Fix secondary tag encoding. See #449449_secondary_tag_url_enc

The second exclamation mark was getting url encoded, when it
should be preserved. Added a regression test.

4 files changed, 27 insertions, 4 deletions

diff --git a/_test/test_issues.py b/_test/test_issues.py
index 736dccb..dce77e5 100644
--- a/_test/test_issues.py
+++ b/_test/test_issues.py
@@ -862,6 +862,12 @@ class TestIssues:
         """
         d = na_round_trip(inp)  # NOQA
 
+    def test_issue_449(self):
+        inp = """\
+        emoji_index: !!python/name:materialx.emoji.twemoji
+        """
+        d = na_round_trip(inp)  # NOQA
+
 #    @pytest.mark.xfail(strict=True, reason='bla bla', raises=AssertionError)
 #    def test_issue_ xxx(self):
 #        inp = """
diff --git a/_test/test_tag.py b/_test/test_tag.py
index 3fd1e05..ec086f3 100644
--- a/_test/test_tag.py
+++ b/_test/test_tag.py
@@ -64,6 +64,15 @@ class TestIndentFailures:
         language: python
         """)
 
+    def test_spec_6_26_tag_shorthands(self):
+        round_trip("""\
+        %TAG !e! tag:example.com,2000:app/
+        ---
+        - !local foo
+        - !!str bar
+        - !e!tag%21 baz
+        """)
+
 
 class TestRoundTripCustom:
     def test_X1(self):
diff --git a/emitter.py b/emitter.py
index f9611ee..2a0d3ca 100644
--- a/emitter.py
+++ b/emitter.py
@@ -97,6 +97,7 @@ class Emitter:
     DEFAULT_TAG_PREFIXES = {
         '!': '!',
         'tag:yaml.org,2002:': '!!',
+        '!!': '!!',
     }
     # fmt: on
 
@@ -1012,7 +1013,7 @@ class Emitter:
         # type: (Any) -> Any
         if not tag:
             raise EmitterError('tag must not be empty')
-        if tag == '!':
+        if tag == '!' or tag == '!!':
             return tag
         handle = None
         suffix = tag
diff --git a/scanner.py b/scanner.py
index 61cae63..86f7caf 100644
--- a/scanner.py
+++ b/scanner.py
@@ -1098,6 +1098,13 @@ class Scanner:
         srp = self.reader.peek
         start_mark = self.reader.get_mark()
         ch = srp(1)
+        short_handle = '!'
+        if ch == '!':
+            short_handle = '!!'
+            self.reader.forward()
+            srp = self.reader.peek
+            ch = srp(1)
+
         if ch == '<':
             handle = None
             self.reader.forward(2)
@@ -1112,7 +1119,7 @@ class Scanner:
             self.reader.forward()
         elif ch in _THE_END_SPACE_TAB:
             handle = None
-            suffix = '!'
+            suffix = short_handle
             self.reader.forward()
         else:
             length = 1
@@ -1123,11 +1130,11 @@ class Scanner:
                     break
                 length += 1
                 ch = srp(length)
-            handle = '!'
+            handle = short_handle
             if use_handle:
                 handle = self.scan_tag_handle('tag', start_mark)
             else:
-                handle = '!'
+                handle = short_handle
                 self.reader.forward()
             suffix = self.scan_tag_uri('tag', start_mark)
         ch = srp()