Introduce fish_wcstoi_partial

fish_wcstoi_partial is like fish_wcstoi: it converts from a string to an
int optionally inferring the radix. fish_wcstoi_partial also returns the
number of characters consumed.

fish-rust/src/builtins/bg.rs

@@ -99,12 +99,12 @@ pub fn bg(parser: &mut parser_t, streams: &mut io_streams_t, args: &mut [&wstr])
     let mut retval = STATUS_CMD_OK;
     let pids: Vec<i64> = args[opts.optind..]
         .iter()
-        .map(|arg| {
-            fish_wcstoi(arg.chars()).unwrap_or_else(|_| {
+        .map(|&arg| {
+            fish_wcstoi(arg).unwrap_or_else(|_| {
                 streams.err.append(wgettext_fmt!(
                     "%ls: '%ls' is not a valid job specifier\n",
                     cmd,
-                    *arg
+                    arg
                 ));
                 retval = STATUS_INVALID_ARGS;
                 0

fish-rust/src/builtins/random.rs

@@ -73,7 +73,7 @@ pub fn random(
         cmd: &wstr,
         num: &wstr,
     ) -> Result<T, wutil::Error> {
-        let res = fish_wcstoi_radix_all(num.chars(), None, true);
+        let res = fish_wcstoi_radix_all(num, None, true);
         if res.is_err() {
             streams
                 .err

fish-rust/src/builtins/return.rs

@@ -117,7 +117,7 @@ pub fn parse_return_value(
     if optind == args.len() {
         Ok(parser.get_last_status().into())
     } else {
-        match fish_wcstoi(args[optind].chars()) {
+        match fish_wcstoi(args[optind]) {
             Ok(i) => Ok(i),
             Err(_e) => {
                 streams

fish-rust/src/builtins/wait.rs

@@ -196,7 +196,7 @@ pub fn wait(
     for i in w.woptind..argc {
         if iswnumeric(argv[i]) {
             // argument is pid
-            let mpid: Result<pid_t, wutil::Error> = fish_wcstoi(argv[i].chars());
+            let mpid: Result<pid_t, wutil::Error> = fish_wcstoi(argv[i]);
             if mpid.is_err() || mpid.unwrap() <= 0 {
                 streams.err.append(wgettext_fmt!(
                     "%ls: '%ls' is not a valid process id\n",

fish-rust/src/redirection.rs

@@ -112,7 +112,7 @@ impl RedirectionSpec {
 
     /// Attempt to parse target as an fd.
     pub fn get_target_as_fd(&self) -> Option<RawFd> {
-        fish_wcstoi(self.target.as_char_slice().iter().copied()).ok()
+        fish_wcstoi(&self.target).ok()
     }
     fn get_target_as_fd_ffi(&self) -> SharedPtr<i32> {
         match self.get_target_as_fd() {

fish-rust/src/wchar_ext.rs

@@ -116,6 +116,22 @@ impl<'a> IntoCharIter for &'a WString {
     }
 }
 
+// Also support `str.chars()` itself.
+impl<'a> IntoCharIter for std::str::Chars<'a> {
+    type Iter = Self;
+    fn chars(self) -> Self::Iter {
+        self
+    }
+}
+
+// Also support `wstr.chars()` itself.
+impl<'a> IntoCharIter for CharsUtf32<'a> {
+    type Iter = Self;
+    fn chars(self) -> Self::Iter {
+        self
+    }
+}
+
 /// \return true if \p prefix is a prefix of \p contents.
 fn iter_prefixes_iter<Prefix, Contents>(prefix: Prefix, mut contents: Contents) -> bool
 where

fish-rust/src/wutil/wcstoi.rs

@@ -1,22 +1,39 @@
 pub use super::errors::Error;
+use crate::wchar::IntoCharIter;
 use num_traits::{NumCast, PrimInt};
-use std::iter::Peekable;
+use std::iter::{Fuse, Peekable};
 use std::result::Result;
 
 struct ParseResult {
     result: u64,
     negative: bool,
     consumed_all: bool,
+    consumed: usize,
 }
 
-/// Helper to get the current char, or \0.
-fn current<Chars>(chars: &mut Peekable<Chars>) -> char
-where
-    Chars: Iterator<Item = char>,
-{
-    match chars.peek() {
-        Some(c) => *c,
-        None => '\0',
+struct CharsIterator<Iter: Iterator<Item = char>> {
+    chars: Peekable<Fuse<Iter>>,
+    consumed: usize,
+}
+
+impl<Iter: Iterator<Item = char>> CharsIterator<Iter> {
+    /// Get the current char, or \0.
+    fn current(&mut self) -> char {
+        self.peek().unwrap_or('\0')
+    }
+
+    /// Get the current char, or None.
+    fn peek(&mut self) -> Option<char> {
+        self.chars.peek().copied()
+    }
+
+    /// Get the next char, incrementing self.consumed.
+    fn next(&mut self) -> Option<char> {
+        let res = self.chars.next();
+        if res.is_some() {
+            self.consumed += 1;
+        }
+        res
     }
 }
 
@@ -26,17 +43,22 @@ where
 ///   - Leading 0 means 8.
 ///   - Otherwise 10.
 /// The parse result contains the number as a u64, and whether it was negative.
-fn fish_parse_radix<Chars>(ichars: Chars, mradix: Option<u32>) -> Result<ParseResult, Error>
-where
-    Chars: Iterator<Item = char>,
-{
+fn fish_parse_radix<Iter: Iterator<Item = char>>(
+    iter: Iter,
+    mradix: Option<u32>,
+) -> Result<ParseResult, Error> {
     if let Some(r) = mradix {
         assert!((2..=36).contains(&r), "fish_parse_radix: invalid radix {r}");
     }
-    let chars = &mut ichars.peekable();
+
+    // Construct a CharsIterator to keep track of how many we consume.
+    let mut chars = CharsIterator {
+        chars: iter.fuse().peekable(),
+        consumed: 0,
+    };
 
     // Skip leading whitespace.
-    while current(chars).is_whitespace() {
+    while chars.current().is_whitespace() {
         chars.next();
     }
 
@@ -46,9 +68,9 @@ where
 
     // Consume leading +/-.
     let mut negative;
-    match current(chars) {
+    match chars.current() {
         '-' | '+' => {
-            negative = current(chars) == '-';
+            negative = chars.current() == '-';
             chars.next();
         }
         _ => negative = false,
@@ -57,9 +79,9 @@ where
     // Determine the radix.
     let radix = if let Some(radix) = mradix {
         radix
-    } else if current(chars) == '0' {
+    } else if chars.current() == '0' {
         chars.next();
-        match current(chars) {
+        match chars.current() {
             'x' | 'X' => {
                 chars.next();
                 16
@@ -71,6 +93,7 @@ where
                     result: 0,
                     negative: false,
                     consumed_all: chars.peek().is_none(),
+                    consumed: chars.consumed,
                 });
             }
         }
@@ -79,19 +102,19 @@ where
     };
 
     // Compute as u64.
-    let mut consumed1 = false;
+    let start_consumed = chars.consumed;
     let mut result: u64 = 0;
-    while let Some(digit) = current(chars).to_digit(radix) {
+    while let Some(digit) = chars.current().to_digit(radix) {
         result = result
             .checked_mul(radix as u64)
             .and_then(|r| r.checked_add(digit as u64))
             .ok_or(Error::Overflow)?;
         chars.next();
-        consumed1 = true;
     }
 
-    // Did we consume at least one char?
-    if !consumed1 {
+    // Did we consume at least one char after the prefix?
+    let consumed = chars.consumed;
+    if consumed == start_consumed {
         return Err(Error::InvalidChar);
     }
 
@@ -104,6 +127,7 @@ where
         result,
         negative,
         consumed_all,
+        consumed,
     })
 }
 
@@ -112,6 +136,7 @@ fn fish_wcstoi_impl<Int, Chars>(
     src: Chars,
     mradix: Option<u32>,
     consume_all: bool,
+    out_consumed: &mut usize,
 ) -> Result<Int, Error>
 where
     Chars: Iterator<Item = char>,
@@ -125,8 +150,9 @@ where
         result,
         negative,
         consumed_all,
-        ..
+        consumed,
     } = fish_parse_radix(src, mradix)?;
+    *out_consumed = consumed;
 
     if !signed && negative {
         Err(Error::InvalidChar)
@@ -158,20 +184,20 @@ where
 ///  - Leading + is supported.
 pub fn fish_wcstoi<Int, Chars>(src: Chars) -> Result<Int, Error>
 where
-    Chars: Iterator<Item = char>,
+    Chars: IntoCharIter,
     Int: PrimInt,
 {
-    fish_wcstoi_impl(src, None, false)
+    fish_wcstoi_impl(src.chars(), None, false, &mut 0)
 }
 
 /// Convert the given wide string to an integer using the given radix.
 /// Leading whitespace is skipped.
 pub fn fish_wcstoi_radix<Int, Chars>(src: Chars, radix: u32) -> Result<Int, Error>
 where
-    Chars: Iterator<Item = char>,
+    Chars: IntoCharIter,
     Int: PrimInt,
 {
-    fish_wcstoi_impl(src, Some(radix), false)
+    fish_wcstoi_impl(src.chars(), Some(radix), false, &mut 0)
 }
 
 pub fn fish_wcstoi_radix_all<Int, Chars>(
@@ -180,10 +206,24 @@ pub fn fish_wcstoi_radix_all<Int, Chars>(
     consume_all: bool,
 ) -> Result<Int, Error>
 where
-    Chars: Iterator<Item = char>,
+    Chars: IntoCharIter,
     Int: PrimInt,
 {
-    fish_wcstoi_impl(src, radix, consume_all)
+    fish_wcstoi_impl(src.chars(), radix, consume_all, &mut 0)
+}
+
+/// Convert the given wide string to an integer.
+/// The semantics here match wcstol():
+///  - Leading whitespace is skipped.
+///  - 0 means octal, 0x means hex
+///  - Leading + is supported.
+/// The number of consumed characters is returned in out_consumed.
+pub fn fish_wcstoi_partial<Int, Chars>(src: Chars, out_consumed: &mut usize) -> Result<Int, Error>
+where
+    Chars: IntoCharIter,
+    Int: PrimInt,
+{
+    fish_wcstoi_impl(src.chars(), None, false, out_consumed)
 }
 
 #[cfg(test)]
@@ -205,8 +245,14 @@ mod tests {
         assert_eq!(run1("0"), Ok(0));
         assert_eq!(run1("-0"), Ok(0));
         assert_eq!(run1("+0"), Ok(0));
+        assert_eq!(run1("+00"), Ok(0));
+        assert_eq!(run1("-00"), Ok(0));
+        assert_eq!(run1("+0x00"), Ok(0));
+        assert_eq!(run1("-0x00"), Ok(0));
         assert_eq!(run1("+-0"), Err(Error::InvalidChar));
         assert_eq!(run1("-+0"), Err(Error::InvalidChar));
+        assert_eq!(run1("5"), Ok(5));
+        assert_eq!(run1("-5"), Ok(-5));
         assert_eq!(run1("123"), Ok(123));
         assert_eq!(run1("+123"), Ok(123));
         assert_eq!(run1("-123"), Ok(-123));
@@ -236,4 +282,26 @@ mod tests {
         test_min_max(std::u32::MIN, std::u32::MAX);
         test_min_max(std::u64::MIN, std::u64::MAX);
     }
+
+    #[test]
+    fn test_partial() {
+        let run1 = |s: &str| -> (i32, usize) {
+            let mut consumed = 0;
+            let res =
+                fish_wcstoi_partial(s.chars(), &mut consumed).expect("Should have parsed an int");
+            (res, consumed)
+        };
+
+        assert_eq!(run1("0"), (0, 1));
+        assert_eq!(run1("-0"), (0, 2));
+        assert_eq!(run1(" -1  "), (-1, 3));
+        assert_eq!(run1(" +1  "), (1, 3));
+        assert_eq!(run1("  345  "), (345, 5));
+        assert_eq!(run1(" -345  "), (-345, 5));
+        assert_eq!(run1("  0345  "), (229, 6));
+        assert_eq!(run1("  +0345  "), (229, 7));
+        assert_eq!(run1("  -0345  "), (-229, 7));
+        assert_eq!(run1(" 0x345  "), (0x345, 6));
+        assert_eq!(run1(" -0x345  "), (-0x345, 7));
+    }
 }