From 4bb62b3d88b1a5b4a83efd0d094ecd1e9d9185f6 Mon Sep 17 00:00:00 2001 From: Ivan Kozik Date: Thu, 11 Apr 2024 02:49:55 +0000 Subject: [PATCH] Add take-wanted --- take-wanted | 24 ++++++++++++++++++++++++ 1 file changed, 24 insertions(+) create mode 100755 take-wanted diff --git a/take-wanted b/take-wanted new file mode 100755 index 0000000..0274fc6 --- /dev/null +++ b/take-wanted @@ -0,0 +1,24 @@ +#!/usr/bin/env python3 + +# Take input filenames over stdin, write filenames to stdout only if they're for a wanted video +# +# Args: file containing just the wanted video IDs + +import re +import sys + +wanted_video_ids_fname = sys.argv[1] +wanted_video_ids = set() +with open(wanted_video_ids_fname, "rb") as f: + for line in f: + video_id = line.rstrip(b"\r\n") + assert len(video_id) == 11, video_id + wanted_video_ids.add(video_id) + del video_id + +for line in sys.stdin.buffer: + filename = line.rstrip(b"\r\n") + video_id = re.findall(rb"-[-_A-Za-z0-9]{11}\.", filename)[-1][1:-1] + assert len(video_id) == 11, video_id + if video_id in wanted_video_ids: + sys.stdout.buffer.write(filename + b"\n")