Whine on first non-ASCII byte-sequence and use it to guess Latin-1 vs…

… UTF-8

lib/Pod/Simple/BlackBox.pm

@@ -123,6 +123,9 @@ sub parse_lines {             # Usage: $parser->parse_lines(@lines)
       }
     }
 
+    if(!$self->{'encoding'}) {
+      $self->_try_encoding_guess($line)
+    }
 
     DEBUG > 5 and print "# Parsing line: [$line]\n";
 
@@ -395,6 +398,24 @@ sub _handle_encoding_second_level {
   return;
 }
 
+sub _try_encoding_guess {
+  my ($self,$line) = @_;
+
+  return unless $line =~ /[^\x00-\x7f]/;  # Look for non-ASCII byte
+
+  my $encoding = $line =~ /[\xC0-\xFD][\x80-\xBF]/ ? 'UTF-8' : 'ISO8859-1';
+  $self->_handle_encoding_line( "=encoding $encoding" );
+  $self->{'_transcoder'} && $self->{'_transcoder'}->($line);
+
+  my ($word) = $line =~ /(\S*[^\x00-\x7f]\S*)/;
+
+  $self->whine(
+    $self->{'line_count'},
+    "Non-ASCII character seen before =encoding in '$word'. Assuming $encoding"
+  );
+
+}
+
 #~`~`~`~`~`~`~`~`~`~`~`~`~`~`~`~`~`~`~`~`~`~`~`~`~`~`~`~`~`~`~`~`~`~`~`~`~`
 
 {

t/corpus/encwarn01.txt

@@ -0,0 +1,11 @@
+
+=head1 NAME
+
+Encoding Warning 1 - implicitly Latin-1
+
+=head2 DESCRIPTION
+
+This line should warn that the word caf� contains a non-ASCII character.
+
+But ch�teau should not generate a warning - once is enough.
+

t/corpus/encwarn01.xml

@@ -0,0 +1,38 @@
+<Document start_line="2">
+  <head1 start_line="2">
+    NAME
+  </head1>
+  <Para start_line="4">
+    Encoding Warning 1 - implicitly Latin-1
+  </Para>
+  <head2 start_line="6">
+    DESCRIPTION
+  </head2>
+  <Para start_line="8">
+    This line should warn that the word caf&#233; contains a
+    non-ASCII character.
+  </Para>
+  <Para start_line="10">
+    But ch&#226;teau should not generate a warning - once is
+    enough.
+  </Para>
+  <head1 errata="1" start_line="-321">
+    POD ERRORS
+  </head1>
+  <Para errata="1" start_line="-321">
+    Hey! 
+    <B>
+      The above document had some coding errors, which are explained
+      below:
+    </B>
+  </Para>
+  <over-text errata="1" indent="4" start_line="-321">
+    <item-text start_line="-321">
+      Around line 8:
+    </item-text>
+    <Para start_line="-321">
+      Non-ASCII character seen before =encoding in &#39;caf&#233;&#39;.
+      Assuming ISO8859-1
+    </Para>
+  </over-text>
+</Document>

t/corpus/encwarn02.txt

@@ -0,0 +1,11 @@
+
+=head1 NAME
+
+Encoding Warning 1 - implicitly UTF-8
+
+=head2 DESCRIPTION
+
+This line should warn that the price €9.99 contains a non-ASCII character.
+
+But château should not generate a warning - once is enough.
+

t/corpus/encwarn02.xml

@@ -0,0 +1,38 @@
+<Document start_line="2">
+  <head1 start_line="2">
+    NAME
+  </head1>
+  <Para start_line="4">
+    Encoding Warning 1 - implicitly UTF-8
+  </Para>
+  <head2 start_line="6">
+    DESCRIPTION
+  </head2>
+  <Para start_line="8">
+    This line should warn that the price &#8364;9.99 contains
+    a non-ASCII character.
+  </Para>
+  <Para start_line="10">
+    But ch&#226;teau should not generate a warning - once is
+    enough.
+  </Para>
+  <head1 errata="1" start_line="-321">
+    POD ERRORS
+  </head1>
+  <Para errata="1" start_line="-321">
+    Hey! 
+    <B>
+      The above document had some coding errors, which are explained
+      below:
+    </B>
+  </Para>
+  <over-text errata="1" indent="4" start_line="-321">
+    <item-text start_line="-321">
+      Around line 8:
+    </item-text>
+    <Para start_line="-321">
+      Non-ASCII character seen before =encoding in &#39;&#8364;9.99&#39;.
+      Assuming UTF-8
+    </Para>
+  </over-text>
+</Document>

t/corpus/lat1frim.xml

@@ -67,4 +67,23 @@
   <Para start_line="33">
     [end]
   </Para>
+  <head1 errata="1" start_line="-321">
+    POD ERRORS
+  </head1>
+  <Para errata="1" start_line="-321">
+    Hey! 
+    <B>
+      The above document had some coding errors, which are explained
+      below:
+    </B>
+  </Para>
+  <over-text errata="1" indent="4" start_line="-321">
+    <item-text start_line="-321">
+      Around line 11:
+    </item-text>
+    <Para start_line="-321">
+      Non-ASCII character seen before =encoding in &#39;s&#233;parant&#39;.
+      Assuming ISO8859-1
+    </Para>
+  </over-text>
 </Document>