#!/usr/bin/env python # vim: set sw=4 et: import argparse import os import sys import logging import brozzler import kombu import yaml import json arg_parser = argparse.ArgumentParser(prog=os.path.basename(__file__), description="brozzler-new-job - queue new job with brozzler", formatter_class=argparse.ArgumentDefaultsHelpFormatter) arg_parser.add_argument('job_conf_file', metavar='JOB_CONF_FILE', help='brozzler job configuration file in yaml') arg_parser.add_argument('-u', '--amqp-url', dest='amqp_url', default='amqp://guest:guest@localhost:5672/%2f', help='URL identifying the amqp server to talk to') arg_parser.add_argument("-v", "--verbose", dest="log_level", action="store_const", default=logging.INFO, const=logging.DEBUG) arg_parser.add_argument("--version", action="version", version="brozzler {} - {}".format(brozzler.version, os.path.basename(__file__))) args = arg_parser.parse_args(args=sys.argv[1:]) logging.basicConfig(stream=sys.stdout, level=args.log_level, format="%(asctime)s %(process)d %(levelname)s %(threadName)s %(name)s.%(funcName)s(%(filename)s:%(lineno)d) %(message)s") def merge(a, b): if isinstance(a, dict) and isinstance(b, dict): merged = dict(a) b_tmp = dict(b) for k in a: merged[k] = merge(a[k], b_tmp.pop(k, None)) merged.update(b_tmp) return merged elif isinstance(a, list) and isinstance(b, list): return a + b else: return a logging.info("loading %s", args.job_conf_file) with open(args.job_conf_file) as f: job_conf = yaml.load(f) logging.info("job_conf=%s", job_conf) seeds = job_conf.pop("seeds") # logging.info("=== global settings ===\n%s", yaml.dump(job_conf)) sites = [] for seed_conf in seeds: if "id" in seed_conf: seed_conf.pop("id") merged_conf = merge(seed_conf, job_conf) # XXX check for unknown settings, invalid url, etc # logging.info("merge(%s, %s) = %s", seed_conf, global_conf, merged_conf) # logging.info("=== seed_conf ===\n%s", yaml.dump(seed_conf)) # logging.info("=== merged_conf ===\n%s", yaml.dump(merged_conf)) extra_headers = None if "warcprox_meta" in merged_conf: warcprox_meta = json.dumps(merged_conf["warcprox_meta"], separators=(',', ':')) extra_headers = {"Warcprox-Meta":warcprox_meta} site = brozzler.Site(seed=merged_conf["url"], scope=merged_conf.get("scope"), time_limit=merged_conf.get("time_limit"), proxy=merged_conf.get("proxy"), ignore_robots=merged_conf.get("ignore_robots"), enable_warcprox_features=merged_conf.get("enable_warcprox_features"), extra_headers=extra_headers) sites.append(site) with kombu.Connection(args.amqp_url) as conn: q = conn.SimpleQueue("brozzler.sites.new") for site in sites: d = site.to_dict() logging.info("feeding amqp queue %s with %s", repr(q.queue.name), repr(d)) q.put(d)