docs: move extensions and incompatibilities to docs/src/extensions.md

Start an mdBook under docs/, laid out as in coreutils, and move the list of
extensions and incompatibilities out of the README into it. The README keeps
a pointer.

A Documentation workflow builds the book with mdbook 0.5.0, the version the
uutils website uses, whenever docs/ changes, and uploads it as an artifact.
mdbook 0.5 rejects the `multilingual` field coreutils still carries (the
website strips it before building), so the book leaves it out.

Cargo.lock

@@ -1103,7 +1103,7 @@ dependencies = [
 
 [[package]]
 name = "sed"
-version = "0.3.0"
+version = "0.2.0"
 dependencies = [
  "assert_fs",
  "chrono",

Cargo.toml

@@ -1,12 +1,11 @@
 # sed (uutils)
 # * see the repository LICENSE, README, and CONTRIBUTING files for more information
 
-# spell-checker:ignore (libs) mangen sed binstall chrono codspeed libc memchr memmap rlimit rustix textwrap uucore uutests xattr
-# spell-checker:ignore (misc) clippy codegen msvc startswith uninlined uutils
+# spell-checker:ignore (libs) mangen sed
 
 [package]
 name = "sed"
-version = "0.3.0"
+version = "0.2.0"
 license = "MIT"
 description = "sed ~ implemented as universal (cross-platform) utils, written in Rust"
 default-run = "sed"

fuzz/Cargo.lock

@@ -897,7 +897,7 @@ checksum = "b39cdef0fa800fc44525c84ccb54a029961a8215f9619753635a9c0d2538d46d"
 
 [[package]]
 name = "sed"
-version = "0.3.0"
+version = "0.2.0"
 dependencies = [
  "clap",
  "clap_complete",

src/sed/compiler.rs

@@ -359,7 +359,7 @@ fn compile_address_range(
         let is_step_match = line.current() == '~'; // E.g. 0~2: Pick even-numbered lines
         line.advance();
         line.eat_spaces();
-        let is_step_end = if !line.eol() && line.current() == '~' {
+        let is_step_end = if line.current() == '~' {
             // E.g. /foo/,~10: Start at foo, include all lines until multiple of 10 is reached.
             line.advance();
             line.eat_spaces();
@@ -373,46 +373,39 @@ fn compile_address_range(
             return compilation_error(lines, line, "~step is invalid in POSIX mode");
         }
 
-        // Look for second address; a missing ~step counts as 0.
-        let has_addr2 = !line.eol() && (is_address_char(line.current()) || line.current() == '+');
-        if !(has_addr2 || is_step_match || is_step_end) {
-            return compilation_error(lines, line, "unexpected `,'");
-        }
-        let addr2 = if has_addr2 {
-            compile_address(lines, line, context)?
-        } else {
-            Address::Line(0)
-        };
-        // Set step_n to the number specified in the (required numeric) address.
-        let step_n = if is_step_match || is_step_end {
-            match addr2 {
-                Address::Line(n) => n,
-                _ => {
-                    return compilation_error(
-                        lines,
-                        line,
-                        "~step can only be specified through numeric values",
-                    );
+        // Look for second address.
+        if !line.eol() {
+            let addr2 = compile_address(lines, line, context)?;
+            // Set step_n to the number specified in the (required numeric) address.
+            let step_n = if is_step_match || is_step_end {
+                match addr2 {
+                    Address::Line(n) => n,
+                    _ => {
+                        return compilation_error(
+                            lines,
+                            line,
+                            "~step can only be specified through numeric values",
+                        );
+                    }
                 }
+            } else {
+                0 // dummy, not used
+            };
+
+            if is_line0 && !matches!(addr2, Address::Re(_)) && !is_step_match {
+                return compilation_error(lines, line, ERR_ADDRESS_0_USAGE);
             }
-        } else {
-            0 // dummy, not used
-        };
 
-        // Address 0 needs a regex end or a positive ~step (0~0 selects no line).
-        if is_line0 && !matches!(addr2, Address::Re(_)) && !(is_step_match && step_n > 0) {
-            return compilation_error(lines, line, ERR_ADDRESS_0_USAGE);
+            // If needed, transform Address::Line into Address::Step*.
+            cmd.addr2 = if is_step_match {
+                Some(Address::StepMatch(step_n))
+            } else if is_step_end {
+                Some(Address::StepEnd(step_n))
+            } else {
+                Some(addr2)
+            };
+            n_addr += 1;
         }
-
-        // If needed, transform Address::Line into Address::Step*.
-        cmd.addr2 = if is_step_match {
-            Some(Address::StepMatch(step_n))
-        } else if is_step_end {
-            Some(Address::StepEnd(step_n))
-        } else {
-            Some(addr2)
-        };
-        n_addr += 1;
     }
 
     // Zero-address read command check
@@ -503,7 +496,7 @@ fn compile_address(
             let number = parse_number(lines, line, true)?.unwrap();
             Ok(Address::Line(number))
         }
-        _ => compilation_error(lines, line, "expected context address"),
+        _ => panic!("invalid context address"),
     }
 }
 

src/sed/processor.rs

@@ -129,7 +129,7 @@ fn applies(
                     Address::Line(n) if linenum >= *n => {
                         context.last_address = true;
                     }
-                    Address::RelLine(0) | Address::StepMatch(0) | Address::StepEnd(0) => {
+                    Address::RelLine(n) if *n == 0 => {
                         context.last_address = true;
                     }
                     _ => {

src/sed/script_line_provider.rs

@@ -40,12 +40,7 @@ enum State {
         input_name: String,       // Input description (path or script string)
         line_number: usize,       // Current line number
     },
-    // All scripts have been processed; keeps the last position for errors
-    // raised at the end of the script.
-    Done {
-        input_name: String,
-        line_number: usize,
-    },
+    Done, // All scripts have been processed
 }
 
 impl ScriptLineProvider {
@@ -60,18 +55,16 @@ impl ScriptLineProvider {
     /// Return the currently processed script line number.
     pub fn get_line_number(&self) -> usize {
         match &self.state {
-            State::Active { line_number, .. } | State::Done { line_number, .. } => *line_number,
-            State::NotStarted => 0,
+            State::Active { line_number, .. } => *line_number,
+            _ => 0,
         }
     }
 
     /// Return the currently processed script descriptive name.
     pub fn get_input_name(&self) -> &str {
         match &self.state {
-            State::Active { input_name, .. } | State::Done { input_name, .. } => {
-                input_name.as_str()
-            }
-            State::NotStarted => "",
+            State::Active { input_name, .. } => input_name.as_str(),
+            _ => "",
         }
     }
 
@@ -101,7 +94,7 @@ impl ScriptLineProvider {
                         return Ok(Some(line));
                     }
                 }
-                State::Done { .. } => {
+                State::Done => {
                     return Ok(None);
                 }
             };
@@ -115,23 +108,7 @@ impl ScriptLineProvider {
     // Move to the next available script source.
     fn advance_source(&mut self, next_index: usize) -> UResult<()> {
         if next_index >= self.sources.len() {
-            let (input_name, line_number) =
-                match std::mem::replace(&mut self.state, State::NotStarted) {
-                    State::Active {
-                        input_name,
-                        line_number,
-                        ..
-                    }
-                    | State::Done {
-                        input_name,
-                        line_number,
-                    } => (input_name, line_number),
-                    State::NotStarted => (String::new(), 0),
-                };
-            self.state = State::Done {
-                input_name,
-                line_number,
-            };
+            self.state = State::Done;
             return Ok(());
         }
 
@@ -174,14 +151,7 @@ impl fmt::Debug for State {
     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
         match self {
             State::NotStarted => f.debug_struct("NotStarted").finish(),
-            State::Done {
-                input_name,
-                line_number,
-            } => f
-                .debug_struct("Done")
-                .field("input_name", input_name)
-                .field("line_number", line_number)
-                .finish(),
+            State::Done => f.debug_struct("Done").finish(),
             State::Active {
                 index,
                 input_name,

tests/by-util/test_sed.rs

@@ -2487,20 +2487,7 @@ fn test_incomplete_test_command_posix() {
         .args(&["--posix", "i\\"])
         .fails()
         .code_is(1)
-        .stderr_is("sed: <script argument 1>:1:3: error: incomplete command\n");
-}
-
-#[test]
-fn test_error_at_end_of_script_keeps_its_position() {
-    // The script is exhausted when the error is found: report its last line,
-    // as GNU sed does ("-e expression #1").
-    new_ucmd!()
-        .args(&["s/a/b\\"])
-        .fails()
-        .code_is(1)
-        .stderr_is(
-            "sed: <script argument 1>:1:7: error: unterminated substitute replacement (unexpected EOF)\n",
-        );
+        .stderr_is("sed: :0:3: error: incomplete command\n");
 }
 
 #[test]
@@ -2559,74 +2546,6 @@ fn test_step_end_non_posix() {
         .stderr_is("sed: <script argument 1>:1:4: error: ~step is invalid in POSIX mode\n");
 }
 
-#[test]
-fn test_step_without_number_is_zero() {
-    // `N~` and `N,~` select line N alone, like a step of 0.
-    for script in ["3~=", "3,~=", "3,~0="] {
-        new_ucmd!()
-            .args(&["-n", script])
-            .pipe_in("q\nr\ns\nt\n")
-            .succeeds()
-            .stdout_is("3\n");
-    }
-}
-
-#[test]
-fn test_step_match_zero_closes_range() {
-    // `3~0` selects line 3 alone, so `c` prints its text there.
-    new_ucmd!()
-        .args(&["3~0c\\X"])
-        .pipe_in("1\n2\n3\n4\n5\n")
-        .succeeds()
-        .stdout_is("1\n2\nX\n4\n5\n");
-}
-
-#[test]
-fn test_step_non_numeric() {
-    for (script, column) in [("1~/x/p", 6), ("1,~/x/p", 7)] {
-        new_ucmd!()
-            .args(&[script])
-            .fails()
-            .code_is(1)
-            .stderr_is(format!(
-                "sed: <script argument 1>:1:{column}: error: ~step can only be specified through numeric values\n"
-            ));
-    }
-}
-
-#[test]
-fn test_step_without_number_at_end_of_line() {
-    new_ucmd!()
-        .args(&["7~"])
-        .fails()
-        .code_is(1)
-        .stderr_is("sed: <script argument 1>:1:3: error: command expected\n");
-}
-
-#[test]
-fn test_addr0_step_zero() {
-    for script in ["0~0p", "0~p"] {
-        new_ucmd!()
-            .args(&[script])
-            .fails()
-            .code_is(1)
-            .stderr_contains("address 0 can only be used with ~step");
-    }
-}
-
-#[test]
-fn test_comma_without_second_address() {
-    for (script, column) in [("5,", 3), ("5,=", 3), ("5, x", 4)] {
-        new_ucmd!()
-            .args(&[script])
-            .fails()
-            .code_is(1)
-            .stderr_is(format!(
-                "sed: <script argument 1>:1:{column}: error: unexpected `,'\n"
-            ));
-    }
-}
-
 // The following test diverse ways in which regexes are matched.
 // Search for 'regex\.' to find them in the code.
 #[test]