===================================================================
RCS file: /cvs/pta/pta_import.pl,v
retrieving revision 1.3
retrieving revision 1.30
diff -u -p -r1.3 -r1.30
--- pta/pta_import.pl	2020/11/02 19:36:53	1.3
+++ pta/pta_import.pl	2021/05/19 21:43:47	1.30
@@ -1,4 +1,4 @@
-#!/usr/bin/perl 
+#!/usr/bin/perl
 
 # Copyright (c) 2020 Freda Bundchen
 
@@ -17,68 +17,189 @@
 use warnings;
 use strict;
 
+use Getopt::Std qw(getopts);
+
+our ($opt_I);
+
+my %date_formats = (
+    'MM/DD/YYYY' => 's#(\d+)/(\d+)/(\d+)#$3$1$2#',
+    'MM/DD/YY'   => 's#(\d+)/(\d+)/(\d+)#20$3$1$2#',
+    'YYYY-MM-DD' => 's#(\d+)-(\d+)-(\d+)#$1$2$3#',
+    'DD.MM.YY'   => 's#(\d+)\.(\d+)\.(\d+)#20$3$2$1#',
+);
+
 # === SUBROUTINES  =====================================================
 
-sub import_chase_credit_card {
-	my $csv_account = "1234";	# DEBUG. hard-coded for now
-	my $fn = 'import_chase_credit.txt';
-	my (@compiled, %bookings, %accounts);
-	open my $in, $fn or die "$fn: $!";
-	while (<$in>) {
-		chomp;
-		next if /^(?:#|$)/;
-		my $line = $_;
-		s/^(.*),\s+(\d+)\s+(\S+)// or 
-		    die "$fn import parse error: $line";
-		my ($reg, $account, $booking) = ($1, $2, $3);
-		$reg =~ s/,,/,.*?/g;
-		$reg = qr/$reg/;
-		push @compiled, $reg;
-		$bookings{$reg} = $booking;
-		$accounts{$reg} = $account;
+sub usage () {
+	printf STDERR "usage: %s -I accountname csvfilename\n", $0;
+	exit 1;
+}
+
+# === MAIN PROGRAM =====================================================
+
+getopts 'I:' or usage;
+unless ($opt_I) {
+	warn "The option -I is required.";
+	usage;
+}
+my $account_name = $opt_I;
+
+# Parse the configuration file.
+my $fn = "import/" . $account_name . ".txt";
+open my $in, '<', $fn or die "$fn: $!";
+my ($cost_center_field, $csv_account, $date_field, $date_regex,
+    $delim, $header, $quantity_field, $quote, @amount_fields,
+    @compiled, @description_fields, @ignored);
+while (<$in>) {
+	chomp;
+	s/\s+$//;
+	next if /^(?:#|$)/;
+	my $line = $_;
+	if (s/^ACCOUNT\s+//) {
+		$csv_account and die "duplicate ACCOUNT line: $_";
+		/^(\d+)$/ or die "ACCOUNT parse error: $_";
+		$csv_account = $1;
+		next;
 	}
-	close $in;
-	while (<>) {
-		chomp;
-		s#(\d+)/(\d+)/(\d+)#$3$1$2#g; 
-		my $line = $_;
-		my ($trans_date, $post_date, $description, $category, 
-		    $type, $amount) = split /,/;
-		$description =~ s/#//g;
-		my ($debit, $credit);
-		my $booking = "";
-		my $account = "";
-		foreach my $regex (@compiled) {
-			if ($line =~ /$regex/) {
-				$account = $accounts{$regex};
-				$booking = $bookings{$regex};
-				next;
-			}
+	if (s/^AMOUNT\s+//) {
+		@amount_fields and die "duplicate AMOUNT line: $_";
+		push @amount_fields, $1 - 1 while s/(\d+)\s*//;
+		$_ eq '' or die "trailing garbage: AMOUNT ... $_";
+		next;
+	}
+	if (s/^COSTCENTER\s+//) {
+		$cost_center_field and
+		    die "duplicate COSTCENTER line: $_";
+		/^(\d+)$/ or die "COSTCENTER parse error: $_";
+		$cost_center_field = $1 - 1;
+		next;
+	}
+	if (s/^DATE\s+//) {
+		$date_field || $date_regex and
+		    die "duplicate DATE line: $_";
+		s/^(\d+)\s+(\S+)$//;
+		$date_field = $1 or
+		    die "DATE date_field parse error: $_";
+		$date_field -= 1;
+		$date_regex = $date_formats{$2}
+		    or die "unknown date format: $2";
+		next;
+	}
+	if (s/^DELIM\s+//) {
+		$delim and die "duplicate DELIM line: $1";
+		/^([^|\^\$\*\+\?\(\)\[\]\{\}\\])$/ or
+		    die "DELIM parse error: $_";
+		$delim = $1;
+		next;
+	}
+	if (s/^DESCRIPTION\s+//) {
+		@description_fields and die "duplicate DESCRIPTION line: $_";
+		push @description_fields, $1 - 1 while s/(\d+)\s*//;
+		$_ eq '' or die "trailing garbage: DESCRIPTION ... $_";
+		next;
+	}
+	if (s/^HEADER\s+//) {
+		$header and die "duplicate HEADER line: $1";
+		$header = $_;
+		next;
+	}
+	if (s/^IGNORE\s+//) {
+		push @ignored, qr/$_/;
+		next;
+	}
+	if (s/^QUANTITY\s+//) {
+		$quantity_field and die "duplicate QUANTITY line: $1";
+		/^(\d+)$/ or die "QUANTITY parse error: $_";
+		$quantity_field = $1 - 1;
+		next;
+	}
+	if (s/^QUOTE\s+//) {
+		$quote and die "duplicate QUOTE line: $1";
+		/^([^|\^\$\*\+\?\(\)\[\]\{\}\\])$/ or
+		die "QUOTE parse error: $_";
+		$quote = $1;
+		next;
+	}
+	$delim or die "no DELIM line in $fn";
+	s/^(.*)$delim\s+(\d+)\s+(\S+)// or
+	    die "$fn import parse error: $line";
+	push @compiled, {
+		re => [map { qr/$_/ } split /$delim/, $1],
+		ac => $2,
+		id => $3,
+	};
+}
+close $in;
+$csv_account or die "no ACCOUNT line in $fn";
+
+# Parse the CSV file from the bank.
+if (@ARGV) {
+	open STDIN, '<', $ARGV[0] or die "$ARGV[0]: $!";
+}
+LINE: while (<STDIN>) {
+	chomp;
+	s/\s*$//;
+	next if (/^$/);
+	foreach my $ignore (@ignored) {
+		next LINE if /$ignore/;
+	}
+	my $line = $_;
+	if ($header) {
+		$line eq $header
+		    or die "expected HEADER $header\nbut got $line";
+		undef $header;
+		next;
+	}
+	my $copy_line = $line;
+	my @fields;
+	if ($quote) {
+		push @fields, $1 while $copy_line =~ s/$quote([^$quote]*)$quote$delim?//;
+	} else {
+		@fields = split /$delim/, $line;
+	}
+	my $matches = 0;
+	my ($account, $booking);
+	foreach my $selector (@compiled) {
+		$matches = 1;
+		for (my $i = 0; $i <= $#{$selector->{re}}; $i++) {
+			next if $fields[$i] =~ $selector->{re}[$i];
+			$matches = 0;
+			last;
 		}
-		if ($booking eq "" || $account eq "") {
-			die "$fn import parse error: $line";
+		if ($matches) {
+			$account = $selector->{ac};
+			$booking = $selector->{id};
+			last;
 		}
-		if ($amount <= 0) {
-			$amount = substr $amount, 1;
-			$credit = $csv_account;	
-			$debit = $account;
-		} else {
-			$debit = $csv_account;	
-			$credit = $account;
+	}
+	$matches or die "unmatched CSV line: $line";
+	my $date = $fields[$date_field] or
+	    die "date parse error: $line";
+	eval '$date =~ ' . $date_regex;
+	foreach my $i (@amount_fields) {
+		if (defined($fields[$i])) {
+			$fields[$i] =~ s/,/./;
+			$fields[$i] =~ s/\$//;
+			$fields[$i] = "-$1" if $fields[$i] =~ /^\((\d+\.\d+)\)/;
 		}
-		print "$post_date $booking $debit $credit $amount " .
-		    "$description\n";
 	}
-}
-
-# === MAIN PROGRAM =====================================================
-
-my ($chase_credit_card_header, $firstLine);
-$chase_credit_card_header = "Transaction Date,Post Date," .
-    "Description,Category,Type,Amount";
-chomp($firstLine = <>);
-if ($firstLine eq $chase_credit_card_header) {
-	import_chase_credit_card;
-} else {
-	print "Undefined header\n";
+	my $debit = $csv_account;
+	my $credit = $account;
+	my $amount = $fields[$amount_fields[-1]] || -$fields[$amount_fields[0]];
+	if ($amount < 0) {
+		$amount *= -1;
+		$credit = $csv_account;
+		$debit = $account;
+	}
+	$amount = sprintf "%.2f", $amount;
+	my $description = join ' ', @fields[@description_fields];
+	defined($date) && defined($amount) && defined($debit) &&
+	    defined($credit) && defined($description)
+	    or die "import parse error: $line";
+	$description =~ s/#//g;
+	$description .= " [$fields[$cost_center_field]]"
+	    if $cost_center_field;
+	$description .= " quantity $fields[$quantity_field]"
+	    if $quantity_field;
+	print "$date $booking $debit $credit $amount $description\n";
 }