hashsum: Refactor uu_app to isolate non-"GNU Coreutils" options

Several binaries have been added to `hashsum` that have never been part
of GNU Coreutils:

- `sha3*sum` (#869)
- `shake*sum` (#987)
- `b3sum` (#3108 and #3164)

In particular, the `--bits` option, and the `--no-names` option added in
#3361, are not valid for any GNU Coreutils `*sum` binary
(as of Coreutils 9.0).

This commit refactors the argument parsing so that `--bits` and
`--no-names` become invalid options for the binaries intended to match
the GNU Coreutils API, instead of being ignored options. It also
refactors the custom binary name handling to distinguish between
binaries intended to match the GNU Coreutils API, and binaries that
don't have that constraint.

Part of #2930.

build.rs

@@ -104,6 +104,10 @@ pub fn main() {
                 );
 
                 let map_value = format!("({krate}::uumain, {krate}::uu_app_common)", krate = krate);
+                let map_value_shake =
+                    format!("({krate}::uumain, {krate}::uu_app_shake)", krate = krate);
+                let map_value_b3sum =
+                    format!("({krate}::uumain, {krate}::uu_app_b3sum)", krate = krate);
                 phf_map.entry("md5sum", &map_value);
                 phf_map.entry("sha1sum", &map_value);
                 phf_map.entry("sha224sum", &map_value);
@@ -115,10 +119,10 @@ pub fn main() {
                 phf_map.entry("sha3-256sum", &map_value);
                 phf_map.entry("sha3-384sum", &map_value);
                 phf_map.entry("sha3-512sum", &map_value);
-                phf_map.entry("shake128sum", &map_value);
-                phf_map.entry("shake256sum", &map_value);
+                phf_map.entry("shake128sum", &map_value_shake);
+                phf_map.entry("shake256sum", &map_value_shake);
                 phf_map.entry("b2sum", &map_value);
-                phf_map.entry("b3sum", &map_value);
+                phf_map.entry("b3sum", &map_value_b3sum);
                 tf.write_all(
                     format!(
                         "#[path=\"{dir}/test_{krate}.rs\"]\nmod test_{krate};\n",

src/uu/hashsum/src/hashsum.rs

@@ -54,27 +54,6 @@ struct Options {
     output_bits: usize,
 }
 
-fn is_custom_binary(program: &str) -> bool {
-    matches!(
-        program,
-        "md5sum"
-            | "sha1sum"
-            | "sha224sum"
-            | "sha256sum"
-            | "sha384sum"
-            | "sha512sum"
-            | "sha3sum"
-            | "sha3-224sum"
-            | "sha3-256sum"
-            | "sha3-384sum"
-            | "sha3-512sum"
-            | "shake128sum"
-            | "shake256sum"
-            | "b2sum"
-            | "b3sum"
-    )
-}
-
 #[allow(clippy::cognitive_complexity)]
 fn detect_algo(
     program: &str,
@@ -373,11 +352,6 @@ pub fn uu_app_common<'a>() -> Command<'a> {
                 .long("tag")
                 .help("create a BSD-style checksum"),
         )
-        .arg(
-            Arg::new("no-names")
-                .long("no-names")
-                .help("Omits filenames in the output (option not present in GNU/Coreutils)"),
-        )
         .arg(
             Arg::new("text")
                 .short('t')
@@ -408,6 +382,34 @@ pub fn uu_app_common<'a>() -> Command<'a> {
                 .long("warn")
                 .help("warn about improperly formatted checksum lines"),
         )
+        .arg(
+            Arg::new("FILE")
+                .index(1)
+                .multiple_occurrences(true)
+                .value_name("FILE")
+                .value_hint(clap::ValueHint::FilePath)
+                .allow_invalid_utf8(true),
+        )
+}
+
+pub fn uu_app_b3sum<'a>() -> Command<'a> {
+    uu_app_b3sum_opts(uu_app_common())
+}
+
+fn uu_app_b3sum_opts<'a>(command: Command<'a>) -> Command<'a> {
+    command.arg(
+        Arg::new("no-names")
+            .long("no-names")
+            .help("Omits filenames in the output (option not present in GNU/Coreutils)"),
+    )
+}
+
+pub fn uu_app_shake<'a>() -> Command<'a> {
+    uu_app_shake_opts(uu_app_common())
+}
+
+fn uu_app_shake_opts<'a>(command: Command<'a>) -> Command<'a> {
+    command
         // Needed for variable-length output sums (e.g. SHAKE)
         .arg(
             Arg::new("bits")
@@ -418,18 +420,10 @@ pub fn uu_app_common<'a>() -> Command<'a> {
                 // XXX: should we actually use validators?  they're not particularly efficient
                 .validator(is_valid_bit_num),
         )
-        .arg(
-            Arg::new("FILE")
-                .index(1)
-                .multiple_occurrences(true)
-                .value_name("FILE")
-                .value_hint(clap::ValueHint::FilePath)
-                .allow_invalid_utf8(true),
-        )
 }
 
 pub fn uu_app_custom<'a>() -> Command<'a> {
-    let mut command = uu_app_common();
+    let mut command = uu_app_b3sum_opts(uu_app_shake_opts(uu_app_common()));
     let algorithms = &[
         ("md5", "work with MD5"),
         ("sha1", "work with SHA1"),
@@ -463,10 +457,26 @@ pub fn uu_app_custom<'a>() -> Command<'a> {
 // hashsum is handled differently in build.rs, therefore this is not the same
 // as in other utilities.
 fn uu_app<'a>(binary_name: &str) -> Command<'a> {
-    if !is_custom_binary(binary_name) {
-        uu_app_custom()
-    } else {
-        uu_app_common()
+    match binary_name {
+        // These all support the same options.
+        "md5sum" | "sha1sum" | "sha224sum" | "sha256sum" | "sha384sum" | "sha512sum" => {
+            uu_app_common()
+        }
+        // b2sum supports the md5sum options plus -l/--length.
+        "b2sum" => uu_app_common(), // TODO: Implement -l/--length
+        // These have never been part of GNU Coreutils, but can function with the same
+        // options as md5sum.
+        "sha3sum" | "sha3-224sum" | "sha3-256sum" | "sha3-384sum" | "sha3-512sum" => {
+            uu_app_common()
+        }
+        // These have never been part of GNU Coreutils, and require an additional --bits
+        // option to specify their output size.
+        "shake128sum" | "shake256sum" => uu_app_shake(),
+        // b3sum has never been part of GNU Coreutils, and has a --no-names option in
+        // addition to the b2sum options.
+        "b3sum" => uu_app_b3sum(),
+        // We're probably just being called as `hashsum`, so give them everything.
+        _ => uu_app_custom(),
     }
 }