def index_articles(self, titles=[], start=0, num=1000000):
"""
@param titles: list of url_titles to retrieve.
@param start: int Index of list at which to start.
@param num: int Number of articles to index, beginning at start.
@returns: (int num posted, list Articles that could not be posted)
"""
posted = 0
could_not_post = []
for n,title in enumerate(titles):
if (posted < num) and (n > start):
logging.debug('%s/%s %s' % (n, len(titles), title))
page = Proxy().page(title)
if (page.published or settings.MEDIAWIKI_SHOW_UNPUBLISHED):
page_sources = [source['encyclopedia_id'] for source in page.sources]
for source in page.sources:
logging.debug(' %s' % source['encyclopedia_id'])
docstore.post(
settings.DOCSTORE_HOSTS, settings.DOCSTORE_INDEX, 'sources',
source['encyclopedia_id'], source
)
page.sources = page_sources
docstore.post(
settings.DOCSTORE_HOSTS, settings.DOCSTORE_INDEX, 'articles',
title, page.__dict__
)
posted = posted + 1
logging.debug('posted %s' % posted)
else:
could_not_post.append(page)
if could_not_post:
logging.debug('Could not post these: %s' % could_not_post)
return posted,could_not_post
def index_authors(self, titles=[]):
"""
@param titles: list of url_titles to retrieve.
"""
for n,title in enumerate(titles):
logging.debug('%s/%s %s' % (n, len(titles), title))
page = Proxy().page(title)
docstore.post(
settings.DOCSTORE_HOSTS, settings.DOCSTORE_INDEX, 'authors',
title, page.__dict__
)
def index_topics(self, json_text=None, url=settings.DDR_TOPICS_SRC_URL):
"""Upload topics.json; used for Encyc->DDR links on article pages.
url = 'http://partner.densho.org/vocab/api/0.2/topics.json'
models.Elasticsearch().index_topics(url)
@param json_text: unicode Raw topics.json file text.
@param url: URL of topics.json
"""
if url and not json_text:
r = requests.get(url)
if r.status_code == 200:
json_text = r.text
docstore.post(
settings.DOCSTORE_HOSTS, settings.DOCSTORE_INDEX, 'vocab',
'topics', json.loads(json_text),
)
