Detect copyright with parens

Signed-off-by: Philippe Ombredanne <pombredanne@aboutcode.org>

Author: pombredanne

src/cluecode/copyrights.py

@@ -429,7 +429,8 @@ def get_tokens(numbered_lines, splitter=re.compile(r'[\t =;]+').split):
 
         # when there is a an openning parens in the middle of a word, add a
         # space before in some cases: exclude (digit , (s , (c , (-
-        line = re.sub(pattern=r'(\([^rsc\-\d])', repl=' \g<1>', string=line, flags=re.IGNORECASE)
+        # this allows to recover from words like KISA(Korean
+        line = re.sub(pattern=r'(\([^rsc\-\d])', repl=' \g<1>', string=line)
         for tok in splitter(line):
             # strip trailing quotes+comma
             if tok.endswith("',"):
@@ -1236,6 +1237,10 @@ def build_detection_from_node(
     (r'^param$', 'JUNK'),
     (r'^which$', 'JUNK'),
 
+    # github/gitlab usernames
+    (r'^Hiroki11x$', 'NNP'),
+    (r'^goldze$', 'NNP'),
+
     # "Es6ToEs3ClassSideInheritance. and related names
     (r"^[A-Z]([a-zA-Z]*[0-9]){2,}[a-zA-Z]+[\.,]?", 'JUNK'),
 
@@ -1300,7 +1305,7 @@ def build_detection_from_node(
     # '!(r)'
     (r'^!\(r\)$', 'JUNK'),
 
-    # vars with curly braces
+    # vars with parens
     (r'\(var', 'JUNK'),
 
     # tags
@@ -1895,7 +1900,7 @@ def build_detection_from_node(
 
     # rare form of trailing punct in name: Ian Robertson).
     (r'^Robert.*', 'NNP'),
-
+    
     ############################################################################
     # Named entities: companies, groups, universities, etc
     ############################################################################
@@ -2233,6 +2238,7 @@ def build_detection_from_node(
     (r'^Xiph.Org$', 'NNP'),
     (r'^iClick,?$', 'NNP'),
     (r'^electronics?$', 'NNP'),
+    (r'^semiconductors?[\.,]?$', 'NNP'),
 
     # proper nouns with digits
     (r'^([A-Z][a-z0-9]+){1,2}[\.,]?$', 'NNP'),
@@ -2370,8 +2376,6 @@ def build_detection_from_node(
     # some punctuation combos
     (r'^(?:=>|->|<-|<=)$', 'JUNK'),
 
-    (r'^semiconductors?[\.,]?$', 'NNP'),
-
     ############################################################################
     # catch all other as Nouns
     ############################################################################

tests/cluecode/data/copyrights/copyrights_with_parens/copy-with-parens-1.txt.yml

@@ -1,4 +1,5 @@
 what:
   - copyrights
   - holders
-  - holders_summary
+copyrights:
+  - Copyright 2014

tests/cluecode/data/copyrights/copyrights_with_parens/copy-with-parens-10.txt.yml

@@ -1,4 +1,7 @@
 what:
   - copyrights
   - holders
-  - holders_summary
+copyrights:
+  - Copyright 2018 dunamu (ju)
+holders:
+  - dunamu (ju)

tests/cluecode/data/copyrights/copyrights_with_parens/copy-with-parens-11.txt.yml

@@ -1,4 +1,7 @@
 what:
   - copyrights
   - holders
-  - holders_summary
+copyrights:
+  - Copyright (c) 2023 Wolken (zwolken@foxmail.com zharry@web.de)
+holders:
+  - Wolken

tests/cluecode/data/copyrights/copyrights_with_parens/copy-with-parens-12.txt.yml

@@ -1,4 +1,7 @@
 what:
   - copyrights
   - holders
-  - holders_summary
+copyrights:
+  - Copyright (c) 2002 Obviex (TM)
+holders:
+  - Obviex (TM)

tests/cluecode/data/copyrights/copyrights_with_parens/copy-with-parens-13.txt.yml

@@ -1,4 +1,7 @@
 what:
   - copyrights
   - holders
-  - holders_summary
+copyrights:
+  - Copyright (c) 2018 qiuri Qiu Ri
+holders:
+  - qiuri Qiu Ri

tests/cluecode/data/copyrights/copyrights_with_parens/copy-with-parens-14.txt.yml

@@ -1,4 +1,7 @@
 what:
   - copyrights
   - holders
-  - holders_summary
+copyrights:
+  - Copyright 1996 V (DL)2 inc.
+holders:
+  - V (DL)2 inc.

tests/cluecode/data/copyrights/copyrights_with_parens/copy-with-parens-15.txt.yml

@@ -1,4 +1,7 @@
 what:
   - copyrights
   - holders
-  - holders_summary
+copyrights:
+  - Xfrom Elsevier, Copyright 2016
+holders:
+  - Xfrom Elsevier

tests/cluecode/data/copyrights/copyrights_with_parens/copy-with-parens-16.txt.yml

@@ -1,4 +1,7 @@
 what:
   - copyrights
   - holders
-  - holders_summary
+copyrights:
+  - Copyright 2018 Hiroki11x (Hiroki Naganuma)
+holders:
+  - Hiroki11x (Hiroki Naganuma)

tests/cluecode/data/copyrights/copyrights_with_parens/copy-with-parens-17.txt.yml

@@ -1,4 +1,3 @@
 what:
   - copyrights
   - holders
-  - holders_summary