Generated: Sun Apr 15 11:46:53 2012 from xmlreline.pl 2011/12/20 9.9 KB.
#!/usr/bin/perl -w # NAME: xmlreline.pl # AIM: Simple read and re-line an XML file input, write to output, if one given # 20/12/2011 - Review only - no major changes use strict; use warnings; use File::Basename; # split path ($name,$dir,$ext) = fileparse($file [, qr/\.[^.]*/] ) use Cwd; my $perl_dir = 'C:\GTools\perl'; unshift(@INC, $perl_dir); require 'lib_utils.pl' or die "Unable to load 'lib_utils.pl' Check paths in \@INC...\n"; # log file stuff our ($LF); my $pgmname = $0; if ($pgmname =~ /(\\|\/)/) { my @tmpsp = split(/(\\|\/)/,$pgmname); $pgmname = $tmpsp[-1]; } my $outfile = $perl_dir."\\temp.$pgmname.txt"; open_log($outfile); # user variables my $VERS = "0.0.2 2011-12-20"; my $keep_spaces = 0; # preserve spaces in text, or NOT my $load_log = 0; my $in_file = ''; my $out_xml = ''; my $verbosity = 0; my $debug_on = 0; # my $def_file = 'C:\FG\29\Atlas-0.4.9\projects\msvc\temp2\fgxcaps.xml'; my $def_file = 'C:\DTEMP\osm-1153716.gpx'; # ;C:\DTEMP>np osm-1153716.gpx my $def_out = $perl_dir."\\tempout.xml"; ### program variables my @warnings = (); my $cwd = cwd(); my $os = $^O; 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" ); } } 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 ($line,$inc,$lnn); my ($len,$ch,$pc,$nc,$i,$i2); my ($tag,$txt,$intag,$inquot,$qc,$ntag,$ltag,@arr,$indent,$scnt); my $nline = 0; my @tagstack = (); $lnn = 0; $ch = ''; $intag = 0; $inquot = 0; $tag = ''; $txt = ''; $ntag = ''; $scnt = 0; $indent = ''; my $xml_msg = ''; my $on_new_line = 1; foreach $line (@lines) { chomp $line; $lnn++; next if ($line =~ /^\s*$/); $len = length($line); for ($i = 0; $i < $len; $i++) { $pc = $ch; $ch = substr($line,$i,1); $i2 = $i + 1; $nc = ($i2 < $len) ? substr($line,$i2,1) : ""; if ($intag) { $tag .= $ch; if ($inquot) { $inquot = 0 if ($ch eq $qc); } else { if ($ch eq '>') { $intag = 0; $txt = trim_tailing($txt) if (!$keep_spaces); if (length($txt)) { if ($on_new_line) { $indent = ' ' x $scnt; prt($indent.$txt) if (VERB1()); $xml_msg .= $indent.$txt; } else { prt($txt) if (VERB1()); $xml_msg .= $txt; } $on_new_line = 0; # now NOT on a new line } if ($tag =~ /^<\//) { $nline = 1; $ntag =~ s/^\///; # closing a tag if (@tagstack) { $ltag = pop @tagstack; if ($ltag ne $ntag) { prtw("WARNING: $lnn: Closing tag [$ntag], not last [$ntag]! [$line]\n"); } $scnt = scalar @tagstack; } else { prtw("WARNING: $lnn: Closing tag [$ntag], but no stack! [$line]\n"); } } elsif ($tag =~ /\/>$/) { $nline = 1; } elsif ($tag =~ /^<\?/) { $nline = 1; $tag = trim_all($tag); } elsif ($tag =~ /^<!/) { $nline = 1; $tag = trim_all($tag); } else { $nline = 0; if ($ntag =~ /\s/) { # contains a space, thus maybe has attributes @arr = split(/\s/,$ntag); $ntag = $arr[0]; } push(@tagstack,$ntag); #$scnt = scalar @tagstack; } $indent = ''; if (length($txt) == 0) { $indent = ' ' x $scnt; } prt($indent."$tag") if (VERB1()); $xml_msg .= $indent."$tag"; $on_new_line = 0; if ($nline) { prt("\n") if (VERB1()); $xml_msg .= "\n"; $on_new_line = 1; } $scnt = scalar @tagstack; $tag = ''; $txt = ''; } else { # continuing an IN TAG $ntag .= $ch; if (($ch eq '"')||($ch eq "'")) { $qc = $ch; $inquot = 1; } elsif ($ch eq '[') { $qc = ']'; $inquot = 1; } } } } else { # awaiting start of tag if ($ch eq '<') { $tag = $ch; $ntag = ''; $intag = 1; if ((length($txt) == 0)&&($nline == 0)) { prt("\n") if (VERB1()); $xml_msg .= "\n"; $on_new_line = 1; } # prt("$lnn: Begin tag\n"); } else { if ($ch =~ /\s/) { if ($keep_spaces) { $txt .= $ch if (length($txt)); } elsif (length($txt)) { $txt .= $ch if ($txt =~ /\S$/); } } else { $txt .= $ch; } } } } # for length of each line } if ((length($out_xml))&&(length($xml_msg))) { write2file($xml_msg,$out_xml); prt("Written XML to [$out_xml]\n"); } elsif (length($xml_msg)) { prt("No output file given, so xml not written...\n"); } } ######################################### ### MAIN ### parse_args(@ARGV); ### prt( "$pgmname: in [$cwd]: Hello, World...\n" ); process_in_file($in_file); pgm_exit(0,""); ######################################## 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. ($outfile)\n"); prt(" --out <file> (-o) = Write output to this file.\n"); } sub need_arg { my ($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/) { $load_log = 1; } elsif ($sarg =~ /^o/) { need_arg(@av); shift @av; $sarg = $av[0]; $out_xml = $sarg; prt("Set output to [$out_xml]\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 ((length($in_file) == 0) && $debug_on) { $in_file = $def_file; $out_xml = $def_out; } if (length($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"); } } # eof - xmlreline.pl