|
| 1 | +#!/usr/bin/env python3 |
| 2 | + |
| 3 | +import sys |
| 4 | +import re |
| 5 | +import time |
| 6 | +import gzip |
| 7 | +from urllib.parse import urljoin |
| 8 | +import requests |
| 9 | +import json |
| 10 | + |
| 11 | +re.compile |
| 12 | + |
| 13 | +isCommit = re.compile("^[0-9a-fA-F]{40}$") |
| 14 | + |
| 15 | +HEADERS = { |
| 16 | + 'User-Agent': 'cwlviewer-load/0.0.1', |
| 17 | + 'Accept': 'application/json' |
| 18 | +} |
| 19 | + |
| 20 | + |
| 21 | +def parse_gitinfos(sourceFp): |
| 22 | + doc = json.load(sourceFp) |
| 23 | + for wf in doc["content"]: |
| 24 | + yield wf["retrievedFrom"] |
| 25 | + |
| 26 | +def is_not_commit(gitinfo): |
| 27 | + return not isCommit.match(gitinfo["branch"]) |
| 28 | + |
| 29 | +def make_requests(gitinfos): |
| 30 | + for git in gitinfos: |
| 31 | + # dict_keys(['repoUrl', 'branch', 'path', 'packedId', 'url', 'rawUrl', 'type']) |
| 32 | + req = { |
| 33 | + "url": git["repoUrl"], |
| 34 | + "branch": git["branch"], |
| 35 | + "path": git["path"] |
| 36 | + } |
| 37 | + if git["packedId"]: |
| 38 | + req["packedId"] = git["packedId"] |
| 39 | + yield req |
| 40 | + |
| 41 | +def send(base, req): |
| 42 | + url = urljoin(base, "/workflows") |
| 43 | + r = requests.post(url, data=req, allow_redirects=False, headers=HEADERS) |
| 44 | + print("Posted: %s" % req) |
| 45 | + if r.status_code == 202: |
| 46 | + location = urljoin(url, r.headers["Location"]) |
| 47 | + print(" queued: %s" % location) |
| 48 | + # need to check later |
| 49 | + return location |
| 50 | + if r.status_code == 303: |
| 51 | + print (" done: %s" % r.headers["Location"]) |
| 52 | + return None # Already there, all OK |
| 53 | + print("Unhandled HTTP status code: %s %s" % |
| 54 | + (r.status_code, r.text)) |
| 55 | + |
| 56 | +def send_requests(base, requests): |
| 57 | + for req in requests: |
| 58 | + yield send(base, req) |
| 59 | + |
| 60 | +def is_running(location): |
| 61 | + if not location: |
| 62 | + return True |
| 63 | + queued = requests.get(location, allow_redirects=False, headers=HEADERS) |
| 64 | + if queued.status_code == 303: |
| 65 | + # Done! |
| 66 | + return False |
| 67 | + j = queued.json() |
| 68 | + if j["cwltoolStatus"] == "RUNNING": |
| 69 | + return True |
| 70 | + elif j["cwltoolStatus"] == "ERROR": |
| 71 | + print("Failed %s: %s" % (location, j["message"])) |
| 72 | + return False |
| 73 | + else: |
| 74 | + raise Exception("Unhandled queue status: %s %s" % (queued.status_code, queued.text)) |
| 75 | + |
| 76 | +MAX_CONCURRENT=6 # Maximum number in queue |
| 77 | +SLEEP=0.5 # wait SLEEP seconds if queue is full |
| 78 | + |
| 79 | +def trim_queue(queue): |
| 80 | + new_queue = [] |
| 81 | + for q in queue: |
| 82 | + if is_running(q): |
| 83 | + #print("Still running %s" % q) |
| 84 | + new_queue.append(q) |
| 85 | + print("Trimmed queue from %s to %s" % (len(queue), len(new_queue))) |
| 86 | + return new_queue |
| 87 | + |
| 88 | +def main(jsonfile="-", base="http://view.commonwl.org:8082/", *args): |
| 89 | + if jsonfile == "-": |
| 90 | + source = sys.stdin |
| 91 | + elif jsonfile.endswith(".gz"): |
| 92 | + source = gzip.open(jsonfile, "rb") |
| 93 | + else: |
| 94 | + source = open(jsonfile, "rb") |
| 95 | + |
| 96 | + gitinfos = parse_gitinfos(source) |
| 97 | + if "--no-commits" in args: |
| 98 | + gitinfos = filter(is_not_commit, gitinfos) |
| 99 | + |
| 100 | + requests = make_requests(gitinfos) |
| 101 | + queued = [] |
| 102 | + for q in send_requests(base, requests): |
| 103 | + if q: |
| 104 | + queued.append(q) |
| 105 | + while len(queued) >= MAX_CONCURRENT: |
| 106 | + time.sleep(SLEEP) |
| 107 | + queued = trim_queue(queued) |
| 108 | + # Finish the rest of the queue |
| 109 | + while queued: |
| 110 | + queued = trim_queue(queued) |
| 111 | + |
| 112 | +if __name__ == "__main__": |
| 113 | + if "-h" in sys.argv: |
| 114 | + print("load.py [jsonfile] [baseurl] [--no-commits]") |
| 115 | + sys.exit(1) |
| 116 | + |
| 117 | + main(*sys.argv[1:]) |
| 118 | + |
0 commit comments