Improve more copyrights from linux

Signed-off-by: Philippe Ombredanne <pombredanne@nexb.com>

Author: pombredanne

src/cluecode/copyrights.py

@@ -815,6 +815,9 @@ def build_detection_from_node(
     (r'^@?link:?$', 'JUNK'),
     (r'@license:?$', 'JUNK'),
 
+    # hex is JUNK 0x3fc3/0x7cff
+    (r'^0x[a-fA-F0-9]+', 'JUNK'),
+
     # found in crypto certificates and LDAP
     (r'^O=$', 'JUNK'),
     (r'^OU=?$', 'JUNK'),
@@ -1045,7 +1048,7 @@ def build_detection_from_node(
     (r'^Owner$', 'JUNK'),
     (r'^behalf$', 'JUNK'),
     (r'^know-how$', 'JUNK'),
-    (r'^interfaces?,?$', 'JUNK'),
+    (r'^[Ii]nterfaces?,?$', 'JUNK'),
     (r'^than$', 'JUNK'),
     (r'^whom$', 'JUNK'),
     (r'^However,?$', 'JUNK'),
@@ -1206,6 +1209,10 @@ def build_detection_from_node(
     (r'^False.?$', 'JUNK'),
     (r'^True.?$', 'JUNK'),
 
+    (r'^high$', 'JUNK'),
+    (r'^low$', 'JUNK'),
+    (r'^on$', 'JUNK'),
+
     (r'^imports?$', 'JUNK'),
     (r'^[Ww]arnings?$', 'JUNK'),
     (r'^[Ww]hether$', 'JUNK'),
@@ -1303,7 +1310,7 @@ def build_detection_from_node(
     (r'^Code$', 'NN'),
     (r'^Collators?$', 'NN'),
     (r'^Commercial', 'NN'),
-    (r'^Commons$', 'NN'),
+    (r'^Commons?$', 'NN'),
     # TODO: Compilation could be JUNK?
     (r'^Compilation', 'NN'),
     (r'^Contact', 'NN'),
@@ -1357,6 +1364,16 @@ def build_detection_from_node(
     (r'^With$', 'NN'),
     (r'^Tick$', 'NN'),
     (r'^Dynamic$', 'NN'),
+    (r'^Battery$', 'NN'),
+    (r'^Charger$', 'NN'),
+    (r'^Dynamic$', 'NN'),
+    (r'^Bugfixes?$', 'NN'),
+    (r'^Likes?$', 'NN'),
+    (r'^STA$', 'NN'),
+
+    (r'^Interrupt$', 'NN'),
+    (r'^cleanups?$', 'JUNK'),
+    (r'^Tape$', 'NN'),
 
     (r'^When$', 'NN'),
     # (r'^Owner$', 'NN'),
@@ -1384,8 +1401,8 @@ def build_detection_from_node(
     (r'^Gaim$', 'NN'),
     (r'^Generated', 'NN'),
     (r'^Glib$', 'NN'),
-    (r'^GPLd', 'NN'),
-    (r'^GPL\'d', 'NN'),
+    (r'^GPLd?\.?$', 'NN'),
+    (r'^GPL\'d$', 'NN'),
     (r'^Gnome$', 'NN'),
     (r'^GnuPG$', 'NN'),
     (r'^Government.', 'NNP'),
@@ -1477,7 +1494,7 @@ def build_detection_from_node(
     (r'^POSIX$', 'NN'),
     (r'^Possible', 'NN'),
     (r'^Powered$', 'NN'),
-    (r'^defined$', 'NN'),
+    (r'^defined?$', 'JUNK'),
     (r'^Predefined$', 'NN'),
     (r'^Promise$', 'NN'),
     (r'^Products?\.?$', 'NN'),
@@ -1576,7 +1593,7 @@ def build_detection_from_node(
     (r'^Branched$', 'NN'),
 
     (r'^Improved$', 'NN'),
-    (r'^Designed$', 'NN'),
+    (r'^Designe[dr]$', 'NN'),
     (r'^Organised$', 'NN'),
     (r'^Re-organised$', 'NN'),
     (r'^Swap$', 'NN'),
@@ -1897,6 +1914,9 @@ def build_detection_from_node(
     # et al.
     (r'^al\.$', 'AUTHDOT'),
 
+    # in Linux LKMs
+    (r'^MODULEAUTHOR$', 'AUTH'),
+
     # Contributor(s)
     (r'^[Cc]ontributors[,\.]?$', 'CONTRIBUTORS'),
     (r'^Contributor[,\.]?$', 'NN'),
@@ -3548,7 +3568,8 @@ def refine_names(s, prefixes):
     r'^\(c\) Object c$',
     r'^copyright headers?',
     r'Copyright \(c\) 2021 Dot',
-    r'^\(c\) \(c\) B$'
+    r'^\(c\) \(c\) B$',
+    r'^\(c\) group$',
 ]
 
 # a collection of junk junk matcher callables
@@ -3591,6 +3612,7 @@ def is_junk_copyryright(s, patterns=COPYRIGHTS_JUNK_PATTERN_MATCHERS):
         'mailto:',
         "name'",
         "a",
+        "moduleauthor",
     ])
 ))
 
@@ -4301,6 +4323,8 @@ def prepare_text_line(line, dedeb=True, to_ascii=True):
         # C and C++ style comment markers
         .replace('/*', ' ').replace('*/', ' ')
         .strip().strip('/*#')
+        # in rst
+        .replace('|copy|', ' (c) ')
         # un common pipe chars in some ascii art
         .replace('|', ' ')
 

src/cluecode/copyrights_hint.py

@@ -48,6 +48,8 @@
 statement_markers = (
     '©',
     '(c)',
+    # C sign in Restructured Text:
+    '|copy|',
     '&#169',
     '&#xa9',
     '169',

tests/cluecode/data/copyrights/misco4/linux2/bugfix.txt

@@ -0,0 +1,10 @@
+ Qua Jun 27 15:02:26 BRT 2001
+ * [27-JUN-2001] Arnaldo Carvalho de Melo <acme@conectiva.com.br> - cleanups
+ *
+ * Parts (C) 1999 David Airlie, airlied@linux.ie
+ * [07-SEP-99] Bugfixes
+ *
+ * [06-Jan-2002] Russell King <rmk@arm.linux.org.uk>
+ * Converted to new serial core
+ */
+

tests/cluecode/data/copyrights/misco4/linux2/bugfix.txt.yml

@@ -0,0 +1,8 @@
+what:
+  - copyrights
+  - holders
+  - authors
+copyrights:
+  - Parts (c) 1999 David Airlie, airlied@linux.ie
+holders:
+  - David Airlie

tests/cluecode/data/copyrights/misco4/linux2/c-group.txt

@@ -0,0 +1 @@
+ (c)group based bandwidth management,

tests/cluecode/data/copyrights/misco4/linux2/c-group.txt.yml

@@ -0,0 +1,4 @@
+what:
+  - copyrights
+  - holders
+  - authors

tests/cluecode/data/copyrights/misco4/linux2/clean.txt

@@ -0,0 +1,10 @@
+ Qua Jun 27 15:02:26 BRT 2001
+ * [27-JUN-2001] Arnaldo Carvalho de Melo <acme@conectiva.com.br> - cleanups
+ *
+ * Parts (C) 1999 David Airlie, airlied@linux.ie
+ * [07-SEP-99] Bugfixes
+ *
+ * [06-Jan-2002] Russell King <rmk@arm.linux.org.uk>
+ * Converted to new serial core
+ */
+

tests/cluecode/data/copyrights/misco4/linux2/clean.txt.yml

@@ -0,0 +1,8 @@
+what:
+  - copyrights
+  - holders
+  - authors
+copyrights:
+  - Parts (c) 1999 David Airlie, airlied@linux.ie
+holders:
+  - David Airlie

tests/cluecode/data/copyrights/misco4/linux2/copy-rst.rst

@@ -0,0 +1,2 @@
+**Copyright** |copy| 1999-2016 : LinuxTV Developers
+

tests/cluecode/data/copyrights/misco4/linux2/copy-rst.rst.yml

@@ -0,0 +1,8 @@
+what:
+  - copyrights
+  - holders
+  - authors
+copyrights:
+  - Copyright (c) 1999-2016 LinuxTV Developers
+holders:
+  - LinuxTV Developers