X-Git-Url: http://wagner.pp.ru/gitweb/?a=blobdiff_plain;f=Tex2fb2;h=693149fcdc401644ad80bda7cf08af16a61a47eb;hb=90f2749e4d60b2700566294a2dc75370dcf356d2;hp=2fe05809eb85e44e5eadcf554beede9f87138bb3;hpb=37b1eefaa8690036f7e07615a62556746c4e91f4;p=fiction%2FKate-the-Empress.git

diff --git a/Tex2fb2 b/Tex2fb2
old mode 100644
new mode 100755
index 2fe0580..693149f
--- a/Tex2fb2
+++ b/Tex2fb2
@@ -1,4 +1,5 @@
-#!/usr/bin/perl
+#!/usr/bin/perl -CDS
+use utf8;
 # char-level modes
 my $poetry = 0;
 my $verbatim = 0;
@@ -21,10 +22,11 @@ while (<F>) {
 close F;
 
 my $header =1;
+LINE:
 while (<>) {
 $environ = undef;
 if (/\\(begin|end){(\w+)}/) {
-	$environ = $2;
+	$environ=$2;
 	$begin=$1 eq "begin";
 	if ($environ eq 'verbatim') {
 		$verbatim=$begin;
@@ -32,33 +34,56 @@ if (/\\(begin|end){(\w+)}/) {
 		if ($begin) {
 		pushsection("poem",undef);
 		} else {
-		flushsection('poem');
+			add_to_section(tag($buffer,'stanza')."\n") if $buffer;
+			$buffer="";
+			flushsection('poem');
 		}
 		$poetry = $begin;
-	} elsif($environ = 'document' && $begin) {
+	} elsif($environ eq 'document' && $begin) {
 		$header=0;
 	}
 }
 next LINE if $header;
-if (/^$/ && $environ && $buffer) {
+if ((/^$/ || $environ) && $buffer) {
 #output on empty line (p or stanza) depending on poetry mode
-	add_to_section(tag($buffer,$poetry?"stanza":"p"));
+	add_to_section(tag($buffer,$poetry?"stanza":"p")."\n");
 	$buffer="";
 }
 next LINE if $environ;
 # Section headings
-if (/\\(part|chapter|section|subsection|subsubsection){(.*)}/) {
+if (/\\(part|chapter|section|subsection|subsubsection)\*?{(.*)}/) {
+	if ($buffer) {
+		add_to_section(tag($buffer,$poetry?"stanza":"p"));
+		$buffer="";
+	}
 	pushsection($1,$2);
+	next LINE;
+}
+if (/\\vspace{/) {
+	add_to_section("<empty-line />");
+	next LINE;
 }
+next LINE if /\\pagebreak\b/;
 #normal mode: 
 if (!$verbatim) {
 #strip TeX comments 
 s/([^\\])%.*$/$1/;
+s/^%.*$//;
+# strip \sloppy
+s/\\sloppy\s+//g;
+s/\\sloppy{}//g;
+s/\\sloppy([^\w])/$1/g;
+# strip extra space
+s/^\s+//;
+s/\s+$//;
+s/(\s)\s+/$1/g;
 #replace TeX ligatures ~ --- << >> \% with appropriate unicode symbols
 s/~/\xA0/g;
 s/---/-/g;
 s/<</«/g;
 s/>>/»/g;
+s/\\%/%/g;
+s/\\dots/\x{2026}/g;
 }
 #replace ' and " with entities
 s/&/&amp;/g;
@@ -69,15 +94,17 @@ s/>/&gt;/g;
 
 if ($poetry) {
 	chomp;
-  $buffer.=tag($_,'v');
+	s/\s*\\\\$//;
+  $buffer.=tag($_,'v')."\n";
 } elsif ($verbatim) {
 	add_to_section(tag(tag($_,"code"),"p"));
 } else {
-  $buffer.=$_;
+  $buffer.=" ".$_;
 }
 }
 if ($buffer) {
 	add_to_section(tag($buffer,"p"));
+	$buffer="";
 }
 
 while (@sections) {
@@ -102,7 +129,7 @@ sub flushsection {
 	}
 	$content .=  $str->{data};
 	if ($#sections >=0) {
-		add_to_section(tag($content,$tag));
+		add_to_section(tag($content,$tag)."\n");
 	} else {
 		print tag($content,$tag);
 	}
@@ -128,5 +155,6 @@ sub pushsection {
 
 sub tag {
 	my ($content,$name) = @_;
+	return "" if $content =~ /^\s*$/s;
 	return "<$name>$content</$name>";
 }