#!/pro/bin/perl

# csv-check: Check validity of CSV file and report
#	   (m)'17 [24 Nov 2017] Copyright H.M.Brand 2007-2019

# This code requires the defined-or feature and PerlIO

use 5.12.0;
use warnings;

use Data::Peek;
use Encode qw( decode );

our $VERSION = "2.01";	# 2017-11-24
my $cmd = $0; $cmd =~ s{.*/}{};

sub usage {
    my $err = shift and select STDERR;
    print <<"EOU";
usage: $cmd [-s <sep>] [-q <quot>] [-e <esc>] [-u] [--pp] [file.csv]
       -s <sep>   use <sep>   as seperator char. Auto-detect, default = ','
                  The string "tab" is allowed.
       -e <esc>   use <sep>   as seperator char. Auto-detect, default = ','
                  The string "undef" is allowed.
       -q <quot>  use <quot>  as quotation char. Default = '"'
                  The string "undef" will disable quotation.
       -u         check if all fields are valid unicode
       -E <enc>   open file with encoding
       -h         check with header (implies BOM)
       -b         check with BOM (no header)
       -f         do not check formula's

       --pp       use Text::CSV_PP instead (cross-check)
EOU
    exit $err;
    } # usage

use Getopt::Long qw(:config bundling);
my $sep;	# Set after reading first line in a flurry attempt to auto-detect
my $quo = '"';
my $esc = '"';
my $opt_u = 0;
my $opt_p = 0;
my $opt_h = 0;
my $opt_b = 0;
my $opt_f = 0;
my $enc;
GetOptions (
    "help|?"		=> sub { usage (0); },
    "V|version"		=> sub { print "$cmd [$VERSION]\n"; exit 0; },

    "c|s|sep=s"		=> \$sep,
    "q|quo|quote=s"	=> \$quo,
    "e|esc|escape=s"	=> \$esc,
    "u|utf|utf8|utf-8!"	=> \$opt_u,
    "E|enc|encoding=s"	=> \$enc,
    "h|hdr|header!"	=> \$opt_h,
    "b|bom!"		=> \$opt_b,
    "f|skip-formula!"	=> \$opt_f,

    "pp!"		=> \$opt_p,
    ) or usage (1);

my  $csvmod = "Text::CSV_XS";
if ($opt_p) {
    require Text::CSV_PP;
    $csvmod = "Text::CSV_PP";
    }
else {
    require Text::CSV_XS;
    }
$csvmod->import ();

my $fn   = $ARGV[0] // "-";
my $data = do { local $/; <> } or die "No data to analyze\n";
my @warn;

my ($bin, $rows, $eol, %cols) = (0, 0, undef);
unless ($sep) { # No sep char passed, try to auto-detect;
    my ($first_line) = ($data =~ m/\A(.*?)(?:\r\n|\n|\r)/);
    $first_line ||= $data; # if no EOL at all, use whole set
    $sep = $first_line =~ m/["\d],["\d,]/ ? ","  :
	   $first_line =~ m/["\d];["\d;]/ ? ";"  :
	   $first_line =~ m/["\d]\t["\d]/ ? "\t" :
	   # If neither, then for unquoted strings
	   $first_line =~ m/\w,[\w,]/     ? ","  :
	   $first_line =~ m/\w;[\w;]/     ? ";"  :
	   $first_line =~ m/\w\t[\w]/     ? "\t" : ",";
    $data =~ m/([\r\n]+)\Z/ and $eol = DDisplay "$1";
    }

my $csv = $csvmod->new ({
    sep_char       => $sep eq "tab"   ? "\t"  : $sep,
    quote_char     => $quo eq "undef" ? undef : $quo,
    escape_char    => $esc eq "undef" ? undef : $esc,
    binary         => 1,
    keep_meta_info => 1,
    auto_diag      => 1,
    formula        => $opt_f ? "none" : "diag",
    });

sub done {
    my $file = $ARGV // "STDIN";
    (my $pv = "$]0000000") =~ s{^([0-9]+)\.([0-9]{3})([0-9]{3})[0-9]*}
			       {sprintf "%d.%d.%d",$1,$2,$3}e;
    my $uv = eval {
	no warnings;
	(my $cv = $]) =~ s/0+$//;
	eval { require Unicode::UCD;     Unicode::UCD::UnicodeVersion () } ||
	eval { require Module::CoreList; $Module::CoreList::version{$cv}{Unicode} };
	} || "unknown";
    print "Checked $file with $cmd $VERSION\nusing $csvmod @{[$csvmod->VERSION]} with perl $pv and Unicode $uv\n";
    my @diag = $csv->error_diag;
    if ($diag[0] == 2012 && $csv->eof) {
	my @coll = sort { $a <=> $b } keys %cols;
	local $" = ", ";
	my $cols = @coll == 1 ? $coll[0] : "(@coll)";
	$eol //= $csv->eol || "--unknown--";
	print "OK: rows: $rows, columns: $cols\n";
	print "    sep = <$sep>, quo = <$quo>, bin = <$bin>, eol = <$eol>\n";
	print "    encoding = $csv->{ENCODING}\n" if $csv->{ENCODING};
	if (@coll > 1) {
	    print "multiple column lengths:\n";
	    printf " %6d line%s with %4d field%s\n",
		$cols{$_}, $cols{$_} == 1 ? " " : "s",
		$_,        $_        == 1 ? ""  : "s"
		    for @coll;
	    }
	$diag[0] = 0;
	}
    elsif ($diag[2]) {
	print "$ARGV record $diag[3] at line $./$diag[2] - $diag[0] - $diag[1]\n";
	my $ep  = $diag[2] - 1; # diag[2] is 1-based
	my $err = $csv->error_input . "         ";
	substr $err, $ep + 1, 0, "*";
	substr $err, $ep,     0, "*";
	($err = substr $err, $ep - 5, 12) =~ s/ +$//;
	print "    |$err|\n";
	}
    else {
	print "$ARGV line $. - $diag[1]\n";
	}
    print for @warn;
    exit $diag[0];
    } # done

sub stats {
    my $r = shift;
    $cols{scalar @$r}++;
    grep { $_ & 0x0002 } $csv->meta_info and $bin = 1;
    if ($opt_u) {
	my @r = @$r;
	foreach my $x (0 .. $#r) {
	    utf8::is_utf8 ($r[$x]) and next;

	    local $SIG{__WARN__} = sub {
		(my $msg = shift) =~ s{ at /\S+Encode.pm.*}{};
		my @h = $csv->column_names;
		push @warn, sprintf "Field %d%s in record %d - '%s'\t- %s",
		    $x + 1, @h ? " (column: '$h[$x]')" : "", $rows,
		    DPeek ($r[$x]), $msg;
		};
	    my $oct = decode ("utf-8", $r[$x], Encode::FB_WARN);
	    }
	}
    } # stats

my $mode = $enc ? "<:encoding($enc)" : "<";
open my $fh, $mode, \$data or die "$fn: $!\n";
if ($opt_h) {
    $csv->header ($fh);
    }
elsif ($opt_b) {
    my @hdr = $csv->header ($fh, { detect_bom => 1, set_column_names => 0 });
    stats \@hdr;
    }

local $SIG{__WARN__} = sub { push @warn, @_; };
while (my $row = $csv->getline ($fh)) {
    $rows++;
    stats $row;
    }
done;