From e0b53d9c865e44b38777eb55abc7472f66ec44cf Mon Sep 17 00:00:00 2001 From: Holger Hans Peter Freyther Date: Sun, 22 Nov 2009 17:19:41 +0100 Subject: [PATCH] [mirror] Add a utilitiy to get the content of a given file Be able to extract one file from the content. --- host-tools/mirror/get_file.py | 23 +++++++++++++++++++++++ 1 file changed, 23 insertions(+) create mode 100755 host-tools/mirror/get_file.py diff --git a/host-tools/mirror/get_file.py b/host-tools/mirror/get_file.py new file mode 100755 index 0000000..9bbdbe6 --- /dev/null +++ b/host-tools/mirror/get_file.py @@ -0,0 +1,23 @@ +#!/usr/bin/env python + +import sqlite3, sys, optparse + +def parse(): + parser = optparse.OptionParser(version = "Print the content of a file to stdout", + usage = "%prog [options] URL") + parser.add_option("-d", "--db", help = "The crawl database to be used", + action = "store", dest = "db_name", default="crawl_db.db") + parser.add_option("-o", "--output", help = "Write the file to output", + action = "store", dest = "output", default=None) + return parser.parse_args(sys.argv) + +(opts, args) = parse() +connection = sqlite3.connect(opts.db_name) + +out = sys.stdout +if opts.output: + out = open(opts.output, "w") + +cursor = connection.execute("SELECT data FROM responses WHERE url like ?", [args[1]]) +for row in cursor: + print >> out, row[0] -- 2.1.4