jjjake’s gists

jjjake / upload_prelinger.py

Created October 24, 2012 19:07

	#!/home/jake/.virtualenvs/ia-wrapper/bin/python
	import logging
	from datetime import datetime
	import re
	import threading, Queue

	import archive


	## Logging!

jjjake / count_derives.py

Last active December 10, 2015 13:28

Count how many times an archive.org item has been derived.

	#!/usr/bin/env python
	import sys
	import os
	import json

	import requests


	def get_tasks(identifier, params={}):

jjjake / ait_derive.php

Created January 8, 2013 20:11

Return a list of AIT items without CDXs, older than one week (for handoff to auto_submit.php perpetual loop).

	<?
	function get_data($url) {
	$ch = curl_init();
	$timeout = 5;
	curl_setopt($ch, CURLOPT_URL, $url);
	curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
	curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, $timeout);
	$data = curl_exec($ch);
	curl_close($ch);
	return $data;

jjjake / wcd_daily_derive.php

Last active December 10, 2015 20:18

Return a list of wcd items without ogg's, no older than two days (for handoff to auto_submit.php perpetual loop).

	<?
	function get_data($url) {
	$ch = curl_init();
	$timeout = 5;
	curl_setopt($ch, CURLOPT_URL, $url);
	curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
	curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, $timeout);
	$data = curl_exec($ch);
	curl_close($ch);
	return $data;

jjjake / no_cdx.py

Created January 11, 2013 02:22

	#!/usr/bin/env python
	from datetime import datetime
	import logging
	import sys

	import ujson

	# parallel_md_get.py available here: https://gist.github.com/3784845
	from parallel_md_get import metadata_record_iterator

jjjake / get_underived_thumper_items.py

Last active December 10, 2015 23:08

	#!/usr/bin/env python
	import datetime
	import logging
	import sys

	import ujson

	# parallel_md_get.py available here: https://gist.github.com/3784845
	from parallel_md_get import metadata_record_iterator

jjjake / get_underived_thumper_items-2.py

Last active December 10, 2015 23:59

	#!/usr/bin/env python
	import datetime
	import logging
	import sys

	import ujson

	# parallel_md_get.py available here: https://gist.github.com/3784845
	from parallel_md_get import metadata_record_iterator

jjjake / mapi.py

Created January 14, 2013 21:25

	def write_metadata(identifier, metadata={}, target='metadata'):
	log_in_cookies = {'logged-in-sig': os.environ['LOGGED_IN_SIG'],
	'logged-in-user': os.environ['LOGGED_IN_USER']}
	url = 'http://archive.org/metadata/%s' % identifier
	src = requests.get(url).json().get(target, {})
	dest = dict((src.items() + metadata.items()))
	json_patch = jsonpatch.make_patch(src, dest).patch
	patch = [{p['op']: p['path'], 'value': p['value']} for p in json_patch]
	if patch == []:
	return 'no changes made to Metadata.'

jjjake / update_metamgr.py

Created January 16, 2013 20:58

	#!/usr/bin/env python
	"""Update a single items record in metamgr.

	Usage:
	./update_metamgr.py $identifier

	Using parallel:
	parallel ./update_metamgr.py < itemlist.txt

	"""

jjjake / irs.py

Created February 4, 2013 20:46

	#!/usr/bin/env python
	import sys
	import ujson

	# Script/Library created by Mike McCabe to do multiple
	# metadata fetches in parallel
	# Available here: https://gist.github.com/3784845
	from parallel_md_get import metadata_record_iterator