Finally generating code

vasm/src/ast.rs

@@ -56,4 +56,10 @@ impl<'a> VASMLine<'a> {
             _ => None,
         }
     }
+    pub fn zero_length(&self) -> bool {
+        matches!(
+            self,
+            VASMLine::Org(_, _) | VASMLine::Equ(_, _) | VASMLine::LabelDef(_)
+        )
+    }
 }

vasm/src/vasm_assembler.rs

@@ -9,6 +9,7 @@ pub enum AssembleError<'a> {
     EquDuplicateError(i32, &'a str),
     OrgResolveError(i32, EvalError<'a>),
     ArgError(i32, EvalError<'a>),
+    NoCode,
 }
 
 impl<'a> Display for AssembleError<'a> {
@@ -26,6 +27,9 @@ impl<'a> Display for AssembleError<'a> {
             AssembleError::ArgError(line, err) => {
                 write!(f, "Cannot calculate argument on line {}: {}", line, err)
             }
+            AssembleError::NoCode => {
+                write!(f, "No output would be generated by this code")
+            }
         }
     }
 }
@@ -178,6 +182,106 @@ fn calculate_args<'a>(
     Ok(args)
 }
 
+fn poke_word(code: &mut Vec<u8>, at: usize, word: i32) {
+    let [low, mid, high, _] = word.to_le_bytes();
+    code[at] = low;
+    code[at + 1] = mid;
+    code[at + 2] = high;
+}
+
+/// Find the lower and upper bounds where this program will place memory
+fn code_bounds<'a>(
+    lines: &[VASMLine<'a>],
+    line_addresses: &LineAddresses,
+    line_lengths: &LineLengths,
+) -> Result<(usize, usize), AssembleError<'a>> {
+    let mut actual_lines = lines
+        .iter()
+        .enumerate()
+        .filter(|(_, line)| !line.zero_length());
+    let (first_idx, _) = actual_lines.next().ok_or(AssembleError::NoCode)?;
+    let start = line_addresses[&(first_idx as i32 + 1)] as usize;
+
+    let actual_lines = lines
+        .iter()
+        .enumerate()
+        .filter(|(_, line)| !line.zero_length());
+    let (last_idx, _) = actual_lines.last().unwrap();
+    let end = line_addresses[&(last_idx as i32 + 1)] as usize;
+    let end_length = line_lengths[&(last_idx as i32 + 1)];
+
+    Ok((start, end + end_length - 1))
+}
+
+/// At this point all lines have addresses and lengths, and all arguments are reduced to
+/// numeric constants. It's time to generate code.
+///
+/// - Make an array of zeroes, length (end - start)
+/// - Go through the list of instructions, generating code for them:
+/// - .db instructions turn into byte values starting at `address - start`
+/// - Opcodes turn into instruction bytes at `address - start` followed (maybe) by
+///   arguments.
+/// - Any other directive is skipped (even .orgs, we already have the addresses calculated)
+///
+/// The instruction bytes are formed of six bits defining the instruction followed by two
+/// bits denoting how many bytes of argument follow it.
+///
+/// Vulcan is a little-endian architecture: multi-byte arguments / .dbs will store the
+/// least-significant byte at the lowest address, then the more significant bytes following.
+fn generate_code<'a>(
+    lines: &[VASMLine<'a>],
+    scope: Scope,
+    line_addresses: LineAddresses,
+    line_lengths: LineLengths,
+) -> Result<Vec<u8>, AssembleError<'a>> {
+    let line_args = calculate_args(lines, scope, &line_addresses)?;
+    let (start, end) = code_bounds(lines, &line_addresses, &line_lengths)?;
+    let mut code = vec![0u8; end - start + 1];
+    let mut current_addr = start;
+
+    for (line_idx, line) in lines.iter().enumerate() {
+        let line_num = (line_idx + 1) as i32;
+        match line {
+            VASMLine::Instruction(_, opcode, None) => {
+                code[current_addr] = u8::from(*opcode) << 2;
+                current_addr += 1;
+            }
+            VASMLine::Instruction(_, opcode, Some(_)) => {
+                let arg = line_args[&line_num];
+                let len = arg_length(arg);
+                let instr = (u8::from(*opcode) << 2) + len as u8;
+                code[current_addr - start] = instr;
+                let [low, mid, high, _] = arg.to_le_bytes();
+                code[current_addr - start + 1] = low;
+                if len > 1 {
+                    code[current_addr - start + 2] = mid
+                }
+                if len > 2 {
+                    code[current_addr - start + 3] = high
+                }
+                current_addr += len + 1;
+            }
+            VASMLine::Db(_, _) => {
+                let arg = line_args[&line_num];
+                poke_word(&mut code, current_addr - start, arg);
+                current_addr += 3;
+            }
+            VASMLine::StringDb(_, string) => {
+                for ch in string.as_bytes() {
+                    code[current_addr - start] = *ch;
+                    current_addr += 1;
+                }
+            }
+            VASMLine::Org(_, _) => {
+                current_addr = line_addresses[&(line_num + 1)] as usize;
+            }
+            VASMLine::Equ(_, _) | VASMLine::LabelDef(_) => {}
+        }
+    }
+
+    Ok(code)
+}
+
 #[cfg(test)]
 mod test {
     use super::AssembleError::*;
@@ -203,11 +307,24 @@ mod test {
     }
 
     fn calculate_args_pass<'a>(lines: &'a [&str]) -> Result<LineArgs, AssembleError<'a>> {
+        let (line_addresses, scope) = place_labels_pass(lines)?;
+        let lines = parse(lines);
+        calculate_args(&lines, scope, &line_addresses)
+    }
+
+    fn bounds<'a>(lines: &'a [&str]) -> Result<(usize, usize), AssembleError<'a>> {
+        let (line_addresses, scope) = place_labels_pass(lines)?;
+        let lines = parse(lines);
+        let lengths = measure_instructions(&lines, &scope);
+        code_bounds(&lines, &line_addresses, &lengths)
+    }
+
+    fn generate_code_pass<'a>(lines: &'a [&str]) -> Result<Vec<u8>, AssembleError<'a>> {
         let lines = parse(lines);
         let scope = solve_equs(&lines).unwrap();
         let lengths = measure_instructions(&lines, &scope);
         let (line_addresses, scope) = place_labels(&lines, scope, &lengths)?;
-        calculate_args(&lines, scope, &line_addresses)
+        generate_code(&lines, scope, line_addresses, lengths)
     }
 
     #[test]
@@ -347,4 +464,38 @@ mod test {
             ))
         );
     }
+
+    #[test]
+    fn test_bounds() {
+        assert_eq!(bounds(&["add"]), Ok((0, 0)));
+        assert_eq!(bounds(&["add -4"]), Ok((0, 3)));
+        assert_eq!(bounds(&["add 7"]), Ok((0, 1)));
+        assert_eq!(bounds(&[".org 0x400", "add"]), Ok((1024, 1024)));
+        assert_eq!(
+            bounds(&["start: .equ 1024", ".org start", "add"]),
+            Ok((1024, 1024))
+        );
+        assert_eq!(
+            bounds(&[".org 0x400", "add", ".org 0x800"]),
+            Ok((1024, 1024))
+        );
+        assert_eq!(
+            bounds(&[".org 0x400", "add", ".org 0x800", ".db 5", "blah:"]),
+            Ok((1024, 2050))
+        );
+        assert_eq!(bounds(&[]), Err(AssembleError::NoCode));
+        assert_eq!(bounds(&[".org 0x400"]), Err(AssembleError::NoCode));
+        assert_eq!(
+            bounds(&["foo: .equ 3", ".org 0x400"]),
+            Err(AssembleError::NoCode)
+        );
+    }
+
+    #[test]
+    fn test_generate_code() {
+        assert_eq!(generate_code_pass(&["add"]), Ok(vec![4]));
+        assert_eq!(generate_code_pass(&[".org 0x400", "add 7"]), Ok(vec![5, 7]));
+        assert_eq!(generate_code_pass(&[".db 57"]), Ok(vec![57, 0, 0]));
+        assert_eq!(generate_code_pass(&[".db \"AZ\0\""]), Ok(vec![65, 90, 0]));
+    }
 }