feat: support zero-address file insertion

Signed-off-by: PranavRJoshi <pranavrjoshi1@gmail.com>

Author: PranavRJoshi

src/sed/compiler.rs

@@ -403,8 +403,15 @@ fn compile_address_range(
         }
     }
 
+    // zero-address r 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.
+        let next_cmd = if line.eol() { '\0' } else { line.current() };
+        if !matches!(next_cmd, 'r') {
+            return compilation_error(lines, line, "address 0 requires a second address");
+        }
     }
 
     Ok(n_addr)
@@ -1804,6 +1811,51 @@ 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("address 0 requires a second addres")
+        );
+    }
+
+    // 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("address 0 requires a second addres")
+        );
+    }
+
     #[test]
     fn test_compile_sequence_empty_input() {
         let mut provider = make_line_provider(&[]);

src/sed/processor.rs

@@ -423,6 +423,23 @@ fn process_file(
     output: &mut OutputBuffer,
     context: &mut ProcessingContext,
 ) -> UResult<()> {
+    // Prescan for zero-address which must produce output
+    // before any input line is read.
+    {
+        let mut current = commands.clone();
+        while let Some(cmd_rc) = current {
+            let cmd = cmd_rc.borrow();
+            if matches!(cmd.code, 'r')
+                && matches!(&cmd.addr1, Some(Address::Line(0)))
+                && cmd.addr2.is_none()
+            {
+                let path = extract_variant!(cmd, Path);
+                output.copy_file(path)?;
+            }
+            current = cmd.next.clone();
+        }
+    }
+
     // 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

@@ -774,6 +774,14 @@ 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<()> {

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