Typo fixes

[fiction/Kate-the-Empress.git] / Tex2fb2
diff --git a/Tex2fb2 b/Tex2fb2

index 30988082287d8c4345292a34f511a10b60ca4e66..abc38bba4320338ddd5275b28ab8f23b90609fff 100755 (executable)
--- a/Tex2fb2
+++ b/Tex2fb2
@@ -1,6 +1,8 @@
  #!/usr/bin/perl -CDS
+
  use utf8;
  use POSIX qw(strftime);
+use MIME::Base64;
  # char-level modes
  my $poetry = 0;
  my $verbatim = 0;
@@ -19,6 +21,7 @@ print "<FictionBook xmlns=\"http://www.gribuser.ru/xml/fictionbook/2.0\"
  xmlns:l=\"http://www.w3.org/1999/xlink\">\n";
  my $metadata = shift @ARGV;
  open F,"<",$metadata;
+my $pics = "";
  while (<F>) {
         # Replace empty date with current date
         if (/<date\s+\/>/ || /<date>\s*<\/date>/) {
@@ -30,6 +33,20 @@ while (<F>) {
                 my $ver = $1+time()/1E10;
                 $_=tag(sprintf("%g",$ver),"version")."\n";
         }
+       # Если  существует cover.png, добавляем coverpage
+       if (/<coverpage \/>/) {
+               $_="";
+               COVER:
+               for $file ("cover.jpg","cover.png") {
+                       if (-f $file) { 
+                               my $id = $file;
+                               $id=~tr/./_/;
+                               $_ = "<coverpage>\n<image l:href=\"#$id\" />\n</coverpage>\n";
+                               $pics .= mkbinary($file,$id);
+                               last COVER;
+                       }
+               }       
+       }
         print $_;
  }
  close F;
@@ -64,7 +81,7 @@ if ((/^$/ || $environ) && $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(flushbuffer($buffer),$poetry?"stanza":"p"));
                 $buffer="";
@@ -72,7 +89,7 @@ if (/\\(part|chapter|section|subsection|subsubsection)\*?{(.*)}/) {
         pushsection($1,tag($2,"p"));
         next LINE;
  }
-if (/\\vspace{/) {
+if (/\\vspace\{/) {
         add_to_section("<empty-line />");
         next LINE;
  }
@@ -89,22 +106,16 @@ if (!$verbatim) {
  s/([^\\])%.*$/$1/;
  s/^%.*$//;
  # strip \sloppy
+s/\\(\w+)\\sloppy/\\$1/g;
  s/\\sloppy\s+//g;
-s/\\sloppy{}//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/\\-/\xAD/g;
-s/---/—/g;
-s/<</«/g;
-s/>>/»/g;
-s/\\%/%/g;
-s/\\dots/\x{2026}/g;
-s/\\verb(.)(.*)\1/<code>$2<\/code>/;
+$_ = fix_ligatures($_);
  }
  
  if ($poetry) {
@@ -117,9 +128,8 @@ if ($poetry) {
                 $poetic_buffer=$_;
                 next LINE;
         }
-       s/\\footnote{(.*)}/push_footnote($1)/e;
         s/\s*\\\\$//;
-  $buffer.=tag($_,'v')."\n";
+  $buffer.=tag(flushbuffer($_),'v')."\n";
  } elsif ($verbatim) {
         add_to_section(tag(tag($_,"code"),"p"));
  } else {
@@ -137,6 +147,7 @@ while (@sections) {
  print "</body>\n";
  ## print footnotes
  print "<body>\n$footnotes\n</body>" if $footnotes;
+print $pics;
  print "</FictionBook>";
  
  sub add_to_section {
@@ -182,9 +193,21 @@ sub pushsection {
         while (scalar(@sections) > $found) {
                 flushsection;
         }
-       push @sections,{level=>$level,title=>$title,data=>""};
+       push @sections,{level=>$level,title=>fix_ligatures($title),data=>""};
  }
  
+sub fix_ligatures {
+       local $_=shift;
+       s/~/\xA0/g;
+       s/\\-/\xAD/g;
+       s/---/—/g;
+       s/<</«/g;
+       s/>>/»/g;
+       s/\\%/%/g;
+       s/\\dots/\x{2026}/g;
+       s/\\verb(.)(.*)\1/<code>$2<\/code>/;
+       return $_;
+}      
  sub tag {
         my ($content,$name) = @_;
         $content =~s/^\s+//s;
@@ -194,9 +217,12 @@ sub tag {
  
  sub flushbuffer {
         local $_ = shift;
-       s/{\\(em|it|bf)(?:\s+|{})([^{}]+)}/<emphasis>$2<\/emphasis>/g;
-       s/\\(emph|textit|textbf){([^{}]+)}/<emphasis>$2<\/emphasis>/g;
-       s/\\footnote{(.*)}/push_footnote($1)/e;
+
+       s/\\footnote\{(.*)\}/push_footnote($1)/e;
+       $_.="}" if (/^\s*\{.*?[^}]$/) ;
+       s/\{\\(em|it|bf)(?:\s+| \{\})([^{}]+)}/<emphasis>$2<\/emphasis>/g;
+       s/\{\\(tt)(?:\s+|\{\})([^{}]+)}/<code>$2<\/code>/g;
+       s/\\(emph|textit|textbf)\{([^{}]+)\}/<emphasis>$2<\/emphasis>/g;
         s/[{}]//g;
         return $_;
  }
@@ -204,6 +230,17 @@ sub flushbuffer {
  
  sub push_footnote {
         my $id = "note_".(++$idseq);
-       $footnotes.="<section id=\"$id\">".tag(shift,'p')."</section>\n";
+       $footnotes.="<section id=\"$id\">".tag(flushbuffer(shift),'p')."</section>\n";
         return "<a l:href=\"#$id\" type=\"note\">$idseq</a>";
  }
+
+sub mkbinary {
+       my ($filename,$id) = @_;
+       my $f;
+       open $f,"<",$filename;
+       binmode $f;
+       local $/;
+       my $data = encode_base64(<$f>);
+       return "<binary id=\"$id\" content-type=\"image/png\">$data</binary>\n";
+       close $f;
+}