Merge branch 'master' into range-builtin

Author: HomerMcMillan

Cargo.lock

@@ -65,9 +65,9 @@ pub enum Statement {
         value: Expression,
     },
     AugAssign {
-        target: Expression,
+        target: Box<Expression>,
         op: Operator,
-        value: Expression,
+        value: Box<Expression>,
     },
     Expression {
         expression: Expression,

parser/src/ast.rs

@@ -64,10 +64,7 @@ pub struct Location {
 
 impl Location {
     pub fn new(row: usize, column: usize) -> Self {
-        Location {
-            row: row,
-            column: column,
-        }
+        Location { row, column }
     }
 
     pub fn get_row(&self) -> usize {
@@ -126,8 +123,7 @@ pub type Spanned<Tok> = Result<(Location, Tok, Location), LexicalError>;
 pub fn make_tokenizer<'a>(source: &'a str) -> impl Iterator<Item = Spanned<Tok>> + 'a {
     let nlh = NewlineHandler::new(source.chars());
     let lch = LineContinationHandler::new(nlh);
-    let lexer = Lexer::new(lch);
-    lexer
+    Lexer::new(lch)
 }
 
 // The newline handler is an iterator which collapses different newline
@@ -144,7 +140,7 @@ where
 {
     pub fn new(source: T) -> Self {
         let mut nlh = NewlineHandler {
-            source: source,
+            source,
             chr0: None,
             chr1: None,
         };
@@ -200,7 +196,7 @@ where
 {
     pub fn new(source: T) -> Self {
         let mut nlh = LineContinationHandler {
-            source: source,
+            source,
             chr0: None,
             chr1: None,
         };
@@ -313,7 +309,7 @@ where
         if keywords.contains_key(&name) {
             Ok((start_pos, keywords.remove(&name).unwrap(), end_pos))
         } else {
-            Ok((start_pos, Tok::Name { name: name }, end_pos))
+            Ok((start_pos, Tok::Name { name }, end_pos))
         }
     }
 
@@ -358,7 +354,7 @@ where
 
         let end_pos = self.get_pos();
         let value = BigInt::from_str_radix(&value_text, radix).unwrap();
-        Ok((start_pos, Tok::Int { value: value }, end_pos))
+        Ok((start_pos, Tok::Int { value }, end_pos))
     }
 
     fn lex_normal_number(&mut self) -> Spanned<Tok> {
@@ -410,26 +406,19 @@ where
                 ))
             } else {
                 let end_pos = self.get_pos();
-                Ok((start_pos, Tok::Float { value: value }, end_pos))
+                Ok((start_pos, Tok::Float { value }, end_pos))
             }
         } else {
             // Parse trailing 'j':
             if self.chr0 == Some('j') {
                 self.next_char();
                 let end_pos = self.get_pos();
                 let imag = f64::from_str(&value_text).unwrap();
-                Ok((
-                    start_pos,
-                    Tok::Complex {
-                        real: 0.0,
-                        imag: imag,
-                    },
-                    end_pos,
-                ))
+                Ok((start_pos, Tok::Complex { real: 0.0, imag }, end_pos))
             } else {
                 let end_pos = self.get_pos();
                 let value = value_text.parse::<BigInt>().unwrap();
-                Ok((start_pos, Tok::Int { value: value }, end_pos))
+                Ok((start_pos, Tok::Int { value }, end_pos))
             }
         }
     }
@@ -548,33 +537,33 @@ where
             }
         };
 
-        return Ok((start_pos, tok, end_pos));
+        Ok((start_pos, tok, end_pos))
     }
 
     fn is_char(&self) -> bool {
         match self.chr0 {
-            Some('a'...'z') | Some('A'...'Z') | Some('_') | Some('0'...'9') => return true,
-            _ => return false,
+            Some('a'...'z') | Some('A'...'Z') | Some('_') | Some('0'...'9') => true,
+            _ => false,
         }
     }
 
     fn is_number(&self, radix: u32) -> bool {
         match radix {
             2 => match self.chr0 {
-                Some('0'...'1') => return true,
-                _ => return false,
+                Some('0'...'1') => true,
+                _ => false,
             },
             8 => match self.chr0 {
-                Some('0'...'7') => return true,
-                _ => return false,
+                Some('0'...'7') => true,
+                _ => false,
             },
             10 => match self.chr0 {
-                Some('0'...'9') => return true,
-                _ => return false,
+                Some('0'...'9') => true,
+                _ => false,
             },
             16 => match self.chr0 {
-                Some('0'...'9') | Some('a'...'f') | Some('A'...'F') => return true,
-                _ => return false,
+                Some('0'...'9') | Some('a'...'f') | Some('A'...'F') => true,
+                _ => false,
             },
             x => unimplemented!("Radix not implemented: {}", x),
         }

parser/src/lexer.rs

@@ -7,6 +7,7 @@ extern crate num_traits;
 pub mod ast;
 pub mod lexer;
 pub mod parser;
+#[cfg_attr(rustfmt, rustfmt_skip)]
 mod python;
 pub mod token;
 

parser/src/lib.rs

@@ -105,7 +105,11 @@ ExpressionStatement: ast::LocatedStatement = {
         let rhs = e2.into_iter().next().unwrap();
         ast::LocatedStatement {
             location: loc,
-            node: ast::Statement::AugAssign { target: expr, op: op, value: rhs },
+            node: ast::Statement::AugAssign {
+                target: Box::new(expr),
+                op,
+                value: Box::new(rhs)
+            },
         }
     },
 };

parser/src/python.lalrpop

@@ -70,7 +70,7 @@ fn main() {
 }
 
 fn _run_string(vm: &mut VirtualMachine, source: &str, source_path: Option<String>) -> PyResult {
-    let code_obj = compile::compile(vm, source, compile::Mode::Exec, source_path)?;
+    let code_obj = compile::compile(vm, source, &compile::Mode::Exec, source_path)?;
     // trace!("Code object: {:?}", code_obj.borrow());
     let builtins = vm.get_builtin_scope();
     let vars = vm.context().new_scope(Some(builtins)); // Keep track of local variables
@@ -115,7 +115,7 @@ fn run_script(vm: &mut VirtualMachine, script_file: &str) -> PyResult {
 }
 
 fn shell_exec(vm: &mut VirtualMachine, source: &str, scope: PyObjectRef) -> bool {
-    match compile::compile(vm, source, compile::Mode::Single, None) {
+    match compile::compile(vm, source, &compile::Mode::Single, None) {
         Ok(code) => {
             match vm.run_code_obj(code, scope) {
                 Ok(_value) => {

src/main.rs

@@ -45,6 +45,8 @@
 assert type(a) is complex
 assert type(a + a) is complex
 
+a = 1
+assert a.conjugate() == a
 
 a = 12345
 

tests/snippets/basic_types.py

@@ -46,3 +46,5 @@ def __bool__(self):
 assert False * 7 == 0
 assert True > 0
 assert int(True) == 1
+assert True.conjugate() == 1
+assert isinstance(True.conjugate(), int)

tests/snippets/bools.py

@@ -6,6 +6,8 @@
 y = (1,)
 assert y[0] == 1
 
+assert x + y == (1, 2, 1)
+
 assert x * 3 == (1, 2, 1, 2, 1, 2)
 # assert 3 * x == (1, 2, 1, 2, 1, 2)
 assert x * 0 == ()

tests/snippets/tuple.py

@@ -0,0 +1,33 @@
+def test_slice_bounds(s):
+    # End out of range
+    assert s[0:100] == s
+    assert s[0:-100] == '' 
+    # Start out of range
+    assert s[100:1] == ''
+    # Out of range both sides
+    # This is the behaviour in cpython
+    # assert s[-100:100] == s
+
+def expect_index_error(s, index):
+    try:
+        s[index]
+    except IndexError:
+        pass
+    else:
+        assert False
+
+unicode_str = "∀∂"
+assert unicode_str[0] == "∀"
+assert unicode_str[1] == "∂"
+assert unicode_str[-1] == "∂"
+
+test_slice_bounds(unicode_str)
+expect_index_error(unicode_str, 100)
+expect_index_error(unicode_str, -100)
+
+ascii_str = "hello world"
+test_slice_bounds(ascii_str)
+assert ascii_str[0] == "h"
+assert ascii_str[1] == "e"
+assert ascii_str[-1] == "d"
+

tests/snippets/unicode_slicing.py

@@ -17,4 +17,5 @@ serde_json = "1.0.26"
 byteorder = "1.2.6"
 regex = "1"
 statrs = "0.10.0"
-caseless = "0.2.1"
+caseless = "0.2.1"
+unicode-segmentation = "1.2.1"

vm/Cargo.toml

@@ -151,7 +151,7 @@ fn builtin_compile(vm: &mut VirtualMachine, args: PyFuncArgs) -> PyResult {
 
     let filename = objstr::get_value(filename);
 
-    compile::compile(vm, &source, mode, Some(filename))
+    compile::compile(vm, &source, &mode, Some(filename))
 }
 
 fn builtin_delattr(vm: &mut VirtualMachine, args: PyFuncArgs) -> PyResult {
@@ -203,6 +203,8 @@ fn builtin_enumerate(vm: &mut VirtualMachine, args: PyFuncArgs) -> PyResult {
     Ok(vm.ctx.new_list(new_items))
 }
 
+/// Implements `eval`.
+/// See also: https://docs.python.org/3/library/functions.html#eval
 fn builtin_eval(vm: &mut VirtualMachine, args: PyFuncArgs) -> PyResult {
     arg_check!(
         vm,
@@ -222,7 +224,7 @@ fn builtin_eval(vm: &mut VirtualMachine, args: PyFuncArgs) -> PyResult {
         let source = objstr::get_value(source);
         // TODO: fix this newline bug:
         let source = format!("{}\n", source);
-        compile::compile(vm, &source, mode, None)?
+        compile::compile(vm, &source, &mode, None)?
     } else {
         return Err(vm.new_type_error("code argument must be str or code object".to_string()));
     };
@@ -249,6 +251,8 @@ fn builtin_eval(vm: &mut VirtualMachine, args: PyFuncArgs) -> PyResult {
     vm.run_code_obj(code_obj.clone(), scope)
 }
 
+/// Implements `exec`
+/// https://docs.python.org/3/library/functions.html#exec
 fn builtin_exec(vm: &mut VirtualMachine, args: PyFuncArgs) -> PyResult {
     arg_check!(
         vm,
@@ -266,7 +270,7 @@ fn builtin_exec(vm: &mut VirtualMachine, args: PyFuncArgs) -> PyResult {
         let source = objstr::get_value(source);
         // TODO: fix this newline bug:
         let source = format!("{}\n", source);
-        compile::compile(vm, &source, mode, None)?
+        compile::compile(vm, &source, &mode, None)?
     } else if objtype::isinstance(source, &vm.ctx.code_type()) {
         source.clone()
     } else {