Interpolation grammar: allow pipe | in unquoted strings

news/799.feature

@@ -0,0 +1 @@
+Enable the use of the pipe symbol `|` in unquoted strings when parsing interpolations.

omegaconf/grammar/OmegaConfGrammarLexer.g4

@@ -65,7 +65,7 @@ BOOL:
 
 NULL: [Nn][Uu][Ll][Ll];
 
-UNQUOTED_CHAR: [/\-\\+.$%*@?];  // other characters allowed in unquoted strings
+UNQUOTED_CHAR: [/\-\\+.$%*@?|];  // other characters allowed in unquoted strings
 ID: (CHAR|'_') (CHAR|DIGIT|'_')*;
 ESC: (ESC_BACKSLASH | '\\(' | '\\)' | '\\[' | '\\]' | '\\{' | '\\}' |
       '\\:' | '\\=' | '\\,' | '\\ ' | '\\\t')+;

omegaconf/grammar/OmegaConfGrammarParser.g4

@@ -70,7 +70,7 @@ primitive:
       | INT                                    // 0, 10, -20, 1_000_000
       | FLOAT                                  // 3.14, -20.0, 1e-1, -10e3
       | BOOL                                   // true, TrUe, false, False
-      | UNQUOTED_CHAR                          // /, -, \, +, ., $, %, *, @
+      | UNQUOTED_CHAR                          // /, -, \, +, ., $, %, *, @, ?, |
       | COLON                                  // :
       | ESC                                    // \\, \(, \), \[, \], \{, \}, \:, \=, \ , \\t, \,
       | WS                                     // whitespaces
@@ -84,7 +84,7 @@ dictKey:
       | INT                                    // 0, 10, -20, 1_000_000
       | FLOAT                                  // 3.14, -20.0, 1e-1, -10e3
       | BOOL                                   // true, TrUe, false, False
-      | UNQUOTED_CHAR                          // /, -, \, +, ., $, %, *, @
+      | UNQUOTED_CHAR                          // /, -, \, +, ., $, %, *, @, ?, |
       | ESC                                    // \\, \(, \), \[, \], \{, \}, \:, \=, \ , \\t, \,
       | WS                                     // whitespaces
     )+;

omegaconf/grammar_parser.py

@@ -27,14 +27,17 @@
 _node_inter = f"\\${{\\s*{_node_path}\\s*}}"  # node interpolation ${foo.bar}
 _id = "[a-zA-Z_]\\w*"  # foo, foo_bar, abc123
 _resolver_name = f"({_id}(\\.{_id})*)?"  # foo, ns.bar3, ns_1.ns_2.b0z
-_arg = "[a-zA-Z_0-9/\\-\\+.$%*@]+"  # string representing a resolver argument
+_arg = "[a-zA-Z_0-9/\\-\\+.$%*@?|]+"  # string representing a resolver argument
 _args = f"{_arg}(\\s*,\\s*{_arg})*"  # list of resolver arguments
 _resolver_inter = f"\\${{\\s*{_resolver_name}\\s*:\\s*{_args}?\\s*}}"  # ${foo:bar}
 _inter = f"({_node_inter}|{_resolver_inter})"  # any kind of interpolation
 _outer = "([^$]|\\$(?!{))+"  # any character except $ (unless not followed by {)
 SIMPLE_INTERPOLATION_PATTERN = re.compile(
     f"({_outer})?({_inter}({_outer})?)+$", flags=re.ASCII
 )
+# NOTE: SIMPLE_INTERPOLATION_PATTERN must not generate false positive matches:
+# it must not accept anything that isn't a valid interpolation (per the
+# interpolation grammar defined in `omegaconf/grammar/*.g4`).
 
 
 class OmegaConfErrorListener(ErrorListener):  # type: ignore

tests/test_grammar.py

@@ -29,6 +29,7 @@
 
 # Characters that are not allowed by the grammar in config key names.
 INVALID_CHARS_IN_KEY_NAMES = r"""\{}()[].:"' """
+UNQUOTED_SPECIAL = r"/-\+.$%*@?|"  # special characters allowed in unquoted strings
 
 # A fixed config that may be used (but not modified!) by tests.
 BASE_TEST_CFG = OmegaConf.create(
@@ -106,7 +107,11 @@
     ("float_minus_nan", "-nan", math.nan),
     # Unquoted strings.
     # Note: raw strings do not allow trailing \, adding a space and stripping it.
-    ("str_legal", r" a/-\+.$*@?\\ ".strip(), r" a/-\+.$*@?\ ".strip()),
+    (
+        "str_legal",
+        (r" a" + UNQUOTED_SPECIAL + r"\\ ").strip(),
+        (r" a" + UNQUOTED_SPECIAL + r"\ ").strip(),
+    ),
     ("str_illegal_1", "a,=b", GrammarParseError),
     ("str_illegal_2", f"{chr(200)}", GrammarParseError),
     ("str_illegal_3", f"{chr(129299)}", GrammarParseError),
@@ -128,8 +133,8 @@
     ("str_esc_illegal_1", r"\#", GrammarParseError),
     ("str_esc_illegal_2", r""" \'\" """.strip(), GrammarParseError),
     # Quoted strings.
-    ("str_quoted_single", "'!@#$%^&*()[]:.,\"'", '!@#$%^&*()[]:.,"'),
-    ("str_quoted_double", '"!@#$%^&*()[]:.,\'"', "!@#$%^&*()[]:.,'"),
+    ("str_quoted_single", "'!@#$%^&*|()[]:.,\"'", '!@#$%^&*|()[]:.,"'),
+    ("str_quoted_double", '"!@#$%^&*|()[]:.,\'"', "!@#$%^&*|()[]:.,'"),
     ("str_quoted_outer_ws_single", "'  a \t'", "  a \t"),
     ("str_quoted_outer_ws_double", '"  a \t"', "  a \t"),
     ("str_quoted_int", "'123'", "123"),
@@ -181,8 +186,10 @@
     ),
     (
         "dict_unquoted_key",
-        fr"{{a0-null-1-3.14-NaN- {TAB}-true-False-/\+.$%*@\(\)\[\]\{{\}}\:\=\ \{TAB}\,:0}}",
-        {fr"a0-null-1-3.14-NaN- {TAB}-true-False-/\+.$%*@()[]{{}}:= {TAB},": 0},
+        fr"{{a0-null-1-3.14-NaN- {TAB}-true-False-{UNQUOTED_SPECIAL}\(\)\[\]\{{\}}\:\=\ \{TAB}\,:0}}",
+        {
+            fr"a0-null-1-3.14-NaN- {TAB}-true-False-{UNQUOTED_SPECIAL}()[]{{}}:= {TAB},": 0
+        },
     ),
     (
         "dict_quoted",
@@ -364,7 +371,11 @@
     ("str_top_middle_quote_double", 'I"d like ${str}', 'I"d like hi'),
     ("str_top_middle_quotes_single", "I like '${str}'", "I like 'hi'"),
     ("str_top_middle_quotes_double", 'I like "${str}"', 'I like "hi"'),
-    ("str_top_any_char", r"${str} !@\#$%^&*})][({,/?;", r"hi !@\#$%^&*})][({,/?;"),
+    (
+        "str_top_any_char",
+        r"${str} " + UNQUOTED_SPECIAL + r"^!#&})][({,;",
+        r"hi " + UNQUOTED_SPECIAL + r"^!#&})][({,;",
+    ),
     ("str_top_esc_inter", r"Esc: \${str}", "Esc: ${str}"),
     ("str_top_esc_inter_wrong_1", r"Wrong: $\{str\}", r"Wrong: $\{str\}"),
     ("str_top_esc_inter_wrong_2", r"Wrong: \${str\}", r"Wrong: ${str\}"),
@@ -602,7 +613,7 @@ def visit() -> Any:
         "$ ${foo} ${bar} ${boz} $",
         "${foo:bar}",
         "${foo : bar, baz, boz}",
-        "${foo:bar,0,a-b+c*d/$.%@}",
+        "${foo:bar,0,a-b+c*d/$.%@?|}",
         r"\${foo}",
         "${foo.bar:boz}",
         "${$foo.bar$.x$y}",
@@ -729,7 +740,7 @@ def callback(inter_key: Any, memo: Optional[Set[int]]) -> Any:
 
 
 def test_custom_resolver_param_supported_chars() -> None:
-    supported_chars = r"abc123_/:-\+.$%*@"
+    supported_chars = r"abc123_:" + UNQUOTED_SPECIAL
     c = OmegaConf.create({"dir1": "${copy:" + supported_chars + "}"})
 
     OmegaConf.register_new_resolver("copy", lambda x: x)