===================================================================
RCS file: /cvs/pta/pta_import.pl,v
retrieving revision 1.10
retrieving revision 1.26
diff -u -p -r1.10 -r1.26
--- pta/pta_import.pl	2020/11/12 21:54:23	1.10
+++ pta/pta_import.pl	2020/11/30 14:38:35	1.26
@@ -1,4 +1,4 @@
-#!/usr/bin/perl 
+#!/usr/bin/perl
 
 # Copyright (c) 2020 Freda Bundchen
 
@@ -21,119 +21,19 @@ use Getopt::Std qw(getopts);
 
 our ($opt_I);
 
+my %banks = (
+    "bbva_usa"              => \&import_bbva_usa,
+    "capital_one_360"       => \&import_capital_one_360,
+    "capital_one_credit"    => \&import_capital_one_credit,
+    "chase_credit"          => \&import_chase_credit,
+    "etrade_ira"            => \&import_etrade_ira,		
+    "optum_hsa"             => \&import_optum_hsa,
+    "sparkasse_camt"        => \&import_sparkasse_camt,
+    "wellsfargo"            => \&import_wellsfargo,
+);
+
 # === SUBROUTINES  =====================================================
 
-sub import_account ($$$$$) {
-	my ($account_name, $accounts_ref, $bookings_ref,
-	    $compiled_ref, $csv_account) = @_;
-	my ($account, $amount, $booking, $credit, $date, $debit,
-	    $description);
-	chomp(my $header = <>);
-	while (<>) {
-		next if (/^$/);
-		chomp;
-		my $line = $_;
-		foreach my $regex (@$compiled_ref) {
-			if ($line =~ /$regex/) {
-				$account = %$accounts_ref{$regex};
-				$booking = %$bookings_ref{$regex};
-				last;
-			}
-		}
-		if ($account_name eq "chase_credit") {
-			($date, $amount, $debit, $credit,
-			    $description) =
-			    import_chase_credit($account,
-			    $csv_account, $line);
-		} elsif ($account_name eq "capital_one_credit") {
-			($date, $amount, $debit, $credit,
-			    $description) =
-			    import_capital_one_credit($account,
-			    $csv_account, $line);
-		} elsif ($account_name eq "optum_hsa") {
-			($date, $amount, $debit, $credit,
-			    $description) =
-			    import_optum_hsa($account,
-			    $csv_account, $line);
-		} else {
-			die "undefined format: $account_name";
-		}
-		unless ($date && $booking && $debit && $credit &&
-		    $amount && $description) {
-			die "import parse error: $line";
-		}
-		$description =~ s/#//g;
-		print "$date $booking $debit $credit $amount " .
-		    "$description\n";
-	}
-}
-
-sub import_chase_credit ($$$) {
-	my ($account, $csv_account, $line) = @_;
-	my ($trans_date, $post_date, $description, $category,
-	    $type, $amount) = split /,/, $line;
-	my ($debit, $credit);
-	$post_date =~ s#(\d+)/(\d+)/(\d+)#$3$1$2#; 
-	($amount, $debit, $credit) =
-	    get_accounts_by_amount_sign($amount, $account,
-	    $csv_account);
-	return ($post_date, $amount, $debit, $credit, $description);
-}
-
-sub import_capital_one_credit ($$$) {
-	my ($account, $csv_account, $line) = @_;
-	my ($trans_date, $post_date, $card_num,
-	    $description, $category, $csv_debit,
-	    $csv_credit) = split /,/, $line;
-	$post_date =~ s/(\d+)-(\d+)-(\d+)/$1$2$3/; 
-	my ($amount, $debit, $credit) = 
-	    get_accounts_by_csv_col($account, $csv_account,
-	    $csv_debit, $csv_credit);
-	return ($post_date, $amount, $debit, $credit, $description);
-}
-
-sub import_optum_hsa ($$$) {
-        my ($account, $csv_account, $line) = @_;
-        my ($date, $description, $amount,
-            $type) = split /,/, $line;
-	my ($debit, $credit);
-	$date =~ s/(\d+)-(\d+)-(\d+)/$1$2$3/;
-	$amount =~ s/\$//;
-	($amount, $debit, $credit) =
-	    get_accounts_by_amount_sign($amount, $account,
-	    $csv_account);
-	return ($date, $amount, $debit, $credit, $description);
-}
-
-sub get_accounts_by_amount_sign ($$$) {
-	my ($amount, $account, $csv_account) = @_;
-	my ($debit, $credit);
-	if ($amount <= 0) {
-		$amount = substr $amount, 1;
-		$credit = $csv_account;	
-		$debit = $account;
-	} else {
-		$debit = $csv_account;	
-		$credit = $account;
-	}
-	return ($amount, $debit, $credit);
-}
-
-sub get_accounts_by_csv_col ($$$$) {
-	my ($account, $csv_account, $csv_debit, $csv_credit) = @_;
-	my ($amount, $debit, $credit);
-	if ($csv_debit eq "") {
-		$amount = $csv_credit;
-		$credit = $account;
-		$debit = $csv_account;
-	} else {
-		$amount = $csv_debit;
-		$credit = $csv_account;
-		$debit = $account;
-	}
-	return ($amount, $debit, $credit);
-}
-
 sub usage () {
 	printf STDERR "usage: %s -I accountname csvfilename\n", $0;
 	exit 1;
@@ -141,34 +41,159 @@ sub usage () {
 
 # === MAIN PROGRAM =====================================================
 
-my ($csv_account, $fn, $in, $account_name, %accounts, %bookings, @compiled);
 getopts 'I:' or usage;
-if ($opt_I) {
-	$account_name = $opt_I;
-	$fn = "import_" . $account_name . ".txt";
-	open $in, $fn or die "$fn: $!";
-} else {
+unless ($opt_I) {
+	warn "The option -I is required.";
 	usage;
 }
+my $account_name = $opt_I;
+$banks{$account_name} or die "unknown accountname: $account_name";
+
+# 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_format,
+    $delim, $header, $quantity_field, $quote, @amount_fields,
+    @compiled, @description_fields, @ignored);
 while (<$in>) {
 	chomp;
 	next if /^(?:#|$)/;
 	my $line = $_;
-	if (/^ACCOUNT\s+(\d+)$/) {
+	if (/^ACCOUNT\s+(\S+)$/) {
 		$csv_account and die "duplicate ACCOUNT line: $1";
 		$csv_account = $1;
 		next;
 	}
-	s/^(.*),\s+(\d+)\s+(\S+)// or
+	if (s/^AMOUNT\s+//) {
+		push @amount_fields, $1 - 1 while s/(\d+)\s*//;
+		$_ eq '' or die "trailing garbage: AMOUNT ... $_";
+		next;
+	}
+	if (/^COSTCENTER\s+(\d+)/) {
+		$cost_center_field = $1 - 1;
+		next;
+	}
+	if (/^DATE\s+(\d+)\s+(\S+)/) {
+		$date_field = $1 - 1;
+		$date_format = $2;
+		next;
+	}
+	if (/^DELIM\s+(\S)$/) {
+		$delim and die "duplicate DELIM line: $1";
+		$delim = $1;
+		next;
+	}
+	if (s/^DESCRIPTION\s+//) {
+	    push @description_fields, $1 - 1 while s/(\d+)\s*//;
+	    $_ eq '' or die "trailing garbage: DESCRIPTION ... $_";
+	    next;
+	}
+	if (/^HEADER\s+(.*)$/) {
+		$header and die "duplicate HEADER line: $1";
+		$header = $1;
+		next;
+	}
+	if (/^IGNORE\s+(.*)/) {
+		push @ignored, qr/$1/;
+		next;
+	}
+	if (/^QUANTITY\s+(\d+)/) {
+		$quantity_field = $1 - 1;
+		next;
+	}
+	if (/^QUOTE\s+(\S)$/) {
+		$quote and die "duplicate QUOTE line: $1";
+		$quote = $1;
+		next;
+	}
+	$delim or die "no DELIM line in $fn";
+	s/^(.*)$delim\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} = $account;
+	push @compiled, {
+		re => [map { qr/$_/ } split /$delim/, $1],
+		ac => $2,
+		id => $3,
+	};
 }
 close $in;
 $csv_account or die "no ACCOUNT line in $fn";
-import_account($account_name, \%accounts, \%bookings,
-    \@compiled, $csv_account);
+
+# Parse the CSV file from the bank.
+if (@ARGV) {
+	open STDIN, '<', $ARGV[0] or die "$ARGV[0]: $!";
+}
+LINE: while (<STDIN>) {
+	chomp;
+	next if (/^$/);
+	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];
+			$matches = 0;
+			last;
+		}
+		if ($matches) {
+			$account = $selector->{ac};
+			$booking = $selector->{id};
+			last;
+		}
+	}
+	$matches or die "unmatched CSV line: $line";
+	my $date = $fields[$date_field] or
+	    die "date parse error: $line";
+	if ($date_format eq "MM/DD/YYYY") {
+		$date =~ s#(\d+)/(\d+)/(\d+)#$3$1$2#;
+	} elsif ($date_format eq "MM/DD/YY") {
+		$date =~ s#(\d+)/(\d+)/(\d+)#20$3$1$2#;
+	} elsif ($date_format eq "YYYY-MM-DD") {
+		$date =~ s#(\d+)-(\d+)-(\d+)#$1$2$3#;
+	} elsif ($date_format eq "DD.MM.YY") {
+		$date =~ s#(\d+)\.(\d+)\.(\d+)#20$3$2$1#;
+	} else {
+		die "unhandled date format: $line";
+	}
+	foreach my $i (@amount_fields) {
+		if (defined($fields[$i])) {
+			$fields[$i] =~ s/,/./;
+			$fields[$i] =~ s/\$//;
+			$fields[$i] = "-$1" if $fields[$i] =~ /^\((\d+\.\d+)\)/;
+		}
+	}
+	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];
+	$date && $amount && $debit && $credit && $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";
+}