changesetcomments.py

#!/usr/bin/env python

import argparse

# results = []
from requests_xml import XMLSession
import re
import datetime
from collections import Counter

session = XMLSession()


def main(cohort):
    now = datetime.datetime.utcnow().replace(microsecond=0).isoformat()
    hotr = re.compile("#hotosm-project-\d+")
    since="2018-06-11T00:00:00"
    allhot = Counter()
    
    for name in cohort:
        name = name.rstrip()
        all_comments = getcomments(name, since, now)
        hot_changesets = list(filter(hotr.match, all_comments))
        hot_activities = Counter(map(lambda x: hotr.findall(x)[0][16:], hot_changesets))
        if len(hot_activities) == 0:
            hot_activities = "none"
        else:
            allhot.update(hot_activities)
        results = {'name': name, 'since': since, 'changesets': len(all_comments), 'hot': len(hot_changesets), 'hot_activities': hot_activities}
        print(results)
    print(sorted(allhot.items()))

def getcomments(name, since, now):
    r = session.get('https://api.openstreetmap.org/api/0.6/changesets?display_name='+name+'&time='+since+','+now+'"')
    comments = [item[0] for item in r.xml.search('<tag k="comment" v="{}"/>')]
    if len(comments) == 100: #recurse for more
        oldest = r.xml.xpath('(//osm/changeset)[last()]', first=True)
        now = oldest.attrs['created_at']
        comments.extend(getcomments(name, since, now))
    return(comments)

if __name__ == "__main__":
    parser = argparse.ArgumentParser()
    parser.add_argument('cohort', type=argparse.FileType('r'))
    args = parser.parse_args()
    main(args.cohort.readlines())