Merge pull request #1567 from HyeockJinKim/issue1527

Modify str to float

tests/snippets/floats.py

@@ -304,3 +304,15 @@ assert float(1e-5).__repr__() == "1e-05"
 assert float(1e-4).__repr__() == "0.0001"
 assert float(1.2345678901234567890).__repr__() == "1.2345678901234567"
 assert float(1.2345678901234567890e308).__repr__() == "1.2345678901234567e+308"
+
+assert float('0_0') == 0.0
+assert float('.0') == 0.0
+assert float('0.') == 0.0
+
+assert_raises(ValueError, lambda: float('0._0'))
+assert_raises(ValueError, lambda: float('0_.0'))
+assert_raises(ValueError, lambda: float('._0'))
+assert_raises(ValueError, lambda: float('0_'))
+assert_raises(ValueError, lambda: float('0._'))
+assert_raises(ValueError, lambda: float('_.0'))
+assert_raises(ValueError, lambda: float('._0'))

vm/src/obj/objfloat.rs

@@ -597,31 +597,56 @@ impl PyFloat {
     }
 }
 
+fn str_to_float(vm: &VirtualMachine, literal: &str) -> PyResult<f64> {
+    if literal.starts_with('_') || literal.ends_with('_') {
+        return Err(invalid_convert(vm, literal));
+    }
+
+    let mut buf = String::with_capacity(literal.len());
+    let mut last_tok: Option<char> = None;
+    for c in literal.chars() {
+        if !(c.is_ascii_alphanumeric() || c == '_' || c == '+' || c == '-' || c == '.') {
+            return Err(invalid_convert(vm, literal));
+        }
+
+        if !c.is_ascii_alphanumeric() {
+            if let Some(l) = last_tok {
+                if !l.is_ascii_alphanumeric() {
+                    return Err(invalid_convert(vm, literal));
+                }
+            }
+        }
+
+        if c != '_' {
+            buf.push(c);
+        }
+        last_tok = Some(c);
+    }
+
+    if let Ok(f) = lexical::parse(buf.as_str()) {
+        Ok(f)
+    } else {
+        Err(invalid_convert(vm, literal))
+    }
+}
+
+fn invalid_convert(vm: &VirtualMachine, literal: &str) -> PyObjectRef {
+    vm.new_value_error(format!("could not convert string to float: '{}'", literal))
+}
+
 fn to_float(vm: &VirtualMachine, obj: &PyObjectRef) -> PyResult<f64> {
     let value = if objtype::isinstance(&obj, &vm.ctx.float_type()) {
         get_value(&obj)
     } else if objtype::isinstance(&obj, &vm.ctx.int_type()) {
         objint::get_float_value(&obj, vm)?
     } else if objtype::isinstance(&obj, &vm.ctx.str_type()) {
-        match lexical::parse(objstr::get_value(&obj).trim()) {
-            Ok(f) => f,
-            Err(_) => {
-                let arg_repr = vm.to_pystr(obj)?;
-                return Err(vm.new_value_error(format!(
-                    "could not convert string to float: '{}'",
-                    arg_repr
-                )));
-            }
-        }
+        str_to_float(vm, objstr::get_value(&obj).trim())?
     } else if objtype::isinstance(&obj, &vm.ctx.bytes_type()) {
         match lexical::parse(objbytes::get_value(&obj).as_slice()) {
             Ok(f) => f,
             Err(_) => {
                 let arg_repr = vm.to_pystr(obj)?;
-                return Err(vm.new_value_error(format!(
-                    "could not convert string to float: '{}'",
-                    arg_repr
-                )));
+                return Err(invalid_convert(vm, arg_repr.as_str()));
             }
         }
     } else {