From e1626b8c64202d70f8269aafce885a952007671f Mon Sep 17 00:00:00 2001
From: Michael Kefeder <m.kefeder@gmail.com>
Date: Fri, 12 Mar 2021 21:52:37 +0100
Subject: [PATCH] ptx: ignore empty regex (#1808)

* ptx: ignore empty -W regex

* ptx: test to verify empty -W regex is ignored

* tests/ptx: verify word regexp from cmd-line is used
---
 src/uu/ptx/src/ptx.rs                         | 19 ++++++++++++++-----
 tests/by-util/test_ptx.rs                     | 16 ++++++++++++++++
 ...roff_no_ref_word_regexp_exc_space.expected |  7 +++++++
 3 files changed, 37 insertions(+), 5 deletions(-)
 create mode 100644 tests/fixtures/ptx/gnu_ext_disabled_roff_no_ref_word_regexp_exc_space.expected

diff --git a/src/uu/ptx/src/ptx.rs b/src/uu/ptx/src/ptx.rs
index 8421e52bc..bfcd6699f 100644
--- a/src/uu/ptx/src/ptx.rs
+++ b/src/uu/ptx/src/ptx.rs
@@ -95,12 +95,21 @@ impl WordFilter {
         if matches.opt_present("b") {
             crash!(1, "-b not implemented yet");
         }
-        let reg = if matches.opt_present("W") {
-            matches.opt_str("W").expect("parsing options failed!")
-        } else if config.gnu_ext {
-            "\\w+".to_owned()
+        // Ignore empty string regex from cmd-line-args
+        let arg_reg: Option<String> = if matches.opt_present("W") {
+            matches.opt_str("W").filter(|reg| !reg.is_empty())
         } else {
-            "[^ \t\n]+".to_owned()
+            None
+        };
+        let reg = match arg_reg {
+            Some(arg_reg) => arg_reg,
+            None => {
+                if config.gnu_ext {
+                    "\\w+".to_owned()
+                } else {
+                    "[^ \t\n]+".to_owned()
+                }
+            }
         };
         WordFilter {
             only_specified: o,
diff --git a/tests/by-util/test_ptx.rs b/tests/by-util/test_ptx.rs
index 77117c5c0..e44943bfa 100644
--- a/tests/by-util/test_ptx.rs
+++ b/tests/by-util/test_ptx.rs
@@ -8,6 +8,22 @@ fn gnu_ext_disabled_roff_no_ref() {
         .stdout_only_fixture("gnu_ext_disabled_roff_no_ref.expected");
 }
 
+#[test]
+fn gnu_ext_disabled_roff_no_ref_empty_word_regexp() {
+    new_ucmd!()
+        .args(&["-G", "-R", "-W", "", "input"])
+        .succeeds()
+        .stdout_only_fixture("gnu_ext_disabled_roff_no_ref.expected");
+}
+
+#[test]
+fn gnu_ext_disabled_roff_no_ref_word_regexp_exc_space() {
+    new_ucmd!()
+        .args(&["-G", "-R", "-W", "[^\t\n]+", "input"])
+        .succeeds()
+        .stdout_only_fixture("gnu_ext_disabled_roff_no_ref_word_regexp_exc_space.expected");
+}
+
 #[test]
 fn gnu_ext_disabled_roff_input_ref() {
     new_ucmd!()
diff --git a/tests/fixtures/ptx/gnu_ext_disabled_roff_no_ref_word_regexp_exc_space.expected b/tests/fixtures/ptx/gnu_ext_disabled_roff_no_ref_word_regexp_exc_space.expected
new file mode 100644
index 000000000..141e6c9f8
--- /dev/null
+++ b/tests/fixtures/ptx/gnu_ext_disabled_roff_no_ref_word_regexp_exc_space.expected
@@ -0,0 +1,7 @@
+.xx "" "" """quotes"", for roff" ""
+.xx "" "" "and some other like %a, b#, c$c" ""
+.xx "" "" "hello world!" ""
+.xx "" "" "let's check special characters:" ""
+.xx "" "" "maybe also~or^" ""
+.xx "" "" "oh, and back\slash" ""
+.xx "" "" "{brackets} for tex" ""