Auto merge of rust-lang#115253 - donno2048:patch-1, r=albertlarsan68

Implement `PROBLEMATIC_CONSTS` generalization

You forgot that `A≈4`, `B≈8`, and `E≈3` and some more constants.

The new `PROBLEMATIC_CONSTS` was generated using this code:

```py
from functools import reduce
def generate_problems(consts: list, letter_digit: dict):
    for const in consts:
        problem = reduce(lambda string, rep: string.replace(*reversed(rep)), ['%X' % const, *letter_digit.items()])
        indexes = [index for index, c in enumerate(problem) if c in letter_digit.keys()]
        for i in range(1 << len(indexes)):
            yield int(''.join(letter_digit[c] if index in indexes and (i >> indexes.index(index)) & 1 else c for index, c in enumerate(problem)), 0x10)

problems = generate_problems(
    [
        # Old PROBLEMATIC_CONSTS:
        184594741, 2880289470, 2881141438, 2965027518, 2976579765, 3203381950, 3405691582, 3405697037,
        3735927486, 3735932941, 4027431614, 4276992702,
        # More of my own:
        195934910, 252707358, 762133, 179681982, 173390526
    ],
    {
    'A': '4',
    'B': '8',
    'E': '3',
    }
)

# print(list(problems)) # won't use that to print formatted

from itertools import islice
while len(cur_problems := list(islice(problems, 8))):
    print('    ', end='')
    print(*cur_problems, sep=', ', end='')
    print(',')
```

Author: bors

Diff for: Cargo.lock

@@ -5598,6 +5598,7 @@ dependencies = [
  "lazy_static",
  "miropt-test-tools",
  "regex",
+ "rustc-hash",
  "semver",
  "termcolor",
  "walkdir",

Diff for: src/tools/tidy/Cargo.toml

@@ -13,7 +13,9 @@ walkdir = "2"
 ignore = "0.4.18"
 semver = "1.0"
 termcolor = "1.1.3"
+rustc-hash = "1.1.0"
 
 [[bin]]
 name = "rust-tidy"
 path = "src/main.rs"
+test = false

Diff for: src/tools/tidy/src/style.rs

@@ -19,8 +19,12 @@
 
 use crate::walk::{filter_dirs, walk};
 use regex::{Regex, RegexSet};
+use rustc_hash::FxHashMap;
 use std::{ffi::OsStr, path::Path};
 
+#[cfg(test)]
+mod tests;
+
 /// Error code markdown is restricted to 80 columns because they can be
 /// displayed on the console with --example.
 const ERROR_CODE_COLS: usize = 80;
@@ -65,12 +69,56 @@ const ANNOTATIONS_TO_IGNORE: &[&str] = &[
     "//@ normalize-stderr-test",
 ];
 
+fn generate_problems<'a>(
+    consts: &'a [u32],
+    letter_digit: &'a FxHashMap<char, char>,
+) -> impl Iterator<Item = u32> + 'a {
+    consts.iter().flat_map(move |const_value| {
+        let problem =
+            letter_digit.iter().fold(format!("{:X}", const_value), |acc, (key, value)| {
+                acc.replace(&value.to_string(), &key.to_string())
+            });
+        let indexes: Vec<usize> = problem
+            .chars()
+            .enumerate()
+            .filter_map(|(index, c)| if letter_digit.contains_key(&c) { Some(index) } else { None })
+            .collect();
+        (0..1 << indexes.len()).map(move |i| {
+            u32::from_str_radix(
+                &problem
+                    .chars()
+                    .enumerate()
+                    .map(|(index, c)| {
+                        if let Some(pos) = indexes.iter().position(|&x| x == index) {
+                            if (i >> pos) & 1 == 1 { letter_digit[&c] } else { c }
+                        } else {
+                            c
+                        }
+                    })
+                    .collect::<String>(),
+                0x10,
+            )
+            .unwrap()
+        })
+    })
+}
+
 // Intentionally written in decimal rather than hex
-const PROBLEMATIC_CONSTS: &[u32] = &[
+const ROOT_PROBLEMATIC_CONSTS: &[u32] = &[
     184594741, 2880289470, 2881141438, 2965027518, 2976579765, 3203381950, 3405691582, 3405697037,
-    3735927486, 3735932941, 4027431614, 4276992702,
+    3735927486, 3735932941, 4027431614, 4276992702, 195934910, 252707358, 762133, 179681982,
+    173390526, 721077,
 ];
 
+fn generate_problematic_strings(
+    consts: &[u32],
+    letter_digit: &FxHashMap<char, char>,
+) -> Vec<String> {
+    generate_problems(consts, letter_digit)
+        .flat_map(|v| vec![v.to_string(), format!("{:x}", v), format!("{:X}", v)])
+        .collect()
+}
+
 const INTERNAL_COMPILER_DOCS_LINE: &str = "#### This error code is internal to the compiler and will not be emitted with normal Rust code.";
 
 /// Parser states for `line_is_url`.
@@ -267,11 +315,10 @@ pub fn check(path: &Path, bad: &mut bool) {
         // We only check CSS files in rustdoc.
         path.extension().map_or(false, |e| e == "css") && !is_in(path, "src", "librustdoc")
     }
-
-    let problematic_consts_strings: Vec<String> = (PROBLEMATIC_CONSTS.iter().map(u32::to_string))
-        .chain(PROBLEMATIC_CONSTS.iter().map(|v| format!("{:x}", v)))
-        .chain(PROBLEMATIC_CONSTS.iter().map(|v| format!("{:X}", v)))
-        .collect();
+    let problematic_consts_strings = generate_problematic_strings(
+        ROOT_PROBLEMATIC_CONSTS,
+        &[('A', '4'), ('B', '8'), ('E', '3')].iter().cloned().collect(),
+    );
     let problematic_regex = RegexSet::new(problematic_consts_strings.as_slice()).unwrap();
 
     walk(path, skip, &mut |entry, contents| {

Diff for: src/tools/tidy/src/style/tests.rs

@@ -0,0 +1,17 @@
+use super::*;
+
+#[test]
+fn test_generate_problematic_strings() {
+    let problematic_regex = RegexSet::new(
+        generate_problematic_strings(
+            ROOT_PROBLEMATIC_CONSTS,
+            &[('A', '4'), ('B', '8'), ('E', '3'), ('0', 'F')].iter().cloned().collect(), // use "futile" F intentionally
+        )
+        .as_slice(),
+    )
+    .unwrap();
+    assert!(problematic_regex.is_match("786357")); // check with no "decimal" hex digits - converted to integer
+    assert!(problematic_regex.is_match("589701")); // check with "decimal" replacements - converted to integer
+    assert!(problematic_regex.is_match("8FF85")); // check for hex display
+    assert!(!problematic_regex.is_match("1193046")); // check for non-matching value
+}