Implement keyword suggestion routine

`suggestions.rs` is almost porting of implementation of
[this](python/cpython#16856) and
[this](python/cpython#25397).

Signed-off-by: snowapril <[email protected]>

Author: Snowapril

common/src/str.rs

@@ -1,7 +1,9 @@
 use ascii::AsciiString;
 use once_cell::unsync::OnceCell;
+use std::cell::RefCell;
 use std::fmt;
 use std::ops::{Bound, RangeBounds};
+use std::thread_local;
 
 #[cfg(not(target_arch = "wasm32"))]
 #[allow(non_camel_case_types)]
@@ -98,6 +100,97 @@ pub const fn bytes_is_ascii(x: &str) -> bool {
     true
 }
 
+fn substitution_cost(mut a: u8, mut b: u8) -> usize {
+    if (a & 31) != (b & 31) {
+        return 2; // MOVE COST
+    }
+    if a == b {
+        return 0;
+    }
+    if (b'A'..=b'Z').contains(&a) {
+        a += b'a' - b'A';
+    }
+    if (b'A'..=b'Z').contains(&b) {
+        b += b'a' - b'A';
+    }
+    if a == b {
+        1 // CASE COST
+    } else {
+        2 // MOVE COST
+    }
+}
+
+pub fn levenshtein_distance(a: &str, b: &str, max_cost: usize) -> usize {
+    const MAX_STRING_SIZE: usize = 40;
+    thread_local! {
+        static BUFFER: RefCell<[usize; MAX_STRING_SIZE]> = RefCell::new([0usize; MAX_STRING_SIZE]);
+    }
+
+    if a == b {
+        return 0;
+    }
+
+    let (mut a_bytes, mut b_bytes) = (a.as_bytes(), b.as_bytes());
+    let (mut a_begin, mut a_end) = (0usize, a.len());
+    let (mut b_begin, mut b_end) = (0usize, b.len());
+
+    while a_end > 0 && b_end > 0 && (a_bytes[a_begin] == b_bytes[b_begin]) {
+        a_begin += 1;
+        b_begin += 1;
+        a_end -= 1;
+        b_end -= 1;
+    }
+    while a_end > 0 && b_end > 0 && (a_bytes[a_begin + a_end - 1] == b_bytes[b_begin + b_end - 1]) {
+        a_end -= 1;
+        b_end -= 1;
+    }
+    if a_end == 0 || b_end == 0 {
+        return (a_end + b_end) * 2;
+    }
+    if a_end > MAX_STRING_SIZE || b_end > MAX_STRING_SIZE {
+        return max_cost + 1;
+    }
+
+    if b_end < a_end {
+        std::mem::swap(&mut a_bytes, &mut b_bytes);
+        std::mem::swap(&mut a_begin, &mut b_begin);
+        std::mem::swap(&mut a_end, &mut b_end);
+    }
+
+    if (b_end - a_end) * 2 > max_cost {
+        return max_cost + 1;
+    }
+
+    BUFFER.with(|buffer| {
+        let mut buffer = buffer.borrow_mut();
+        for i in 0..a_end {
+            buffer[i] = (i + 1) * 2;
+        }
+
+        let mut result = 0usize;
+        for (b_index, b_code) in b_bytes[b_begin..(b_begin + b_end)].iter().enumerate() {
+            result = b_index * 2;
+            let mut distance = result;
+            let mut minimum = usize::MAX;
+            for (a_index, a_code) in a_bytes[a_begin..(a_begin + a_end)].iter().enumerate() {
+                let substitute = distance + substitution_cost(*b_code, *a_code);
+                distance = buffer[a_index];
+                let insert_delete = usize::min(result, distance) + 2;
+                result = usize::min(insert_delete, substitute);
+
+                buffer[a_index] = result;
+                if result < minimum {
+                    minimum = result;
+                }
+            }
+            if minimum > max_cost {
+                return max_cost + 1;
+            }
+        }
+        result
+    })
+}
+
 #[macro_export]
 macro_rules! ascii {
     ($x:literal) => {{

vm/src/builtins/code.rs

@@ -256,7 +256,7 @@ impl PyRef<PyCode> {
     }
 
     #[pyproperty]
-    fn co_varnames(self, vm: &VirtualMachine) -> PyTupleRef {
+    pub fn co_varnames(self, vm: &VirtualMachine) -> PyTupleRef {
         let varnames = self
             .code
             .varnames

vm/src/lib.rs

@@ -70,6 +70,7 @@ mod sequence;
 mod signal;
 pub mod sliceable;
 pub mod stdlib;
+pub mod suggestion;
 pub mod types;
 pub mod utils;
 pub mod version;

vm/src/suggestion.rs

@@ -0,0 +1,75 @@
+use crate::{
+    builtins::{PyStr, PyStrRef},
+    exceptions::types::PyBaseExceptionRef,
+    sliceable::PySliceableSequence,
+    IdProtocol, PyObjectRef, TypeProtocol, VirtualMachine,
+};
+use rustpython_common::str::levenshtein_distance;
+use std::iter::ExactSizeIterator;
+
+const MAX_CANDIDATE_ITEMS: usize = 750;
+
+fn calculate_suggestions<'a>(
+    dir_iter: impl ExactSizeIterator<Item = &'a PyObjectRef>,
+    name: &PyObjectRef,
+) -> Option<PyStrRef> {
+    if dir_iter.len() >= MAX_CANDIDATE_ITEMS {
+        return None;
+    }
+
+    let mut suggestion: Option<&PyStrRef> = None;
+    let mut suggestion_distance = usize::MAX;
+    let name = name.downcast_ref::<PyStr>()?;
+
+    for item in dir_iter {
+        let item_name = item.downcast_ref::<PyStr>()?;
+        if name.as_str() == item_name.as_str() {
+            continue;
+        }
+        // No more than 1/3 of the characters should need changed
+        let max_distance = usize::min(
+            (name.len() + item_name.len() + 3) / 3,
+            suggestion_distance - 1,
+        );
+        let current_distance =
+            levenshtein_distance(name.as_str(), item_name.as_str(), max_distance);
+        if current_distance > max_distance {
+            continue;
+        }
+        if suggestion.is_none() || current_distance < suggestion_distance {
+            suggestion = Some(item_name);
+            suggestion_distance = current_distance;
+        }
+    }
+    suggestion.cloned()
+}
+
+pub fn offer_suggestions(exc: &PyBaseExceptionRef, vm: &VirtualMachine) -> Option<PyStrRef> {
+    if exc.class().is(&vm.ctx.exceptions.attribute_error) {
+        let name = exc.as_object().clone().get_attr("name", vm).unwrap();
+        let obj = exc.as_object().clone().get_attr("obj", vm).unwrap();
+
+        calculate_suggestions(vm.dir(Some(obj)).ok()?.borrow_vec().iter(), &name)
+    } else if exc.class().is(&vm.ctx.exceptions.name_error) {
+        let name = exc.as_object().clone().get_attr("name", vm).unwrap();
+        let mut tb = exc.traceback().unwrap();
+        while let Some(traceback) = tb.next.clone() {
+            tb = traceback;
+        }
+
+        let varnames = tb.frame.code.clone().co_varnames(vm);
+        if let Some(suggestions) = calculate_suggestions(varnames.as_slice().iter(), &name) {
+            return Some(suggestions);
+        };
+
+        let globals = vm.extract_elements(tb.frame.globals.as_object()).ok()?;
+        if let Some(suggestions) = calculate_suggestions(globals.as_slice().iter(), &name) {
+            return Some(suggestions);
+        };
+
+        let builtins = vm.extract_elements(tb.frame.builtins.as_object()).ok()?;
+        calculate_suggestions(builtins.as_slice().iter(), &name)
+    } else {
+        None
+    }
+}