Skip to main content

Browse the archive

Revision fe303085b0c8f132dfa07ce7ab5b38941a1a8d86 authored by Matthias Bernt on 26 August 2022, 10:55:08 UTC, committed by Matthias Bernt on 26 August 2022, 13:53:40 UTC 
bed, interval: set default for strandCol
 
bed and interval datatypes `check element_is_set` before overwriting
metadata. since the having `default=None` and `no_value=0` makes
`element_is_set` return True (since `None!=0`). Therefore metadata
values are never set
1 parent 4e23fbb
Raw File
parse_builds_3_sites.py 
#!/usr/bin/env python
"""
Connects to sites and determines which builds are available at each.
"""

import xml.etree.ElementTree as ElementTree

import requests

sites = ['http://genome.ucsc.edu/cgi-bin/',
         'http://archaea.ucsc.edu/cgi-bin/',
         'http://genome-test.gi.ucsc.edu/cgi-bin/']
names = ['main', 'archaea', 'test']


def main():
    for i in range(len(sites)):
        site = sites[i] + "das/dsn"
        trackurl = sites[i] + "hgTracks?"
        builds = []
        try:
            text = requests.get(site).text
        except Exception:
            print("#Unable to connect to " + site)
            continue

        try:
            tree = ElementTree.fromstring(text)
        except Exception:
            print("#Invalid xml passed back from " + site)
            continue
        print("#Harvested from", site)

        for dsn in tree:
            build = dsn.find("SOURCE").attrib['id']
            builds.append(build)
            build_dict = {}
        for build in builds:
            build_dict[build] = 0
            builds = list(build_dict.keys())
        yield [names[i], trackurl, builds]


if __name__ == "__main__":
    for site in main():
        print(site[0] + "\t" + site[1] + "\t" + ",".join(site[2]))
back to top