Make String#{strip,lstrip}{,!} strip leading NUL bytes

The documentation already specifies that they strip whitespace
and defines whitespace to include null.

This wraps the new behavior in the appropriate guards in the specs,
but does not specify behavior for previous versions, because this
is a bug that could be backported.

Fixes [Bug #17467]

4 files changed, 40 insertions, 20 deletions

diff --git a/spec/ruby/core/string/lstrip_spec.rb b/spec/ruby/core/string/lstrip_spec.rb
index 12c7b13200..6b40189500 100644
--- a/spec/ruby/core/string/lstrip_spec.rb
+++ b/spec/ruby/core/string/lstrip_spec.rb
@@ -7,11 +7,13 @@ describe "String#lstrip" do
    "  hello world  ".lstrip.should == "hello world  "
    "\n\r\t\n\v\r hello world  ".lstrip.should == "hello world  "
    "hello".lstrip.should == "hello"
-   "\000 \000hello\000 \000".lstrip.should == "\000 \000hello\000 \000"
   end
 
-  it "does not strip leading \\0" do
-   "\x00hello".lstrip.should == "\x00hello"
+  ruby_version_is '3.1' do
+    it "strips leading \\0" do
+     "\x00hello".lstrip.should == "hello"
+     "\000 \000hello\000 \000".lstrip.should == "hello\000 \000"
+    end
   end
 
   ruby_version_is ''...'2.7' do
@@ -28,10 +30,14 @@ describe "String#lstrip!" do
     a = "  hello  "
     a.lstrip!.should equal(a)
     a.should == "hello  "
+  end
 
-    a = "\000 \000hello\000 \000"
-    a.lstrip!
-    a.should == "\000 \000hello\000 \000"
+  ruby_version_is '3.1' do
+    it "strips leading \\0" do
+      a = "\000 \000hello\000 \000"
+      a.lstrip!
+      a.should == "hello\000 \000"
+    end
   end
 
   it "returns nil if no modifications were made" do
diff --git a/spec/ruby/core/string/strip_spec.rb b/spec/ruby/core/string/strip_spec.rb
index 252d4a9cc3..463a9fedf3 100644
--- a/spec/ruby/core/string/strip_spec.rb
+++ b/spec/ruby/core/string/strip_spec.rb
@@ -6,11 +6,12 @@ describe "String#strip" do
     "   hello   ".strip.should == "hello"
     "   hello world   ".strip.should == "hello world"
     "\tgoodbye\r\v\n".strip.should == "goodbye"
-    "\x00 goodbye \x00".strip.should == "\x00 goodbye"
   end
 
-  it "returns a copy of self with trailing NULL bytes and whitespace" do
-    " \x00 goodbye \x00 ".strip.should == "\x00 goodbye"
+  ruby_version_is '3.1' do
+    it "returns a copy of self without leading and trailing NULL bytes and whitespace" do
+      " \x00 goodbye \x00 ".strip.should == "goodbye"
+    end
   end
 
   ruby_version_is ''...'2.7' do
@@ -31,11 +32,6 @@ describe "String#strip!" do
     a = "\tgoodbye\r\v\n"
     a.strip!
     a.should == "goodbye"
-
-    a = "\000 goodbye \000"
-    a.strip!
-    a.should == "\000 goodbye"
-
   end
 
   it "returns nil if no modifications where made" do
@@ -44,10 +40,12 @@ describe "String#strip!" do
     a.should == "hello"
   end
 
-  it "modifies self removing trailing NULL bytes and whitespace" do
-    a = " \x00 goodbye \x00 "
-    a.strip!
-    a.should == "\x00 goodbye"
+  ruby_version_is '3.1' do
+    it "removes leading and trailing NULL bytes and whitespace" do
+      a = "\000 goodbye \000"
+      a.strip!
+      a.should == "goodbye"
+    end
   end
 
   it "raises a FrozenError on a frozen instance that is modified" do
diff --git a/string.c b/string.c
index dbff149840..7605c225b9 100644
--- a/string.c
+++ b/string.c
@@ -9263,14 +9263,14 @@ lstrip_offset(VALUE str, const char *s, const char *e, rb_encoding *enc)
 
     /* remove spaces at head */
     if (single_byte_optimizable(str)) {
-	while (s < e && ascii_isspace(*s)) s++;
+        while (s < e && (*s == '\0' || ascii_isspace(*s))) s++;
     }
     else {
 	while (s < e) {
 	    int n;
 	    unsigned int cc = rb_enc_codepoint_len(s, e, &n, enc);
 
-	    if (!rb_isspace(cc)) break;
+            if (cc && !rb_isspace(cc)) break;
 	    s += n;
 	}
     }
diff --git a/test/ruby/test_string.rb b/test/ruby/test_string.rb
index 4814872789..29ad98beaf 100644
--- a/test/ruby/test_string.rb
+++ b/test/ruby/test_string.rb
@@ -1852,6 +1852,7 @@ CODE
   def test_strip
     assert_equal(S("x"), S("      x        ").strip)
     assert_equal(S("x"), S(" \n\r\t     x  \t\r\n\n      ").strip)
+    assert_equal(S("x"), S("\x00x\x00").strip)
 
     assert_equal("0b0 ".force_encoding("UTF-16BE"),
                  "\x00 0b0 ".force_encoding("UTF-16BE").strip)
@@ -1870,6 +1871,10 @@ CODE
     assert_equal(S("x"), a.strip!)
     assert_equal(S("x"), a)
 
+    a = S("\x00x\x00")
+    assert_equal(S("x"), a.strip!)
+    assert_equal(S("x"), a)
+
     a = S("x")
     assert_nil(a.strip!)
     assert_equal(S("x") ,a)
@@ -2703,6 +2708,7 @@ CODE
   def test_rstrip
     assert_equal("  hello", "  hello  ".rstrip)
     assert_equal("\u3042", "\u3042   ".rstrip)
+    assert_equal("\u3042", "\u3042\u0000".rstrip)
     assert_raise(Encoding::CompatibilityError) { "\u3042".encode("ISO-2022-JP").rstrip }
   end
 
@@ -2723,12 +2729,17 @@ CODE
     assert_equal(nil, s4.rstrip!)
     assert_equal("\u3042", s4)
 
+    s5 = S("\u3042\u0000")
+    assert_equal("\u3042", s5.rstrip!)
+    assert_equal("\u3042", s5)
+
     assert_raise(Encoding::CompatibilityError) { "\u3042".encode("ISO-2022-JP").rstrip! }
   end
 
   def test_lstrip
     assert_equal("hello  ", "  hello  ".lstrip)
     assert_equal("\u3042", "   \u3042".lstrip)
+    assert_equal("hello  ", "\x00hello  ".lstrip)
   end
 
   def test_lstrip_bang
@@ -2747,6 +2758,11 @@ CODE
     s4 = S("\u3042")
     assert_equal(nil, s4.lstrip!)
     assert_equal("\u3042", s4)
+
+    s5 = S("\u0000\u3042")
+    assert_equal("\u3042", s5.lstrip!)
+    assert_equal("\u3042", s5)
+
   end
 
   def test_delete_prefix