Polish the handling of 0r

- Use common constant for error message.
- Simplify command check at compile and runtime.
- Allow for future R command.
- Separate runtime handling into a function.

src/sed/compiler.rs

@@ -29,6 +29,9 @@ use uucore::error::{UResult, USimpleError};
 
 const DEFAULT_OUTPUT_WIDTH: usize = 60;
 
+const ERR_ADDRESS_0_USAGE: &str =
+    "address 0 can only be used with ~step, a second regular expression, or a read command";
+
 // Handling required after processing a command
 #[derive(Debug, Clone, Copy, PartialEq, Eq)]
 enum CommandHandling {
@@ -380,11 +383,7 @@ fn compile_address_range(
             };
 
             if is_line0 && !matches!(addr2, Address::Re(_)) && !is_step_match {
-                return compilation_error(
-                    lines,
-                    line,
-                    "address 0 can only be used with a regular expression or ~step",
-                );
+                return compilation_error(lines, line, ERR_ADDRESS_0_USAGE);
             }
 
             // If needed, transform Address::Line into Address::Step*.
@@ -399,8 +398,14 @@ fn compile_address_range(
         }
     }
 
+    // Zero-address read command check
     if is_line0 && n_addr == 1 {
-        return compilation_error(lines, line, "address 0 requires a second address");
+        // After retrieval of first address, subsequent spaces
+        // are consumed unconditionally. By now, the position
+        // must be in non-whitespace character or EOL.
+        if line.eol() || !matches!(line.current(), 'r' | 'R') {
+            return compilation_error(lines, line, ERR_ADDRESS_0_USAGE);
+        }
     }
 
     Ok(n_addr)
@@ -1849,6 +1854,51 @@ mod tests {
         ScriptCharProvider::new("")
     }
 
+    #[test]
+    fn test_zero_addr_r_accepted() {
+        for input in ["0r", "0  r"] {
+            let (lines, mut chars) = make_providers(input);
+            let mut cmd = Rc::new(RefCell::new(Command::default()));
+            let n_addr = compile_address_range(&lines, &mut chars, &mut cmd, &ctx()).unwrap();
+
+            assert_eq!(n_addr, 1);
+            assert!(matches!(cmd.borrow().addr1, Some(Address::Line(0))));
+            assert_eq!(chars.current(), 'r');
+        }
+    }
+
+    // Zero-address with no commands
+    #[test]
+    fn test_zero_addr_no_commands() {
+        let (lines, mut chars) = make_providers("0");
+        let mut cmd = Rc::new(RefCell::new(Command::default()));
+        let result = compile_address_range(&lines, &mut chars, &mut cmd, &ctx());
+
+        assert!(result.is_err());
+        assert!(
+            result
+                .unwrap_err()
+                .to_string()
+                .contains(ERR_ADDRESS_0_USAGE)
+        );
+    }
+
+    // Zero-address with a command other than 'r' must still be rejected.
+    #[test]
+    fn test_zero_addr_non_r_rejected() {
+        let (lines, mut chars) = make_providers("0p");
+        let mut cmd = Rc::new(RefCell::new(Command::default()));
+        let result = compile_address_range(&lines, &mut chars, &mut cmd, &ctx());
+
+        assert!(result.is_err());
+        assert!(
+            result
+                .unwrap_err()
+                .to_string()
+                .contains(ERR_ADDRESS_0_USAGE)
+        );
+    }
+
     #[test]
     fn test_compile_sequence_empty_input() {
         let mut provider = make_line_provider(&[]);

src/sed/processor.rs

@@ -415,6 +415,35 @@ fn list(output: &mut OutputBuffer, line: &IOChunk, max_width: usize) -> UResult<
     Ok(())
 }
 
+/// Handle address 0 read at the beginning of each file.
+fn address_0_process(
+    commands: Option<Rc<RefCell<Command>>>,
+    output: &mut OutputBuffer,
+) -> UResult<()> {
+    // Prescan for zero-address which must produce output
+    // before any input line is read.
+    {
+        let mut current = commands;
+        while let Some(cmd_rc) = current {
+            let next = {
+                let cmd = cmd_rc.borrow();
+
+                if cmd.code == 'r'
+                    && matches!(cmd.addr1, Some(Address::Line(0)))
+                    && cmd.addr2.is_none()
+                {
+                    let path = extract_variant!(cmd, Path);
+                    output.copy_file(path)?;
+                }
+
+                cmd.next.clone()
+            };
+            current = next;
+        }
+    }
+    Ok(())
+}
+
 #[allow(clippy::cognitive_complexity)]
 /// Process a single input file
 fn process_file(
@@ -423,6 +452,8 @@ fn process_file(
     output: &mut OutputBuffer,
     context: &mut ProcessingContext,
 ) -> UResult<()> {
+    address_0_process(commands.clone(), output)?;
+
     // Loop over the input lines as pattern space.
     'lines: while let Some(mut pattern) = reader.get_line()? {
         context.line_number += 1;

tests/by-util/test_sed.rs

@@ -973,6 +973,14 @@ check_output!(
 check_output!(read_ok, [format!("4r {LINES2}"), LINES1.to_string()]);
 check_output!(read_missing, ["5r /xyzzyxyzy42", LINES1]);
 check_output!(read_empty, ["6r input/empty", LINES1]);
+check_output!(
+    cmd_read_zero_addr,
+    [format!("0r {LINES2}"), LINES1.to_string()]
+);
+check_output!(
+    cmd_read_one_addr,
+    [format!("1r {LINES2}"), LINES1.to_string()]
+);
 
 #[test]
 fn write_single_file() -> std::io::Result<()> {
@@ -1339,7 +1347,7 @@ fn test_addr0_second_required() {
         .args(&["0p"])
         .fails()
         .code_is(1)
-        .stderr_is("sed: <script argument 1>:1:2: error: address 0 requires a second address\n");
+        .stderr_is("sed: <script argument 1>:1:2: error: address 0 can only be used with ~step, a second regular expression, or a read command\n");
 }
 
 #[test]
@@ -1348,7 +1356,7 @@ fn test_addr0_second_re_only() {
         .args(&["0,4p"])
         .fails()
         .code_is(1)
-        .stderr_is("sed: <script argument 1>:1:4: error: address 0 can only be used with a regular expression or ~step\n");
+        .stderr_is("sed: <script argument 1>:1:4: error: address 0 can only be used with ~step, a second regular expression, or a read command\n");
 }
 
 #[test]

tests/fixtures/sed/output/cmd_read_one_addr

@@ -0,0 +1,23 @@
+l1_1
+l2_1
+l2_2
+l2_3
+l2_4
+l2_5
+l2_6
+l2_7
+l2_8
+l2_9
+l1_2
+l1_3
+l1_4
+l1_5
+l1_6
+l1_7
+l1_8
+l1_9
+l1_10
+l1_11
+l1_12
+l1_13
+l1_14

tests/fixtures/sed/output/cmd_read_zero_addr

@@ -0,0 +1,23 @@
+l2_1
+l2_2
+l2_3
+l2_4
+l2_5
+l2_6
+l2_7
+l2_8
+l2_9
+l1_1
+l1_2
+l1_3
+l1_4
+l1_5
+l1_6
+l1_7
+l1_8
+l1_9
+l1_10
+l1_11
+l1_12
+l1_13
+l1_14