Get prompts for Austalk items stored in Alveo using a SPARQL query

get_prompts.py

"""
Author: Steve Cassidy
Date: 2/11/2017

Query the Alveo SPARQL API for information about the prompt
for an item list taken from the Austalk collection.

This should be in the item metadata but due to an error in the ingest
it was lost, however it is still stored in the RDF metadata since
there is a link from each item to it's 'prototype' in the protocol
and this has a record of the prompt. Consequently we just need to
run a SPARQL query to find the relevant information.

"""

from __future__ import print_function
import pyalveo


PREFIXES = """
PREFIX dcterms:<http://purl.org/dc/terms/>
PREFIX austalk:<http://ns.austalk.edu.au/>
"""


def find_prompt(client, item):
    """Get the prompt string for an Austalk item, return
    a tuple (id, prompt) if found, None if not"""

    query = PREFIXES + """
    select ?id ?prompt {
        BIND (<%s> as ?item)
        ?item dcterms:identifier ?id .
        ?item austalk:prototype ?prot .
        ?prot austalk:prompt ?prompt .
    }
    """ % item

    result = client.sparql_query('austalk', query)

    b = result['results']['bindings']

    # should be exactly one result
    if len(b) == 1:
        return b[0]['id']['value'], b[0]['prompt']['value']
    else:
        return None


if __name__=='__main__':
    import sys

    client = pyalveo.Client()

    # item list URL is the only argument
    listurl = sys.argv[1]

    itemlist = client.get_item_list(listurl)
    for item in itemlist:
        id, prompt = find_prompt(client, item)
        print(id, prompt)