From cac125bd1cbae94f0e8b2e16a740c680fe4b9a5e Mon Sep 17 00:00:00 2001
From: Natapat Samutpong <natapat.samutpong@gmail.com>
Date: Thu, 27 Jan 2022 10:43:46 +0700
Subject: [PATCH] feat: simpler instruction parser

---
 blspc/example/if.bbb                  | 19 ------
 blspc/src/vm/instr.rs                 | 30 ++++-----
 blspc/src/vm/parser.rs                | 95 ++++++++++++++-------------
 {blspc/example => example}/hello.blsp |  0
 {blspc/example => example}/if.blsp    |  0
 {blspc/example => example}/math.blsp  |  0
 {blspc/example => example}/s.blsp     |  0
 7 files changed, 64 insertions(+), 80 deletions(-)
 delete mode 100644 blspc/example/if.bbb
 rename {blspc/example => example}/hello.blsp (100%)
 rename {blspc/example => example}/if.blsp (100%)
 rename {blspc/example => example}/math.blsp (100%)
 rename {blspc/example => example}/s.blsp (100%)

diff --git a/blspc/example/if.bbb b/blspc/example/if.bbb
deleted file mode 100644
index af8719a..0000000
--- a/blspc/example/if.bbb
+++ /dev/null
@@ -1,19 +0,0 @@
-; ------------------------------
-; (if true (print (+ 34 35)) (print "False"))
-; ------- Condition block ------
-1: STORE r1 $true              ; r1 = True
-2: JUMP_IF_FALSE r1 9          ; if
-; --------- True block ---------
-3: STORE r2 $34                ; r2 = 34
-4: STORE r3 $35                ; r3 = 35
-5: IADD r2 r3 r4               ; r2 + r3 -> r4
-6: STORE r5 $1                 ; r5 = 1 (function calling)
-7: CALL r5 r4                  ; call 1 (print) 34 (r4)
-8: JUMP 12                     
-; -------- False block ---------
-9: STORE r6 $"False"           ; r6 = "False"
-10: STORE r7 $1                ; r7 = 1 (function calling)
-11: CALL r7 r6                 ; call 1 (print) "False" (r6)
-12: STORE r8 $0                ; r8 = 0 (return value)
-13: RETURN r8                  ; return 0 (exit code)
-; ------------ End -------------
\ No newline at end of file
diff --git a/blspc/src/vm/instr.rs b/blspc/src/vm/instr.rs
index ac95168..707fdf2 100644
--- a/blspc/src/vm/instr.rs
+++ b/blspc/src/vm/instr.rs
@@ -12,9 +12,9 @@ pub enum Type {
 impl Display for Type {
     fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
         match self {
-            Type::Int(i)     => write!(f, "${}", i),
-            Type::Float(fl)  => write!(f, "${}", fl),
-            Type::Boolean(b) => write!(f, "${}", b),
+            Type::Int(i)     => write!(f, ":{}", i),
+            Type::Float(fl)  => write!(f, ":{}", fl),
+            Type::Boolean(b) => write!(f, ":{}", b),
             Type::String(s)  => write!(f, "$\"{}\"", s),
         }
     }
@@ -24,7 +24,7 @@ impl FromStr for Type {
     type Err = String;
 
     fn from_str(s: &str) -> Result<Self, Self::Err> {
-        if !s.starts_with("$") {
+        if !s.starts_with("$") && !s.starts_with(":") {
             return Err(format!("Invalid literal: {}", s));
         }
 
@@ -92,17 +92,17 @@ impl Display for Instr {
     fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
         match self {
             // Instr::Load { address, label }         => write!(f, "{}: LOAD {}", label, address),
-            Instr::Store { address, value , label} => write!(f, "{}: STORE {} {}", label, address, value),
-            Instr::Call { address, args, label }   => write!(f, "{}: CALL {} {}", label, address, args),
-            Instr::Push { value, label }           => write!(f, "{}: PUSH {}", label, value),
-            Instr::Pop { address, label }          => write!(f, "{}: POP {}", label, address),
-            Instr::Add { label }                   => write!(f, "{}: ADD", label),
-            Instr::Sub { label }                   => write!(f, "{}: SUB", label),
-            Instr::Mul { label }                   => write!(f, "{}: MUL", label),
-            Instr::Div { label }                   => write!(f, "{}: DIV", label),
-            Instr::Jump { to, label }              => write!(f, "{}: JUMP {}", label, to),
-            Instr::PopJumpIfFalse { to, label }    => write!(f, "{}: POP_JUMP_IF_FALSE {}", label, to),
-            Instr::Return { value, label }         => write!(f, "{}: RETURN {}", label, value),
+            Instr::Store { address, value , label} => write!(f, "{} STORE {} {}", label, address, value),
+            Instr::Call { address, args, label }   => write!(f, "{} CALL {} {}", label, address, args),
+            Instr::Push { value, label }           => write!(f, "{} PUSH {}", label, value),
+            Instr::Pop { address, label }          => write!(f, "{} POP {}", label, address),
+            Instr::Add { label }                   => write!(f, "{} ADD", label),
+            Instr::Sub { label }                   => write!(f, "{} SUB", label),
+            Instr::Mul { label }                   => write!(f, "{} MUL", label),
+            Instr::Div { label }                   => write!(f, "{} DIV", label),
+            Instr::Jump { to, label }              => write!(f, "{} JMP {}", label, to),
+            Instr::PopJumpIfFalse { to, label }    => write!(f, "{} POP_JUMP_IF_FALSE {}", label, to),
+            Instr::Return { value, label }         => write!(f, "{} RETURN {}", label, value),
         }
     }
 }
\ No newline at end of file
diff --git a/blspc/src/vm/parser.rs b/blspc/src/vm/parser.rs
index c81fd0b..1bc597b 100644
--- a/blspc/src/vm/parser.rs
+++ b/blspc/src/vm/parser.rs
@@ -1,57 +1,60 @@
+use regex::Regex;
+
 use crate::vm::instr::*;
 
+const REGEX: &str = r###"[^\s\$";]+|\$"[^"]*"|;.*"###;
+
+macro_rules! value    { ($s:expr) => { $s.parse::<Type>().unwrap() }; }
+macro_rules! register { ($s:expr) => { $s.parse::<Register>().unwrap() }; }
+macro_rules! label    { ($s:expr) => { $s.parse::<usize>().unwrap() }; }
+
 pub fn parse_instr(src: &str) -> Vec<Instr> {
+    let regex = Regex::new(REGEX).unwrap();
     let mut result = Vec::new();
 
     for line in src.lines() {
-        // <label>: <instr> <arg>+
-        let mut parts = line.split_whitespace();
+        let tokens = regex.find_iter(line).map(|m| m.as_str()).collect::<Vec<_>>();
+        if tokens[0].starts_with(";") { continue; }
 
-        let label = parts.next();
-        if label == Some(";") { continue; }
-        let instr = parts.next();
-        
-        let mut args = Vec::new();
-        let mut in_quote = false;
-        let mut str = String::new();
-        while let Some(part) = parts.next() { match in_quote {
-                true => {
-                    if part.ends_with("\"") {
-                        str.push_str(&format!(" {}", part.trim_end_matches("\"")));
-                        args.push(str);
-                        str = String::new();
-                        in_quote = false;
-                    } else { str.push_str(&format!(" {}", part)); }
-                },
-                false => {
-                    if part.starts_with("$\"") {
-                        str.push_str(&part.trim_start_matches("$\""));
-                        in_quote = true;
-                    } else { args.push(part.to_string()); }
-                }
-            }
+        let label = label!(tokens[0]);
+
+        match tokens[1] {
+            "STORE" => { result.push(Instr::Store {
+                address: register!(tokens[2]),
+                value: value!(tokens[3]),
+                label,
+            }); },
+            "CALL" => { result.push(Instr::Call {
+                address: register!(tokens[2]),
+                args: register!(tokens[3]),
+                label,
+            }); },
+            "PUSH" => { result.push(Instr::Push {
+                value: value!(tokens[2]),
+                label,
+            }); },
+            "POP" => { result.push(Instr::Pop {
+                address: register!(tokens[2]),
+                label,
+            }); },
+            "ADD" => { result.push(Instr::Add { label }); },
+            "SUB" => { result.push(Instr::Sub { label }); },
+            "MUL" => { result.push(Instr::Mul { label }); },
+            "DIV" => { result.push(Instr::Div { label }); },
+            "JMP" => { result.push(Instr::Jump {
+                to: label!(tokens[2]),
+                label,
+            }); },
+            "POP_JUMP_IF_FALSE" => { result.push(Instr::PopJumpIfFalse {
+                to: label!(tokens[2]),
+                label,
+            }); },
+            "RETURN" => { result.push(Instr::Return {
+                value: register!(tokens[2]),
+                label,
+            }); },
+            _ => panic!("Unknown instruction: {}", tokens[1]),
         }
-
-        result.push(match instr {
-            Some("STORE") => {
-                let address = args[0].parse::<Register>().unwrap();
-                let value = args[1].parse::<Type>().unwrap();
-                let label = label.map(|l| l[..1].parse::<usize>().unwrap()).unwrap();
-                Instr::Store { address, value, label }
-            },
-            Some("CALL") => {
-                let address = args[0].parse::<Register>().unwrap();
-                let args = args[1].parse::<Register>().unwrap();
-                let label = label.map(|l| l[..1].parse::<usize>().unwrap()).unwrap();
-                Instr::Call { address, args, label }
-            },
-            Some("RETURN") => {
-                let value = args[0].parse::<Register>().unwrap();
-                let label = label.map(|l| l[..1].parse::<usize>().unwrap()).unwrap();
-                Instr::Return { value, label }
-            },
-            _ => panic!("Unknown instruction: {}", instr.unwrap())
-        });
     }
 
     result
diff --git a/blspc/example/hello.blsp b/example/hello.blsp
similarity index 100%
rename from blspc/example/hello.blsp
rename to example/hello.blsp
diff --git a/blspc/example/if.blsp b/example/if.blsp
similarity index 100%
rename from blspc/example/if.blsp
rename to example/if.blsp
diff --git a/blspc/example/math.blsp b/example/math.blsp
similarity index 100%
rename from blspc/example/math.blsp
rename to example/math.blsp
diff --git a/blspc/example/s.blsp b/example/s.blsp
similarity index 100%
rename from blspc/example/s.blsp
rename to example/s.blsp