#!/usr/bin/env python

# dois2metadata.py - given csv file containing a column named 'doi', save metadata describing them

# Eric Lease Morgan <emorgan@nd.edu>
# (c) University of Notre Dame; distributed under GNU Public License

# January 19, 2024 - first cut; in the middle of a creative fit
# January 26, 2024 - tweaking for a specific project


# configure
CACHE    = './etc/unpaywall.cache'
COLUMN   = 'doi'
METADATA = './etc/metadata.csv'

# require
from unpywall       import Unpywall
from unpywall.cache import UnpywallCache
import sys
import pandas as pd

# get input
if len( sys.argv ) != 2 : sys.exit( 'Usage: ' + sys.argv[ 0 ] + " <csv>" )
csv = sys.argv[ 1 ]

# initialize
Unpywall.init_cache( UnpywallCache( CACHE ) )

# read dois, process them, output, and done
dois     = list( pd.read_csv( csv )[ COLUMN ] )
metadata = Unpywall.doi( dois=dois, progress=True, errors='ignore' )      

# output the tiniest bit of statistics
sys.stderr.write( str( metadata.oa_status.value_counts(normalize=True) ) + '\n' )

# output the metadata and done
with open( METADATA, 'w' ) as handle : handle.write( metadata.to_csv( index=False ) )
exit()