gen-ignores, ignore_patterns: update to exclude unsupported Perl syntax, backreferences

1 files changed, 5 insertions, 1 deletions

diff --git a/gen-ignores.py b/gen-ignores.py
index 25b3cac..ede0529 100755
--- a/gen-ignores.py
+++ b/gen-ignores.py
@@ -13,13 +13,16 @@ import os
 import sys
 
 archivebot_ignore_path = sys.argv[1]
+unsupported = ['\\1', '\\2', '(?!', '(?=']
+
 print 'package crawl\n\nvar defaultIgnorePatterns = []string{'
+
 for fn in glob.glob(os.path.join(archivebot_ignore_path, '*.json')):
     try:
         with open(fn) as fd:
             print '\n\t// %s' % os.path.basename(fn)
             for p in json.load(fd)['patterns']:
-                if '\\\\1' in p or '(?!' in p:
+                if any(x in p for x in unsupported):
                     # RE2 does not support backreferences or other
                     # fancy PCRE constructs. This excludes <10
                     # patterns from the ignore list.
@@ -27,5 +30,6 @@ for fn in glob.glob(os.path.join(archivebot_ignore_path, '*.json')):
                 print '\t%s,' % json.dumps(p)
     except Exception, e:
         print >>sys.stderr, 'error in %s: %s' % (fn, e)
+
 print '}'