Add instruction dumper

A vector of instructions and constants can now be decompiled as text on
the terminal to give an idea of what the VM is doing.

Signed-off-by: Alek Ratzloff <alekratz@gmail.com>

src/bin/not.rs

@@ -32,13 +32,14 @@ fn main() -> Result<()> {
             return Err("errors reported, exiting".into());
         }
     }.unwrap();
-    //println!("{:#?}", ast);
+    println!("{:#?}", ast);
 
     let mut compile = Compile::new();
     let (inst, const_pool) = compile.compile(&ast)?;
 
-    println!("{:#?}", inst);
+    //println!("{:#?}", inst);
-    println!("{:#?}", const_pool);
+    //println!("{:#?}", const_pool);
+    not_python::vm::inst::dump_inst_body(&inst, &const_pool);
 
     Ok(())
 }

src/compile/mod.rs

@@ -11,6 +11,7 @@ pub struct Compile {
     const_data: ConstData,
     globals: BTreeMap<Sym, Local>,
     locals: Vec<BTreeMap<Sym, Local>>,
+    names: Vec<Sym>,
     next_local: Local,
 }
 
@@ -63,6 +64,15 @@ impl Compile {
             .copied()
     }
 
+    /// Looks up a variable name, or creates a global name if it doesn't exist.
+    pub fn lookup_scope_or_create_global(&mut self, sym: Sym) -> Local {
+        if let Some(local) = self.lookup_scope(sym) {
+            local
+        } else {
+            self.create_global(sym)
+        }
+    }
+
     /// Creates a new local variable if it does not exist in the current local scope.
     pub(crate) fn create_local(&mut self, sym: Sym) -> Local {
         let locals = self.locals.last_mut().expect("scope");
@@ -73,6 +83,8 @@ impl Compile {
             let local = self.next_local;
             self.next_local = self.next_local.next();
             locals.insert(sym, local);
+            self.names.push(sym);
+            assert_eq!(self.names.len(), self.next_local.index());
             local
         }
     }
@@ -85,6 +97,8 @@ impl Compile {
             let global = self.next_local;
             self.next_local = self.next_local.next();
             self.globals.insert(sym, global);
+            self.names.push(sym);
+            assert_eq!(self.names.len(), self.next_local.index());
             global
         }
     }

src/compile/thunk.rs

@@ -264,8 +264,9 @@ impl Visit for CompileBody<'_> {
                 let attr = global_sym(expr.access.to_string());
                 thunk.push(Inst::SetAttr(attr));
             }
-            LhsExpr::Local(local) => {
+            LhsExpr::Local(local_name) => {
-                let local = global_sym(local.to_string());
+                let sym = global_sym(local_name.to_string());
+                let local = self.compile.lookup_scope_or_create_global(sym);
                 thunk = Inst::Pop(Some(local)).into();
             }
         }
@@ -320,6 +321,7 @@ impl Visit for CompileBody<'_> {
         for arg in expr.args.iter() {
             thunk.push_thunk(self.visit_expr(&arg)?);
         }
+        thunk.push(Inst::Call(expr.args.len()));
         Ok(thunk)
     }
 
@@ -347,13 +349,7 @@ impl Visit for CompileBody<'_> {
         let thunk = match atom {
             Atom::Ident(ident) => {
                 let sym = global_sym(ident.to_string());
-                let local = if let Some(local) = self.compile.lookup_scope(sym) {
+                let local = self.compile.lookup_scope_or_create_global(sym);
-                    local
-                } else {
-                    // create a global that gets looked up instead, since nothing with this name
-                    // has been declared/assigned in this scope
-                    self.compile.create_global(sym)
-                };
                 // get local
                 Inst::PushLocal(local).into()
             }

src/obj/sym.rs

@@ -1,6 +1,6 @@
 use crate::obj::{intern::Interner, names::*, prelude::*};
 use once_cell::sync::Lazy;
-use std::{collections::BTreeMap, sync::Mutex};
+use std::{collections::BTreeMap, sync::{Arc, Mutex}};
 
 //
 // struct Sym
@@ -93,6 +93,12 @@ pub fn global_sym(s: String) -> Sym {
     table.insert(s)
 }
 
+/// Gets the pointer to a name from a global symbol.
+pub fn global_sym_lookup(sym: Sym) -> Option<Arc<String>> {
+    let table = SYMS.lock().unwrap();
+    table.lookup(sym)
+}
+
 //
 // SymTable interner types
 //

src/vm/inst.rs

@@ -1,4 +1,5 @@
-use crate::{obj::prelude::*, vm::consts::ConstHandle};
+use crate::{obj::prelude::*, vm::consts::*};
+use std::fmt::{self, Display, Formatter};
 
 #[derive(Debug, PartialEq, Clone, Copy)]
 pub enum Inst {
@@ -12,7 +13,7 @@ pub enum Inst {
     PushLocal(Local),
 
     /// Pop a value from the stack, possibly into a local symbol.
-    Pop(Option<Sym>),
+    Pop(Option<Local>),
 
     /// Pops a symbol value and an object reference.
     ///
@@ -109,3 +110,125 @@ pub enum Inst {
     /// pushing the result to the stack.
     BinOr,
 }
+
+//
+// impl Inst
+//
+impl Inst {
+    pub fn name(&self) -> &'static str {
+        match self {
+            Inst::PushSym(_) => "PUSH_SYM",
+            Inst::PushConst(_) => "PUSH_CONST",
+            Inst::PushLocal(_) => "PUSH_LOCAL",
+            Inst::Pop(_) => "POP",
+            Inst::GetAttr(_) => "GET_ATTR",
+            Inst::SetAttr(_) => "SET_ATTR",
+            Inst::Jump(_) => "JUMP",
+            Inst::JumpTrue(_) => "JUMP_TRUE",
+            Inst::Call(_) => "CALL",
+            Inst::Index => "INDEX",
+            Inst::Return => "RETURN",
+            Inst::UnNeg => "UN_NEG",
+            Inst::UnPos => "UN_POS",
+            Inst::BinPlus => "BIN_PLUS",
+            Inst::BinMinus => "BIN_MINUS",
+            Inst::BinMul => "BIN_MUL",
+            Inst::BinDiv => "BIN_DIV",
+            Inst::BinEq => "BIN_EQ",
+            Inst::BinNeq => "BIN_NEQ",
+            Inst::BinLt => "BIN_LT",
+            Inst::BinLe => "BIN_LE",
+            Inst::BinGt => "BIN_GT",
+            Inst::BinGe => "BIN_GE",
+            Inst::BinAnd => "BIN_AND",
+            Inst::BinOr => "BIN_OR",
+        }
+    }
+}
+
+pub fn dump_inst_body(insts: &Vec<Inst>, const_pool: &ConstPool) {
+    println!("{}", InstFormatter::new(insts, const_pool))
+}
+
+//
+// struct InstFormatter
+//
+
+pub struct InstFormatter<'i, 'c> {
+    insts: &'i Vec<Inst>,
+    const_pool: &'c ConstPool,
+}
+
+impl<'i, 'c> InstFormatter<'i, 'c> {
+    pub fn new(insts: &'i Vec<Inst>, const_pool: &'c ConstPool) -> Self {
+        Self { insts, const_pool, }
+    }
+}
+
+impl Display for InstFormatter<'_, '_> {
+    fn fmt(&self, fmt: &mut Formatter) -> fmt::Result {
+        // column widths
+        let addr_w = num_digits(self.insts.len(), 16).max(4);
+        let inst_col = 16 - addr_w;
+        let inst_w = 16;
+
+        for (addr, inst) in self.insts.iter().enumerate() {
+            let (param_val, mut param_name) = match inst {
+                Inst::PushSym(sym) | Inst::GetAttr(sym) | Inst::SetAttr(sym) => (
+                    sym.index().to_string(),
+                    global_sym_lookup(*sym).unwrap().to_string(),
+                ),
+                Inst::PushConst(hdl) => (
+                    hdl.index().to_string(),
+                    {
+                        let obj_ref = self.const_pool.get(*hdl);
+                        read_obj!(let obj = obj_ref);
+                        // XXX weirdness with coercion, can't deref as a &dyn Obj because
+                        // RwReadLockGuard is not Obj - but using Deref::deref works
+                        let obj: &dyn Obj = std::ops::Deref::deref(obj);
+                        format!("{:?}", obj)
+                    },
+                ),
+                Inst::PushLocal(local) => (
+                    local.index().to_string(),
+                    "TODO: local name".to_string(),
+                ),
+                Inst::Pop(local) => (
+                    format!("{:?}", local),
+                    "TODO: local name".to_string(),
+                ),
+                Inst::Jump(addr) | Inst::JumpTrue(addr) => (
+                    addr.to_string(),
+                    String::new(),
+                ),
+                Inst::Call(argc) => (
+                    argc.to_string(),
+                    String::new(),
+                ),
+                _ => (String::new(), String::new()),
+            };
+
+            if !param_name.is_empty() {
+                param_name = format!("({})", param_name);
+            }
+
+            writeln!(
+                fmt,
+                "{:0addr_w$x}{space: >inst_col$}{: <inst_w$}{} {}",
+                addr,
+                inst.name(),
+                param_val,
+                param_name,
+                space = "",
+                addr_w = addr_w,
+                inst_w = inst_w,
+                inst_col = inst_col,
+            )?;
+        }
+        Ok(())
+    }
+}
+
+fn num_digits(n: usize, radix: usize) -> usize {
+    ((n as f64) + 1.0).log(radix as f64).ceil() as usize
+}