Print changes in match list; find COPYING files in subdirs too

src/main.py

@@ -6,15 +6,13 @@ from difflib import SequenceMatcher
 import os
 import re
 
-DEFAULT_SIZE_DIFF = 0.1
 DEFAULT_MIN_RATIO = 0.8
 DEFAULT_N = 5
 
 
 class Tester(object):
-	def __init__(self, licenses_dir, max_size_diff=DEFAULT_SIZE_DIFF, min_ratio=DEFAULT_MIN_RATIO):
+	def __init__(self, licenses_dir, min_ratio=DEFAULT_MIN_RATIO):
 		self.licenses_dir = licenses_dir
-		self.max_size_diff = max_size_diff
 		self.min_ratio = min_ratio
 
 		self._licenses = None
@@ -39,18 +37,29 @@ class Tester(object):
 		matches = []
 
 		with open(src, 'r') as f:
-			src_content = '\n'.join([x.strip() for x in f])
+			src_content = [x.strip() for x in f]
 
 		for license_file, license_size in self.licenses:
-			if abs(src_size - license_size) > src_size * self.max_size_diff:
-				continue
-
 			with open(license_file, 'r') as l:
-				lic_content = '\n'.join([x.strip() for x in l])
-				s = SequenceMatcher(lambda x: x in " \t", src_content, lic_content)
+				lic_content = [x.strip() for x in l]
+
+				s = SequenceMatcher(lambda x: x in " \t", lic_content, src_content)
 
 				if s.quick_ratio() > self.min_ratio:
-					matches.append((s.ratio(), license_file))
+					ops = s.get_grouped_opcodes()
+
+					adds, deletes, edits = 0, 0, 0
+					for x in ops:
+						for y in x:
+							action, ll, lh, rl, rh = y
+							if action == "insert":
+								adds += 1
+							if action == "delete":
+								deletes += 1
+							if action == "replace":
+								edits += 1
+
+					matches.append((s.ratio(), license_file, (adds, deletes, edits)))
 
 		return src, matches
 
@@ -79,7 +88,8 @@ class Finder(object):
 			l.append(os.path.join(src, 'COPYING'))
 
 		for root, dirs, files in os.walk(src):
-			l.extend([os.path.join(root, x) for x in files if re.match("^.*license(\.(txt|md|rst))?", x, re.IGNORECASE)])
+			l.extend([os.path.join(root, x) for x in files if re.match("^.*license(\.(txt|md|rst))?$", x, re.IGNORECASE)])
+			l.extend([os.path.join(root, x) for x in files if x == "COPYING"])
 
 		return l
 
@@ -89,8 +99,6 @@ if __name__ == "__main__":
 	parser.add_argument('--licenses-dir', default='/usr/portage/licenses',
 						help="Directory to look for licenses (default: %(default)s)")
 
-	parser.add_argument('--size-diff', default=DEFAULT_SIZE_DIFF, type=float,
-						help="Ignore license files which diff by size*x bytes (default: %(default)s)")
 	parser.add_argument('--min-ratio', default=DEFAULT_MIN_RATIO, type=float,
 						help="Minimum ratio to consider the file as a match (default: %(default)s)")
 	parser.add_argument('-n', default=DEFAULT_N, type=int, help="Show at most x matching licenses")
@@ -98,7 +106,7 @@ if __name__ == "__main__":
 
 	args = parser.parse_args()
 
-	tester = Tester(args.licenses_dir, args.size_diff, args.min_ratio)
+	tester = Tester(args.licenses_dir, args.min_ratio)
 
 	for f_or_d in args.files_or_dirs:
 		for filename, matches in tester.test(Finder.find(f_or_d)):
@@ -106,7 +114,9 @@ if __name__ == "__main__":
 			print("-" * (len(filename)+1))
 
 			if matches:
-				for ratio, license_name in sorted(matches, reverse=True)[:args.n]:
-					print("%8.3f %%: %s" % (ratio*100.0, license_name))
+				for ratio, license_name, changes in sorted(matches, reverse=True)[:args.n]:
+					print("%8.3f %%: %s (+%s -%s ~%s)" % ((ratio*100.0, license_name) + changes))
 			else:
 				print("No license matched :(")
+
+			print("")