ptx: match GNU default behavior by skipping non-alphabetic index tokens (#10919)

* ptx: skip non-alphabetic tokens in default word matching

* Simplify no output operation

Co-authored-by: Daniel Hofstetter <daniel.hofstetter@42dh.com>

* fix: Keyword must start at first alphabetic char

* Simplify find first alphabet

---------

Co-authored-by: Daniel Hofstetter <daniel.hofstetter@42dh.com>

Author: Xylphy

src/uu/ptx/src/ptx.rs

@@ -356,7 +356,19 @@ fn create_word_set(config: &Config, filter: &WordFilter, file_map: &FileMap) ->
             };
             // match words with given regex
             for mat in reg.find_iter(line) {
-                let (beg, end) = (mat.start(), mat.end());
+                let (mut beg, end) = (mat.start(), mat.end());
+
+                // GNU-compatible default behavior:
+                // with default regexp, keyword must start at first alphabetic char.
+                if filter.word_regex == Config::default().context_regex {
+                    let matched = &line[beg..end];
+                    if let Some(pos) = matched.find(|c: char| c.is_alphabetic()) {
+                        beg += pos;
+                    } else {
+                        continue;
+                    }
+                }
+
                 if config.input_ref && ((beg, end) == (ref_beg, ref_end)) {
                     continue;
                 }

tests/by-util/test_ptx.rs

@@ -300,6 +300,11 @@ fn test_gnu_mode_dumb_format() {
     new_ucmd!().pipe_in("a b").succeeds().stdout_only(
         "                                       a b\n                                   a   b\n",
     );
+
+    new_ucmd!()
+        .pipe_in("2a")
+        .succeeds()
+        .stdout_only(format!("{}2   a\n", " ".repeat(35)));
 }
 
 #[test]
@@ -330,6 +335,15 @@ fn test_unicode_padding_alignment() {
         .stdout_only("        a\n        é\n");
 }
 
+#[test]
+fn test_gnu_compat_numeric_token_with_emoji_produces_no_index() {
+    // GNU ptx produces no output for this input in default mode.
+    new_ucmd!()
+        .pipe_in("012345678901234567890123456789🛠\n")
+        .succeeds()
+        .no_output();
+}
+
 #[test]
 fn test_unicode_truncation_alignment() {
     new_ucmd!()