Whitelist valid identifier characters

Instead of making up a list of characters that end an identifier, do the
other way around since it's far less cumbersome and it prevents from
silly bugs such as "var+1" being considered a single identifier.

Signed-off-by: Miquel Sabaté Solà <mikisabate@gmail.com>

2 files changed, 27 insertions, 8 deletions

diff --git a/lib/xixanta/src/assembler.rs b/lib/xixanta/src/assembler.rs
index e189b08..5e94070 100644
--- a/lib/xixanta/src/assembler.rs
+++ b/lib/xixanta/src/assembler.rs
@@ -2385,15 +2385,24 @@ mod tests {
             r#"
     Variable = 4
     adc Variable
+    adc Variable + 1
+    adc Variable+1
     "#,
         );
 
-        assert_eq!(res.len(), 1);
+        assert_eq!(res.len(), 3);
+
+        assert_eq!(res[0].size, 2);
+        assert_eq!(res[0].bytes[0], 0x65);
+        assert_eq!(res[0].bytes[1], 0x04);
 
-        let instr = res.first().unwrap();
-        assert_eq!(instr.size, 2);
-        assert_eq!(instr.bytes[0], 0x65);
-        assert_eq!(instr.bytes[1], 0x04);
+        assert_eq!(res[1].size, 2);
+        assert_eq!(res[1].bytes[0], 0x65);
+        assert_eq!(res[1].bytes[1], 0x05);
+
+        assert_eq!(res[2].size, 2);
+        assert_eq!(res[2].bytes[0], 0x65);
+        assert_eq!(res[2].bytes[1], 0x05);
     }
 
     #[test]
diff --git a/lib/xixanta/src/parser.rs b/lib/xixanta/src/parser.rs
index 50f2af1..e47f115 100644
--- a/lib/xixanta/src/parser.rs
+++ b/lib/xixanta/src/parser.rs
@@ -239,8 +239,18 @@ impl Parser {
             .chars()
             .peekable();
         while let Some(c) = chars.next() {
-            // Check for characters that end an identifier.
-            if c.is_whitespace() || c == ':' || c == '(' || c == ')' || c == '=' {
+            // Check for the end of the identifier. For this, it's easier to
+            // simply list what is allowed and negate it.
+            if !(c.is_ascii_alphanumeric()
+                || c == '.'
+                || c == '#'
+                || c == '$'
+                || c == '%'
+                || c == '@'
+                || c == '_'
+                || c == '\''
+                || c == '"')
+            {
                 // This next match looks scarier than what it actually is. To
                 // sum things up, the ':' character is quite troublesome, since
                 // it can mean three things depending on the context.
@@ -2160,7 +2170,7 @@ mod tests {
         assert!(node.right.is_none());
 
         let literal = &node.left.clone().unwrap();
-        assert_node(literal, NodeType::Literal, line, "#<NUM_SPRITES");
+        assert_node(literal, NodeType::Literal, line, "#");
 
         let op = &literal.left.clone().unwrap();
         assert_eq!(op.node_type, NodeType::Operation(OperationType::LoByte));