Merge pull request #427 from rustcoreutils/updates

Updates

Author: jgarzik

awk/Cargo.toml

@@ -13,6 +13,7 @@ libc.workspace = true
 pest = "2.7"
 pest_derive = "2.7"
 lexical = { version = "6.1", features = ["format"] }
+plib = { path = "../plib" }
 rand = { version = "0.8", default-features = false, features = ["small_rng"] }
 
 [dev-dependencies]

awk/regex.rs

@@ -7,36 +7,14 @@
 // SPDX-License-Identifier: MIT
 //
 
+use plib::regex::{Match, Regex as PlibRegex, RegexFlags};
 use std::ffi::CString;
-use std::ptr;
-
-fn regex_compilation_result(
-    status_integer: libc::c_int,
-    regex: &libc::regex_t,
-) -> Result<(), String> {
-    if status_integer != 0 {
-        let mut error_buffer = vec![b'\0'; 128];
-        unsafe {
-            libc::regerror(
-                status_integer,
-                ptr::from_ref(regex),
-                error_buffer.as_mut_ptr() as *mut libc::c_char,
-                128,
-            )
-        };
-        let error = CString::from_vec_with_nul(error_buffer)
-            .expect("error message returned from `libc::regerror` is an invalid CString");
-        Err(error
-            .into_string()
-            .expect("error message from `libc::regerror' contains invalid utf-8"))
-    } else {
-        Ok(())
-    }
-}
 
+/// A regex wrapper that provides CString-compatible API for AWK.
+/// Internally uses plib::regex for POSIX ERE support.
 pub struct Regex {
-    raw_regex: libc::regex_t,
-    regex_string: CString,
+    inner: PlibRegex,
+    pattern_string: String,
 }
 
 #[cfg_attr(test, derive(Debug))]
@@ -46,95 +24,95 @@ pub struct RegexMatch {
     pub end: usize,
 }
 
+impl From<Match> for RegexMatch {
+    fn from(m: Match) -> Self {
+        RegexMatch {
+            start: m.start,
+            end: m.end,
+        }
+    }
+}
+
+/// Iterator over regex matches in a string.
+/// Owns the input CString to preserve lifetimes.
 pub struct MatchIter<'re> {
-    string: CString,
+    // Store the string as owned String to avoid lifetime issues
+    string: String,
     next_start: usize,
     regex: &'re Regex,
 }
 
 impl Iterator for MatchIter<'_> {
     type Item = RegexMatch;
     fn next(&mut self) -> Option<Self::Item> {
-        if self.next_start >= self.string.as_bytes().len() {
-            return None;
-        }
-        let mut match_range = libc::regmatch_t {
-            rm_so: -1,
-            rm_eo: -1,
-        };
-        let exec_status = unsafe {
-            libc::regexec(
-                ptr::from_ref(&self.regex.raw_regex),
-                self.string.as_ptr().add(self.next_start),
-                1,
-                ptr::from_mut(&mut match_range),
-                0,
-            )
-        };
-        if exec_status == libc::REG_NOMATCH {
+        if self.next_start >= self.string.len() {
             return None;
         }
+
+        // Find match starting from current offset
+        let substring = &self.string[self.next_start..];
+        let m = self.regex.inner.find(substring)?;
+
         let result = RegexMatch {
-            start: self.next_start + match_range.rm_so as usize,
-            end: self.next_start + match_range.rm_eo as usize,
+            start: self.next_start + m.start,
+            end: self.next_start + m.end,
         };
-        self.next_start += match_range.rm_eo as usize;
+
+        // Move past this match for next iteration
+        // Ensure we make progress even on zero-width matches
+        self.next_start = if m.end > 0 {
+            self.next_start + m.end
+        } else {
+            self.next_start + 1
+        };
+
         Some(result)
     }
 }
 
 impl Regex {
     pub fn new(regex: CString) -> Result<Self, String> {
-        let mut raw = unsafe { std::mem::zeroed::<libc::regex_t>() };
-        let compilation_status =
-            unsafe { libc::regcomp(ptr::from_mut(&mut raw), regex.as_ptr(), libc::REG_EXTENDED) };
-        regex_compilation_result(compilation_status, &raw)?;
+        let pattern = regex.to_str().map_err(|e| e.to_string())?;
+        let inner = PlibRegex::new(pattern, RegexFlags::ere()).map_err(|e| e.to_string())?;
         Ok(Self {
-            raw_regex: raw,
-            regex_string: regex,
+            inner,
+            pattern_string: pattern.to_string(),
         })
     }
 
+    /// Returns an iterator over all match locations in the string.
+    /// Takes ownership of the CString.
     pub fn match_locations(&self, string: CString) -> MatchIter {
+        let s = string.into_string().unwrap_or_default();
         MatchIter {
             next_start: 0,
             regex: self,
-            string,
+            string: s,
         }
     }
 
     pub fn matches(&self, string: &CString) -> bool {
-        let exec_status = unsafe {
-            libc::regexec(
-                ptr::from_ref(&self.raw_regex),
-                string.as_ptr(),
-                0,
-                ptr::null_mut(),
-                0,
-            )
-        };
-        exec_status != libc::REG_NOMATCH
+        let s = string.to_str().unwrap_or("");
+        self.inner.is_match(s)
     }
 }
 
 impl Drop for Regex {
     fn drop(&mut self) {
-        unsafe {
-            libc::regfree(ptr::from_mut(&mut self.raw_regex));
-        }
+        // plib::regex handles cleanup internally
     }
 }
 
 #[cfg(test)]
 impl core::fmt::Debug for Regex {
     fn fmt(&self, f: &mut core::fmt::Formatter<'_>) -> core::fmt::Result {
-        writeln!(f, "/{}/", self.regex_string.to_str().unwrap())
+        writeln!(f, "/{}/", self.pattern_string)
     }
 }
 
 impl PartialEq for Regex {
     fn eq(&self, other: &Self) -> bool {
-        self.regex_string == other.regex_string
+        self.pattern_string == other.pattern_string
     }
 }
 

display/Cargo.toml

@@ -12,6 +12,7 @@ clap.workspace = true
 clap.features = ["env"]
 gettext-rs.workspace = true
 libc.workspace = true
+plib = { path = "../plib" }
 termion = "4.0"
 thiserror = "1.0"
 

display/more.rs

@@ -9,18 +9,15 @@
 
 use clap::Parser;
 use gettextrs::{bind_textdomain_codeset, gettext, setlocale, textdomain, LocaleCategory};
-use libc::{
-    getegid, getgid, getuid, regcomp, regex_t, regexec, setgid, setuid, REG_ICASE, REG_NOMATCH,
-};
+use libc::{getegid, getgid, getuid, setgid, setuid};
+use plib::regex::{Regex, RegexFlags};
 use std::collections::HashMap;
-use std::ffi::CString;
 use std::fs::File;
 use std::io::{stdout, BufRead, BufReader, Cursor, Read, Seek, SeekFrom, Write};
 use std::ops::{Not, Range};
 use std::os::fd::AsRawFd;
 use std::path::PathBuf;
 use std::process::{exit, ExitStatus};
-use std::ptr;
 use std::str::FromStr;
 use std::sync::mpsc::{channel, Receiver, TryRecvError};
 use std::sync::Mutex;
@@ -942,8 +939,8 @@ struct SourceContext {
     /// Current search pattern
     current_pattern: String,
     /// Last search settings
-    last_search: Option<(regex_t, bool, Direction)>,
-    /// Storage for marks that were set durring current [`Source`] processing
+    last_search: Option<(Regex, bool, Direction)>,
+    /// Storage for marks that were set during current [`Source`] processing
     marked_positions: HashMap<char, usize>,
     /// Flag that [`true`] if input files count is more that 1
     is_many_files: bool,
@@ -1206,7 +1203,7 @@ impl SourceContext {
     pub fn search(
         &mut self,
         count: Option<usize>,
-        pattern: regex_t,
+        pattern: Regex,
         is_not: bool,
         direction: Direction,
     ) -> Result<(), MoreError> {
@@ -1222,21 +1219,10 @@ impl SourceContext {
                     Direction::Backward => haystack + &last_string,
                 };
             }
-            let c_input = CString::new(haystack)
-                .map_err(|_| MoreError::StringParse(self.current_source.name()))?;
-            let has_match = unsafe {
-                regexec(
-                    &pattern as *const regex_t,
-                    c_input.as_ptr(),
-                    0,
-                    ptr::null_mut(),
-                    0,
-                )
-            };
             let has_match = if is_not {
-                has_match == REG_NOMATCH
+                !pattern.is_match(&haystack)
             } else {
-                has_match != REG_NOMATCH
+                pattern.is_match(&haystack)
             };
             if has_match {
                 let Some((rows, _)) = self.terminal_size else {
@@ -1291,7 +1277,7 @@ impl SourceContext {
             } else {
                 direction.clone()
             };
-            self.search(count, *pattern, *is_not, direction)
+            self.search(count, pattern.clone(), *is_not, direction)
         } else {
             Err(MoreError::SourceContext(
                 SourceContextError::MissingLastSearch,
@@ -1647,39 +1633,19 @@ impl Prompt {
     }
 }
 
-/// Compiles [`pattern`] as [`regex_t`]
-fn compile_regex(pattern: String, ignore_case: bool) -> Result<regex_t, MoreError> {
-    #[cfg(target_os = "macos")]
-    let mut pattern = pattern.replace("\\\\", "\\");
-    #[cfg(all(unix, not(target_os = "macos")))]
+/// Compiles [`pattern`] as a POSIX BRE regex
+fn compile_regex(pattern: String, ignore_case: bool) -> Result<Regex, MoreError> {
+    // Normalize backslash escapes
     let pattern = pattern.replace("\\\\", "\\");
-    let mut cflags = 0;
-    if ignore_case {
-        cflags |= REG_ICASE;
-    }
 
-    // macOS version of [regcomp](regcomp) from `libc` provides additional check
-    // for empty regex. In this case, an error
-    // [REG_EMPTY](https://developer.apple.com/library/archive/documentation/System/Conceptual/ManPages_iPhoneOS/man3/regcomp.3.html)
-    // will be returned. Therefore, an empty pattern is replaced with ".*".
-    #[cfg(target_os = "macos")]
-    {
-        pattern = if pattern == "" {
-            String::from(".*")
-        } else {
-            pattern
-        };
-    }
-
-    let c_pattern =
-        CString::new(pattern.clone()).map_err(|_| MoreError::StringParse(pattern.clone()))?;
-    let mut regex = unsafe { std::mem::zeroed::<regex_t>() };
-
-    if unsafe { regcomp(&mut regex, c_pattern.as_ptr(), cflags) } == 0 {
-        Ok(regex)
+    let flags = if ignore_case {
+        RegexFlags::bre().ignore_case()
     } else {
-        Err(MoreError::StringParse(pattern))
-    }
+        RegexFlags::bre()
+    };
+
+    // plib::regex handles macOS empty pattern workaround internally
+    Regex::new(&pattern, flags).map_err(|_| MoreError::StringParse(pattern))
 }
 
 /// More state