From: epg <>
Date: Sat, 3 Apr 2004 05:11:30 +0000 (+0000)
Subject: Collapse the spam-checking and filtering passes into a single pass and
X-Git-Url: https://diplodocus.org/git/minc/commitdiff_plain/95a97f883e0e32b7e78b58ff11424fe8a4d4630e

Collapse the spam-checking and filtering passes into a single pass and
adapt the diagnostic output to make sense in the new single pass.

(store_message): Don't print anything.

(preprocess): Move the message sorting to MAIN and spam processing to
filter_mail.

(filter_mail): Do spam filtering in this loop and do all status
printing here.

(%FOLDERS): This should not have been global; move it into
filter_mail.

(MAIN): Sort messages before calling filter_mail.  If any SPAM
messages were detected, scan(1) them.
---

diff --git a/minc b/minc
index 1b5100f..8d8b6a3 100755
--- a/minc
+++ b/minc
@@ -390,12 +390,6 @@ sub get_highest_msgnum {
     return $highest;
 }
 
-# We want to print the name of each list that has new mail only once,
-# so use this hash to store the lists that have already been printed.
-# Start the list out with SPAM already in it, since we don't care when
-# new messages are added to it.
-my %FOLDERS = ('SPAM'=>1);
-
 sub store_message {
     my $msg = shift;
     my $mhfolder = shift;
@@ -479,11 +473,6 @@ sub store_message {
             }
         }
     }
-
-    if (not $FOLDERS{$mhfolder}) {
-        print("+$mhfolder\n");
-        $FOLDERS{$mhfolder} = 1;
-    }
 }
 
 # Parse a message file into a structure describing the headers.  The
@@ -535,47 +524,6 @@ sub get_headers {
     return %headers;
 }
 
-###############################################################################
-# Spam handling, sorting, etc.
-
-sub preprocess {
-    my @msglist = @_;
-    my @baton;
-    my $msg;
-    my $i;
-    my $st;
-    my @result;
-
-    @baton = spam_start_hook();
-
-    foreach $msg (@msglist) {
-        printf('%sChecking for spam... %6d/%d',
-               "\r", ++$i, scalar(@msglist));
-        if ($printfilenames) {
-            print($msg);
-        }
-        if (spam_check($msg, @baton)) {
-            print(" SPAM\n");
-            store_message($msg, 'SPAM');
-        } else {
-            push(@result, $msg);
-        }
-    }
-    print("\nDone: ", scalar(@result), " survivors\n");
-
-    spam_stop_hook(@baton);
-
-    return (map { $_->[1] }
-            sort { $a->[0] <=> $b->[0] }
-            map {
-                if (not ($st = stat($_))) {
-                    die("stat($_): $!");
-                }
-                [$st->mtime, $_]
-            }
-            @result);
-}
-
 
 ###############################################################################
 # Filtering
@@ -644,18 +592,47 @@ sub find_mh_folder {
 
 sub filter_mail {
     my @msglist = @_;
+    my $msgcount = scalar(@msglist);
+    my @baton;
     my $msg;
     my $mhfolder;
+    my $spam = 0;
+    my $saved = 0;
+    my %FOLDERS = ('SPAM'=>1);
 
+    @baton = spam_start_hook();
+
+    print("$msgcount messages...");
     foreach $msg (@msglist) {
-        $mhfolder = find_mh_folder($msg);
+        if (spam_check($msg, @baton)) {
+            $mhfolder = 'SPAM';
+            $spam = scalar(@SPAM)+ 1;
+        } else {
+            $mhfolder = find_mh_folder($msg);
+            $saved++;
+        }
+
         store_message($msg, $mhfolder);
+
+        print("\r");
+        if (not $FOLDERS{$mhfolder}) {
+            print(' ' x length($msgcount));
+            print("                                \r+$mhfolder\n");
+            $FOLDERS{$mhfolder} = 1;
+        }
+
+        printf('%6d SPAM %6d saved %6d/%1d',
+               $spam, $saved, $spam + $saved, $msgcount);
     }
+    print("\n");
+
+    spam_stop_hook(@baton);
 }
 
 
 MAIN: {
     my @msglist;
+    my $st;
 
     if ($dumpfilters) {
         $Data::Dumper::Indent = 1;
@@ -663,15 +640,20 @@ MAIN: {
         exit;
     }
 
-    @msglist = getfiles("$MAILDIR/new");
+    @msglist = (
+                map { $_->[1] }
+                sort { $a->[0] <=> $b->[0] }
+                map {
+                    if (not ($st = stat($_))) {
+                        die("stat($_): $!");
+                    }
+                    [$st->mtime, $_]
+                }
+                getfiles("$MAILDIR/new"));
 
-    if (not $filteronly) {
-        @msglist = preprocess(@msglist);
-    }
+    filter_mail(@msglist);
 
-    if (not $spamonly) {
-        filter_mail(@msglist);
-    }
+    @SPAM and (exec('scan', '+SPAM', @SPAM) or die);
 }