cvs2cl.pl

java写的html的解析器parser

          $suppress_line_start_match = 1;

          last;
        }
        else
        {
          $idx--;
        }
      }

      if ($idx == 0)
      {
        # We bottomed out because the line is longer than the
        # available space.  But that could be because the space is
        # small, or because the line is longer than even the maximum
        # possible space.  Handle both cases below.

        if ($length_remaining == ($max_line_length - (length ($user_indent))))
        {
          # The line is simply too long -- there is no hope of ever
          # breaking it nicely, so just insert it verbatim, with
          # appropriate padding.
          $this_line = "\n${left_pad_str}${this_line}";
        }
        else
        {
          # Can't break it here, but may be able to on the next round...
          unshift (@lines, $this_line);
          $length_remaining = $max_line_length - (length ($user_indent));
          $this_line = "\n${left_pad_str}";
        }
      }
    }
    else  # $this_len < $length_remaining, so tack on what we can.
    {
      # Leave a note for the next iteration.
      $length_remaining = $length_remaining - $this_len;

      if ($this_line =~ /\.$/)
      {
        $this_line .= "  ";
        $length_remaining -= 2;
      }
      else  # not a sentence end
      {
        $this_line .= " ";
        $length_remaining -= 1;
      }
    }

    # Unconditionally indicate that loop has run at least once.
    $first_time = 0;

    $wrapped_text .= "${user_indent}${this_line}";
  }

  # One last bit of padding.
  $wrapped_text .= "\n";

  return $wrapped_text;
}

sub xml_escape ()
{
  my $txt = shift;
  $txt =~ s/&/&amp;/g;
  $txt =~ s/</&lt;/g;
  $txt =~ s/>/&gt;/g;
  return $txt;
}

sub maybe_read_user_map_file ()
{
  my %expansions;

  if ($User_Map_File)
  {
    open (MAPFILE, "<$User_Map_File")
        or die ("Unable to open $User_Map_File ($!)");

    while (<MAPFILE>)
    {
      next if /^\s*#/;  # Skip comment lines.
      next if not /:/;  # Skip lines without colons.

      # It is now safe to split on ':'.
      my ($username, $expansion) = split ':';
      chomp $expansion;
      $expansion =~ s/^'(.*)'$/$1/;
      $expansion =~ s/^"(.*)"$/$1/;

      # If it looks like the expansion has a real name already, then
      # we toss the username we got from CVS log.  Otherwise, keep
      # it to use in combination with the email address.

      if ($expansion =~ /^\s*<{0,1}\S+@.*/) {
        # Also, add angle brackets if none present
        if (! ($expansion =~ /<\S+@\S+>/)) {
          $expansions{$username} = "$username <$expansion>";
        }
        else {
          $expansions{$username} = "$username $expansion";
        }
      }
      else {
        $expansions{$username} = $expansion;
      }
    } # fi ($User_Map_File)

    close (MAPFILE);
  }

  if (defined $User_Passwd_File)
  {
    if ( ! defined $Mail_Domain ) {
      if ( -e MAILNAME ) {
        chomp($Mail_Domain = slurp_file(MAILNAME));
      } else {
      MAILDOMAIN_CMD:
        for ([qw(hostname -d)], 'dnsdomainname', 'domainname') {
          my ($text, $exit, $sig, $core) = run_ext($_);
          if ( $exit == 0 && $sig == 0 && $core == 0 ) {
            chomp $text;
            if ( length $text ) {
              $Mail_Domain = $text;
              last MAILDOMAIN_CMD;
            }
          }
        }
      }
    }

    die "No mail domain found\n"
      unless defined $Mail_Domain;

    open (MAPFILE, "<$User_Passwd_File")
        or die ("Unable to open $User_Passwd_File ($!)");
    while (<MAPFILE>)
    {
      # all lines are valid
      my ($username, $pw, $uid, $gid, $gecos, $homedir, $shell) = split ':';
      my $expansion = '';
      ($expansion) = split (',', $gecos)
        if defined $gecos && length $gecos;

      $expansions{$username} = "$expansion <$username\@$Mail_Domain>";
    }
    close (MAPFILE);
  }

  return %expansions;
}

sub parse_options ()
{
  # Check this internally before setting the global variable.
  my $output_file;

  # If this gets set, we encountered unknown options and will exit at
  # the end of this subroutine.
  my $exit_with_admonishment = 0;

  while (my $arg = shift (@ARGV))
  {
    if ($arg =~ /^-h$|^-help$|^--help$|^--usage$|^-?$/) {
      $Print_Usage = 1;
    }
    elsif ($arg =~ /^--delta$/) {
      my $narg = shift(@ARGV) || die "$arg needs argument.\n";
      if ($narg =~ /^([A-Za-z][A-Za-z0-9_\-]*):([A-Za-z][A-Za-z0-9_\-]*)$/) {
	$Delta_From = $1;
	$Delta_To = $2;
	$Delta_Mode = 1;
      } else {
	die "--delta FROM_TAG:TO_TAG is what you meant to say.\n";
      }
    }
    elsif ($arg =~ /^--debug$/) {        # unadvertised option, heh
      $Debug = 1;
    }
    elsif ($arg =~ /^--version$/) {
      $Print_Version = 1;
    }
    elsif ($arg =~ /^-g$|^--global-opts$/) {
      my $narg = shift (@ARGV) || die "$arg needs argument.\n";
      # Don't assume CVS is called "cvs" on the user's system:
      $Log_Source_Command =~ s/(^\S*)/$1 $narg/;
    }
    elsif ($arg =~ /^-l$|^--log-opts$/) {
      my $narg = shift (@ARGV) || die "$arg needs argument.\n";
      $Log_Source_Command .= " $narg";
    }
    elsif ($arg =~ /^-f$|^--file$/) {
      my $narg = shift (@ARGV) || die "$arg needs argument.\n";
      $output_file = $narg;
    }
    elsif ($arg =~ /^--accum$/) {
      $Cumulative = 1;
    }
    elsif ($arg =~ /^--update$/) {
      $Update = 1;
    }
    elsif ($arg =~ /^--fsf$/) {
      $FSF_Style = 1;
    }
    elsif ($arg =~ /^--FSF$/) {
      $Show_Times = 0;
      $Common_Dir = 0;
    }
    elsif ($arg =~ /^--rcs/) {
      my $narg = shift (@ARGV) || die "$arg needs argument.\n";
      $RCS_Root = $narg;
      $RCS_Mode = 1;
    }
    elsif ($arg =~ /^-U$|^--usermap$/) {
      my $narg = shift (@ARGV) || die "$arg needs argument.\n";
      $User_Map_File = $narg;
    }
    elsif ($arg =~ /^--gecos$/) {
      $Gecos = 1;
    }
    elsif ($arg =~ /^--domain$/) {
      my $narg = shift (@ARGV) || die "$arg needs argument.\n";
      $Domain = $narg;
    }
    elsif ($arg =~ /^--passwd$/) {
      my $narg = shift (@ARGV) || die "$arg needs argument.\n";
      $User_Passwd_File = $narg;
    }
    elsif ($arg =~ /^--mailname$/) {
      my $narg = shift (@ARGV) || die "$arg needs argument.\n";
      $Mail_Domain = $narg;
    }
    elsif ($arg =~ /^-W$|^--window$/) {
      defined(my $narg = shift (@ARGV)) || die "$arg needs argument.\n";
      $Max_Checkin_Duration = $narg;
    }
    elsif ($arg =~ /^--chrono$/) {
      $Chronological_Order = 1;
    }
    elsif ($arg =~ /^-I$|^--ignore$/) {
      my $narg = shift (@ARGV) || die "$arg needs argument.\n";
      push (@Ignore_Files, $narg);
    }
    elsif ($arg =~ /^-C$|^--case-insensitive$/) {
      $Case_Insensitive = 1;
    }
    elsif ($arg =~ /^-R$|^--regexp$/) {
      my $narg = shift (@ARGV) || die "$arg needs argument.\n";
      $Regexp_Gate = $narg;
    }
    elsif ($arg =~ /^--stdout$/) {
      $Output_To_Stdout = 1;
    }
    elsif ($arg =~ /^--version$/) {
      $Print_Version = 1;
    }
    elsif ($arg =~ /^-d$|^--distributed$/) {
      $Distributed = 1;
    }
    elsif ($arg =~ /^-P$|^--prune$/) {
      $Prune_Empty_Msgs = 1;
    }
    elsif ($arg =~ /^-S$|^--separate-header$/) {
      $After_Header = "\n\n";
    }
    elsif ($arg =~ /^--no-wrap$/) {
      $No_Wrap = 1;
    }
    elsif ($arg =~ /^--summary$/) {
      $Summary = 1;
      $After_Header = "\n\n"; # Summary implies --separate-header
    }
    elsif ($arg =~ /^--gmt$|^--utc$/) {
      $UTC_Times = 1;
    }
    elsif ($arg =~ /^-w$|^--day-of-week$/) {
      $Show_Day_Of_Week = 1;
    }
    elsif ($arg =~ /^--no-times$/) {
      $Show_Times = 0;
    }
    elsif ($arg =~ /^-r$|^--revisions$/) {
      $Show_Revisions = 1;
    }
    elsif ($arg =~ /^--show-dead$/) {
      $Show_Dead = 1;
    }
    elsif ($arg =~ /^-t$|^--tags$/) {
      $Show_Tags = 1;
    }
    elsif ($arg =~ /^-T$|^--tagdates$/) {
      $Show_Tag_Dates = 1;
    }
    elsif ($arg =~ /^-b$|^--branches$/) {
      $Show_Branches = 1;
    }
    elsif ($arg =~ /^-F$|^--follow$/) {
      my $narg = shift (@ARGV) || die "$arg needs argument.\n";
      push (@Follow_Branches, $narg);
    }
    elsif ($arg =~ /^--stdin$/) {
      $Input_From_Stdin = 1;
    }
    elsif ($arg =~ /^--header$/) {
      my $narg = shift (@ARGV) || die "$arg needs argument.\n";
      $ChangeLog_Header = &slurp_file ($narg);
      if (! defined ($ChangeLog_Header)) {
        $ChangeLog_Header = "";
      }
    }
    elsif ($arg =~ /^--xml-encoding$/) {
      my $narg = shift (@ARGV) || die "$arg needs argument.\n";
      $XML_Encoding = $narg ;
    }
    elsif ($arg =~ /^--xml$/) {
      $XML_Output = 1;
    }
    elsif ($arg =~ /^--hide-filenames$/) {
      $Hide_Filenames = 1;
      $After_Header = "";
    }
    elsif ($arg =~ /^--no-common-dir$/) {
      $Common_Dir = 0;
    }
    elsif ($arg =~ /^--ignore-tag$/ ) {
      die "$arg needs argument.\n"
        unless @ARGV;
      $ignore_tags{shift @ARGV} = 1;
    }
    elsif ($arg =~ /^--show-tag$/ ) {
      die "$arg needs argument.\n"
        unless @ARGV;
      $show_tags{shift @ARGV} = 1;
    }
    elsif ( lc ($arg) eq '--test-code' ) {
      # Deliberately undocumented.  This is not a public interface,
      # and may change/disappear at any time.
      die "$arg needs argument.\n"
        unless @ARGV;
      $TestCode = shift @ARGV;
    }
    elsif ($arg =~ /^--no-ancestors$/) {
      $No_Ancestors = 1;
    }
    else {
      # Just add a filename as argument to the log command
      $Log_Source_Command .= " '$arg'";
    }
  }

  ## Check for contradictions...

  if ($Output_To_Stdout && $Distributed) {
    print STDERR "cannot pass both --stdout and --distributed\n";
    $exit_with_admonishment = 1;
  }

  if ($Output_To_Stdout && $output_file) {
    print STDERR "cannot pass both --stdout and --file\n";
    $exit_with_admonishment = 1;
  }

  if ($XML_Output && $Cumulative) {
    print STDERR "cannot pass both --xml and --accum\n";
    $exit_with_admonishment = 1;
  }

  # Or if any other error message has already been printed out, we
  # just leave now:
  if ($exit_with_admonishment) {
    &usage ();
    exit (1);
  }
  elsif ($Print_Usage) {
    &usage ();
    exit (0);
  }
  elsif ($Print_Version) {
    &version ();
    exit (0);
  }

  ## Else no problems, so proceed.

  if ($output_file) {
    $Log_File_Name = $output_file;
  }
}

sub slurp_file ()
{
  my $filename = shift || die ("no filename passed to slurp_file()");
  my $retstr;

  open (SLURPEE, "<${filename}") or die ("unable to open $filename ($!)");
  my $saved_sep = $/;
  undef $/;
  $retstr = <SLURPEE>;
  $/ = $saved_sep;
  close (SLURPEE);
  return $retstr;
}

sub debug ()
{
  if ($Debug) {
    my $msg = shift;
    print STDERR $msg;
  }
}

sub version ()
{
  print "cvs2cl.pl version ${VERSION}; distributed under the GNU GPL.\n";
}

sub usage ()
{
  &version ();
  print <<'END_OF_INFO';
Generate GNU-style ChangeLogs in CVS working copies.

Notes about the output format(s):

   The default output of cvs2cl.pl is designed to be