From 989178f438812ddbab23ef5bfd157e9ddda95b6f Mon Sep 17 00:00:00 2001
From: Ben Wiederhake <BenWiederhake.GitHub@gmx.de>
Date: Fri, 16 Feb 2024 03:03:38 +0100
Subject: [PATCH] shuf: extract minimal required interface between data and
 core loop

---
 src/uu/shuf/src/shuf.rs | 79 +++++++++++++++++++++++++++++++++--------
 1 file changed, 65 insertions(+), 14 deletions(-)

diff --git a/src/uu/shuf/src/shuf.rs b/src/uu/shuf/src/shuf.rs
index bab328e2e..6f33492eb 100644
--- a/src/uu/shuf/src/shuf.rs
+++ b/src/uu/shuf/src/shuf.rs
@@ -3,14 +3,14 @@
 // For the full copyright and license information, please view the LICENSE
 // file that was distributed with this source code.
 
-// spell-checker:ignore (ToDO) cmdline evec seps rvec fdata
+// spell-checker:ignore (ToDO) cmdline evec seps shufable rvec fdata
 
 use clap::{crate_version, Arg, ArgAction, Command};
 use memchr::memchr_iter;
 use rand::prelude::SliceRandom;
 use rand::RngCore;
 use std::fs::File;
-use std::io::{stdin, stdout, BufReader, BufWriter, Read, Write};
+use std::io::{stdin, stdout, BufReader, BufWriter, Error, Read, Write};
 use uucore::display::Quotable;
 use uucore::error::{FromIo, UResult, USimpleError, UUsageError};
 use uucore::{format_usage, help_about, help_usage};
@@ -116,18 +116,18 @@ pub fn uumain(args: impl uucore::Args) -> UResult<()> {
         Mode::Echo(args) => {
             let mut evec = args.iter().map(String::as_bytes).collect::<Vec<_>>();
             find_seps(&mut evec, options.sep);
-            shuf_bytes(&mut evec, options)?;
+            shuf_exec(&mut evec, options)?;
         }
         Mode::InputRange((b, e)) => {
             let rvec = (b..e).map(|x| format!("{x}")).collect::<Vec<String>>();
             let mut rvec = rvec.iter().map(String::as_bytes).collect::<Vec<&[u8]>>();
-            shuf_bytes(&mut rvec, options)?;
+            shuf_exec(&mut rvec, options)?;
         }
         Mode::Default(filename) => {
             let fdata = read_input_file(&filename)?;
             let mut fdata = vec![&fdata[..]];
             find_seps(&mut fdata, options.sep);
-            shuf_bytes(&mut fdata, options)?;
+            shuf_exec(&mut fdata, options)?;
         }
     }
 
@@ -251,7 +251,62 @@ fn find_seps(data: &mut Vec<&[u8]>, sep: u8) {
     }
 }
 
-fn shuf_bytes(input: &mut Vec<&[u8]>, opts: Options) -> UResult<()> {
+trait Shufable {
+    type Item: Writable;
+    fn is_empty(&self) -> bool;
+    fn choose(&self, rng: &mut WrappedRng) -> Self::Item;
+    // This type shouldn't even be known. However, because we want to support
+    // Rust 1.70, it is not possible to return "impl Iterator".
+    // TODO: When the MSRV is raised, rewrite this to return "impl Iterator".
+    type PartialShuffleIterator<'b>: Iterator<Item = Self::Item>
+    where
+        Self: 'b;
+    fn partial_shuffle<'b>(
+        &'b mut self,
+        rng: &'b mut WrappedRng,
+        amount: usize,
+    ) -> Self::PartialShuffleIterator<'b>;
+}
+
+impl<'a> Shufable for Vec<&'a [u8]> {
+    type Item = &'a [u8];
+    fn is_empty(&self) -> bool {
+        (**self).is_empty()
+    }
+    fn choose(&self, rng: &mut WrappedRng) -> Self::Item {
+        // Note: "copied()" only copies the reference, not the entire [u8].
+        // Returns None if the slice is empty. We checked this before, so
+        // this is safe.
+        (**self).choose(rng).unwrap()
+    }
+    type PartialShuffleIterator<'b> = std::iter::Copied<std::slice::Iter<'b, &'a [u8]>> where Self: 'b;
+    fn partial_shuffle<'b>(
+        &'b mut self,
+        rng: &'b mut WrappedRng,
+        amount: usize,
+    ) -> Self::PartialShuffleIterator<'b> {
+        // Note: "copied()" only copies the reference, not the entire [u8].
+        (**self).partial_shuffle(rng, amount).0.iter().copied()
+    }
+}
+
+trait Writable {
+    fn write_all_to(&self, output: &mut impl Write) -> Result<(), Error>;
+}
+
+impl<'a> Writable for &'a [u8] {
+    fn write_all_to(&self, output: &mut impl Write) -> Result<(), Error> {
+        output.write_all(self)
+    }
+}
+
+impl Writable for usize {
+    fn write_all_to(&self, output: &mut impl Write) -> Result<(), Error> {
+        output.write_all(format!("{self}").as_bytes())
+    }
+}
+
+fn shuf_exec(input: &mut impl Shufable, opts: Options) -> UResult<()> {
     let mut output = BufWriter::new(match opts.output {
         None => Box::new(stdout()) as Box<dyn Write>,
         Some(s) => {
@@ -276,22 +331,18 @@ fn shuf_bytes(input: &mut Vec<&[u8]>, opts: Options) -> UResult<()> {
 
     if opts.repeat {
         for _ in 0..opts.head_count {
-            // Returns None is the slice is empty. We checked this before, so
-            // this is safe.
-            let r = input.choose(&mut rng).unwrap();
+            let r = input.choose(&mut rng);
 
-            output
-                .write_all(r)
+            r.write_all_to(&mut output)
                 .map_err_context(|| "write failed".to_string())?;
             output
                 .write_all(&[opts.sep])
                 .map_err_context(|| "write failed".to_string())?;
         }
     } else {
-        let (shuffled, _) = input.partial_shuffle(&mut rng, opts.head_count);
+        let shuffled = input.partial_shuffle(&mut rng, opts.head_count);
         for r in shuffled {
-            output
-                .write_all(r)
+            r.write_all_to(&mut output)
                 .map_err_context(|| "write failed".to_string())?;
             output
                 .write_all(&[opts.sep])