view release on metacpan or  search on metacpan

bin/html2plain  view on Meta::CPAN

GetOptions('help|?'=>\$PrintHelp, 
	   'man'=>\$PrintManual,
	   'warnings!'=>\$Warnings,
	   'html-ext=s'=>\$HTMLSuffix,
	   'assert-html!'=>\$AssertHTML,
	   'symbolic-char-entities-to-chars!'=>\$ConvCharEnts,
	   'numerical-char-entities-to-chars!'=>\$ConvNumEnts,
	   'source-encoding=s'=>\$SourceEncoding,
	   'clean-whitespace!'=>\$CleanWS,
	   'assert-assumptions!'=>\$AssertAss,
	   'out-dir=s'=>\$ODir,
	   'out-ext=s'=>\$OutSuffix,
	   'N-per-out-dir=s'=>\$NPerOurDir) or 
    pod2usage(2);
pod2usage(1) if $PrintHelp;
pod2usage(-exitstatus => 0, -verbose => 2) if $PrintManual;
pod2usage(1) if (@ARGV!=1);

my $SDir=shift @ARGV;

$|=1;

my $C=Alvis::HTML->new(alvisKeep=>1,
		       alvisRemove=>1,
		       obsolete=>1,
		       proprietary=>1,
		       xhtml=>1,
		       wml=>1,
		       keepAll=>0,
		       assertHTML=>$AssertHTML,
		       convertCharEnts=>$ConvCharEnts,
		       convertNumEnts=>$ConvNumEnts,
		       sourceEncoding=>$SourceEncoding,
		       cleanWhitespace=>$CleanWS,
		       assertSourceAssumptions=>$AssertAss
		       );

my %Seen;
my $outputN=0;
if (!&_convert_collection($SDir,{htmlSuffix=>$HTMLSuffix}))
{
    die("Conversion failed. " . $C->errmsg());
}

sub _parse_entries
{
    my $entries=shift;
    my $options=shift;
    my $html_entries=shift;
    
    for my $e (@$entries)
    {
	if ($Seen{$e})
	{
	    next;
	}
	
	$Seen{$e}=1;
	if (-d $e)
	{
	    my @entries=glob("$e/*");;
	    &_parse_entries(\@entries,$options,$html_entries);
	    next;
	}

	my ($basename,$suffix);
	if ($e=~/^(.*)\.([^\.]+)$/)
	{
	    $basename=$1;
	    $suffix=$2;
	}
	else
	{
	    warn "Skipping non-suffixed non-directory entry \"$e\".\n" if 
		$Warnings;
	    next;
	}
	
	if ($suffix eq $options->{htmlSuffix})
	{
	    $html_entries->{$basename}{htmlF}=$e;
	}
    }
}

sub _convert_collection
{
    my $root_dir=shift;
    my $options=shift;

    my @entries=glob("$root_dir/*");
    my %html_entries=();
    %Seen=();
    print "Parsing the source directory entries...\r";
    &_parse_entries(\@entries,$options,\%html_entries);	
    print "                                       \r";

    for my $base_name (keys %html_entries)
    {
	my ($html_txt,$plain_txt,$header);

	if (exists($html_entries{$base_name}{htmlF}))
	{
	    my $html_txt;
	    if (!defined(open(F,"<$html_entries{$base_name}{htmlF}")))
	    {
		warn "Unable to open \"$html_entries{$base_name}{htmlF}\".\n";
		next;
	    }
	    while (my $l=<F>)
	    {
		$html_txt.=$l;
	    }
	    close(F);

	    ($plain_txt,$header)=$C->clean($html_txt);
	    if (!defined($plain_txt))
	    {
		warn "Getting the plain text for basename \"$base_name\" failed. " .
		    $C->errmsg() if 
		    $Warnings;
		$C->clearerr();
		next;
	    }
	}
	else
	{
	     warn "No HTML file for basename \"$base_name\".\n" if 
		$Warnings;
	     next;
	}

	if (!&_output_plain($plain_txt))
	{
	    warn "Outputting the Alvis records for base name \"$base_name\" failed. " . $C->errmsg() if 
		$Warnings;
	    $C->clearerr();
	    next;
	}
    }

    return 1;
}

sub _output_plain
{
    my $plain_txt=shift;

    my $out_f;
    my $dir=$ODir . '/' . 
	int($outputN / $NPerOurDir);