From c84b2dffddc0f31ebdd2e95ac95d0dce65eb3921 Mon Sep 17 00:00:00 2001
From: Warren Toomey <wkt@tuhs.org>
Date: Fri, 26 Feb 2016 07:35:36 +1000
Subject: [PATCH] I've changed the assembler so that it groks the <ab> string
 literal syntax. I think it's OK, but it probably needs more testing.
 write_test.s was changed to exercise the string parsing.

---
 tools/as7          | 30 +++++++++++++++++++++++++++++-
 tools/write_test.s |  2 +-
 2 files changed, 30 insertions(+), 2 deletions(-)

diff --git a/tools/as7 b/tools/as7
index 835f830..1d235b0 100755
--- a/tools/as7
+++ b/tools/as7
@@ -198,7 +198,7 @@ sub parse_statement {
 
 # Parse an expression and return either:
 #  a single value which is a PDP-7 word
-#  a single value which is undef, as this was a statement
+#  a single value which is undef, as this was unrecognised
 sub parse_expression {
     my $expression = shift;
 
@@ -346,6 +346,34 @@ sub parse_expression {
     #printf( "o>%s<o o>%s<o o>%s<o\n",
     #			$word1 || "", $word2 || "", $word3 || "" );
 
+    # This is a 2-character string literal, put each in separate nine bits
+    if ($word1=~ m{^<(.)(.)>$}) {
+      print("String literal <$1 and $2>\n") if ($debug);
+      return((ord($1) << 9) | ord($2));
+    }
+
+    # This is a <n string literal which might have been followed by a numeric literal
+    if ($word1=~ m{^<(.)}) {
+      print("String literal <$1 and $word2\n") if ($debug);
+      my $msb= $1;
+      my $lsb= (defined($word2)) ? parse_expression($word2) : 0;
+      return((ord($msb) << 9) | $lsb);
+    }
+
+    # This is a n> string literal preceded by a numeric literal
+    if (defined($word2) && ($word2=~ m{(.)>$})) {
+      print("String literal $word1 and $1>\n") if ($debug);
+      my $msb= parse_expression($word1);
+      my $lsb= $1;
+      return(($msb << 9) | ord($lsb));
+    }
+
+    # This is a n> string literal not preceded by a numeric literal
+    if ($word1=~ m{(.)>$}) {
+      print("String literal $1>\n") if ($debug);
+      return(ord($1));
+    }
+
     # This a defined instruction
     if ( defined( $inst{$word1} ) ) {
         #printf( "Found the instruction %s: 0%o\n", $word1, $inst{$word1} );
diff --git a/tools/write_test.s b/tools/write_test.s
index 375b0bc..6a1ea8e 100644
--- a/tools/write_test.s
+++ b/tools/write_test.s
@@ -47,7 +47,7 @@ in: 023
 out: 0
 
 " Hello, world\n, two ASCII chars per word
-hello: 0110145; 0154154; 0157054; 040; 0167157; 0162154; 0144012
+hello: <He>; <l 0154; 0157 ,>; 040; <wo>; <rl>; <d 012
 helloptr: hello
 
 " fred as a string, NUL terminated