add test for 3-seq and extra-words-spans

Signed-off-by: Alok Kumar <[email protected]>

Author: alok1304

src/licensedcode/detection.py

@@ -1095,7 +1095,7 @@ def is_correct_detection(license_matches):
     ]
 
     return (
-        all(matcher in ("1-hash", "1-spdx-id", "2-aho") for matcher in matchers)
+        all(matcher in ("1-hash", "1-spdx-id", "2-aho", "3-seq") for matcher in matchers)
         and all(is_match_coverage_perfect)
     )
 
@@ -1156,6 +1156,17 @@ def has_extra_words(license_matches):
     )
 
 
+def has_extra_words_spans(license_matches):
+    """
+    Return True if all of the matches rules in ``license_matches`` (a list of LicenseMatch)
+    has `extra_phrase` marker and also have matcher `3-seq`.
+    """
+    return all(
+        match.matcher == '3-seq' and match.rule.extra_phrase_spans
+        for match in license_matches
+    )
+
+
 def has_low_rule_relevance(license_matches):
     """
     Return True if all on the matches in ``license_matches`` List of LicenseMatch

tests/licensedcode/test_detection_validate.py

@@ -18,6 +18,7 @@
 from licensedcode import cache
 from licensedcode import models
 from licensedcode.detection import is_correct_detection
+from licensedcode.detection import has_extra_words_spans
 from licensedcode.models import licenses_data_dir
 from licensedcode.models import rules_data_dir
 from licensedcode.models import License
@@ -99,7 +100,7 @@ def check_rule_or_license_can_be_detected_exactly(licensish):
         assert results == expected
 
     icm = is_correct_detection(matches)
-    if not icm:
+    if not icm and not has_extra_words_spans(matches):
         expected.append(f'file://{licensish.rule_file()}')
         assert results == expected