From b2a38518f3be458181d7ad75900d8305cdb9a297 Mon Sep 17 00:00:00 2001
From: AlDanial <al.danial@gmail.com>
Date: Fri, 16 Feb 2024 17:15:51 -0800
Subject: [PATCH] XML start block comment on line w/code #811

---
 Unix/cloc                                     | 20 ++++++++++++++----
 Unix/t/01_opts.t                              |  7 +++++++
 cloc                                          | 21 +++++++++++++++----
 tests/inputs/issues/811/inline_comment.xml    |  6 ++++++
 .../issues/811/inline_comment.xml.yaml        | 21 +++++++++++++++++++
 5 files changed, 67 insertions(+), 8 deletions(-)
 create mode 100644 tests/inputs/issues/811/inline_comment.xml
 create mode 100644 tests/outputs/issues/811/inline_comment.xml.yaml

diff --git a/Unix/cloc b/Unix/cloc
index 78ef6c1c..b20ff37a 100755
--- a/Unix/cloc
+++ b/Unix/cloc
@@ -7160,6 +7160,7 @@ sub remove_matches {                         # {{{1
         $pattern , # in   Perl regular expression (case insensitive)
        ) = @_;
     print "-> remove_matches(pattern=$pattern)\n" if $opt_v > 2;
+    print "[", join("]\n[", @{$ra_lines}), "]\n" if $opt_v > 4;
 
     my @save_lines = ();
     foreach (@{$ra_lines}) {
@@ -7171,6 +7172,7 @@ sub remove_matches {                         # {{{1
         push @save_lines, $_;
     }
 
+    print "[", join("]\n[", @{$ra_lines}), "]\n" if $opt_v > 4;
     print "<- remove_matches\n" if $opt_v > 2;
 #print "remove_matches returning\n   ", join("\n   ", @save_lines), "\n";
     return @save_lines;
@@ -7181,6 +7183,7 @@ sub remove_matches_2re {                     # {{{1
         $pattern2, # in Perl regex 2 (case insensitive) to not match prev line
        ) = @_;
     print "-> remove_matches_2re(pattern=$pattern1,$pattern2)\n" if $opt_v > 2;
+    print "[", join("]\n[", @{$ra_lines}), "]\n" if $opt_v > 4;
 
     my @save_lines = ();
     for (my $i = 0; $i < scalar @{$ra_lines}; $i++) {
@@ -7197,6 +7200,7 @@ sub remove_matches_2re {                     # {{{1
         push @save_lines, $ra_lines->[$i];
     }
 
+    print "[", join("]\n[", @{$ra_lines}), "]\n" if $opt_v > 4;
     print "<- remove_matches_2re\n" if $opt_v > 2;
 #print "remove_matches_2re returning\n   ", join("\n   ", @save_lines), "\n";
     return @save_lines;
@@ -7308,6 +7312,7 @@ sub remove_between {                         # {{{1
     # '{}'  '()'  '[]'  or  '<>'
 
     print "-> remove_between(marker=$marker)\n" if $opt_v > 2;
+    print "[", join("]\n[", @{$ra_lines}), "]\n" if $opt_v > 4;
     my %acceptable = ('{}'=>1,  '()'=>1,  '[]'=>1,  '<>'=>1, );
     die "remove_between:  invalid delimiter '$marker'\n",
         "the delimiter must be one of these four pairs:\n",
@@ -7326,6 +7331,7 @@ sub remove_between {                         # {{{1
         $all_lines =~ s/$1//g;
     }
 
+    print "[", join("]\n[", split(/\n/, $all_lines)), "]\n" if $opt_v > 4;
     print "<- remove_between\n" if $opt_v > 2;
     return split("\n", $all_lines);
 } # 1}}}
@@ -7337,6 +7343,7 @@ sub rm_comments_in_strings {                 # {{{1
     print "-> rm_comments_in_strings(string_marker=$string_marker, " .
           "start_comment=$start_comment, end_comment=$end_comment)\n"
         if $opt_v > 2;
+    print "[", join("]\n[", @{$ra_lines}), "]\n" if $opt_v > 4;
 
     my @save_lines = ();
     my $in_ml_string = 0;
@@ -7407,6 +7414,7 @@ sub rm_comments_in_strings {                 # {{{1
         push @save_lines, $new_line;
     }
 
+    print "[", join("]\n[", @save_lines), "]\n" if $opt_v > 4;
     print "<- rm_comments_in_strings\n" if $opt_v > 2;
     return @save_lines;
 } # 1}}}
@@ -7539,7 +7547,7 @@ sub replace_between_regex  {                 # {{{1
         push @save_lines, $_;
     }
 
-    print "[", join("][", @{$ra_lines}), "]\n" if $opt_v > 4;
+    print "[", join("]\n[", @{$ra_lines}), "]\n" if $opt_v > 4;
     print "<- replace_between_regex\n" if $opt_v > 2;
     return @save_lines;
 } # 1}}}
@@ -7656,7 +7664,11 @@ sub remove_html_comments {                   # {{{1
         }
         next if /^\s*$/;
         $in_comment = 1 if /^(.*?)<!\-\-/;
-        next if defined $1 and $1 =~ /^\s*$/;
+        if (defined $1 and $1 !~ /^\s*$/) {
+            # has both code and comment
+            push @save_lines, "$1\n";
+            next;
+        }
         next if ($in_comment);
         push @save_lines, $_;
     }
@@ -7937,7 +7949,7 @@ sub docstring_rm_comments {                  # {{{1
         }
     }
 
-    print "[", join("][", @{$ra_lines}), "]\n" if $opt_v > 4;
+    print "[", join("]\n[", @{$ra_lines}), "]\n" if $opt_v > 4;
     print "<- docstring_rm_comments\n" if $opt_v > 2;
     return @{$ra_lines};
 } # 1}}}
@@ -11514,7 +11526,7 @@ sub call_regexp_common {                     # {{{1
     # a bogus use of %RE to avoid:
     # Name "main::RE" used only once: possible typo at cloc line xx.
     print scalar keys %RE if $opt_v < -20;
-    print "[", join("][", @{$ra_lines}), "]\n" if $opt_v > 4;
+    print "[", join("]\n[", @{$ra_lines}), "]\n" if $opt_v > 4;
     print "<- call_regexp_common\n" if $opt_v > 2;
     return split("\n", $all_lines);
 } # 1}}}
diff --git a/Unix/t/01_opts.t b/Unix/t/01_opts.t
index c1063d91..c361c4bc 100755
--- a/Unix/t/01_opts.t
+++ b/Unix/t/01_opts.t
@@ -781,6 +781,13 @@ my @Tests = (
                     'ref'  => '../tests/outputs/issues/806/results.yaml',
                 },
 
+                {
+                    'name' => 'XML start block comment on line w/code #811',
+                    'cd'   => '../tests/inputs/issues/811',
+                    'args' => 'inline_comment.xml',
+                    'ref'  => '../tests/outputs/issues/811/inline_comment.xml.yaml',
+                },
+
             );
 
 # Special cases:
diff --git a/cloc b/cloc
index e57162c1..1855e52a 100755
--- a/cloc
+++ b/cloc
@@ -7175,6 +7175,7 @@ sub remove_matches {                         # {{{1
         $pattern , # in   Perl regular expression (case insensitive)
        ) = @_;
     print "-> remove_matches(pattern=$pattern)\n" if $opt_v > 2;
+    print "[", join("]\n[", @{$ra_lines}), "]\n" if $opt_v > 4;
 
     my @save_lines = ();
     foreach (@{$ra_lines}) {
@@ -7186,6 +7187,7 @@ sub remove_matches {                         # {{{1
         push @save_lines, $_;
     }
 
+    print "[", join("]\n[", @{$ra_lines}), "]\n" if $opt_v > 4;
     print "<- remove_matches\n" if $opt_v > 2;
 #print "remove_matches returning\n   ", join("\n   ", @save_lines), "\n";
     return @save_lines;
@@ -7196,6 +7198,7 @@ sub remove_matches_2re {                     # {{{1
         $pattern2, # in Perl regex 2 (case insensitive) to not match prev line
        ) = @_;
     print "-> remove_matches_2re(pattern=$pattern1,$pattern2)\n" if $opt_v > 2;
+    print "[", join("]\n[", @{$ra_lines}), "]\n" if $opt_v > 4;
 
     my @save_lines = ();
     for (my $i = 0; $i < scalar @{$ra_lines}; $i++) {
@@ -7212,6 +7215,7 @@ sub remove_matches_2re {                     # {{{1
         push @save_lines, $ra_lines->[$i];
     }
 
+    print "[", join("]\n[", @{$ra_lines}), "]\n" if $opt_v > 4;
     print "<- remove_matches_2re\n" if $opt_v > 2;
 #print "remove_matches_2re returning\n   ", join("\n   ", @save_lines), "\n";
     return @save_lines;
@@ -7323,6 +7327,7 @@ sub remove_between {                         # {{{1
     # '{}'  '()'  '[]'  or  '<>'
 
     print "-> remove_between(marker=$marker)\n" if $opt_v > 2;
+    print "[", join("]\n[", @{$ra_lines}), "]\n" if $opt_v > 4;
     my %acceptable = ('{}'=>1,  '()'=>1,  '[]'=>1,  '<>'=>1, );
     die "remove_between:  invalid delimiter '$marker'\n",
         "the delimiter must be one of these four pairs:\n",
@@ -7341,6 +7346,7 @@ sub remove_between {                         # {{{1
         $all_lines =~ s/$1//g;
     }
 
+    print "[", join("]\n[", split(/\n/, $all_lines)), "]\n" if $opt_v > 4;
     print "<- remove_between\n" if $opt_v > 2;
     return split("\n", $all_lines);
 } # 1}}}
@@ -7352,6 +7358,7 @@ sub rm_comments_in_strings {                 # {{{1
     print "-> rm_comments_in_strings(string_marker=$string_marker, " .
           "start_comment=$start_comment, end_comment=$end_comment)\n"
         if $opt_v > 2;
+    print "[", join("]\n[", @{$ra_lines}), "]\n" if $opt_v > 4;
 
     my @save_lines = ();
     my $in_ml_string = 0;
@@ -7422,6 +7429,7 @@ sub rm_comments_in_strings {                 # {{{1
         push @save_lines, $new_line;
     }
 
+    print "[", join("]\n[", @save_lines), "]\n" if $opt_v > 4;
     print "<- rm_comments_in_strings\n" if $opt_v > 2;
     return @save_lines;
 } # 1}}}
@@ -7554,7 +7562,7 @@ sub replace_between_regex  {                 # {{{1
         push @save_lines, $_;
     }
 
-    print "[", join("][", @{$ra_lines}), "]\n" if $opt_v > 4;
+    print "[", join("]\n[", @{$ra_lines}), "]\n" if $opt_v > 4;
     print "<- replace_between_regex\n" if $opt_v > 2;
     return @save_lines;
 } # 1}}}
@@ -7671,7 +7679,11 @@ sub remove_html_comments {                   # {{{1
         }
         next if /^\s*$/;
         $in_comment = 1 if /^(.*?)<!\-\-/;
-        next if defined $1 and $1 =~ /^\s*$/;
+        if (defined $1 and $1 !~ /^\s*$/) {
+            # has both code and comment
+            push @save_lines, "$1\n";
+            next;
+        }
         next if ($in_comment);
         push @save_lines, $_;
     }
@@ -7952,7 +7964,7 @@ sub docstring_rm_comments {                  # {{{1
         }
     }
 
-    print "[", join("][", @{$ra_lines}), "]\n" if $opt_v > 4;
+    print "[", join("]\n[", @{$ra_lines}), "]\n" if $opt_v > 4;
     print "<- docstring_rm_comments\n" if $opt_v > 2;
     return @{$ra_lines};
 } # 1}}}
@@ -8543,6 +8555,7 @@ sub set_constants {                          # {{{1
             'jsf'         => 'JavaServer Faces'      ,
             'jsx'         => 'JSX'                   ,
             'xhtml'       => 'XHTML'                 ,
+            'j2'          => 'Jinja Template'        ,
             'jinja'       => 'Jinja Template'        ,
             'jinja2'      => 'Jinja Template'        ,
             'yyp'         => 'JSON'                  ,
@@ -11528,7 +11541,7 @@ sub call_regexp_common {                     # {{{1
     # a bogus use of %RE to avoid:
     # Name "main::RE" used only once: possible typo at cloc line xx.
     print scalar keys %RE if $opt_v < -20;
-    print "[", join("][", @{$ra_lines}), "]\n" if $opt_v > 4;
+    print "[", join("]\n[", @{$ra_lines}), "]\n" if $opt_v > 4;
     print "<- call_regexp_common\n" if $opt_v > 2;
     return split("\n", $all_lines);
 } # 1}}}
diff --git a/tests/inputs/issues/811/inline_comment.xml b/tests/inputs/issues/811/inline_comment.xml
new file mode 100644
index 00000000..6b28a52f
--- /dev/null
+++ b/tests/inputs/issues/811/inline_comment.xml
@@ -0,0 +1,6 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<foo>
+    <bar/><!-- bar 
+            is not 
+            foo -->
+</foo>
diff --git a/tests/outputs/issues/811/inline_comment.xml.yaml b/tests/outputs/issues/811/inline_comment.xml.yaml
new file mode 100644
index 00000000..304035c7
--- /dev/null
+++ b/tests/outputs/issues/811/inline_comment.xml.yaml
@@ -0,0 +1,21 @@
+---
+# github.com/AlDanial/cloc
+header : 
+  cloc_url           : github.com/AlDanial/cloc
+  cloc_version       : 1.99
+  elapsed_seconds    : 0.00256609916687012
+  n_files            : 1
+  n_lines            : 6
+  files_per_second   : 389.696553005668
+  lines_per_second   : 2338.17931803401
+  report_file        : ../../../outputs/issues/811/inline_comment.xml.yaml
+'XML' :
+  nFiles: 1
+  blank: 0
+  comment: 2
+  code: 4
+SUM: 
+  blank: 0
+  comment: 2
+  code: 4
+  nFiles: 1