test_api_articlemetaapi.py

# coding: utf-8

import time
import pprint

from opac_proc.web import config
from opac_proc.extractors.source_clients.thrift import am_clients
from articlemeta.client import ThriftClient

COLLECTION = 'spa'
ISSN = '0034-8910'

'''

Evaluation time approximalety

SPA['RSP'](ids):
    Time to get ids journal: 0.13660001754760742 seconds.
    Time to get ids issue: 2.8168540000915527 seconds.
    Time to get ids article: 4.821263074874878 seconds.

SPA['RSP'](objects):
    Time to get ids journal: 0.21854615211486816 seconds.
    Time to get ids issue: 14.316105127334595 seconds.
    Time to get ids article: 1258.500862121582 seconds.
'''


def main(cl, articlemeta):

    # print("Init ids evaluation")

    # ids_evaluation(cl)

    # print("Finish ids evaluation")

    print("Init objects evaluation")

    objects_evaluation(cl, articlemeta)

    print("Finish objects evaluation")


def objects_evaluation(cl, articlemeta):

    # Start time
    journal_start = time.time()

    journals = cl.journals(issn=ISSN, collection=COLLECTION)

    journal_list = [j for j in journals]

    # End Time
    journal_end = time.time()

    pprint.pprint(journal_list)

    print("Journal total: %s" % len(journal_list))

    # Start time
    issue_start = time.time()

    issues = cl.issues(issn=ISSN, collection=COLLECTION)

    issue_list = [i for i in issues]

    # End Time
    issue_end = time.time()

    pprint.pprint(issue_list)

    print("Issue total: %s" % len(issue_list))

    article_id_list = articlemeta.get_article_identifiers(
                        collection=COLLECTION,
                        issn=ISSN)

    # Start time
    article_start = time.time()

    for article_id in article_id_list:

        article = cl.document(code=article_id, collection=COLLECTION, body=True)

    # article_list = [d for d in articles]
        print(article.publisher_id)

    # End Time
    article_end = time.time()

    # pprint.pprint(article_list)

    # print("Article total: %s" % len(article_list))

    print(80 * '*')

    print("Time to get object journal: {0} seconds.".format(journal_end-journal_start))

    print("Time to get object issue: {0} seconds.".format(issue_end-issue_start))

    print("Time to get object article: {0} seconds.".format(article_end-article_start))

    print(80 * '*')


def ids_evaluation(cl, articlemeta):

    # Start time
    journal_start = time.time()

    journals = cl.journals(issn=ISSN, only_identifiers=True, collection=COLLECTION)

    journal_list = [j for j in journals]

    # End Time
    journal_end = time.time()

    pprint.pprint(journal_list)

    print("Journal total: %s" % len(journal_list))

    # Start time
    issue_start = time.time()

    issues = cl.issues(issn=ISSN, only_identifiers=True, collection=COLLECTION)

    issue_list = [i for i in issues]

    # End Time
    issue_end = time.time()

    pprint.pprint(issue_list)

    print("Issue total: %s" % len(issue_list))

    # Start time
    article_start = time.time()

    articles = cl.documents(issn=ISSN, collection=COLLECTION, only_identifiers=True)

    article_list = [d for d in articles]

    # End Time
    article_end = time.time()

    pprint.pprint(article_list)

    print("Article total: %s" % len(article_list))

    print(80 * '*')

    print("Time to get ids journal: {0} seconds.".format(journal_end-journal_start))

    print("Time to get ids issue: {0} seconds.".format(issue_end-issue_start))

    print("Time to get ids article: {0} seconds.".format(article_end-article_start))

    print(80 * '*')


if __name__ == "__main__":

    cl = ThriftClient()

    articlemeta = am_clients.ArticleMeta(
            config.ARTICLE_META_THRIFT_DOMAIN,
            config.ARTICLE_META_THRIFT_PORT)

    main(cl, articlemeta)