util.py

from itertools import takewhile
import numpy as np
import pandas as pd

def invalidate_records(frame, check_name):
    for index,percent in frame.ix[frame[check_name] > 0, check_name].iteritems():
        invalidate_record(frame, index, check_name, percent)
    return frame

def invalidate_record(frame, index, check_name, value):
    frame.ix[frame.index[index], check_name] = value
    corrupt_view = frame.iloc[index].sceneMirrorPosition
    if corrupt_view in [ord('H'),ord('A')]:

        # Corrupt calibration view, must also invalidate neighboring scenes
        _idx = index + 1
        while _idx < len(frame):
            if frame.sceneMirrorPosition.iloc[_idx] == corrupt_view:
                # Made one cycle
                break
            elif frame.sceneMirrorPosition.iloc[_idx] in [ord('H'), ord('A')]:
                # Skip opposite calibration views
                _idx += 1
                continue
            else:
                # Invalidate non-calibration views
                frame.ix[frame.index[_idx], check_name] = value
                _idx += 1

        _idx = index - 1
        while _idx >= 0:
            if frame.sceneMirrorPosition.iloc[_idx] == corrupt_view:
                # Made one cycle
                break
            elif frame.sceneMirrorPosition.iloc[_idx] in [ord('H'), ord('A')]:
                # Skip opposite calibration views
                _idx -= 1
                continue
            else:
                # Invalidate non-calibration views
                frame.ix[frame.index[_idx], check_name] = value
                _idx -= 1

    return frame

class BaseCheckList:

    def __init__(self, *args, **kwargs):
        self.check_results = {}
        self.parameters = {}

    def set_params(self, parameters):
        self.parameters = parameters

    def update_qc_percent(self, frame):
        for check_func in self.checks:
            name = check_func.__name__
            if name in frame.columns:
                results = frame[name].fillna(0)
                # Compute P(A U B)
                previous_percent = frame['qc_percent']
                frame['qc_percent'] = previous_percent + results - previous_percent*results
        return frame

    def compute(self, frame):
        # Filter bad records from previous level
        filtered_frame = frame.ix[frame.qc_percent < 1].copy()
        for check in self.checks:
            filtered_frame = check(filtered_frame, self.parameters)
        return self.update_qc_percent(frame.combine_first(filtered_frame))