mht: Handle contiguous elements (like Qp)

Qp is a quoted paragraph.  A subsequent Qp should be put within the
<blockquote> tag of the previous Qp.

1 files changed, 52 insertions, 10 deletions

diff --git a/mht b/mht
index 070c121..1591730 100755
--- a/mht
+++ b/mht
@@ -12,22 +12,28 @@ use Text::ParseWords qw/quotewords/;
 # 1.1  Global program state
 
 my $empty = ''; # currently buffered empty lines
-my $close = ''; # buffered closing tag for currently opened element
+my $close = ''; # buffered closing tag for currently opened block element
+my $postclose = ''; # buffered extra closing tag for opened contiguous element
+my $opel = ''; # currently opened block element
 
 
 # 1.2  Elements
 
-# 1.2.1  Block elements
+# 1.2.1  Contiguous elements
+my %ctels = (
+	Qp_0 => '<blockquote>{<p>%</p>}</blockquote>',
+);
+
+# 1.2.2  Block elements
 my %blels = (
 	Pp_0 => '<p>%</p>',
-	Qp_0 => '<blockquote><p>%</p></blockquote>',
 	Pr_0 => '<pre>%</pre>',
 	Sh_0 => '<h3>%</h3>',
 	Sh_1 => '<h\\$1>%</h\\$1>',
 	Ti_0 => '<title>%</title>',
 );
 
-# 1.2.2  Inline elements
+# 1.2.3  Inline elements
 my %inels = (
 	Au_1 => '<meta name="author" content="\\$1"/>',
 	Bd_0 => '\\$3<b>\\$1</b>\\$2',
@@ -53,15 +59,33 @@ sub request {
 	my $n = @argv;
 	my $elkey = "${el}_$n";
 
-	if (exists $blels{$elkey}) {
+	if (exists $blels{$elkey} or exists $ctels{$elkey}) {
 		# Clear empty line buffer
 		$empty = '';
 
 		# Close currently open block element, open new
+		my ($base, $prestart, $newpostclose) = ('', '', '');
+		if (exists $ctels{$elkey}) {
+			$base = $ctels{$elkey};
+			$prestart = prestart($base);
+			$newpostclose = postclose($base);
+			$base = inner($base);
+		} else {
+			$base = $blels{$elkey};
+		}
+		my $start = start($base);
+		my $newclose = interpol(_close($base), @argv) . "\n";
+
 		print $close;
-		$close = interpol(end($blels{$elkey}), @argv) . "\n";
+		$close = $newclose;
 
-		print interpol(start($blels{$elkey}), @argv) . "\n";
+		print $postclose if $el ne $opel;
+		$postclose = $newpostclose;
+
+		print $prestart if $el ne $opel;
+		print interpol($start, @argv) . "\n";
+
+		$opel = $el;
 	} elsif (exists $inels{$elkey}) {
 		print interpol($inels{$elkey}, @argv) . "\n";
 	} else {
@@ -82,16 +106,34 @@ sub interpol {
 	return $s;
 }
 
-# 1.3.3  Retrieve first portion of block element string
+# 1.3.3  Retrieve opening tag of block element string
 sub start {
 	return (split '%', shift)[0];
 }
 
-# 1.3.4  Retrieve second portion of block element string
-sub end {
+# 1.3.4  Retrieve closing tag of block element string
+sub _close {
 	return (split '%', shift)[1];
 }
 
+# 1.3.5  Retrieve extra opening tag of block element string
+sub prestart {
+	return (split '{', shift)[0];
+}
+
+# 1.3.6  Retrieve extra closing tag of block element string
+sub postclose {
+	return (split '}', shift)[-1];
+}
+
+# 1.3.7  Retrieve inner block of contiguous tag
+sub inner {
+	my $s = shift;
+	$s =~ s/^.*?\{//;
+	$s =~ s/}.*?$//;
+	return $s
+}
+
 
 # 2  Program