ripgrep: better --path-separator error message

This commit improves the error message when --path-separator fails. Namely,
it prints the separator it got and also prints a notice for Windows users
for common failure modes.

Fixes #957

CHANGELOG.md

@@ -87,6 +87,9 @@ Bug fixes:
 * [BUG #955](https://github.com/BurntSushi/ripgrep/issues/955):
   Use buffered writing when not printing to a tty, which fixes a performance
   regression.
+* [BUG #957](https://github.com/BurntSushi/ripgrep/issues/957):
+  Improve the error message shown for `--path separator /` in some Windows
+  shells.
 * [BUG #964](https://github.com/BurntSushi/ripgrep/issues/964):
   Add a `--no-fixed-strings` flag to disable `-F/--fixed-strings`.
 * [BUG #988](https://github.com/BurntSushi/ripgrep/issues/988):

src/args.rs

@@ -22,7 +22,7 @@ use ignore::overrides::{Override, OverrideBuilder};
 use ignore::types::{FileTypeDef, Types, TypesBuilder};
 use ignore;
 use printer::{ColorSpecs, Printer};
-use unescape::unescape;
+use unescape::{escape, unescape};
 use worker::{Worker, WorkerBuilder};
 
 use config;
@@ -749,7 +749,12 @@ impl<'a> ArgMatches<'a> {
                 } else if sep.len() > 1 {
                     Err(From::from(format!(
                         "A path separator must be exactly one byte, but \
-                         the given separator is {} bytes.", sep.len())))
+                         the given separator is {} bytes: {}\n\
+                         In some shells on Windows '/' is automatically \
+                         expanded. Use '//' instead.",
+                         sep.len(),
+                         escape(&sep),
+                    )))
                 } else {
                     Ok(Some(sep[0]))
                 }

src/unescape.rs

@@ -11,6 +11,15 @@ enum State {
     Literal,
 }
 
+/// Escapes an arbitrary byte slice such that it can be presented as a human
+/// readable string.
+pub fn escape(bytes: &[u8]) -> String {
+    use std::ascii::escape_default;
+
+    let escaped = bytes.iter().flat_map(|&b| escape_default(b)).collect();
+    String::from_utf8(escaped).unwrap()
+}
+
 /// Unescapes a string given on the command line. It supports a limited set of
 /// escape sequences:
 ///