hf_dumper

#!/usr/bin/env python

import ROOT
ROOT.gROOT.SetBatch(True)
import yaml
import click
import os
from xml.etree import ElementTree as etree

def get_path(basedir,relpath):
    return '{}/{}'.format(basedir,relpath.split('./',1)[-1])

def parse_histfactory_xml(toplvlxml):
    dirname = os.path.abspath(os.path.dirname(toplvlxml))
    histfithome = dirname.split('/config')[0]

    p = etree.parse(toplvlxml)
    channels =  [etree.parse(open(get_path(histfithome,inpt.text))).findall('.')[0] for inpt in p.findall('Input')]

    parsed_data = {
        'Combination':{
            'Prefix':p.findall('.')[0].attrib['OutputFilePrefix'].split('./',1)[-1],
            'Measurements':[ {'name':x.attrib['Name'] for x in p.findall('Measurement')}]
        }
    }

    channel_info = []
    for input_tag in p.findall('Input'):
        channel_xml = etree.parse(open(get_path(histfithome,input_tag.text)))
        channel_name = channel_xml.findall('.')[0].attrib['Name']
        sample_names = [x.attrib['Name'] for x in channel_xml.findall('Sample')]
        channel_info += [{'name':channel_name,'samples':sample_names}]

    parsed_data['Combination']['Inputs'] = channel_info
    return parsed_data


@click.command()
@click.argument('toplvlxml')
@click.argument('output')
def dump_information(toplvlxml,output):
    parsed_data = parse_histfactory_xml(toplvlxml)
    with open(output,'w') as f:
        f.write(yaml.safe_dump(parsed_data,default_flow_style = False))

if __name__=='__main__':
  dump_information()