Update

changes/77.fix.rst

@@ -0,0 +1 @@
+Fix IN_LIST operation bug - it was ignoring them

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "rita-dsl"
-version = "0.5.6"
+version = "0.5.7"
 description = "DSL for building language rules"
 authors = [
     "Šarūnas Navickas <zaibacu@gmail.com>"

rita/__init__.py

@@ -10,7 +10,7 @@
 
 logger = logging.getLogger(__name__)
 
-__version__ = (0, 5, 6, os.getenv("VERSION_PATCH"))
+__version__ = (0, 5, 7, os.getenv("VERSION_PATCH"))
 
 
 def get_version():

rita/engine/translate_standalone.py

@@ -21,7 +21,8 @@ def apply_operator(syntax, op):
 
 
 def any_of_parse(lst, op=None):
-    return apply_operator(r"(^|\s)(({0})\s?)".format("|".join(sorted(lst, key=lambda x: (-len(x), x)))), op)
+    clause = r"((^|\s)(({0})\s?))".format("|".join(sorted(lst, key=lambda x: (-len(x), x))))
+    return apply_operator(clause, op)
 
 
 def regex_parse(r, op=None):

rita/macros.py

@@ -52,7 +52,7 @@ def ASSIGN(k, v, config, op=None):
 
 def IN_LIST(*args, config, op=None):
     return "any_of", [resolve_value(arg, config=config)
-                      for arg in flatten(args)], None
+                      for arg in flatten(args)], op
 
 
 def PATTERN(*args, config, op=None):

tests/test_rules.py

@@ -273,6 +273,20 @@ def test_multiple_optionals(self):
             ]
         }
 
+    def test_optional_list(self):
+        rules = self.compiler("""
+        elements = {"one", "two"}
+        {IN_LIST(elements)?}->MARK("OPTIONAL_LIST")
+        """)
+
+        print(rules)
+
+        assert len(rules) == 1
+        assert rules[0] == {
+            "label": "OPTIONAL_LIST",
+            "pattern": [{"LOWER": {"REGEX": "^(one|two)$"}, "OP": "?"}]
+        }
+
 
 class TestStandalone(object):
     @property
@@ -311,7 +325,7 @@ def test_multiple_words(self):
         ''')
         print(rules)
         assert len(rules) == 1
-        assert rules[0] == re.compile(r"(?P<MULTI_LABEL>(^|\s)((test1|test2)\s?))", self.flags)
+        assert rules[0] == re.compile(r"(?P<MULTI_LABEL>((^|\s)((test1|test2)\s?)))", self.flags)
 
     def test_simple_pattern(self):
         rules = self.compiler('''
@@ -349,19 +363,19 @@ def test_or_branch_multi_w_single(self):
         print(rules)
         assert len(rules) == 4
         assert rules[0] == re.compile(
-            r"(?P<MULTI_SPLIT_LABEL>(test1\s?)([.,!;?:]\s?)?(^|\s)((three|one|two)\s?)([.,!;?:]\s?)?(test3\s?))",
+            r"(?P<MULTI_SPLIT_LABEL>(test1\s?)([.,!;?:]\s?)?((^|\s)((three|one|two)\s?))([.,!;?:]\s?)?(test3\s?))",
             self.flags
         )
         assert rules[1] == re.compile(
-            r"(?P<MULTI_SPLIT_LABEL>(test2\s?)([.,!;?:]\s?)?(^|\s)((three|one|two)\s?)([.,!;?:]\s?)?(test3\s?))",
+            r"(?P<MULTI_SPLIT_LABEL>(test2\s?)([.,!;?:]\s?)?((^|\s)((three|one|two)\s?))([.,!;?:]\s?)?(test3\s?))",
             self.flags
         )
         assert rules[2] == re.compile(
-            r"(?P<MULTI_SPLIT_LABEL>(test1\s?)([.,!;?:]\s?)?(^|\s)((three|one|two)\s?)([.,!;?:]\s?)?(test4\s?))",
+            r"(?P<MULTI_SPLIT_LABEL>(test1\s?)([.,!;?:]\s?)?((^|\s)((three|one|two)\s?))([.,!;?:]\s?)?(test4\s?))",
             self.flags
         )
         assert rules[3] == re.compile(
-            r"(?P<MULTI_SPLIT_LABEL>(test2\s?)([.,!;?:]\s?)?(^|\s)((three|one|two)\s?)([.,!;?:]\s?)?(test4\s?))",
+            r"(?P<MULTI_SPLIT_LABEL>(test2\s?)([.,!;?:]\s?)?((^|\s)((three|one|two)\s?))([.,!;?:]\s?)?(test4\s?))",
             self.flags
         )
 
@@ -372,7 +386,7 @@ def test_branching_list(self):
         ''')
         print(rules)
         assert len(rules) == 2
-        assert rules[0] == re.compile(r"(?P<SPLIT_LIST>(^|\s)((test1|test2|test4)\s?))", self.flags)
+        assert rules[0] == re.compile(r"(?P<SPLIT_LIST>((^|\s)((test1|test2|test4)\s?)))", self.flags)
         assert rules[1] == re.compile(r"(?P<SPLIT_LIST>(test-3\s?))", self.flags)
 
     def test_word_with_accent(self):
@@ -381,7 +395,7 @@ def test_word_with_accent(self):
         ''')
         print(rules)
         assert len(rules) == 1
-        assert rules[0] == re.compile(r"(?P<TWO_WORDS>(^|\s)((Sarunas|Šarūnas)\s?))", self.flags)
+        assert rules[0] == re.compile(r"(?P<TWO_WORDS>((^|\s)((Sarunas|Šarūnas)\s?)))", self.flags)
 
     def test_list_with_accent(self):
         rules = self.compiler('''
@@ -390,7 +404,7 @@ def test_list_with_accent(self):
         ''')
         print(rules)
         assert len(rules) == 1
-        assert rules[0] == re.compile(r"(?P<EXTENDED_LIST>(^|\s)((Sarunas|Šarūnas|Jurgis|Jonas)\s?))", self.flags)
+        assert rules[0] == re.compile(r"(?P<EXTENDED_LIST>((^|\s)((Sarunas|Šarūnas|Jurgis|Jonas)\s?)))", self.flags)
 
     def test_double_op(self):
         rules = self.compiler('''
@@ -415,7 +429,7 @@ def test_prefix_on_list(self):
         ''')
         print(rules)
         assert len(rules) == 1
-        assert rules[0] == re.compile(r"(?P<META_LIST>(^|\s)((metamathematics|metaphysics)\s?))", self.flags)
+        assert rules[0] == re.compile(r"(?P<META_LIST>((^|\s)((metamathematics|metaphysics)\s?)))", self.flags)
 
     def test_prefix_on_unknown_type(self):
         rules = self.compiler('''
@@ -435,3 +449,14 @@ def test_save_and_load_rules_from_file(self):
             engine.save(rules_path)
             engine.load(rules_path)
             engine.execute("Hello world")
+
+    def test_optional_list(self):
+        rules = self.compiler("""
+        elements = {"one", "two"}
+        {IN_LIST(elements)?}->MARK("OPTIONAL_LIST")
+        """)
+
+        print(rules)
+
+        assert len(rules) == 1
+        assert rules[0] == re.compile(r"(?P<OPTIONAL_LIST>((^|\s)((one|two)\s?))?)", self.flags)