Add raw identifier

compiler/erg_compiler/codegen.rs

@@ -1648,11 +1648,6 @@ impl CodeGenerator {
                 self.emit_expr(*tasc.expr);
             }
             Expr::Import(acc) => self.emit_import(acc),
-            other => {
-                CompileError::feature_error(self.cfg.input.clone(), other.loc(), "??", "".into())
-                    .write_to_stderr();
-                self.crash("cannot compile this expression at this time");
-            }
         }
     }
 

compiler/erg_compiler/context/initialize/mod.rs

@@ -1641,6 +1641,11 @@ impl Context {
             NoneType,
         );
         let t_bin = nd_func(vec![kw("n", Int)], None, Str);
+        let t_bytes = nd_func(
+            vec![kw("str", Str), kw("encoding", Str)],
+            None,
+            mono("Bytes"),
+        );
         let t_chr = nd_func(
             vec![kw("i", Type::from(value(0usize)..=value(1_114_111usize)))],
             None,
@@ -1742,6 +1747,7 @@ impl Context {
         self.register_builtin_py_impl("ascii", t_ascii, Immutable, Private, Some("ascii"));
         self.register_builtin_impl("assert", t_assert, Const, Private); // assert casting に悪影響が出る可能性があるため、Constとしておく
         self.register_builtin_py_impl("bin", t_bin, Immutable, Private, Some("bin"));
+        self.register_builtin_py_impl("bytes", t_bytes, Immutable, Private, Some("bytes"));
         self.register_builtin_py_impl("chr", t_chr, Immutable, Private, Some("chr"));
         self.register_builtin_py_impl("classof", t_classof, Immutable, Private, Some("type"));
         self.register_builtin_py_impl("compile", t_compile, Immutable, Private, Some("compile"));

compiler/erg_compiler/context/register.rs

@@ -183,7 +183,7 @@ impl Context {
             self.impl_of(),
             py_name,
         );
-        log!(info "Registered {}::{}: {} {:?}", self.name, ident.name, vi.t, vi.impl_of);
+        log!(info "Registered {}::{}: {}", self.name, ident.name, vi);
         self.locals.insert(ident.name.clone(), vi);
         Ok(())
     }

compiler/erg_compiler/context/tyvar.rs

@@ -841,10 +841,6 @@ impl Context {
                 }
                 Ok(())
             }
-            hir::Expr::Decl(decl) => {
-                decl.t = self.deref_tyvar(mem::take(&mut decl.t), Covariant, decl.loc())?;
-                Ok(())
-            }
             hir::Expr::Def(def) => {
                 // It is not possible to further dereference the quantified type.
                 // TODO: However, it is possible that there are external type variables within the quantified type.

compiler/erg_compiler/hir.rs

@@ -3,6 +3,8 @@ use std::fmt;
 
 use erg_common::dict::Dict as HashMap;
 use erg_common::error::Location;
+#[allow(unused_imports)]
+use erg_common::log;
 use erg_common::traits::{Locational, NestedDisplay, Stream};
 use erg_common::vis::{Field, Visibility};
 use erg_common::Str;
@@ -547,10 +549,22 @@ impl Accessor {
 
     pub fn local_name(&self) -> Option<&str> {
         match self {
-            Self::Ident(ident) => ident.qual_name.as_ref().map(|s| {
-                let mut seps = s.split_with(&[".", "::"]);
-                seps.remove(seps.len() - 1)
-            }),
+            Self::Ident(ident) => ident
+                .qual_name
+                .as_ref()
+                .map(|s| {
+                    let mut seps = s.split_with(&[".", "::"]);
+                    seps.remove(seps.len() - 1)
+                })
+                .or_else(|| {
+                    let mut raw_parts = ident.name.inspect().split_with(&["'"]);
+                    // "'aaa'".split_with(&["'"]) == ["", "aaa", ""]
+                    if raw_parts.len() == 3 || raw_parts.len() == 4 {
+                        Some(raw_parts.remove(1))
+                    } else {
+                        Some(ident.name.inspect())
+                    }
+                }),
             _ => None,
         }
     }
@@ -1810,7 +1824,6 @@ pub enum Expr {
     UnaryOp(UnaryOp),
     Call(Call),
     Lambda(Lambda),
-    Decl(Decl),
     Def(Def),
     ClassDef(ClassDef),
     AttrDef(AttrDef),
@@ -1820,10 +1833,10 @@ pub enum Expr {
     Import(Accessor),
 }
 
-impl_nested_display_for_chunk_enum!(Expr; Lit, Accessor, Array, Tuple, Dict, Record, BinOp, UnaryOp, Call, Lambda, Decl, Def, ClassDef, AttrDef, Code, Compound, TypeAsc, Set, Import);
+impl_nested_display_for_chunk_enum!(Expr; Lit, Accessor, Array, Tuple, Dict, Record, BinOp, UnaryOp, Call, Lambda, Def, ClassDef, AttrDef, Code, Compound, TypeAsc, Set, Import);
 impl_display_from_nested!(Expr);
-impl_locational_for_enum!(Expr; Lit, Accessor, Array, Tuple, Dict, Record, BinOp, UnaryOp, Call, Lambda, Decl, Def, ClassDef, AttrDef, Code, Compound, TypeAsc, Set, Import);
-impl_t_for_enum!(Expr; Lit, Accessor, Array, Tuple, Dict, Record, BinOp, UnaryOp, Call, Lambda, Decl, Def, ClassDef, AttrDef, Code, Compound, TypeAsc, Set, Import);
+impl_locational_for_enum!(Expr; Lit, Accessor, Array, Tuple, Dict, Record, BinOp, UnaryOp, Call, Lambda, Def, ClassDef, AttrDef, Code, Compound, TypeAsc, Set, Import);
+impl_t_for_enum!(Expr; Lit, Accessor, Array, Tuple, Dict, Record, BinOp, UnaryOp, Call, Lambda, Def, ClassDef, AttrDef, Code, Compound, TypeAsc, Set, Import);
 
 impl Default for Expr {
     fn default() -> Self {

compiler/erg_compiler/lib/pystd/io.d.er

@@ -1,4 +1,13 @@
 .StringIO!: ClassType
-.StringIO!.getvalue!: (self: Ref(StringIO!),) => Str
+.StringIO! <: FileLike!
+.StringIO!.read!: (self: RefMut(.StringIO!), ) => Str
+.StringIO!.write!: (self: RefMut(.StringIO!), s: Str) => NoneType
+.StringIO!.getvalue!: (self: Ref(.StringIO!),) => Str
 
 .TextIOWrapper!: ClassType
+
+.BytesIO!: ClassType
+.BytesIO! <: FileLike!
+.BytesIO!.read!: (self: RefMut(.BytesIO!), ) => Bytes
+.BytesIO!.write!: (self: RefMut(.BytesIO!), b: Bytes) => NoneType
+.newBytesIO = 'BytesIO': (bytes: Bytes,) -> .BytesIO!

compiler/erg_compiler/lib/pystd/zipfile.d.er

@@ -1,7 +1,7 @@
 .ZipFile!: ClassType
 .ZipFile! <: FileLike!
 
-.ZipFile!.open!: (path: PathLike or NoneType := NoneType, mode := Str) => .ZipFile!
+.ZipFile!.open!: (path: PathLike or FileLike!, mode := Str) => .ZipFile!
 .ZipFile!.add!: (self: RefMut(.ZipFile!), name: PathLike, arcname: PathLike or NoneType := NoneType, recursive := Bool) => NoneType
 .ZipFile!.close!: (self: .ZipFile!,) => NoneType
 .ZipFile!.extractall!: (self: RefMut(.ZipFile!), path := PathLike, members: [Str; _] or NoneType := NoneType, numeric_owner := Bool) => NoneType

compiler/erg_compiler/link.rs

@@ -121,7 +121,6 @@ impl<'a> Linker<'a> {
                     self.resolve_pymod_path(&mut arg.expr);
                 }
             }
-            Expr::Decl(_decl) => {}
             Expr::Def(def) => {
                 for chunk in def.body.block.iter_mut() {
                     self.resolve_pymod_path(chunk);
@@ -238,7 +237,6 @@ impl<'a> Linker<'a> {
                     }
                 }
             },
-            Expr::Decl(_decl) => {}
             Expr::Def(def) => {
                 for chunk in def.body.block.iter_mut() {
                     self.replace_import(chunk);

compiler/erg_compiler/lower.rs

@@ -1515,9 +1515,11 @@ impl ASTLowerer {
             _ => unreachable!(),
         };
         let id = body.id;
-        self.ctx.assign_var_sig(&sig, found_body_t, id, py_name)?;
+        self.ctx
+            .assign_var_sig(&sig, found_body_t, id, py_name.clone())?;
         let mut ident = hir::Identifier::bare(ident.dot.clone(), ident.name.clone());
         ident.vi.t = found_body_t.clone();
+        ident.vi.py_name = py_name;
         let sig = hir::VarSignature::new(ident);
         let body = hir::DefBody::new(body.op, block, body.id);
         Ok(hir::Def::new(hir::Signature::Var(sig), body))
@@ -1648,6 +1650,10 @@ impl ASTLowerer {
         t: &Type,
         py_name: Str,
     ) -> LowerResult<()> {
+        // .X = 'x': Type
+        if ident.is_raw() {
+            return Ok(());
+        }
         if ident.is_const() {
             let vi = VarInfo::new(
                 t.clone(),
@@ -1693,6 +1699,9 @@ impl ASTLowerer {
     }
 
     fn declare_subtype(&mut self, ident: &ast::Identifier, trait_: &Type) -> LowerResult<()> {
+        if ident.is_raw() {
+            return Ok(());
+        }
         if let Some((_, ctx)) = self.ctx.get_mut_type(ident.inspect()) {
             ctx.register_marker_trait(trait_.clone());
             Ok(())

compiler/erg_compiler/varinfo.rs

@@ -130,8 +130,8 @@ impl fmt::Display for VarInfo {
     fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
         write!(
             f,
-            "VarInfo{{t: {}, muty: {:?}, vis: {:?}, kind: {:?}}}",
-            self.t, self.muty, self.vis, self.kind
+            "VarInfo{{t: {}, muty: {:?}, vis: {:?}, kind: {:?}, py_name: {:?}}}",
+            self.t, self.muty, self.vis, self.kind, self.py_name,
         )
     }
 }

compiler/erg_parser/ast.rs

@@ -2018,6 +2018,10 @@ impl VarName {
             .unwrap_or(false)
     }
 
+    pub fn is_raw(&self) -> bool {
+        self.0.content.starts_with('\'')
+    }
+
     pub const fn token(&self) -> &Token {
         &self.0
     }
@@ -2094,6 +2098,10 @@ impl Identifier {
         self.name.is_const()
     }
 
+    pub fn is_raw(&self) -> bool {
+        self.name.is_raw()
+    }
+
     pub const fn vis(&self) -> Visibility {
         match &self.dot {
             Some(_) => Visibility::Public,

compiler/erg_parser/lex.rs

@@ -795,6 +795,45 @@ impl Lexer /*<'a>*/ {
         ))
     }
 
+    fn lex_raw_ident(&mut self) -> LexResult<Token> {
+        let mut s = "\'".to_string();
+        while let Some(c) = self.peek_cur_ch() {
+            match c {
+                '\n' => {
+                    let token = self.emit_token(Illegal, &s);
+                    return Err(LexError::simple_syntax_error(line!() as usize, token.loc()));
+                }
+                '\'' => {
+                    s.push(self.consume().unwrap());
+                    if self.peek_cur_ch() == Some('!') {
+                        s.push(self.consume().unwrap());
+                    }
+                    let token = self.emit_token(Symbol, &s);
+                    return Ok(token);
+                }
+                _ => {
+                    let c = self.consume().unwrap();
+                    s.push(c);
+                    if Self::is_bidi(c) {
+                        return Err(self._invalid_unicode_character(&s));
+                    }
+                }
+            }
+        }
+        let token = self.emit_token(Illegal, &s);
+        Err(LexError::syntax_error(
+            0,
+            token.loc(),
+            switch_lang!(
+                "japanese" => "raw識別子が'によって閉じられていません",
+                "simplified_chinese" => "raw标识符没有被'关闭",
+                "traditional_chinese" => "raw標誌符沒有被'關閉",
+                "english" => "raw identifier is not closed by '",
+            ),
+            None,
+        ))
+    }
+
     // for single strings and multi-line strings
     fn _invalid_unicode_character(&mut self, s: &str) -> LexError {
         let token = self.emit_token(Illegal, s);
@@ -1117,7 +1156,34 @@ impl Iterator for Lexer /*<'a>*/ {
                 }
             }
             // TODO:
-            Some('\'') => self.deny_feature("'", "raw identifier"),
+            Some('\'') => {
+                let c = self.peek_cur_ch();
+                match c {
+                    None => {
+                        let token = self.emit_token(Illegal, "'");
+                        Some(Err(LexError::syntax_error(
+                            0,
+                            token.loc(),
+                            switch_lang!(
+                                "japanese" => "raw識別子が'によって閉じられていません",
+                                "simplified_chinese" => "raw識別子没被'关闭",
+                                "traditional_chinese" => "raw識別字沒被'關閉",
+                                "english" => "raw identifier is not ended with '",
+                            ),
+                            None,
+                        )))
+                    }
+                    Some(c) => {
+                        if c == '\'' {
+                            self.consume(); // consume second '\''
+                            let token = self.emit_token(Illegal, "\"\"");
+                            Some(Err(LexError::simple_syntax_error(0, token.loc())))
+                        } else {
+                            Some(self.lex_raw_ident())
+                        }
+                    }
+                }
+            }
             // Symbolized operators (シンボル化された演算子)
             // e.g. `-`(l, r) = l + (-r)
             Some('`') => {

examples/raw_ident.er

@@ -0,0 +1,6 @@
+t = 1
+'2t+3' = 2 * t + 3
+assert '2t+3' == 5
+
+'echo'! = print!
+# 'echo' = print! is not permitted

tests/test.rs

@@ -82,6 +82,11 @@ fn exec_quantified() -> Result<(), ()> {
     expect_success("examples/quantified.er")
 }
 
+#[test]
+fn exec_raw_ident() -> Result<(), ()> {
+    expect_success("examples/raw_ident.er")
+}
+
 #[test]
 fn exec_rec() -> Result<(), ()> {
     // this script is valid but the current code generating process has a bug.