Merge VIVO resoruces.

merge_resources.py

"""
Script to login to the VIVO interface as use the 
'Merge Resources' tool under 'Ingest Tools' in the admin interface.

An incoming CSV file without headers is expected. 
 - the primary resource URI is in the first column
 - the duplicate resource URI is in the second column

This script relies on the python requests package:
http://docs.python-requests.org/en/latest/user/install/

Run as:

$python merge_resources.py duplicate_file.csv

Use with caution as it merges data in your system.  
"""

import csv
import os
import sys

import requests

#Before running export the following environment variables
#or change the user, password, and base_url variables below.
#export VIVO_USER=user
#export VIVO_PASS=password
#export VIOV_USER=http://vivo.school.edu/

user = os.getenv('VIVO_USER')
password = os.getenv('VIVO_PASS')
base_url = os.getenv('VIVO_URL')

if not base_url.endswith('/'):
    base_url += '/'

if (user is None) or (password is None) or (base_url is None):
    raise Exception("""
        VIVO credentials not found.
        Environment variables need to be set.
        $export VIVO_USER=user
        $export VIVO_PASS=password
        $export VIVO_URL=http://vivo.edu/
        """)

payload = {
        'loginName': user,
        'loginPassword': password,
        'loginForm': 'Log in'
}

#VIVO session
session = requests.session()

#login
login = session.post(
    base_url + 'authenticate',
    data=payload,
    verify=False
)

if login.status_code != 200:
    raise Exception("VIVO login failed.")

#Open and loop through incoming csv file.
with open(sys.argv[1]) as infile:
    #Not expecting headers in the file.
    #Expecting primary uri in column 0 and secondary uri in column 1
    for index, row in enumerate(csv.reader(infile)):
        uri1 = row[0]
        uri2 = row[1]
        print>>sys.stderr, "Merging", uri1, uri2
        params = {'action': 'mergeResources',
                  'uri1': uri1,
                  'uri2': uri2,
                  'usePrimaryLabelOnly': 'Use Primary Label Only',
                  'submit': 'Merge resources'}
        #Call merge - it's a GET!
        merge = session.get(base_url + 'ingest', params=params)

        if merge.url == base_url + 'authenticate':
            raise Exception("Failed to login to VIVO.")
        if merge.status_code != 200:
            raise Exception("Merge failed; requests output: %s" % merge.content)

logout = session.get(base_url + 'logout')