svn commit: r1929159 - in spamassassin/trunk: . lib/Mail/SpamAssassin t

Author: fkento
Date: Wed Oct 15 19:19:13 2025
New Revision: 1929159

Log:
Add support for nested anchor tags

Added:
   spamassassin/trunk/t/html_nested_anchors.t
Modified:
   spamassassin/trunk/MANIFEST
   spamassassin/trunk/lib/Mail/SpamAssassin/HTML.pm

Modified: spamassassin/trunk/MANIFEST
==============================================================================
--- spamassassin/trunk/MANIFEST	Wed Oct 15 14:28:12 2025	(r1929158)
+++ spamassassin/trunk/MANIFEST	Wed Oct 15 19:19:13 2025	(r1929159)
@@ -555,6 +555,7 @@ t/header.t
 t/header_utf8.t
 t/hashbl.t
 t/html_colors.t
+t/html_nested_anchors.t
 t/html_obfu.t
 t/html_utf8.t
 t/html_visibility.t

Modified: spamassassin/trunk/lib/Mail/SpamAssassin/HTML.pm
==============================================================================
--- spamassassin/trunk/lib/Mail/SpamAssassin/HTML.pm	Wed Oct 15 14:28:12 2025	(r1929158)
+++ spamassassin/trunk/lib/Mail/SpamAssassin/HTML.pm	Wed Oct 15 19:19:13 2025	(r1929159)
@@ -125,6 +125,7 @@ sub new {
   $self->{SA_character_semantics_input} = $character_semantics_input;
   $self->{SA_encode_results} =
     $character_semantics_input && !$character_semantics_output;
+  $self->{anchor_refs} = [];
   $self;
 }
 
@@ -343,6 +344,7 @@ sub html_tag {
     }
     # end tags
     else {
+      pop(@{ $self->{anchor_refs} }) if $tag eq "a";
       $self->{closed_html} = 1 if $tag eq "html";
       $self->{closed_body} = 1 if $tag eq "body";
     }
@@ -859,9 +861,7 @@ sub html_tests {
     }
   }
   if ($tag eq "img" && exists $self->{inside}{a} && $self->{inside}{a} > 0) {
-    my $uri = $self->{anchor_last};
-    utf8::encode($uri) if $self->{SA_encode_results};
-    $self->{uri}->{$uri}->{anchor_text}->[-1] .= "<img>\n";
+    $$_ .= "<img>\n" for (@{$self->{anchor_refs}});
     $self->{anchor}->[-1] .= "<img>\n";
   }
 
@@ -896,11 +896,13 @@ sub html_tests {
 
   # special text delimiters - <a> and <title>
   if ($tag eq "a") {
-    my $uri = $self->{anchor_last} =
+    my $uri =
       (exists $attr->{href} ? $self->canon_uri($attr->{href}) : "");
     utf8::encode($uri) if $self->{SA_encode_results};
     push(@{$self->{uri}->{$uri}->{anchor_text}}, '');
     push(@{$self->{anchor}}, '');
+    # keep track of nested anchors (max depth 10)
+    push(@{$self->{anchor_refs}}, \($self->{uri}->{$uri}->{anchor_text}->[-1])) if @{$self->{anchor_refs}} < 10;
   }
   if ($tag eq "title") {
     $self->{title_index}++;
@@ -977,10 +979,7 @@ sub html_text {
 
   # text that is part of body and also stored separately
   if (exists $self->{inside}{a} && $self->{inside}{a} > 0) {
-    # this doesn't worry about nested anchors
-    my $uri = $self->{anchor_last};
-    utf8::encode($uri) if $self->{SA_encode_results};
-    $self->{uri}->{$uri}->{anchor_text}->[-1] .= $text;
+    $$_ .= $text for (@{$self->{anchor_refs}});
     $self->{anchor}->[-1] .= $text;
   }
   if (exists $self->{inside}{title} && $self->{inside}{title} > 0) {

Added: spamassassin/trunk/t/html_nested_anchors.t
==============================================================================
--- /dev/null	00:00:00 1970	(empty, because file is newly added)
+++ spamassassin/trunk/t/html_nested_anchors.t	Wed Oct 15 19:19:13 2025	(r1929159)
@@ -0,0 +1,45 @@
+#!/usr/bin/perl -T
+use strict;
+use warnings;
+use lib '.'; use lib 't';
+use SATest; sa_t_init("html_nested_anchors");
+use Mail::SpamAssassin::HTML;
+use Test::More;
+
+my @tests = (
+    {
+        html => '<a href="#one">foo<a href="#two">bar</a>baz</a>',
+        uris => {
+            '#one' => {
+                types       => { 'a' => 1 },
+                anchor_text => ['foobarbaz'],
+            },
+            '#two' => {
+                types       => { 'a' => 1 },
+                anchor_text => ['bar'],
+            }
+        }
+    },
+    {
+        html => '<a href="#one">foo<a href="#one">bar</a>baz</a>',
+        uris => {
+            '#one' => {
+                types       => { 'a' => 1 },
+                anchor_text => ['foobarbaz', 'bar'],
+            },
+        }
+    },
+);
+
+plan tests => scalar @tests;
+
+foreach my $test (@tests) {
+    my $html = $test->{html};
+
+    my $html_obj = Mail::SpamAssassin::HTML->new(0, 0);
+    $html_obj->parse($html);
+
+    my $uris = $html_obj->{results}->{uri_detail};
+    is_deeply($uris, $test->{uris}, "URI details match for HTML: $html");
+
+}