root/feedmelinks/bin/import-bookmarks.pl

#!/usr/local/bin/perl
use lib "/home/fml/public_html/lib";
use FeedMeLinks::Environment;
#use strict;

my ($debug, $link, $tag, @tags, @links, @folks, $url, $hideGarbage, $dd, $glc, $mandatory_tag, $path, $filename, $bin, $format, $cmd, $depth, $folksPath, $lastDepth, $lastFolder, $description);

$glc = 0;
$debug = 0;

$link = $tag = undef;
$hideGarbage = 1; 
$glc = 0; 
push @folks, "ROOT";
$mandatory_tag = "imported links";
push @tags, $mandatory_tag;

$dd = "___DELIMETER___";

# extract the user name from the file path (this practice oculd be considered a bit shady...)
$path = $filename = $ARGV[0];
my $parser     = $ARGV[1];
my $link_xslt  = $ARGV[2];
my $tag_xslt   = $ARGV[3];
$path =~ s:\/([^/]+)$::; #strip filename
$bin ="/home/fml/public_html/bin";

my $success_code = "";

my $converted = "/tmp/import-" . int(rand(1000));
# one of: delicious-hybrid, netscape, delicious-API
$format = detect_bookmarks_format( $bin, $filename );

print "format: |$format|\n" if( $debug );

if( $format eq "delicious-hybrid" ) {
        print "converting from hybrid format:\n" if( $debug );
        $cmd = "$bin/hybrid-to-xml.pl $filename > $converted";
        print "cmd: $cmd\n";
        `$cmd`;
        $filename = $converted;
        #print "new converted xml output is in $filename\n";
        my $out = `cat $filename`;
        print "$filename contains: \n</code><form><textarea rows='40' cols='150'>$out</textarea></form><code>\n\n" if( $debug );
        $success_code = "converted from hybrid to stock xml";

        #magic format swap-eroo now that we've converted
        $format = "delicious-API";
}

while( <> ) {
        my $line = $_;

        if( $format eq "delicious-API" ) {

                $cmd = "xsltproc $link_xslt $filename > $path/LINKS.RAW\n";
                print "cmd:<br>$cmd\n" if( $debug );
                my $out = `$cmd`;
                print "cmd output was: $out\n";
                
                $cmd = "xsltproc $tag_xslt $filename | sort | uniq | grep -v '^\$' > $path/TAGS.RAW\n";
                print "cmd: $cmd\n";
                $out = `$cmd`;
                print "cmd output: $out\n";

                # FIXME: 
                print "leaving the xml loop thing\n";
                print qq(<result delimeter="$dd" note="$success_code">SUCCESS</result>\n);
                exit 0;

        } 

        # match lines starting with whitespace that are not just whitespace or link descriptions
        # these lines mean we need to do something with, like links or tags
        if( $line =~ m/^(\s+)<D/ ) {
                $depth = length( $1 ) / 4;
                # this needs to change also, for IE

                $folksPath = get_folks_path();

                if( $line =~ m/<H3.+>(.*)<\/H3>/ ) { # match folder lines and grab the folder name
                #moz: #if( $line =~ m/<H3 ADD_DATE.*>(.*)<\/H3>/ ) { # match folder lines and grab the folder name

                        add_pending_link_or_tag();
                        $tag = $1;
                        if( $depth < $lastDepth ) {
                                d( "close " . ($lastDepth - $depth) . " folders" );
                                d( "found folder '$tag' (child of '" . $folks[ (scalar @folks) - 1 - ($lastDepth - $depth)] .  "')" );
                                pop @folks;
                        } else {
                                d( "found folder '$tag' (child of '$folksPath')" );
                        }
                        push @folks, ($tag ? $tag : "__ROOT__");

                } elsif( $line =~ m/<A HREF="([^"]+)"[^>]*>([^<]+)<\/A>/ ) { # match link lines and grab the link url and name and date
                #moz: } elsif( $line =~ m/<A HREF="([^"]+)" ADD_DATE="([^"]+)" .+>([^<]+)<\/A>/ ) { # match link lines and grab the link url and name and date
                        
                        add_pending_link_or_tag();
                        $url = $1;
                        #moz: $date = $2;
                        #moz: $link = $3;
                        $link = $2;
                        
                        if( $depth < $lastDepth ) {
                                d( "close " . ($lastDepth - $depth) . " folders" );
                                pop @folks;
                        }
                        d( "found link: '$link' ($url)" );
                }

        } elsif( $line =~ m/^<DD>(.*)$/ ) { # start of a description lins and grab the description

                # ignore firefox's (or is it sage's) internal housekeeping no-updated crap
                unless( m/no-updated/ ) {
                        $description = $1;
                        d( "  found description for " . ($link ? "link '$link'" : " tag '$tag'" ) . ": $description" );
                }

        } elsif( $line =~ m/<\/DL><p>/ ) { # no-op; stupid netscape delimiter crap

        } else { # additional description
                if( $description ) {
                        chomp( $line );
                        $description = $description . "\n" . $line;
                        d( "  found description for " . ($link ? "link '$link'" : " tag '$tag'" ) . ": $description" );

                } else {
                        if( ! $line =~ m/^\w*$/ ) { # don't really need to bitch about whitespace lines, just ignore the lies we cannot parse...
                                p( "XXXXXXXXXXXX GARBAGE FOUND: $line" ) unless $hideGarbage;
                        }
                }
        }

        $lastDepth = $depth;
        $lastFolder = $tag;
}

$folksPath = get_folks_path();
add_pending_link_or_tag();
write_raw_files();
unlink( $converted );

exit( 0 );

## END ##

sub detect_bookmarks_format {
        my( $bin, $filename) = @_;
        my $cmd = "$bin/detect-bookmarks-file-format.sh $filename";
        my $out = `$cmd`;
        chomp( $out );
        return $out;
}

sub write_raw_files() {
        @tags = sort @tags;
        open LINKS, ">$path/LINKS.RAW" || die( "can't open $path/LINKS.RAW" );
        foreach( @links ) { print LINKS "$_\n"; }
        close( LINKS );
        open TAGS, ">$path/TAGS.RAW" || die( "can't open $path/TAGS.RAW" );
        foreach( @tags ) { print TAGS "$_\n"; }
        close( TAGS );
        print qq(<result delimeter="$dd" note="$success_code">SUCCESS</result>\n);
}

sub get_folks_path() {
        my $folksPath = "";
        foreach( @folks ) {
                my $f = $_;
                $folksPath = "$folksPath / $f";
        }
        $folksPath =~ s/^ \/ //;
        return $folksPath;
}

sub strip_first {
        my $fp = $_[0];
        $fp =~ s/ \/ /$dd/g;
        $fp =~ s/^ROOT/$mandatory_tag/;
        return $fp;
}

sub add_pending_link_or_tag() {
        my $s;
        if( $link ) {
                my $tag_list = strip_first( $folksPath);
                $s = $url . $dd . $link . $dd . $tag_list;
                push @links, $s;
                $link = $description = undef;
        } elsif( $tag ) {
                $s = $tag;
                push @tags, $s;
                $tag = undef;
        }
}

sub p() {
        ++$glc;
        print $glc;
        print "> " . $_[0] . "\n";
}

sub d() {
        #print "DEBUG> " . $_[0] . "\n";
}