Improve copyright detection

Signed-off-by: Philippe Ombredanne <[email protected]>

src/cluecode/copyrights.py

@@ -581,6 +581,8 @@ def build_detection_from_node(
     '(20[0-3][0-9][\\.,\\-])+20[0-3][0-9]'  # 2001-2012
     '|'
     '(20[0-3][0-9][\\.,\\-])+20[0-3]x'  # 2001-201x
+    '|'
+    '(20[0-3][0-9][\\.,\\-])+20[0-3][0-9]a'  # 2001-2012a
 ')')
 
 _PUNCT = (
@@ -1257,6 +1259,7 @@ def build_detection_from_node(
     (r'^Activation\.?$', 'NN'),
     (r'^Act[\.,]?$', 'NN'),
     (r'^Added$', 'NN'),
+    (r'^added$', 'JUNK'),
     (r'^As$', 'NN'),
     (r'^I$', 'NN'),
     (r'^Additional$', 'NN'),
@@ -1350,6 +1353,11 @@ def build_detection_from_node(
     (r'^Entity$', 'NN'),
     (r'^Example', 'NN'),
     (r'^Except', 'NN'),
+    (r'^Fragments$', 'NN'),
+    (r'^With$', 'NN'),
+    (r'^Tick$', 'NN'),
+    (r'^Dynamic$', 'NN'),
+
     (r'^When$', 'NN'),
     # (r'^Owner$', 'NN'),
     (r'^Specifications?$', 'NN'),
@@ -1567,6 +1575,14 @@ def build_detection_from_node(
     (r'^They$', 'JUNK'),
     (r'^Branched$', 'NN'),
 
+    (r'^Improved$', 'NN'),
+    (r'^Designed$', 'NN'),
+    (r'^Organised$', 'NN'),
+    (r'^Re-organised$', 'NN'),
+    (r'^Swap$', 'NN'),
+    (r'^Adapted$', 'JUNK'),
+    (r'^Thumb$', 'NN'),
+
     # alone this is not enough for an NNP
     (r'^Free$', 'NN'),
 
@@ -1620,6 +1636,7 @@ def build_detection_from_node(
     (r'^Unlike$', 'NN'),
     (r'^Compression$', 'NN'),
     (r'^Letter$', 'NN'),
+    (r'^Moved$', 'NN'),
 
     # dual caps that are not NNP
     (r'^Make[A-Z]', 'JUNK'),
@@ -1799,6 +1816,9 @@ def build_detection_from_node(
     (r'^(S\.?A\.?S?|Sas|sas|A\/S|AG,?|AB|Labs?|[Cc][Oo]|Research|Center|INRIA|Societe|KG)[,\.]?$', 'COMP'),
     # French SARL
     (r'^(SARL|S\.A\.R\.L\.)[\.,\)]*$', 'COMP'),
+    # More company suffix : a.s. in Czechia and otehrs
+    (r'^(a\.s\.|S\.r\.l\.?)$', 'COMP'),
+    (r'^Vertriebsges\.m\.b\.H\.?,?$', 'COMP'),
 
     # company suffix : AS: this is frequent beyond Norway.
     (r'^AS', 'CAPS'),
@@ -2070,6 +2090,9 @@ def build_detection_from_node(
     # proper noun with apostrophe ': d'Itri
     (r"^[a-z]'[A-Z]?[a-z]+[,\.]?$", 'NNP'),
 
+    # exceptions to all CAPS words
+    (r'^[A-Z]{3,4}[0-9]{4},?$', 'NN'),
+
     # all CAPS word, at least 1 char long such as MIT, including an optional trailing comma or dot
     (r'^[A-Z0-9]+,?$', 'CAPS'),
 
@@ -2680,6 +2703,9 @@ def build_detection_from_node(
     # Copyright (c) Ian F. Darwin 1986, 1987, 1989, 1990, 1991, 1992, 1994, 1995.
     COPYRIGHT: {<COPY>+ <NAME|NAME-EMAIL|NAME-YEAR>+ <YR-RANGE>*}        #157999
 
+    # Copyright (c) 2014 Czech Technical University in Prague
+    COPYRIGHT: {<COPYRIGHT>  <NN>  <UNI>  <NAME>}                     #157999-name
+
     COPYRIGHT: {<COPY>+ <CAPS|NNP>+ <CC> <NN> <COPY> <YR-RANGE>?}        #1590
 
     # // (c) (C) → ©

tests/cluecode/data/copyrights/misco4/linux/2011a.txt

@@ -0,0 +1,2 @@
+ * Copyright 2010-2011a Analog Devices Inc.
+

tests/cluecode/data/copyrights/misco4/linux/2011a.txt.yml

@@ -0,0 +1,8 @@
+what:
+  - copyrights
+  - holders
+  - authors
+copyrights:
+  - Copyright 2010-2011a Analog Devices Inc.
+holders:
+  - Analog Devices Inc.

tests/cluecode/data/copyrights/misco4/linux/adapted.txt

@@ -0,0 +1,3 @@
+Copyright (C) 1998 Frederic Rible F1OAT ([email protected])
+Adapted from baycom.c driver written by Thomas Sailer ([email protected])
+

tests/cluecode/data/copyrights/misco4/linux/adapted.txt.yml

@@ -0,0 +1,10 @@
+what:
+  - copyrights
+  - holders
+  - authors
+copyrights:
+  - Copyright (c) 1998 Frederic Rible F1OAT ([email protected])
+holders:
+  - Frederic Rible F1OAT
+authors:
+  - Thomas Sailer ([email protected])

tests/cluecode/data/copyrights/misco4/linux/adapted2.txt

@@ -0,0 +1,2 @@
+(c) 2004 MontaVista Software, Inc.
+ *  - Adapted from gdb/sim/arm/thumbemu.c

tests/cluecode/data/copyrights/misco4/linux/adapted2.txt.yml

@@ -0,0 +1,8 @@
+what:
+  - copyrights
+  - holders
+  - authors
+copyrights:
+  - (c) 2004 MontaVista Software, Inc.
+holders:
+  - MontaVista Software, Inc.

tests/cluecode/data/copyrights/misco4/linux/added.txt

@@ -0,0 +1,3 @@
+additions, Copyright (C) 2003-2004, Paul Clements, SteelEye Technology, Inc.:
+ * - added disk storage for bitmap
+

tests/cluecode/data/copyrights/misco4/linux/added.txt.yml

@@ -0,0 +1,8 @@
+what:
+  - copyrights
+  - holders
+  - authors
+copyrights:
+  - Copyright (c) 2003-2004, Paul Clements, SteelEye Technology, Inc.
+holders:
+  - Paul Clements, SteelEye Technology, Inc.

tests/cluecode/data/copyrights/misco4/linux/ak.txt

@@ -0,0 +1,3 @@
+ * Copyright (C) 2012 AK signal Brno a.s.
+ *               2012 Jiri Prchal <[email protected]>
+

tests/cluecode/data/copyrights/misco4/linux/ak.txt.yml

@@ -0,0 +1,8 @@
+what:
+  - copyrights
+  - holders
+  - authors
+copyrights:
+  - Copyright (c) 2012 AK signal Brno a.s. 2012 Jiri Prchal <[email protected]>
+holders:
+  - AK signal Brno a.s. Jiri Prchal

tests/cluecode/data/copyrights/misco4/linux/caps.txt

@@ -0,0 +1,3 @@
+Copyright (c) 2005-2006 Mauro Carvalho Chehab <[email protected]>
+     SAA7111, SAA7113 and SAA7118 support
+

tests/cluecode/data/copyrights/misco4/linux/caps.txt.yml

@@ -0,0 +1,8 @@
+what:
+  - copyrights
+  - holders
+  - authors
+copyrights:
+  - Copyright (c) 2005-2006 Mauro Carvalho Chehab <[email protected]>
+holders:
+  - Mauro Carvalho Chehab

tests/cluecode/data/copyrights/misco4/linux/ctu.txt

@@ -0,0 +1,2 @@
+Copyright: (c) 2014 Czech Technical University in Prague
+

tests/cluecode/data/copyrights/misco4/linux/ctu.txt.yml

@@ -0,0 +1,8 @@
+what:
+  - copyrights
+  - holders
+  - authors
+copyrights:
+  - Copyright (c) 2014 Czech Technical University in Prague
+holders:
+  - Czech Technical University in Prague

tests/cluecode/data/copyrights/misco4/linux/dynamic.txt

@@ -0,0 +1,30 @@
+Copyright (c) Dynamic Network Services, Inc
+Copyright (c) frisB.com <[email protected]>
+
+
+
+
+
+
+Copyright (c) dynamic-evaluation, https://github.com/benkrause/dynamic-evaluation
+
+
+
+
+
+
+
+
+copyright (c) Dynamic Drive (www.dynamicdrive.com)
+
+
+
+
+
+ Copyright (C) Dynamic System Bars Project
+
+
+
+
+
+# Copyright (c) Dynamic Solutions. All rights reserved.

tests/cluecode/data/copyrights/misco4/linux/dynamic.txt.yml

@@ -0,0 +1,18 @@
+what:
+  - copyrights
+  - holders
+  - authors
+copyrights:
+  - Copyright (c) Dynamic Network Services, Inc
+  - Copyright (c) frisB.com <[email protected]>
+  - Copyright (c) dynamic-evaluation, https://github.com/benkrause/dynamic-evaluation
+  - copyright (c) Dynamic Drive (www.dynamicdrive.com)
+  - Copyright (c) Dynamic System Bars Project
+  - Copyright (c) Dynamic Solutions
+holders:
+  - Dynamic Network Services, Inc
+  - frisB.com
+  - dynamic-evaluation
+  - Dynamic Drive
+  - Dynamic System Bars Project
+  - Dynamic Solutions

tests/cluecode/data/copyrights/misco4/linux/improved.txt

@@ -0,0 +1,3 @@
+*                              Copyright (C) 1998  Andrea Arcangeli
+ *  1999-03-10  Improved NTP compatibility by Ulrich Windl
+

tests/cluecode/data/copyrights/misco4/linux/improved.txt.yml

@@ -0,0 +1,8 @@
+what:
+  - copyrights
+  - holders
+  - authors
+copyrights:
+  - Copyright (c) 1998 Andrea Arcangeli 1999-03-10
+holders:
+  - Andrea Arcangeli

tests/cluecode/data/copyrights/misco4/linux/irq.txt

@@ -0,0 +1,7 @@
+Copyright (C) 1992 Linus Torvalds
+Modifications for ARM processor Copyright (C) 1995-2000 Russell King.
+
+Support for Dynamic Tick Timer Copyright (C) 2004-2005 Nokia Corporation.
+Dynamic Tick Timer written by Tony Lindgren <[email protected]> and
+Tuukka Tikkanen <[email protected]>.
+

tests/cluecode/data/copyrights/misco4/linux/irq.txt.yml

@@ -0,0 +1,14 @@
+what:
+  - copyrights
+  - holders
+  - authors
+copyrights:
+  - Copyright (c) 1992 Linus Torvalds
+  - Copyright (c) 1995-2000 Russell King
+  - Copyright (c) 2004-2005 Nokia Corporation
+holders:
+  - Linus Torvalds
+  - Russell King
+  - Nokia Corporation
+authors:
+  - Tony Lindgren <[email protected]> and Tuukka Tikkanen <[email protected]>

tests/cluecode/data/copyrights/misco4/linux/moved.txt

@@ -0,0 +1,4 @@
+Extra data:
+ *  Copyright (C) 1994-1999 Russell King
+ *  Moved from linux/arch/arm/kernel/debug.S by Ben Dooks
+

tests/cluecode/data/copyrights/misco4/linux/moved.txt.yml

@@ -0,0 +1,8 @@
+what:
+  - copyrights
+  - holders
+  - authors
+copyrights:
+  - Copyright (c) 1994-1999 Russell King
+holders:
+  - Russell King

tests/cluecode/data/copyrights/misco4/linux/reorg.txt

@@ -0,0 +1,2 @@
+ *  Copyright (C) 1991-1998  Linus Torvalds
+ *  Re-organised Feb 1998 Russell King

tests/cluecode/data/copyrights/misco4/linux/reorg.txt.yml

@@ -0,0 +1,8 @@
+what:
+  - copyrights
+  - holders
+  - authors
+copyrights:
+  - Copyright (c) 1991-1998 Linus Torvalds
+holders:
+  - Linus Torvalds

tests/cluecode/data/copyrights/misco4/linux/with-add.txt

@@ -0,0 +1,3 @@
+Copyright (C) 1997 Sigurdur Asgeirsson
+With additional hacking by Jeffrey Kuskin ([email protected])
+

tests/cluecode/data/copyrights/misco4/linux/with-add.txt.yml

@@ -0,0 +1,10 @@
+what:
+  - copyrights
+  - holders
+  - authors
+copyrights:
+  - Copyright (c) 1997 Sigurdur Asgeirsson
+holders:
+  - Sigurdur Asgeirsson
+authors:
+  - Jeffrey Kuskin ([email protected])