Generated: Mon Aug 16 14:14:26 2010 from linkhtml02.pl 2010/04/14 4.5 KB.
#!/perl -w # NAME: linkhtml02.pl # AIM: Given one input HTML file, parse HTML elements, and show each 'link' given in the file... # 2010/04/14 - moved main contents to htmllib.pl # 2010/04/13 - looking good... # 2010/04/12 - geoff mclane - http://geoffair.net/mperl/ use strict; use warnings; use File::Basename; # split path ($name,$dir,$ext) = fileparse($file [, qr/\.[^.]*/] ) use Cwd; unshift(@INC, 'C:\GTools\perl'); require 'logfile.pl' or die "Unable to load logfile.pl ...\n"; require 'htmllib.pl' or die "Unable to load htmlib.pl ...\n"; # log file stuff my ($LF); my $pgmname = $0; if ($pgmname =~ /\w{1}:\\.*/) { my @tmpsp = split(/\\/,$pgmname); $pgmname = $tmpsp[-1]; } my $perl_dir = 'C:\GTools\perl'; my $outfile = $perl_dir."\\temp.$pgmname.txt"; open_log($outfile); # user variables my $load_log = 0; my $def_infile = 'C:/HOMEPAGE/FG/Downloads/scenery-0.9.10.html'; #my $def_infile = 'C:\HOMEPAGE\FG\testpage2.htm'; #my $def_infile = 'C:\HOMEPAGE\FG\Docs\getstart\getstartch9.html'; #my $def_infile = 'C:\HOMEPAGE\FG\Projects\747-JW\index.html'; my $in_file = ''; my $ignore_close_element = 1; # dangerous - ignoring a close element my $dbg_flag = 0; my $options = $ignore_close_element; # 1 = no_ignore_close_element my $out_html = $perl_dir."\\temphtml.htm"; ### program variables my $verbosity = 0; my @warnings = (); my $cwd = cwd(); # debug my $dbg01 = 0; # show each item pushed to the stack my $dbg38 = 0; # prt( "[dbg38] Got [$hr2] = [$txt] [$fil]\n" ) if ($dbg38); my $dbg39 = 0; # prt( "[dbg39] Got [$hr2] = [$txt] - no inverted commas! [$fil]\n" ) if ($dbg39); sub VERB1() { return (($verbosity > 0) ? 1 : 0); } sub VERB5() { return (($verbosity >= 5) ? 1 : 0); } sub VERB9() { return (($verbosity >= 9) ? 1 : 0); } sub pgm_exit($$) { my ($val,$msg) = @_; if (length($msg)) { $msg .= "\n" if (!($msg =~ /\n$/)); prt($msg) } close_log($outfile,$load_log); exit($val); } sub prtw($) { my ($tx) = shift; $tx =~ s/\n$//; prt("$tx\n"); push(@warnings,$tx); } sub show_warnings() { if (@warnings) { prt( "\nGot ".scalar @warnings." WARNINGS...\n" ); foreach my $itm (@warnings) { prt("$itm\n"); } prt("\n"); } else { prt( "\nNo warnings issued.\n\n" ); } } ######################################### ### MAIN ### parse_args(@ARGV); prt( "$pgmname: in [$cwd]: Process $in_file...\n" ); my $hash_ref = get_html_file_hash($in_file,$options,$dbg_flag); list_files_in_hash_ref($hash_ref,$out_html); pgm_exit(0,"Normal exit(0)"); ######################################## sub give_help { prt("$pgmname: version 0.0.1 2010-04-06\n"); prt("Usage: $pgmname [options] in_file_name\n"); prt("Options:\n"); prt(" -h (or -?) = THis help, and exit 0\n"); prt(" -l = Load log file at end.\n"); prt(" -v[num] = Bump, or set verbosity to [num]\n"); prt("Parse input file, and report any problems...\n"); } sub parse_args { my (@av) = @_; my ($arg,$sarg,$ch); while (@av) { $arg = $av[0]; if ($arg =~ /^-/) { $sarg = substr($arg,1); $sarg = substr($sarg,1) while ($sarg =~ /^-/); $ch = substr($sarg,0,1); if ($ch =~ /h/i) { give_help(); pgm_exit(0,"Help exit"); } elsif ($ch =~ /l/i) { $load_log = 1; prt("Set to load log at end\n"); } elsif ($ch =~ /v/i) { $sarg = substr($sarg,1); if (length($sarg)) { if ($sarg =~ /^\d+$/) { $verbosity = $sarg; prt("Set verbosity to [$verbosity]\n"); } else { pgm_exit(1,"Unknown argument [$arg] - verbosity is -v[num]. Try -h for help\n"); } } else { $verbosity++; prt("Bumped verbosity to [$verbosity]\n"); } } else { pgm_exit(1,"Unknown argument [$arg] Try -h for help\n"); } } else { $in_file = $arg; prt("Set input file to [$in_file]\n"); } shift @av; } if (!length($in_file)) { $in_file = $def_infile; $load_log = 1; $verbosity = 9; prt("Set DEFAULT input file to [$in_file], and set load_log=1, and verbosity=$verbosity\n"); } } # eof - template.pl