#!/usr/bin/perl -w
# NAME: tidyattrs.pl
# AIM: VERY SPECIFIC compare attribute enumeration, with attr table, and 
# report order problems...
# tidyenum.h contains enumeration - /** Known HTML attributes
# */
# typedef enum
# {
#  TidyAttr_UNKNOWN,           /**< UNKNOWN= */
#  TidyAttr_ABBR,              /**< ABBR= */
# AND 
# attrs.c contains static const Attribute attribute_defs [] =
# {
#  { TidyAttr_UNKNOWN,                 "unknown!",                NULL         }, 
#  { TidyAttr_ABBR,                    "abbr",                    CH_PCDATA    }, 
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, $perl_dir);
require 'lib_utils.pl' or die "Unable to load 'lib_utils.pl' Check paths in \@INC...\n";
# log file stuff
our ($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 = "0.0.5 2015-01-09";
my $load_log = 0;
my $in_dir = '';
my $verbosity = 0;
my $out_file = '';

# ### DEBUG ###
my $debug_on = 1;
my $def_dir = 'F:\Projects\tidy-html5';

### 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 ($line,$inc,$lnn);
    $lnn = 0;
    foreach $line (@lines) {
        chomp $line;
        $lnn++;
        if ($line =~ /\s*#\s*include\s+(.+)$/) {
            $inc = $1;
            prt("$lnn: $inc\n");
        }
    }
}

sub process_in_dir($) {
    my $dir = shift;
    my $in_enum = $dir.$PATH_SEP."include".$PATH_SEP."tidyenum.h";
    my $in_attr = $dir.$PATH_SEP."src".$PATH_SEP."attrs.c";
    my ($inf);
    if (! -f $in_enum) {
        pgm_exit(1,"Error: Unable to locate '$in_enum'!\n");
    }
    if (! -f $in_attr) {
        pgm_exit(1,"Error: Unable to locate '$in_attr'!\n");
    }
    $inf = $in_enum;
    if (! open INF, "<$inf") {
        pgm_exit(1,"ERROR: Unable to open file [$inf]\n"); 
    }
    my @lines1 = <INF>;
    close INF;
    my $lncnt1 = scalar @lines1;
    prt("Processing $lncnt1 lines, from [$inf]...\n");
    $inf = $in_attr;
    if (! open INF, "<$inf") {
        pgm_exit(1,"ERROR: Unable to open file [$inf]\n"); 
    }
    my @lines2 = <INF>;
    close INF;
    my $lncnt2 = scalar @lines2;
    prt("Processing $lncnt2 lines, from [$inf]...\n");
    my ($line,$lnn,$tline,$len,$incomm);
    my ($i,$ch,$pc);
    $lnn = 0;
    $incomm = 0;
    $ch = '';
    my $attr = '';
    my @attrs1 = ();
    my @attrs2 = ();
    # process the ENUMERATION
    foreach $line (@lines1) {
        chomp $line;
        $lnn++;
        $tline = trim_all($line);
        $len = length($tline);
        next if ($len == 0);
        for ($i = 0; $i < $len; $i++) {
            $pc = $ch;
            $ch = substr($tline,$i,1);
            if ($incomm) {
                if (($pc eq '*') && ($ch eq '/')) {
                    prt("$lnn: End comm '$tline'\n") if (VERB9());
                    $incomm = 0;
                }
            } else {
                if (($pc eq '/') && ($ch eq '*')) {
                    $incomm = 1;
                    prt("$lnn: Bgn comm '$tline'\n") if (VERB9());
                    next;
                }
            }
            if ($ch =~ /\w/) {
                $attr .= $ch;
            } else {
                if ($attr =~ /TidyAttr_(\w+)$/) {
                    prt("$lnn: $attr\n") if (VERB5());
                    push(@attrs1,$attr);
                } elsif ($attr eq 'N_TIDY_ATTRIBS') {
                    prt("$lnn: $attr - LAST\n");
                    push(@attrs1,$attr);
                }
                $attr = '';
            }
        }
        $attr = '';
    }
    # process the src attrs
    $lnn = 0;
    $ch = '';
    my $in_table = 0;
    my $in_zero = 0;
    foreach $line (@lines2) {
        chomp $line;
        $lnn++;
        $tline = trim_all($line);
        $len = length($tline);
        next if ($len == 0);
        if ($in_zero) {
            if ($line =~ /\#\s*else/) {
                $in_zero = 0;
            }
            next;
        } else {
           if ($line =~ /\#\s*if\s+0/) {
               $in_zero = 1;
               next;
           }
        }
        for ($i = 0; $i < $len; $i++) {
            $pc = $ch;
            $ch = substr($tline,$i,1);
            if ($incomm) {
                if (($pc eq '*') && ($ch eq '/')) {
                    prt("$lnn: End comm '$tline'\n") if (VERB9());
                    $incomm = 0;
                }
            } else {
                if (($pc eq '/') && ($ch eq '*')) {
                    $incomm = 1;
                    prt("$lnn: Bgn comm '$tline'\n") if (VERB9());
                    next;
                }
            }
            if ($in_table) {
                if ($ch =~ /\w/) {
                    $attr .= $ch;
                } else {
                    if ($attr =~ /TidyAttr_(\w+)$/) {
                        prt("$lnn: $attr\n") if (VERB5());
                        push(@attrs2,$attr);
                    } elsif ($attr eq 'N_TIDY_ATTRIBS') {
                        prt("$lnn: $attr - LAST\n");
                        push(@attrs2,$attr);
                    }
                    $attr = '';
                }
            }
        }
        if ($in_table) {
            if ($tline =~ /N_TIDY_ATTRIBS/) {
                prt("$lnn: End table '$tline'\n");
                $in_table = 0;
            }

        } else {
            if ($tline =~ /static\s+const\s+Attribute\s+attribute_defs/) {
                prt("$lnn: Begin table '$tline'\n");
                $in_table = 1;
            }
        }
        $attr = '';
    }
    $lncnt1 = scalar @attrs1;
    $lncnt2 = scalar @attrs2;
    prt("Found $lncnt1 and $lncnt2 attributes...\n");
    $lnn = 0;
    foreach $attr (@attrs1) {
        if ($lnn < $lncnt2) {
            $line = $attrs2[$lnn];
            if ($attr ne $line) {
                pgm_exit(1,"Off $lnn NE $attr $line\n");
            }
        }
        $lnn++;
    }

}

#########################################
### MAIN ###
parse_args(@ARGV);
process_in_dir($in_dir);
###process_in_file($in_dir);
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);
    my $verb = VERB2();
    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);
                    }
                }
                $verb = VERB2();
                prt("Verbosity = $verbosity\n") if ($verb);
            } elsif ($sarg =~ /^l/) {
                if ($sarg =~ /^ll/) {
                    $load_log = 2;
                } else {
                    $load_log = 1;
                }
                prt("Set to load log at end. ($load_log)\n") if ($verb);
            } elsif ($sarg =~ /^o/) {
                need_arg(@av);
                shift @av;
                $sarg = $av[0];
                $out_file = $sarg;
                prt("Set out file to [$out_file].\n") if ($verb);
            } else {
                pgm_exit(1,"ERROR: Invalid argument [$arg]! Try -?\n");
            }
        } else {
            $in_dir = $arg;
            prt("Set input to [$in_dir]\n") if ($verb);
        }
        shift @av;
    }

    if ($debug_on) {
        prtw("WARNING: DEBUG is ON!\n");
        if (length($in_dir) ==  0) {
            $in_dir = $def_dir;
            prt("Set DEFAULT input to [$in_dir]\n");
        }
    }
    if (length($in_dir) ==  0) {
        pgm_exit(1,"ERROR: No input files found in command!\n");
    }
    if (! -d $in_dir) {
        pgm_exit(1,"ERROR: Unable to find in file [$in_dir]! 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