lto_habitat.py

#usr/bin/env python
import sys
import ififuncs

csv_file = sys.argv[1]
csv_dict = ififuncs.extract_metadata(csv_file)
habitat_dict = {}
for entry in csv_dict[0]:
    if entry['share'] == 'preservation_storage':
        accession_number = entry['path'].split('/')[1]
        if accession_number[:2] == 'aa':
            raw_tape_list = (entry['stloc_state_method'].split('|'))
            if len(raw_tape_list) > 1:
                if accession_number in habitat_dict:
                    tape_list = habitat_dict[accession_number]
                else:
                    tape_list = []
                tape1 = raw_tape_list[0].split(':')[0].strip()
                if accession_number in habitat_dict:
                    if tape1 not in tape_list:
                        tape_list.append(tape1)
                else:
                    tape_list.append(tape1)

                tape2 = raw_tape_list[1].split(':')[0].strip()
                if accession_number in habitat_dict:
                    if tape2 not in tape_list:
                        tape_list.append(tape2)
                else:
                    tape_list.append(tape2)
                    
                tape3 = raw_tape_list[2].split(':')[0].strip()
                
                if accession_number in habitat_dict:
                    if tape3 not in tape_list:
                        tape_list.append(tape3)
                else:
                    tape_list.append(tape3)

                habitat_dict[accession_number] = tape_list
        
                #habitat_dict[accession_number] = entry['stloc_state_method']
outliers = []
csv_headings = 'Accession Number, Habitat, backup_habitat'
print(csv_headings)
for key,value in sorted(habitat_dict.items()):
    
    string = ''
    
    #print key, " => " , value
    if len(value) == 3:
        primary_habitat = 'preservation_storage, %s, comms_room|' % value[0]
        
        for habitat in value[1:]:
            string += 'preservation_storage, %s, comms_room|' % habitat

        print('%s,%s,%s' % (key, primary_habitat, string))            
    else:
        outliers.append([key, value])
for i in outliers:
    print(i)