From cd225683d4414006c9422ff23076d88638af1794 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?John=20Ankarstr=C3=B6m?= <john@ankarstrom.se>
Date: Wed, 11 Nov 2020 15:13:38 +0100
Subject: add backslash escaping, auto-remove more punctuation

---
 rf   | 71 +++++++++++++++++++++++++++++++++++++++++++++++++-------------------
 rf.1 | 28 +++++++++++++++++++++------
 2 files changed, 73 insertions(+), 26 deletions(-)

diff --git a/rf b/rf
index 754bc70..535affb 100755
--- a/rf
+++ b/rf
@@ -16,11 +16,25 @@ my @lines;
 my $i = 0;
 my @refs;
 
+# Load entire file
+
 while (<>) {
 	# Formats
-	if (/^\.Ff +(.*)/) { $format_full = $1; next; }
-	if (/^\.Fx +(.*)/) { $format_extra = $1; next; }
-	if (/^\.Fl +(.*)/) { $format_list = $1; next; }
+	if (/^\.F([fxl])(\\?) +(.*)/) {
+		my ($type, $join, $line) = ($1, $2, $3);
+		chomp $line;
+		$line = "\n$line" if not $join;
+		while (not eof()) {
+			last if not $line =~ s/\\$//;
+			$line .= "\n" . <>;
+			chomp $line;
+		}
+		$line .= "\n";
+		$format_full = $line if $type eq 'f';
+		$format_list = $line if $type eq 'l';
+		$format_extra = $line if $type eq 'x';
+		next;
+	}
 
 	# Reference definitions
 	if (/^\.R([a-z]) +(.*)/) {
@@ -46,28 +60,33 @@ while (<>) {
 	push @lines, $_;
 }
 
+# Print processed file
+
 $i = -1;
+my $last = '';
 for (@lines) {
-	$i++;
 	# Inline reference
 	if (/^\.R([fx]) +(.*)/) {
 		my ($suffix, $prefix, @points);
-		my ($fld, $def) = ($1, $2);
-		if ($def =~ s/ ([.,:;\])]) ?([\[(])?$//) {
+		my ($fld, $def) = ($1, $2, $3);
+
+		# find potential prefix/suffix and split into words
+		if ($def =~ s/ ([.,?!:;\])]) ?([\[(])?$//) {
 			($suffix, $prefix) = ($1, $2);
 		}
-		my $winner = 0;
 		my @words = split /\s/, $def;
 
-		# replace '' with preceding word
+		# replace '' with last word on preceding line
 		for (@words) {
 			if ($_ eq "''") {
-				$_ = $lines[$i-1];
+				$_ = $last;
 				chomp;
 				$_ =~ s/^.*\s(\S+)\s*$/$1/;
 			}
 		}
 
+		# find matching reference list entry
+		my $winner = 0;
 		for (my $i = 0; $i < scalar @refs; $i++) {
 			$points[$i] = 0 if not defined $points[$i];
 			$points[$i] += 100 if likeness($refs[$i]{a}, @words);
@@ -81,44 +100,54 @@ for (@lines) {
 			print STDERR "Error: Reference '$def' could not be resolved.\n";
 			exit 1;
 		}
+
 		if ($points[$winner] < 150) {
 			print STDERR "Warning: Guessing that reference '$def' refers to " . fmt($format_full, $winner) . " (match = $points[$winner]).\n";
 		}
 
+		# Print formatted reference
+		my $fmt;
 		if ($fld eq 'f') {
-			no warnings;
-			print $prefix . fmt($format_full, $winner) . "$suffix\n";
+			$fmt = fmt($format_full, $winner);
 		} else {
-			no warnings;
-			print $prefix . fmt($format_extra, $winner) . "$suffix\n";
+			$fmt = fmt($format_extra, $winner);
 		}
+		$prefix = '' if not $prefix;
+		$suffix = '' if not $suffix;
+		chomp $last;
+		print $last;
+		$last = $prefix . substr($fmt, 0, index($fmt, "\n")) . "$suffix";
+		$last .= substr($fmt, index($fmt, "\n"));
 		next;
 	}
 
 	# Reference definition
 	if (/^\.R! (\d+)/) {
-		print fmt($format_list, $1, 1) . "\n";
+		print $last;
+		$last = fmt($format_list, $1, 1) . "\n";
 		next;
 	}
 
 	# Non-rf line
-	print "$_";
+	print $last;
+	$last = $_;
 }
 
+# Format a given reference after $fmt
 sub fmt {
 	my ($fmt, $i, $full) = @_;
 	my %ref = %{$refs[$i]};
-	for my $fld (split //, 'acdnpqtwy') {
-		if ($ref{$fld}) {
+	for my $fld (split //, 'Aacdnpqtwy') {
+		if ($ref{lc $fld}) {
 			no warnings;
-			my $val = $ref{$fld};
+			my $val = $ref{lc $fld};
 			$val = fmta($val) if $fld eq 'a' and not $full;
 			$fmt =~ s/\{([^{}%]*)%\Q$fld\E([^{}]*)}/$1$val$2/g;
 			$fmt =~ s/%\Q$fld\E/$val/g;
 		} else {
 			no warnings;
-			$fmt =~ s/[.(]?\{([^{}%]*)%\Q$fld\E([^{}]*)}[.,:;)]?//g;
-			$fmt =~ s/[.(]?%\Q$fld\E[.,:;)]?//g;
+			$fmt =~ s/[.([]?\{([^{}%]*)%\Q$fld\E([^{}]*)}[.,?!:;\])]?//g;
+			$fmt =~ s/[.([]?%\Q$fld\E[.,?!:;\])]?//g;
 		}
 	}
 	$fmt =~ s/ +/ /g;
@@ -129,6 +158,7 @@ sub fmt {
 	return $fmt;
 }
 
+# Remove forenames from a given string of authors
 sub fmta {
 	my ($a) = @_;
 	my $r;
@@ -144,6 +174,7 @@ sub fmta {
 	return $r;
 }
 
+# Calculate a basic likeness value a string and a list of strings
 sub likeness {
 	my ($string, @strings) = @_;
 	my $r = 0;
diff --git a/rf.1 b/rf.1
index d85121b..fbf429f 100644
--- a/rf.1
+++ b/rf.1
@@ -189,16 +189,32 @@ For example:
 \&.Ff (%a{, %y}).   \\" Here, ', ' is removed if there is no year.
 .Ed
 .Pp
-Note that all fields are inserted literally, except
-.Em %a ,
-from which the forenames of the authors are removed.
+You can include a newline in your format by ending the line with a backslash and continuing on the next line.
+If you add a backslash immediately following the macro name, the interpolated format will be joined to the preceding line.
+By using backslashes, you can implement footnote-based references.
+For example, using the ms macro package:
+.Bd -literal -offset indent
+\&.Fx\\ \\**\\
+\&.FS\\
+%A, {\\fI%t\\fR}, {\\*Q%q\\*U}, %y.\\
+\&.FE 
+.Ed
 .Pp
-Note also that you can use troff requests and other macros in your format by starting with a dot. This will expand to a valid request/macro in the output of
+As you can see above, you can use troff requests and other macros in your format.
+This will expand to a valid request/macro in the output of
 .Nm .
 .Pp
-Finally, you usually don't need to worry about excessive spaces and stray punctuation marks, as
+Note that all fields are inserted literally, with one exception:
+when
+.Em %a
+is used in an inline reference format, the forenames of the authors are automatically removed.
+To include the forenames, use
+.Em %A
+instead.
+.Pp
+Finally, you usually don't need to worry about excessive spaces and stray punctuation marks.
 .Nm
-removes many of these these automatically.
+removes common punctuation immediately preceding and following an unsuccessful interpolation.
 .
 .Sh EXAMPLE
 Below is an example of an ms-based troff document using
-- 
cgit v1.2.3