listhrefs.pl to HTML.

index -|- end

Generated: Sun Mar 2 17:20:08 2014 from listhrefs.pl 2013/12/22 8.7 KB. text copy

#!/usr/bin/perl -w
# NAME: listhrefs.pl
# AIM: Given a html file, output a list of hrefs
# 22/12/2013 geoff mclane http://geoffair.net/mperl
use strict;
use warnings;
use File::Basename;  # split path ($name,$dir,$ext) = fileparse($file [, qr/\.[^.]*/] )
use Cwd;
my $os = $^O;
my $perl_dir = '/home/geoff/bin';
my $PATH_SEP = '/';
my $temp_dir = '/tmp';
if ($os =~ /win/i) {
    $perl_dir = 'C:\GTools\perl';
    $temp_dir = $perl_dir;
    $PATH_SEP = "\\";
}
unshift(@INC, $perl_dir);
require 'lib_utils.pl' or die "Unable to load 'lib_utils.pl' Check paths in \@INC...\n";
require 'lib_html.pl' or die "Unable to load 'lib_html.pl' Check paths in \@INC...\n";
# log file stuff
our ($LF);
my $pgmname = $0;
if ($pgmname =~ /(\\|\/)/) {
    my @tmpsp = split(/(\\|\/)/,$pgmname);
    $pgmname = $tmpsp[-1];
}
my $outfile = $temp_dir.$PATH_SEP."temp.$pgmname.txt";
open_log($outfile);

# user variables
my $VERS = "0.0.1 2013-12-22";
my $load_log = 0;
my $in_file = '';
my $verbosity = 0;
my $out_file = '';

# ### DEBUG ###
my $debug_on = 1;
my $def_file = 'F:\data\DEM\map.html';

### program variables
my @warnings = ();
my $cwd = cwd();

sub VERB1() { return $verbosity >= 1; }
sub VERB2() { return $verbosity >= 2; }
sub VERB5() { return $verbosity >= 5; }
sub VERB9() { return $verbosity >= 9; }

sub show_warnings($) {
    my ($val) = @_;
    if (@warnings) {
        prt( "\nGot ".scalar @warnings." WARNINGS...\n" );
        foreach my $itm (@warnings) {
           prt("$itm\n");
        }
        prt("\n");
    } else {
        prt( "\nNo warnings issued.\n\n" ) if (VERB9());
    }
}

sub pgm_exit($$) {
    my ($val,$msg) = @_;
    if (length($msg)) {
        $msg .= "\n" if (!($msg =~ /\n$/));
        prt($msg);
    }
    show_warnings($val);
    close_log($outfile,$load_log);
    exit($val);
}


sub prtw($) {
   my ($tx) = shift;
   $tx =~ s/\n$//;
   prt("$tx\n");
   push(@warnings,$tx);
}

sub process_in_file($) {
    my ($inf) = @_;
    if (! open INF, "<$inf") {
        pgm_exit(1,"ERROR: Unable to open file [$inf]\n"); 
    }
    my @lines = <INF>;
    close INF;
    my $lncnt = scalar @lines;
    prt("Processing $lncnt lines, from [$inf]...\n");
    my ($line,$lnn,$len,$ch,$tag,$intag,$txt,$inquot,$qc,$i,$show,$rha,$hadsp,$attrs,$key,$val,$msg);
    $intag = 0;
    $txt = '';
    $inquot = 0;
    $tag = '';
    $lnn = 0;
    $show = 0;
    $hadsp = 0;
    $attrs = '';
    my @hrefs = ();
    foreach $line (@lines) {
        $lnn++;
        chomp $line;
        $line = trim_all($line);
        $len = length($line);
        for ($i = 0; $i < $len; $i++) {
            $ch = substr($line,$i,1);
            $show = 0;
            if ($intag) {
                if ($ch eq '>') {
                    $show = 1;
                } elsif ($ch eq '<') {
                    prt("Got OPEN before CLOSE!\n");
                    $show = 1;
                } else {
                    if ($hadsp) {
                        $attrs .= $ch;
                    } elsif ($ch =~ /\s/) {
                        $hadsp = 1;
                    } else {
                        $tag .= $ch;
                    }
                }
            } elsif ($ch eq '<') {
                $intag = 1;
                $hadsp = 0;
            } else {
                $txt .= $ch;
            }
            # got a CLOSE
            if ($show) {
                if ($tag =~ /^\//) {
                    # just a close tag
                } elsif ($tag =~ /^AREA$/i) {
                    $msg = '';
                    $msg .= "$txt " if (length($txt));
                    if (length($tag)) {
                        $msg .= "<$tag";
                        if (length($attrs)) {
                            $msg .= " $attrs";
                            $rha = get_attr_refhash($attrs);
                            foreach $key (keys %{$rha}) {
                                if ($key =~ /href/i) {
                                    $val = ${$rha}{$key};
                                    $val = strip_quotes($val);
                                    push(@hrefs,$val);
                                }
                            }
                        }
                        $msg .= ">";
                    }
                    if (length($txt) || length($tag)) {
                        $msg .= "\n";
                    }
                    if (length($msg)) {
                        prt($msg) if (VERB5());
                    }
                }
                $txt = '';
                $tag = '';
                $intag = 0;
                $attrs = '';
                $hadsp = 0;
            }
        }
        if ($intag) {
            if ($hadsp) {
                $attrs .= ' ' if (length($attrs));
            } else {
                $tag .= ' ' if (length($tag));
            }
        } else {
            $txt .= ' ' if (length($txt));
        }
    }
    $len = scalar @hrefs;
    if ($len) {
        @hrefs = sort @hrefs;
        prt(join("\n",@hrefs)."\n");
        prt("Listed $len AREA hrefs in $inf.\n");
    } else {
        prt("No AREA referecnes found in $inf.\n");
    }
    $load_log = 1;
}

sub process_in_file_FAILED($) {
    my ($inf) = @_;
    if (! open INF, "<$inf") {
        pgm_exit(1,"ERROR: Unable to open file [$inf]\n"); 
    }
    my @lines = <INF>;
    close INF;
    my $lncnt = scalar @lines;
    prt("Processing $lncnt lines, from [$inf]...\n");
    my ($line,$ra,$rarea,$rahrefs,$cnt,$i,$item,$txt,$i2,$rha,$key);
    $line = join("",@lines);
    $ra = get_html_refarray($line);
    my $opts = -1;
    $rarea = get_whole_tag_array($ra,'area',$opts);
    $cnt = scalar @{$rarea};
    for ($i = 0; $i < $cnt; $i++) {
        $i2 = $i + 1;
        $rha = ${$rarea}[$i][2];    # get attribute ref hash
        foreach $key (keys %{$rha}) {
            if ($key =~ /^href$/i) {
                $item = ${$rha}{$key};
                prt("$item\n");
            }
        }
    }
    
}



#########################################
### MAIN ###
parse_args(@ARGV);
process_in_file($in_file);
pgm_exit(0,"");
########################################

sub need_arg {
    my ($arg,@av) = @_;
    pgm_exit(1,"ERROR: [$arg] must have a following argument!\n") if (!@av);
}

sub parse_args {
    my (@av) = @_;
    my ($arg,$sarg);
    while (@av) {
        $arg = $av[0];
        if ($arg =~ /^-/) {
            $sarg = substr($arg,1);
            $sarg = substr($sarg,1) while ($sarg =~ /^-/);
            if (($sarg =~ /^h/i)||($sarg eq '?')) {
                give_help();
                pgm_exit(0,"Help exit(0)");
            } elsif ($sarg =~ /^v/) {
                if ($sarg =~ /^v.*(\d+)$/) {
                    $verbosity = $1;
                } else {
                    while ($sarg =~ /^v/) {
                        $verbosity++;
                        $sarg = substr($sarg,1);
                    }
                }
                prt("Verbosity = $verbosity\n") if (VERB1());
            } elsif ($sarg =~ /^l/) {
                if ($sarg =~ /^ll/) {
                    $load_log = 2;
                } else {
                    $load_log = 1;
                }
                prt("Set to load log at end. ($load_log)\n") if (VERB1());
            } elsif ($sarg =~ /^o/) {
                need_arg(@av);
                shift @av;
                $sarg = $av[0];
                $out_file = $sarg;
                prt("Set out file to [$out_file].\n") if (VERB1());
            } else {
                pgm_exit(1,"ERROR: Invalid argument [$arg]! Try -?\n");
            }
        } else {
            $in_file = $arg;
            prt("Set input to [$in_file]\n") if (VERB1());
        }
        shift @av;
    }

    if ($debug_on) {
        prtw("WARNING: DEBUG is ON!\n");
        if ((length($in_file) ==  0) && $debug_on) {
            $in_file = $def_file;
            prt("Set DEFAULT input to [$in_file]\n");
        }
    }
    if (length($in_file) ==  0) {
        pgm_exit(1,"ERROR: No input files found in command!\n");
    }
    if (! -f $in_file) {
        pgm_exit(1,"ERROR: Unable to find in file [$in_file]! Check name, location...\n");
    }
}

sub give_help {
    prt("$pgmname: version $VERS\n");
    prt("Usage: $pgmname [options] in-file\n");
    prt("Options:\n");
    prt(" --help  (-h or -?) = This help, and exit 0.\n");
    prt(" --verb[n]     (-v) = Bump [or set] verbosity. def=$verbosity\n");
    prt(" --load        (-l) = Load LOG at end. ($outfile)\n");
    prt(" --out <file>  (-o) = Write output to this file.\n");
}

# eof - template.pl

index -|- top

checked by tidy  Valid HTML 4.01 Transitional