Refactored some string handling to prevent ICEs and FNs

Author: xFrednet

clippy_lints/src/enum_variants.rs

@@ -2,7 +2,7 @@
 
 use clippy_utils::diagnostics::{span_lint, span_lint_and_help};
 use clippy_utils::source::is_present_in_source;
-use clippy_utils::str_utils;
+use clippy_utils::str_utils::{self, count_match_end, count_match_start};
 use rustc_hir::{EnumDef, Item, ItemKind};
 use rustc_lint::{LateContext, LateLintPass};
 use rustc_session::{declare_tool_lint, impl_lint_pass};
@@ -117,26 +117,6 @@ impl_lint_pass!(EnumVariantNames => [
     MODULE_INCEPTION
 ]);
 
-/// Returns the number of chars that match from the start
-#[must_use]
-fn partial_match(pre: &str, name: &str) -> usize {
-    let mut name_iter = name.chars();
-    let _ = name_iter.next_back(); // make sure the name is never fully matched
-    pre.chars().zip(name_iter).take_while(|&(l, r)| l == r).count()
-}
-
-/// Returns the number of chars that match from the end
-#[must_use]
-fn partial_rmatch(post: &str, name: &str) -> usize {
-    let mut name_iter = name.chars();
-    let _ = name_iter.next(); // make sure the name is never fully matched
-    post.chars()
-        .rev()
-        .zip(name_iter.rev())
-        .take_while(|&(l, r)| l == r)
-        .count()
-}
-
 fn check_variant(
     cx: &LateContext<'_>,
     threshold: u64,
@@ -150,7 +130,7 @@ fn check_variant(
     }
     for var in def.variants {
         let name = var.ident.name.as_str();
-        if partial_match(item_name, &name) == item_name_chars
+        if count_match_start(item_name, &name).char_count == item_name_chars
             && name.chars().nth(item_name_chars).map_or(false, |c| !c.is_lowercase())
             && name.chars().nth(item_name_chars + 1).map_or(false, |c| !c.is_numeric())
         {
@@ -161,7 +141,7 @@ fn check_variant(
                 "variant name starts with the enum's name",
             );
         }
-        if partial_rmatch(item_name, &name) == item_name_chars {
+        if count_match_end(item_name, &name).char_count == item_name_chars {
             span_lint(
                 cx,
                 ENUM_VARIANT_NAMES,
@@ -176,7 +156,7 @@ fn check_variant(
     for var in def.variants {
         let name = var.ident.name.as_str();
 
-        let pre_match = partial_match(pre, &name);
+        let pre_match = count_match_start(pre, &name).byte_count;
         pre = &pre[..pre_match];
         let pre_camel = str_utils::camel_case_until(pre).byte_index;
         pre = &pre[..pre_camel];
@@ -193,8 +173,8 @@ fn check_variant(
             }
         }
 
-        let post_match = partial_rmatch(post, &name);
-        let post_end = post.len() - post_match;
+        let post_match = count_match_end(post, &name);
+        let post_end = post.len() - post_match.byte_count;
         post = &post[post_end..];
         let post_camel = str_utils::camel_case_start(post);
         post = &post[post_camel.byte_index..];
@@ -266,14 +246,16 @@ impl LateLintPass<'_> for EnumVariantNames {
                             );
                         }
                     }
-                    if item.vis.node.is_pub() {
-                        let matching = partial_match(mod_camel, &item_camel);
-                        let rmatching = partial_rmatch(mod_camel, &item_camel);
+                    // The `module_name_repetitions` lint should only trigger if the item has the module in its
+                    // name. Having the same name is accepted.
+                    if item.vis.node.is_pub() && item_camel.len() > mod_camel.len() {
+                        let matching = count_match_start(mod_camel, &item_camel);
+                        let rmatching = count_match_end(mod_camel, &item_camel);
                         let nchars = mod_camel.chars().count();
 
                         let is_word_beginning = |c: char| c == '_' || c.is_uppercase() || c.is_numeric();
 
-                        if matching == nchars {
+                        if matching.char_count == nchars {
                             match item_camel.chars().nth(nchars) {
                                 Some(c) if is_word_beginning(c) => span_lint(
                                     cx,
@@ -284,7 +266,7 @@ impl LateLintPass<'_> for EnumVariantNames {
                                 _ => (),
                             }
                         }
-                        if rmatching == nchars {
+                        if rmatching.char_count == nchars {
                             span_lint(
                                 cx,
                                 MODULE_NAME_REPETITIONS,

clippy_utils/src/str_utils.rs

@@ -99,6 +99,75 @@ pub fn camel_case_start(s: &str) -> StrIndex {
     last_index
 }
 
+/// Dealing with sting comparison can be complicated, this struct ensures that both the
+/// character and byte count are provided for correct indexing.
+#[derive(Debug, Default, PartialEq, Eq)]
+pub struct StrCount {
+    pub char_count: usize,
+    pub byte_count: usize,
+}
+
+impl StrCount {
+    pub fn new(char_count: usize, byte_count: usize) -> Self {
+        Self { char_count, byte_count }
+    }
+}
+
+/// Returns the number of chars that match from the start
+///
+/// ```
+/// assert_eq!(count_match_start("hello_mouse", "hello_penguin"), StrCount::new(6, 6));
+/// assert_eq!(count_match_start("hello_clippy", "bye_bugs"), StrCount::new(0, 0));
+/// assert_eq!(count_match_start("hello_world", "hello_world"), StrCount::new(11, 11));
+/// assert_eq!(count_match_start("T\u{f6}ffT\u{f6}ff", "T\u{f6}ff"), StrCount::new(4, 5));
+/// ```
+#[must_use]
+pub fn count_match_start(str1: &str, str2: &str) -> StrCount {
+    // (char_index, char1)
+    let char_count = str1.chars().count();
+    let iter1 = (0..=char_count).zip(str1.chars());
+    // (byte_index, char2)
+    let iter2 = str2.char_indices();
+
+    iter1
+        .zip(iter2)
+        .take_while(|((_, c1), (_, c2))| c1 == c2)
+        .last()
+        .map_or_else(StrCount::default, |((char_index, _), (byte_index, character))| {
+            StrCount::new(char_index + 1, byte_index + character.len_utf8())
+        })
+}
+
+/// Returns the number of chars and bytes that match from the end
+///
+/// ```
+/// assert_eq!(count_match_end("hello_cat", "bye_cat"), StrCount::new(4, 4));
+/// assert_eq!(count_match_end("if_item_thing", "enum_value"), StrCount::new(0, 0));
+/// assert_eq!(count_match_end("Clippy", "Clippy"), StrCount::new(6, 6));
+/// assert_eq!(count_match_end("MyT\u{f6}ff", "YourT\u{f6}ff"), StrCount::new(4, 5));
+/// ```
+#[must_use]
+pub fn count_match_end(str1: &str, str2: &str) -> StrCount {
+    let char_count = str1.chars().count();
+    if char_count == 0 {
+        return StrCount::default();
+    }
+
+    // (char_index, char1)
+    let iter1 = (0..char_count).rev().zip(str1.chars().rev());
+    // (byte_index, char2)
+    let byte_count = str2.len();
+    let iter2 = str2.char_indices().rev();
+
+    iter1
+        .zip(iter2)
+        .take_while(|((_, c1), (_, c2))| c1 == c2)
+        .last()
+        .map_or_else(StrCount::default, |((char_index, _), (byte_index, _))| {
+            StrCount::new(char_count - char_index, byte_count - byte_index)
+        })
+}
+
 #[cfg(test)]
 mod test {
     use super::*;

tests/ui/crashes/ice-7869.rs

@@ -0,0 +1,7 @@
+enum Tila {
+    TyöAlkoi,
+    TyöKeskeytyi,
+    TyöValmis,
+}
+
+fn main() {}

tests/ui/crashes/ice-7869.stderr

@@ -0,0 +1,15 @@
+error: all variants have the same prefix: `Työ`
+  --> $DIR/ice-7869.rs:1:1
+   |
+LL | / enum Tila {
+LL | |     TyöAlkoi,
+LL | |     TyöKeskeytyi,
+LL | |     TyöValmis,
+LL | | }
+   | |_^
+   |
+   = note: `-D clippy::enum-variant-names` implied by `-D warnings`
+   = help: remove the prefixes and use full paths to the variants instead of glob imports
+
+error: aborting due to previous error
+

tests/ui/enum_variants.stderr

@@ -60,7 +60,7 @@ LL | | }
    |
    = help: remove the prefixes and use full paths to the variants instead of glob imports
 
-error: all variants have the same prefix: `With`
+error: all variants have the same prefix: `WithOut`
   --> $DIR/enum_variants.rs:81:1
    |
 LL | / enum Seallll {

tests/ui/match_ref_pats.rs

@@ -1,5 +1,5 @@
 #![warn(clippy::match_ref_pats)]
-#![allow(clippy::equatable_if_let)]
+#![allow(clippy::equatable_if_let, clippy::enum_variant_names)]
 
 fn ref_pats() {
     {