Get number of times multiple intervals overlap

[jvinolas]

Is there any way with your library using IntervalDict to get atomic intervals while getting as value the number of times it was overlapped with other intervals in list?

I've done this simple script that gets atomic intervals but I don't know how to get as value the units field in dict multiplied by the number of times the new atomic interval is overlapped with original intervals list.

plans = [{
    "start": "2022-03-29T00:00:00+0000" ,
    "end":   "2022-03-29T05:59:59+0000" ,
    "units": 10 ,
},{
    "start": "2022-03-29T01:00:00+0000" ,
    "end":   "2022-03-29T04:59:59+0000" ,
    "units": 10 ,
},{
    "start": "2022-03-29T02:00:00+0000" ,
    "end":   "2022-03-29T03:59:59+0000" ,
    "units": 10 ,
},{
    "start": "2022-03-29T02:29:59+0000" ,
    "end":   "2022-03-29T03:29:59+0000" ,
    "units": 10 ,
},{
    "start": "2022-03-29T06:29:59+0000" ,
    "end":   "2022-03-29T07:29:59+0000" ,
    "units": 10 ,
}]

def get_plans_intersected(plans):
    d = P.IntervalDict()
    i=1
    for plan in plans:
        d[P.closed(plan["start"], plan["end"])] = i*plan["units"]
        i+=1

    items=[]
    for interval, value in d.items():
        for atomic in interval: 
            items.append((atomic, value))
    items.sort()
    return [{"start":P.to_data(item[0])[0][1],"end":P.to_data(item[0])[0][2],"units":item[1]} for item in items]

pprint(get_plans_intersected(plans))

And this is the output:

[{'end': '2022-03-29T01:00:00+0000',
  'start': '2022-03-29T00:00:00+0000',
  'units': 10},
 {'end': '2022-03-29T02:00:00+0000',
  'start': '2022-03-29T01:00:00+0000',
  'units': 20},
 {'end': '2022-03-29T02:29:59+0000',
  'start': '2022-03-29T02:00:00+0000',
  'units': 30},
 {'end': '2022-03-29T03:29:59+0000',
  'start': '2022-03-29T02:29:59+0000',
  'units': 40},
 {'end': '2022-03-29T03:59:59+0000',
  'start': '2022-03-29T03:29:59+0000',
  'units': 30},
 {'end': '2022-03-29T04:59:59+0000',
  'start': '2022-03-29T03:59:59+0000',
  'units': 20},
 {'end': '2022-03-29T05:59:59+0000',
  'start': '2022-03-29T04:59:59+0000',
  'units': 10},
 {'end': '2022-03-29T07:29:59+0000',
  'start': '2022-03-29T06:29:59+0000',
  'units': 50}]

The result units values are just a coincidence of the original plans list data ordering as for example the non overlapping interval gets 50 units.

[AlexandreDecan]

Hi,

I'm not sure I understood exactly what you would like to have at the end, but have a look at IntervalDict.combine. E.g.:

output = P.IntervalDict()
for interval in plans:
    i = P.closed(interval['start'], interval['end'])
    d = P.IntervalDict({i: 1})
    output = output.combine(d, how=lambda x,y: x+y)

Given all the things that happen in combine, do not expect it to work fast on large list of intervals, though.

[jvinolas]

It seems now I'm getting the number of times the atomic interval was overlapped. I understand now the combine should be done between each two intervals to decide what value should I get as a result. That is really powerful.

Thanks a lot for this great library for handling intervals in python!

[AlexandreDecan]

You're welcome :-)