Feature/regex tag

changes/81.feature.rst

@@ -0,0 +1,5 @@
+introduce the TAG element as a module. Needs a new parser for the SpaCy translate.
+Would allow more flexible matching of detailed part-of-speech tag, like all adjectives or nouns: TAG("^NN|^JJ").
+
+Implemented by:
+Roland M. Mueller (https://github.com/rolandmueller)

rita/engine/translate_spacy.py

@@ -74,6 +74,18 @@ def phrase_parse(value, config, op=None):
         yield generic_parse("ORTH", value, config=config, op=None)
 
 
+def tag_parse(r, config, op=None):
+    """
+    For generating POS/TAG patterns based on a Regex
+    e.g. TAG("^NN|^JJ") for adjectives or nouns
+    """
+    d = {"TAG": {"REGEX": r}}
+
+    if op:
+        d["OP"] = op
+    yield d
+
+
 PARSERS = {
     "any_of": any_of_parse,
     "value": partial(generic_parse, "ORTH"),
@@ -84,6 +96,7 @@ def phrase_parse(value, config, op=None):
     "punct": punct_parse,
     "fuzzy": fuzzy_parse,
     "phrase": phrase_parse,
+    "tag": tag_parse,
 }
 
 

rita/modules/tag.py

@@ -0,0 +1,9 @@
+from rita.macros import resolve_value
+
+
+def TAG(name, config, op=None):
+    """
+    For generating POS/TAG patterns based on a Regex
+    e.g. TAG("^NN|^JJ") for nouns or adjectives
+    """
+    return "tag", resolve_value(name, config=config), op

tests/test_rules.py

@@ -287,6 +287,21 @@ def test_optional_list(self):
             "pattern": [{"LOWER": {"REGEX": "^(one|two)$"}, "OP": "?"}]
         }
 
+    def test_tag_module(self):
+        rules = self.compiler("""
+        !IMPORT("rita.modules.tag")
+
+        TAG("^NN|^JJ")->MARK("TEST_TAG")
+        """)
+
+        print(rules)
+
+        assert len(rules) == 1
+        assert rules[0] == {
+            "label": "TEST_TAG",
+            "pattern": [{"TAG": {"REGEX": "^NN|^JJ"}}]
+        }
+
 
 class TestStandalone(object):
     @property