Fix #372 quote simplemrs preds with reserved chars

CHANGELOG.md

@@ -17,6 +17,7 @@
 * Stop using Python's undocumented parse_template re function ([#378])
 * Resolved newline and encoding issues for running tests on Windows ([#379])
 * Capture legacy single-quoted predicates longer than 1 character ([#373])
+* Quote SimpleMRS predicates with reserved characters ([#372])
 
 
 ## [v1.8.1]
@@ -1621,6 +1622,7 @@ information about changes, except for
 [#360]: https://github.com/delph-in/pydelphin/issues/360
 [#364]: https://github.com/delph-in/pydelphin/issues/364
 [#367]: https://github.com/delph-in/pydelphin/issues/367
+[#372]: https://github.com/delph-in/pydelphin/issues/372
 [#373]: https://github.com/delph-in/pydelphin/issues/373
 [#374]: https://github.com/delph-in/pydelphin/issues/374
 [#375]: https://github.com/delph-in/pydelphin/issues/375

delphin/codecs/simplemrs.py

@@ -4,6 +4,7 @@
 
 from pathlib import Path
 from typing import Optional
+import re
 
 from delphin.util import Lexer
 from delphin import predicate
@@ -361,7 +362,7 @@ def _encode_rels(rels, varprops, lnk, indent):
     delim = ('\n  ' + ' ' * len('RELS: < ')) if indent else ' '
     tokens = []
     for rel in rels:
-        pred = rel.predicate
+        pred = _encode_predicate(rel.predicate)
         if lnk:
             pred += str(rel.lnk)
         reltoks = ['[', pred]
@@ -382,6 +383,12 @@ def _encode_rels(rels, varprops, lnk, indent):
     return tokens
 
 
+def _encode_predicate(predicate: str) -> str:
+    if re.search(r"[\s\"':<>[\]]", predicate):
+        return f'"{_escape(predicate)}"'
+    return predicate
+
+
 def _encode_hcons(hcons):
     tokens = ['{} {} {}'.format(hc.hi, hc.relation, hc.lo)
               for hc in hcons]

tests/codecs/simplemrs_test.py

@@ -86,3 +86,24 @@ def test_legacy_single_quote_predicates_issue_373():
     # https://github.com/delph-in/pydelphin/issues/373
     m = simplemrs.decode("[ RELS: < [ 'single+quoted LBL: h0 ] > ]")
     assert m.rels[0].predicate == "single+quoted"
+
+
+def test_quote_reserved_characters_issue_372():
+    # https://github.com/delph-in/pydelphin/issues/372
+
+    def assert_quoted(p: str, escape: bool = False):
+        m = simplemrs.decode(f'[ RELS: < [ "{p}"<1:2> LBL: h0 ] > ]')
+        _p = m.rels[0].predicate
+        assert (_p.replace('"', r'\"') if escape else _p) == p
+        s = simplemrs.encode(m)
+        assert f'"{p}"' in s
+        simplemrs.decode(s)  # confirm it roundtrips without error
+
+    assert_quoted("a space")
+    assert_quoted("a:colon")
+    assert_quoted(r'double\"quotes', escape=True)
+    assert_quoted("single'quotes")
+    assert_quoted("left<angle")
+    assert_quoted("right>angle")
+    assert_quoted("left[bracket")
+    assert_quoted("right]bracket")