RustPython · youknowone · Nov 2, 2021 · Sep 13, 2021 · Sep 13, 2021 · Sep 14, 2021
diff --git a/Cargo.lock b/Cargo.lock
diff --git a/Cargo.toml b/Cargo.toml
@@ -32,7 +32,7 @@ ssl-vendor = ["rustpython-stdlib/ssl-vendor"]
 
 [dependencies]
 log = "0.4"
-env_logger = { version = "0.8", default-features = false, features = ["atty", "termcolor"] }
+env_logger = { version = "0.9.0", default-features = false, features = ["atty", "termcolor"] }
 clap = "2.33"
 rustpython-compiler = { path = "compiler/porcelain", version = "0.1.1" }
 rustpython-parser = { path = "parser", version = "0.1.1" }
@@ -50,8 +50,8 @@ flamescope = { version = "0.1", optional = true }
 rustyline = "9"
 
 [dev-dependencies]
-cpython = "0.6"
-python3-sys = "0.6"
+cpython = "0.7.0"
+python3-sys = "0.7.0"
 criterion = "0.3"
 
 [[bench]]

diff --git a/Lib/test/test_float.py b/Lib/test/test_float.py
@@ -89,8 +89,6 @@ def test_underscores(self):
         # Check that we handle bytes values correctly.
         self.assertRaises(ValueError, float, b'0_.\xff9')
 
-    # TODO: RUSTPYTHON
-    @unittest.expectedFailure
     def test_non_numeric_input_types(self):
         # Test possible non-numeric types for the argument x, including
         # subclasses of the explicitly documented accepted types.
@@ -120,8 +118,6 @@ class CustomByteArray(bytearray): pass
                 with self.assertRaisesRegex(ValueError, "could not convert"):
                     float(f(b'A' * 0x10))
 
-    # TODO: RUSTPYTHON
-    @unittest.expectedFailure
     def test_float_memoryview(self):
         self.assertEqual(float(memoryview(b'12.3')[1:4]), 2.3)
         self.assertEqual(float(memoryview(b'12.3\x00')[1:4]), 2.3)

diff --git a/Lib/test/test_runpy.py b/Lib/test/test_runpy.py
@@ -676,6 +676,7 @@ def test_basic_script_no_suffix(self):
 
     # TODO: RUSTPYTHON
     @unittest.expectedFailure
+    @unittest.skipIf(sys.platform == 'win32', "TODO: RUSTPYTHON; weird panic in lz4-flex")
     def test_script_compiled(self):
         with temp_dir() as script_dir:
             mod_name = 'script'

diff --git a/bytecode/Cargo.toml b/bytecode/Cargo.toml
@@ -9,9 +9,9 @@ license = "MIT"
 
 
 [dependencies]
-bincode =  "1.1"
-bitflags = "1.1"
-lz4_flex = "0.7"
+bincode = "1.3.3"
+bitflags = "1.2.1"
+lz4_flex = "0.9.0"
 num-bigint = { version = "0.4.2", features = ["serde"] }
 num-complex = { version = "0.4.0", features = ["serde"] }
 serde = { version = "1.0", features = ["derive"] }

diff --git a/common/Cargo.toml b/common/Cargo.toml
@@ -13,8 +13,8 @@ parking_lot = { version = "0.11.0", optional = true }
 num-traits = "0.2"
 num-complex = "0.4.0"
 num-bigint = "0.4.2"
-lexical-core = "0.7"
-hexf-parse = "0.1.0"
+lexical-parse-float = { version = "0.8.0", features = ["format"] }
+hexf-parse = "0.2.1"
 cfg-if = "1.0"
 once_cell = "1.4.1"
 siphasher = "0.3"

diff --git a/common/src/float_ops.rs b/common/src/float_ops.rs
@@ -64,36 +64,36 @@ pub fn gt_int(value: f64, other_int: &BigInt) -> bool {
 }
 
 pub fn parse_str(literal: &str) -> Option<f64> {
-    if literal.starts_with('_') || literal.ends_with('_') {
-        return None;
-    }
-
-    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 None;
-        }
+    parse_inner(literal.trim().as_bytes())
+}
 
-        if !c.is_ascii_alphanumeric() {
-            if let Some(l) = last_tok {
-                if !l.is_ascii_alphanumeric() && !(c == '.' && (l == '-' || l == '+')) {
-                    return None;
-                }
-            }
-        }
+pub fn parse_bytes(literal: &[u8]) -> Option<f64> {
+    parse_inner(trim_slice(literal, |b| b.is_ascii_whitespace()))
+}
 
-        if c != '_' {
-            buf.push(c);
-        }
-        last_tok = Some(c);
+fn trim_slice<T>(v: &[T], mut trim: impl FnMut(&T) -> bool) -> &[T] {
+    let mut it = v.iter();
+    // it.take_while_ref(&mut trim).for_each(drop);
+    // hmm.. `&mut slice::Iter<_>` is not `Clone`
+    // it.by_ref().rev().take_while_ref(&mut trim).for_each(drop);
+    while it.clone().next().map_or(false, &mut trim) {
+        it.next();
     }
-
-    if let Ok(f) = lexical_core::parse(buf.as_bytes()) {
-        Some(f)
-    } else {
-        None
+    while it.clone().next_back().map_or(false, &mut trim) {
+        it.next_back();
     }
+    it.as_slice()
+}
+
+fn parse_inner(literal: &[u8]) -> Option<f64> {
+    use lexical_parse_float::{
+        format::PYTHON3_LITERAL, FromLexicalWithOptions, NumberFormatBuilder, Options,
+    };
+    // lexical-core's format::PYTHON_STRING is inaccurate
+    const PYTHON_STRING: u128 = NumberFormatBuilder::rebuild(PYTHON3_LITERAL)
+        .no_special(false)
+        .build();
+    f64::from_lexical_with_options::<PYTHON_STRING>(literal, &Options::new()).ok()
 }
 
 pub fn is_integer(v: f64) -> bool {

diff --git a/compiler/Cargo.toml b/compiler/Cargo.toml
@@ -8,7 +8,7 @@ license = "MIT"
 edition = "2021"
 
 [dependencies]
-indexmap = "1.0"
+indexmap = "1.7.0"
 itertools = "0.10.0"
 rustpython-bytecode = { path = "../bytecode", version = "0.1.1" }
 rustpython-ast = { path = "../ast", features = ["unparse"] }

diff --git a/derive/Cargo.toml b/derive/Cargo.toml
@@ -19,6 +19,6 @@ rustpython-compiler = { path = "../compiler/porcelain", version = "0.1.1" }
 rustpython-bytecode = { path = "../bytecode", version = "0.1.1" }
 maplit = "1.0"
 once_cell = "1.8.0"
-textwrap = "0.13.4"
+textwrap = { version = "0.14.2", default-features = false }
 indexmap = "^1"
 serde_json = "1.0.68"
diff --git a/jit/Cargo.toml b/jit/Cargo.toml
@@ -10,16 +10,16 @@ edition = "2021"
 autotests = false
 
 [dependencies]
-cranelift = "0.72.0"
-cranelift-module = "0.72.0"
-cranelift-jit = "0.72.0"
+cranelift = "0.76.0"
+cranelift-module = "0.76.0"
+cranelift-jit = "0.76.0"
 num-traits = "0.2"
-libffi = "1.0"
+libffi = "2.0.0"
 rustpython-bytecode = { path = "../bytecode", version = "0.1.2" }
 thiserror = "1.0"
 
 [dev-dependencies]
-approx = "0.4.0"
+approx = "0.5.0"
 rustpython-derive = { path = "../derive", version = "0.1.2" }
 
 [[test]]

diff --git a/jit/src/instructions.rs b/jit/src/instructions.rs
@@ -288,12 +288,7 @@ impl<'a, 'b> FunctionCompiler<'a, 'b> {
                         UnaryOperator::Minus => {
                             // Compile minus as 0 - a.
                             let zero = self.builder.ins().iconst(types::I64, 0);
-                            let (out, carry) = self.builder.ins().isub_ifbout(zero, a.val);
-                            self.builder.ins().trapif(
-                                IntCC::Overflow,
-                                carry,
-                                TrapCode::IntegerOverflow,
-                            );
+                            let out = self.compile_sub(zero, a.val);
                             self.stack.push(JitValue {
                                 val: out,
                                 ty: JitType::Int,
@@ -342,12 +337,7 @@ impl<'a, 'b> FunctionCompiler<'a, 'b> {
                             Ok(())
                         }
                         BinaryOperator::Subtract => {
-                            let (out, carry) = self.builder.ins().isub_ifbout(a.val, b.val);
-                            self.builder.ins().trapif(
-                                IntCC::Overflow,
-                                carry,
-                                TrapCode::IntegerOverflow,
-                            );
+                            let out = self.compile_sub(a.val, b.val);
                             self.stack.push(JitValue {
                                 val: out,
                                 ty: JitType::Int,
@@ -397,4 +387,19 @@ impl<'a, 'b> FunctionCompiler<'a, 'b> {
             _ => Err(JitCompileError::NotSupported),
         }
     }
+
+    fn compile_sub(&mut self, a: Value, b: Value) -> Value {
+        // TODO: this should be fine, but cranelift doesn't special-case isub_ifbout
+        // let (out, carry) = self.builder.ins().isub_ifbout(a, b);
+        // self.builder
+        //     .ins()
+        //     .trapif(IntCC::Overflow, carry, TrapCode::IntegerOverflow);
+        // TODO: this shouldn't wrap
+        let neg_b = self.builder.ins().ineg(b);
+        let (out, carry) = self.builder.ins().iadd_ifcout(a, neg_b);
+        self.builder
+            .ins()
+            .trapif(IntCC::Overflow, carry, TrapCode::IntegerOverflow);
+        out
+    }
 }
diff --git a/jit/src/lib.rs b/jit/src/lib.rs
@@ -10,6 +10,7 @@ use rustpython_bytecode as bytecode;
 mod instructions;
 
 use instructions::FunctionCompiler;
+use std::mem::ManuallyDrop;
 
 #[derive(Debug, thiserror::Error)]
 pub enum JitCompileError {
@@ -82,8 +83,12 @@ impl Jit {
             &self.ctx.func.signature,
         )?;
 
-        self.module
-            .define_function(id, &mut self.ctx, &mut codegen::binemit::NullTrapSink {})?;
+        self.module.define_function(
+            id,
+            &mut self.ctx,
+            &mut codegen::binemit::NullTrapSink {},
+            &mut codegen::binemit::NullStackMapSink {},
+        )?;
 
         self.module.clear_context(&mut self.ctx);
 
@@ -105,14 +110,14 @@ pub fn compile<C: bytecode::Constant>(
     Ok(CompiledCode {
         sig,
         code,
-        module: jit.module,
+        module: ManuallyDrop::new(jit.module),
     })
 }
 
 pub struct CompiledCode {
     sig: JitSig,
     code: *const u8,
-    module: JITModule,
+    module: ManuallyDrop<JITModule>,
 }
 
 impl CompiledCode {
@@ -287,7 +292,7 @@ unsafe impl Sync for CompiledCode {}
 impl Drop for CompiledCode {
     fn drop(&mut self) {
         // SAFETY: The only pointer that this memory will also be dropped now
-        unsafe { self.module.free_memory() }
+        unsafe { ManuallyDrop::take(&mut self.module).free_memory() }
     }
 }
 

diff --git a/parser/Cargo.toml b/parser/Cargo.toml
@@ -20,8 +20,8 @@ num-traits = "0.2"
 unic-emoji-char = "0.9"
 unic-ucd-ident  = "0.9"
 unicode_names2 = "0.4"
-phf = { version = "0.9", features = ["macros"] }
+phf = { version = "0.10.0", features = ["macros"] }
 ahash = "0.7.2"
 
 [dev-dependencies]
-insta = "1.5"
+insta = "1.8.0"
diff --git a/stdlib/Cargo.toml b/stdlib/Cargo.toml
@@ -48,9 +48,8 @@ num-integer = "0.1.44"
 
 crossbeam-utils = "0.8.5"
 itertools = "0.10.1"
-lexical-core = "0.7"
+lexical-parse-float = "0.8.0"
 num-traits = "0.2.14"
-crc = "^1.0.0"
 memchr = "2"
 base64 = "0.13"
 csv-core = "0.1"

diff --git a/stdlib/src/array.rs b/stdlib/src/array.rs
@@ -33,7 +33,7 @@ mod array {
     };
     use crossbeam_utils::atomic::AtomicCell;
     use itertools::Itertools;
-    use lexical_core::Integer;
+    use num_traits::ToPrimitive;
     use std::cmp::Ordering;
     use std::{fmt, os::raw};
 
@@ -1411,7 +1411,8 @@ mod array {
                     ))
                 })?
                 .try_to_primitive::<i32>(vm)?
-                .try_u8_or_max()
+                .to_u8()
+                .unwrap_or(u8::MAX)
                 .try_into()
                 .map_err(|_| {
                     vm.new_value_error("third argument must be a valid machine format code.".into())

diff --git a/stdlib/src/binascii.rs b/stdlib/src/binascii.rs
@@ -1,13 +1,14 @@
 pub(crate) use decl::make_module;
 
+pub(super) use decl::crc32;
+
 #[pymodule(name = "binascii")]
 mod decl {
     use crate::vm::{
-        builtins::PyTypeRef,
+        builtins::{PyIntRef, PyTypeRef},
         function::{ArgAsciiBuffer, ArgBytesLike, OptionalArg},
         PyResult, VirtualMachine,
     };
-    use crc::{crc32, Hasher32};
     use itertools::Itertools;
 
     #[pyattr(name = "Error")]
@@ -94,13 +95,14 @@ mod decl {
     }
 
     #[pyfunction]
-    fn crc32(data: ArgBytesLike, value: OptionalArg<u32>) -> u32 {
-        let crc = value.unwrap_or(0);
-
-        let mut digest = crc32::Digest::new_with_initial(crc32::IEEE, crc);
-        data.with_ref(|bytes| digest.write(bytes));
+    pub(crate) fn crc32(data: ArgBytesLike, init: OptionalArg<PyIntRef>) -> u32 {
+        let init = init.map_or(0, |i| i.as_u32_mask());
 
-        digest.sum32()
+        let mut hasher = crc32fast::Hasher::new_with_initial(init);
+        data.with_ref(|bytes| {
+            hasher.update(bytes);
+            hasher.finalize()
+        })
     }
 
     #[derive(FromArgs)]

diff --git a/stdlib/src/zlib.rs b/stdlib/src/zlib.rs
@@ -9,7 +9,6 @@ mod zlib {
         IntoPyRef, PyResult, PyValue, VirtualMachine,
     };
     use adler32::RollingAdler32 as Adler32;
-    use crc32fast::Hasher as Crc32;
     use crossbeam_utils::atomic::AtomicCell;
     use flate2::{
         write::ZlibEncoder, Compress, Compression, Decompress, FlushCompress, FlushDecompress,
@@ -74,13 +73,7 @@ mod zlib {
     /// Compute a CRC-32 checksum of data.
     #[pyfunction]
     fn crc32(data: ArgBytesLike, begin_state: OptionalArg<PyIntRef>) -> u32 {
-        data.with_ref(|data| {
-            let begin_state = begin_state.map_or(0, |i| i.as_u32_mask());
-
-            let mut hasher = Crc32::new_with_initial(begin_state);
-            hasher.update(data);
-            hasher.finalize()
-        })
+        crate::binascii::crc32(data, begin_state)
     }
 
     fn compression_from_int(level: Option<i32>) -> Option<Compression> {