===================================================================
RCS file: /cvs/pta/pta_import.pl,v
retrieving revision 1.21
retrieving revision 1.29
diff -u -p -r1.21 -r1.29
--- pta/pta_import.pl	2020/11/25 12:46:41	1.21
+++ pta/pta_import.pl	2021/05/19 17:10:53	1.29
@@ -21,123 +21,15 @@ use Getopt::Std qw(getopts);
 
 our ($opt_I);
 
-my %banks = (
-    "bbva_usa"              => \&import_bbva_usa,
-    "capital_one_credit"    => \&import_capital_one_credit,
-    "chase_credit"          => \&import_chase_credit,
-    "optum_hsa"             => \&import_optum_hsa,
-    "sparkasse_camt"        => \&import_sparkasse_camt,
-    "wellsfargo"            => \&import_wellsfargo,
+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_bbva_usa ($$$) {
-	my ($account, $csv_account, $line) = @_;
-	$line =~
-	    m#(\d+)/(\d+)/(\d+),(.*?),\S+,(\(?\d+\.\d+\)?),# or
-	    die "parse error: $line";
-	my ($month, $day, $year, $description, $amount) =
-	    ($1, $2, $3, $4, $5);
-	$amount =~ s/\((\d+\.\d+)\)/-$1/;
-	return "$year$month$day",
-	    get_accounts_by_amount_sign($amount, $account, $csv_account),
-	    $description;
-}
-
-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 ($$$) {
-        my ($account, $csv_account, $line) = @_;
-        my ($date, $description, $amount,
-            $type) = split /,/, $line;
-	my ($debit, $credit);
-	$date =~ s/(\d+)-(\d+)-(\d+)/$1$2$3/;
-	$amount =~ s/\$//;
-	($amount, $debit, $credit) =
-	    get_accounts_by_amount_sign($amount, $account,
-	    $csv_account);
-	return ($date, $amount, $debit, $credit, $description);
-}
-
-sub import_sparkasse_camt ($$$) {
-        my ($account, $csv_account, $line) = @_;
-	my @fields;
-	$_ = $line;
-	push @fields, $1 while s/"([^"]*)";?//;
-	$_ eq "" or die "CAMT parse error before $_ in $line";
-	@fields == 17 or die "not 17 but @fields fields: $line";
-	$fields[1] =~ s/^(\d\d)\.(\d\d)\.(\d\d)$/20$3$2$1/
-	    or die "date parse error: $line";
-	$fields[14] =~ s/,/./;
-	return $fields[1],
-	    get_accounts_by_amount_sign($fields[14], $account, $csv_account),
-	    (join ' ', $fields[11], $fields[4]);
-}
-
-sub import_wellsfargo ($$$) {
-	my ($account, $csv_account, $line) = @_;
-	$line =~
-	    m#^"(\d+)/(\d+)/(\d+)","(-?\d+\.\d+)","\*",".*","(.*?)"$# or
-	    die "parse error: $line" ;
-	my ($month, $day, $year, $amount, $description) =
-	    ($1, $2, $3, $4, $5);
-	return "$year$month$day",
-	    get_accounts_by_amount_sign($amount, $account, $csv_account),
-	    $description;
-}
-
-sub get_accounts_by_amount_sign ($$$) {
-	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", $0;
 	exit 1;
@@ -151,40 +43,83 @@ unless ($opt_I) {
 	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";
+my $fn = "import/" . $account_name . ".txt";
 open my $in, '<', $fn or die "$fn: $!";
-my ($csv_account, $delim, $header, $quote, @compiled, @ignored);
+my ($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 = $_;
-	if (/^ACCOUNT\s+(\S+)$/) {
-		$csv_account and die "duplicate ACCOUNT line: $1";
+	if (s/^ACCOUNT\s+//) {
+		$csv_account and die "duplicate ACCOUNT line: $_";
+		/^(\d+)$/ or die "ACCOUNT parse error: $_";
 		$csv_account = $1;
 		next;
 	}
-	if (/^DELIM\s+(\S)$/) {
+	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 = $1;
 		next;
 	}
-	if (/^HEADER\s+(.*)$/) {
+	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 = $1;
+		$header = $_;
 		next;
 	}
-	if (/^QUOTE\s+(\S)$/) {
-		$quote and die "duplicate QUOTE line: $1";
-		$quote = $1;
+	if (s/^IGNORE\s+//) {
+		push @ignored, qr/$_/;
 		next;
 	}
-	if (/^IGNORE\s+(.*)/) {
-		push @ignored, qr/$1/;
+	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: $line";
@@ -203,6 +138,7 @@ if (@ARGV) {
 }
 LINE: while (<STDIN>) {
 	chomp;
+	s/\s*$//;
 	next if (/^$/);
 	foreach my $ignore (@ignored) {
 		next LINE if /$ignore/;
@@ -237,10 +173,32 @@ LINE: while (<STDIN>) {
 		}
 	}
 	$matches or die "unmatched CSV line: $line";
-	my ($date, $amount, $debit, $credit, $description) =
-	    $banks{$account_name}->($account, $csv_account, $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] =~ /^\((\d+\.\d+)\)/;
+		}
+	}
+	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";
 }