env: do not start a word for an unset variable in -S

src/uu/env/src/split_iterator.rs

@@ -95,7 +95,7 @@ impl<'a> SplitIterator<'a> {
         self.expander.get_parser_mut()
     }
 
-    fn substitute_variable<'x>(&'x mut self) -> Result<(), EnvError> {
+    fn substitute_variable<'x>(&'x mut self) -> Result<bool, EnvError> {
         let mut var_parse = VariableParser::<'a, '_> {
             parser: self.get_parser_mut(),
         };
@@ -105,9 +105,10 @@ impl<'a> SplitIterator<'a> {
         let varname_os_str_cow = from_native_int_representation(Cow::Borrowed(name));
         if let Some(value) = std::env::var_os(varname_os_str_cow) {
             self.expander.put_string(value);
+            return Ok(true);
         }
 
-        Ok(())
+        Ok(false)
     }
 
     fn check_and_replace_ascii_escape_code(&mut self, c: char) -> Result<bool, EnvError> {
@@ -152,6 +153,12 @@ impl<'a> SplitIterator<'a> {
                 Some(c) if ASCII_WHITESPACE_CHARS.contains(&c) => {
                     self.skip_one()?;
                 }
+                Some(DOLLAR) => {
+                    // An unset variable must not start a word.
+                    if self.substitute_variable()? {
+                        self.state_unquoted()?;
+                    }
+                }
                 Some(_) => {
                     // Don't consume char. Will be done in unquoted state.
                     self.state_unquoted()?;

tests/by-util/test_env.rs

@@ -692,6 +692,32 @@ fn test_split_string_backslash_a_behavior_matches_gnu_quoting_context() {
         .no_stdout()
         .stderr_contains("invalid sequence '\\a' in -S");
 }
+
+#[cfg(not(windows))] // no printf available
+#[test]
+fn test_split_string_unset_variable_does_not_start_a_word() {
+    let scene = TestScenario::new(util_name!());
+
+    scene
+        .ucmd()
+        .arg("-Sprintf [%s] a ${UU_UNSET}")
+        .succeeds()
+        .stdout_is("[a]");
+
+    scene
+        .ucmd()
+        .arg("-S${UU_UNSET} printf hi")
+        .succeeds()
+        .stdout_is("hi");
+
+    scene
+        .ucmd()
+        .env("UU_EMPTY", "")
+        .arg("-Sprintf [%s] a ${UU_EMPTY}")
+        .succeeds()
+        .stdout_is("[a][]");
+}
+
 #[test]
 #[allow(clippy::cognitive_complexity)] // Ignore clippy lint of too long function sign
 fn test_env_parsing_errors() {
@@ -1464,6 +1490,33 @@ mod tests_split_iterator {
         ]);
     }
 
+    #[test]
+    fn split_unset_variable_on_its_own_adds_no_word() {
+        assert!(std::env::var_os("UU_UNSET").is_none());
+        split_ok(&[
+            ("${UU_UNSET}", &[]),
+            ("a ${UU_UNSET}", &["a"]),
+            ("a ${UU_UNSET} b", &["a", "b"]),
+            ("${UU_UNSET} a", &["a"]),
+            ("a ${UU_UNSET}#c d", &["a"]),
+            (r"a ${UU_UNSET}\_b", &["a", "b"]),
+            (r"a ${UU_UNSET}\c", &["a"]),
+        ]);
+    }
+
+    #[test]
+    fn split_unset_variable_next_to_other_text_keeps_the_word() {
+        assert!(std::env::var_os("UU_UNSET").is_none());
+        split_ok(&[
+            ("x${UU_UNSET}", &["x"]),
+            ("${UU_UNSET}x", &["x"]),
+            ("a ${UU_UNSET}${UU_UNSET}x", &["a", "x"]),
+            (r#""${UU_UNSET}""#, &[""]),
+            (r#"""${UU_UNSET}"#, &[""]),
+            (r#"${UU_UNSET}"""#, &[""]),
+        ]);
+    }
+
     #[test]
     fn test_quote() {
         assert_eq!(quote(""), "''");