block-grep: search for patterns, grouped per block

1 files changed, 78 insertions, 0 deletions

diff --git a/block-grep b/block-grep
new file mode 100755
index 0000000..5d10f72
--- /dev/null
+++ b/block-grep
@@ -0,0 +1,78 @@
+#!/usr/bin/env python
+# Searches block separated by multiple newlines.
+# Originally written for filtering ASAN (LSan) reports.
+#
+# Example usage:
+# ASAN_OPTIONS=log_path=$PWD/leak wireshark
+# block-grep leak.1337 -e dbus -v | block-grep -e Direct
+
+import argparse
+import re
+import sys
+
+def is_leaked(lines):
+    return 'Objects leaked above:' in lines
+
+def parse_file(f, match_func):
+    lines = ''
+    last_matched = ''
+    def handle_lines():
+        nonlocal last_matched
+        if match_func(lines):
+            sys.stdout.write(lines)
+            # Uncomment if you want to see the leaked objects too
+            #last_matched = lines
+        elif last_matched:
+            # If the previous block had a matching "Indirect leak of x bytes"
+            # and this block is its related "Objects leaked above", write it too
+            if  'irect leak of ' in last_matched and is_leaked(lines):
+               sys.stdout.write(lines)
+            last_matched = ''
+    for line in f:
+        lines += line
+        if line.strip('\r\n'):
+            continue
+
+        handle_lines()
+        lines = ''
+    handle_lines()
+
+parser = argparse.ArgumentParser()
+parser.add_argument('-e', '--regexp', dest='pattern')
+parser.add_argument('-i', '--ignore-case', action='store_true')
+parser.add_argument('-v', '--invert-match', action='store_true')
+parser.add_argument('file', nargs='*')
+
+def main():
+    args = parser.parse_args()
+    files = args.file
+    if not args.pattern:
+        if not files:
+            parser.error('Missing pattern option')
+        pattern = files.pop(0)
+    else:
+        pattern = args.pattern
+
+    if not files:
+        files = ['-']
+
+    flags = 0
+    if args.ignore_case:
+        flags |= re.IGNORECASE
+    pattern = re.compile(pattern, flags)
+
+    match_func = lambda lines: pattern.search(lines)
+    if args.invert_match:
+        match_func = lambda lines: not pattern.search(lines) and not is_leaked(lines)
+
+    for filename in files:
+        if filename == '-':
+            parse_file(sys.stdin, match_func)
+        else:
+            parse_file(open(filename, 'r'), match_func)
+
+if __name__ == '__main__':
+    try:
+        main()
+    except (KeyboardInterrupt, BrokenPipeError):
+        pass