# Uses python3
import sys
from collections import namedtuple

Segment = namedtuple('Segment', 'start end')


def fast_count_segments(starts, ends, points):
    # https://www.coursera.org/learn/algorithmic-toolbox/discussions/all/threads/QJ1jK9wNEeWdPBL2iFTrAw/replies/Ihiw4txhEeWK5g7mfcS2Xw/comments/oyAMaeIiEeWyqwpvChh66Q
    # I'm following the algo described above. I've worked it out using Java already
    # l->1, p->2, r->3
    index = 0
    segments = []
    tempPointAndIndex = []
    while index < len(starts):
        segments.append(Segment(starts[index], 1))
        segments.append(Segment(ends[index], 3))
        index = index + 1
    index = 0
    while index < len(points):
        segments.append(Segment(points[index], 2))
        tempPointAndIndex.append(Segment(points[index], index))
        index = index + 1
    segments.sort()
    tempPointAndIndex.sort()
    results = []
    count = 0
    for i in range(len(segments)):
        if segments[i].end == 1:
            count = count + 1
        if segments[i].end == 2:
            results.append(count)
        if segments[i].end == 3:
            count = count - 1
    finalResults = [0] * len(points)
    for i in range(len(results)):
       finalResults[tempPointAndIndex[i].end] = results[i] 
    return finalResults


if __name__ == '__main__':
    input = sys.stdin.read()
    data = list(map(int, input.split()))
    n = data[0]
    m = data[1]
    starts = data[2:2 * n + 2:2]
    ends = data[3:2 * n + 2:2]
    points = data[2 * n + 2:]
    # use fast_count_segments
    cnt = fast_count_segments(starts, ends, points)
    for x in cnt:
        print(x, end=' ')