n" "${msg}" >> "${HAM_LIST}"
+    done
+
+    # Invoking sa-learn.
+    log "Invoking sa-learn"
+    /usr/bin/sa-learn -f "${HAM_LIST}"
+
+    # Saving the ham.
+    log "Saving the ham into a different location"
+    while read -r msg; do
+        log "Copying ${msg} to ${SAVE_HAM_DIR}"
+        cp "${msg}" "${SAVE_HAM_DIR}"
+    done < "${HAM_LIST}"
 }
 
 while [ -n "${1}" ]; do