#!/usr/bin/perl

{  # Undef $/ locally so we can slurp the entire input file at once.
  local $/;
  open(FH,$ARGV[0]);
  $_ = <FH>;
}

print "I'm working on >$_<\n";

  # Our input is a bunch of SGML.  We want to pick out
  # - The US Patent number between <B110> and </B110> tags,
  # - The Japanese title between <B542> and </B542> tags, but be
  #   careful to pick out the Japanese title, not the English one,
  #   identified by a preceeding <B541>JP<\/B541>.
  # - And the Japanese abstract between <SDOAB> and </SDOAB>.
  #   Be aware that inside the Japanese abstract, are multiple paragraphs,
  #   bracketed by <P> and </P>, so assist our loop below by picking
  #   out the first <P> and last </P> tags.
  if (/<B110>([0-9]*)<\/B110>.*<B541>JP<\/B541><B542>(.*?)<\/B542>.*<SDOAB><P>(.*?)(?:<\/P>)?<\/SDOAB>/s)
 {
  $patn = sprintf("US%08s__",$1);
  $japanese_title = $2;
  $raw_japanese_abstract = $3;
# print "The raw japanese abstract is >$raw_japanese_abstract<\n";
  $raw_japanese_abstract =~ s/\r\n/ /g;

  # Parse out each paragraph of the Japanese abstract.
  # Keep track of each with a numeric key.
  $num=0;
  foreach $paragraph (split('</P><P>',$raw_japanese_abstract)) {
    if ($paragraph) {
      $num++;
      $paragraphs{$num} = $paragraph;
      print "Set \$paragraphs{$num} to =>$paragraph<\n";
    }
  }

  print "Aha, I passed my search with patn=>$patn\n";
  print "and japanese title =>$japanese_title<\n";
  print "and japanese abstract\n";

  # Access our paragraphs hash numerically (code to sort numerically
  # lifted from Perl book, page 790 -- the sort function).
  foreach $num (sort {$a<=>$b} keys %paragraphs) {
    print "Paragraph $num=>$paragraphs{$num}<\n";
  };
} else {
  print "Error.  Input file did not match my pattern.\n"
};