Add address range support for a, i and = and check for invalid POSIX flag usage with GNU only extensions

- Add a context parameter in get_verified_cmd_spec function and update the relative test cases
- Add integration tests for address range

Author: LoukasPap

src/sed/command.rs

@@ -70,6 +70,13 @@ pub struct ProcessingContext {
     pub append_elements: Vec<AppendElement>,
 }
 
+impl ProcessingContext {
+    /// Check if POSIX extension is enabled
+    pub fn is_posix_enabled(&self) -> bool {
+        self.posix
+    }
+}
+
 #[derive(Clone, Debug)]
 /// Elements that shall be appended at the end of each command processing cycle
 pub enum AppendElement {

src/sed/compiler.rs

@@ -264,14 +264,14 @@ fn compile_sequence(
         let mut cmd = Rc::new(RefCell::new(Command::at_position(lines, line)));
         let n_addr = compile_address_range(lines, line, &mut cmd, context)?;
         line.eat_spaces();
-        let mut cmd_spec = get_verified_cmd_spec(lines, line, n_addr)?;
+        let mut cmd_spec = get_verified_cmd_spec(lines, line, n_addr, context)?;
 
         // Compile the command according to its specification.
         let mut cmd_mut = cmd.borrow_mut();
         cmd_mut.code = line.current();
         match (cmd_spec.handler)(lines, line, &mut cmd_mut, context)? {
             CommandHandling::GetNext => {
-                cmd_spec = get_verified_cmd_spec(lines, line, n_addr)?;
+                cmd_spec = get_verified_cmd_spec(lines, line, n_addr, context)?;
                 cmd_mut.code = line.current();
                 (cmd_spec.handler)(lines, line, &mut cmd_mut, context)?;
             }
@@ -1175,6 +1175,7 @@ fn get_verified_cmd_spec(
     lines: &ScriptLineProvider,
     line: &ScriptCharProvider,
     n_addr: usize,
+    context: &ProcessingContext,
 ) -> UResult<CommandSpec> {
     if line.eol() {
         return compilation_error(lines, line, "command expected");
@@ -1194,9 +1195,18 @@ fn get_verified_cmd_spec(
         );
     }
 
+    if context.is_posix_enabled() && is_gnu_extension(ch, n_addr) {
+        return compilation_error(lines, line, "command only uses one address");
+    }
+
     Ok(cmd_spec)
 }
 
+/// Check if the command usage is a GNU extension (not allowed in POSIX mode)
+fn is_gnu_extension(cmd_code: char, n_addr: usize) -> bool {
+    matches!(cmd_code, 'a' | 'i' | '=' | 'l' | 'q' | 'r') && n_addr == 2
+}
+
 // Look up a command addresses and handler by its command code.
 fn get_cmd_spec(
     lines: &ScriptLineProvider,
@@ -1209,7 +1219,7 @@ fn get_cmd_spec(
             handler: compile_negation_command,
         }),
         '=' => Ok(CommandSpec {
-            n_addr: 1,
+            n_addr: 2,
             handler: compile_empty_command,
         }),
         ':' => Ok(CommandSpec {
@@ -1225,7 +1235,7 @@ fn get_cmd_spec(
             handler: compile_end_group_command,
         }),
         'a' | 'i' => Ok(CommandSpec {
-            n_addr: 1,
+            n_addr: 2,
             handler: compile_text_command,
         }),
         'b' | 't' => Ok(CommandSpec {
@@ -1382,7 +1392,7 @@ mod tests {
     fn test_missing_command_character() {
         let lines = ScriptLineProvider::with_active_state("test.sed", 1);
         let line = char_provider_from("");
-        let result = get_verified_cmd_spec(&lines, &line, 0);
+        let result = get_verified_cmd_spec(&lines, &line, 0, &ctx());
 
         assert!(result.is_err());
         let msg = result.unwrap_err().to_string();
@@ -1393,7 +1403,7 @@ mod tests {
     fn test_invalid_command_character() {
         let lines = ScriptLineProvider::with_active_state("script.sed", 2);
         let line = char_provider_from("@");
-        let result = get_verified_cmd_spec(&lines, &line, 0);
+        let result = get_verified_cmd_spec(&lines, &line, 0, &ctx());
 
         assert!(result.is_err());
         let msg = result.unwrap_err().to_string();
@@ -1404,7 +1414,7 @@ mod tests {
     fn test_too_many_addresses() {
         let lines = ScriptLineProvider::with_active_state("input.sed", 3);
         let line = char_provider_from("q"); // q takes one address
-        let result = get_verified_cmd_spec(&lines, &line, 2);
+        let result = get_verified_cmd_spec(&lines, &line, 2, &ctx());
 
         assert!(result.is_err());
         let msg = result.unwrap_err().to_string();
@@ -1417,7 +1427,7 @@ mod tests {
     fn test_valid_command_spec() {
         let lines = ScriptLineProvider::with_active_state("input.sed", 4);
         let line = char_provider_from("a"); // valid command
-        let result = get_verified_cmd_spec(&lines, &line, 1);
+        let result = get_verified_cmd_spec(&lines, &line, 1, &ctx());
 
         assert!(result.is_ok());
         let spec = result.unwrap();

tests/by-util/test_sed.rs

@@ -581,6 +581,30 @@ tb"#,
 
 // Check both POSIX and GNU parsing routines.
 
+check_output_posix!(
+    text_simple_insert,
+    [
+        "-e",
+        r#"
+2,4i\
+extra
+"#,
+        LINES1
+    ]
+);
+
+check_output_posix!(
+    text_simple_append,
+    [
+        "-e",
+        r#"
+2,4a\
+extra
+"#,
+        LINES1
+    ]
+);
+
 check_output_posix!(
     text_insert_quit,
     [
@@ -772,6 +796,8 @@ fn write_two_files() -> std::io::Result<()> {
 // =, l commands
 check_output!(number_continuous, ["/l2_/=", LINES1, LINES2]);
 check_output!(number_separate, ["-s", "/l._8/=", LINES1, LINES2]);
+check_output!(number_range, ["-e", "10,12=", LINES1]); // Correct output but test is not passing  
+check_output!(number_range_out_of_bounds, ["-e", "47,60=", LINES1]);
 
 check_output!(list_ascii, ["-n", "l 60", "input/ascii"]);
 check_output!(list_empty, ["-n", "l 60", "input/empty"]);

tests/fixtures/sed/output/number_range

@@ -0,0 +1,17 @@
+l1_1
+l1_2
+l1_3
+l1_4
+l1_5
+l1_6
+l1_7
+l1_8
+l1_9
+10
+l1_10
+11
+l1_11
+12
+l1_12
+l1_13
+l1_14

tests/fixtures/sed/output/number_range_out_of_bounds

@@ -0,0 +1,14 @@
+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

tests/fixtures/sed/output/text_simple_append

@@ -0,0 +1,17 @@
+l1_1
+l1_2
+extra
+l1_3
+extra
+l1_4
+extra
+l1_5
+l1_6
+l1_7
+l1_8
+l1_9
+l1_10
+l1_11
+l1_12
+l1_13
+l1_14

tests/fixtures/sed/output/text_simple_insert

@@ -0,0 +1,17 @@
+l1_1
+extra
+l1_2
+extra
+l1_3
+extra
+l1_4
+l1_5
+l1_6
+l1_7
+l1_8
+l1_9
+l1_10
+l1_11
+l1_12
+l1_13
+l1_14