itproject-code/pipeline/consolidators.py

"""Consolidators for features extracted from media files"""

from abc import ABC

class Consolidator(ABC):
    """Consolidator interface.

    A Consolidator's purpose is to reduce a list of features based on eg adjacency/overlap in time.

    Illustration
    ============

    feature1:
    <---------->
    feature2:
         <------------>
    consolidated:
    <----------------->

    TODO: consider how to ensure we only try to merge features from the same path!! #important
    """
    def __init__(self, features=None):
        if features is None or len(features) == 0:
            raise ValueError("Features must be provided") # TODO: #API -- decide if this is necessary
        # maybe we want to permit [] which would 'consolidate' to []
        self.features = features

        def consolidate(self):
            """Consolidate features by some criterion/criteria (eg overlap, adjacency)"""


class OverlapConsolidator(Consolidator):
    """Consolidator that merges overlapping features.

    An overlap is defined as two features that share a common interval of time,ie:
      interval1.end < interval2.start or
      interval2.start < interval1.end
    """

    def consolidate(self):
        """Consolidate overlapping features"""
        # sort features by start time
        self.features.sort(key=lambda x: x.interval.start)

        # merge overlapping features
        # TODO: check my working here (or write tests...)
        consolidated = []
        current = self.features[0]
        for feature in self.features[1:]:
            if current.interval.end < feature.interval.start:
                # no overlap
                consolidated.append(current)
                current = feature
            else:
                # overlap
                current.interval.end = max(current.interval.end, feature.interval.end)
        consolidated.append(current)

        self.features = consolidated