def authors(report=False, dryrun=False, force=False):
index = set_hosts_index()
logprint('debug', '------------------------------------------------------------------------')
logprint('debug', 'getting mw_authors...')
mw_author_titles = Proxy().authors(cached_ok=False)
mw_articles = Proxy().articles_lastmod()
logprint('debug', 'getting es_authors...')
es_authors = Author.authors()
if force:
logprint('debug', 'forcibly update all authors')
authors_new = [page['title'] for page in es_authors]
authors_delete = []
else:
logprint('debug', 'determining new,delete...')
authors_new,authors_delete = Elasticsearch.authors_to_update(
mw_author_titles, mw_articles, es_authors)
logprint('debug', 'mediawiki authors: %s' % len(mw_author_titles))
logprint('debug', 'authors to add: %s' % len(authors_new))
logprint('debug', 'authors to delete: %s' % len(authors_delete))
if report:
return
logprint('debug', 'deleting...')
for n,title in enumerate(authors_delete):
logprint('debug', '--------------------')
logprint('debug', '%s/%s %s' % (n, len(authors_delete), title))
author = Author.get(title)
if not dryrun:
author.delete()
logprint('debug', 'adding...')
for n,title in enumerate(authors_new):
logprint('debug', '--------------------')
logprint('debug', '%s/%s %s' % (n, len(authors_new), title))
logprint('debug', 'getting from mediawiki')
mwauthor = Proxy().page(title)
try:
existing_author = Author.get(title)
logprint('debug', 'exists in elasticsearch')
except:
existing_author = None
logprint('debug', 'creating author')
author = Author.from_mw(mwauthor, author=existing_author)
if not dryrun:
logprint('debug', 'saving')
author.save()
try:
a = Author.get(title)
except NotFoundError:
logprint('error', 'ERROR: Author(%s) NOT SAVED!' % title)
logprint('debug', 'DONE')
def create_index():
index = set_hosts_index()
logprint('debug', 'creating new index')
index = Index(settings.DOCSTORE_INDEX)
index.create()
logprint('debug', 'creating mappings')
Author.init()
Page.init()
Source.init()
logprint('debug', 'registering doc types')
index.doc_type(Author)
index.doc_type(Page)
index.doc_type(Source)
logprint('debug', 'DONE')
def author(request, url_title, template_name="wikiprox/author.html"):
try:
author = Author.get(url_title)
author.scrub()
except NotFoundError:
raise Http404
return render_to_response(template_name, {"author": author}, context_instance=RequestContext(request))
def author(request, url_title, format=None):
"""DOCUMENTATION GOES HERE.
"""
try:
author = Author.get(url_title)
except NotFoundError:
return Response(status=status.HTTP_404_NOT_FOUND)
articles = [
{
'title': article.title,
'url': reverse('wikiprox-api-page', args=([article.url_title]), request=request),
}
for article in author.articles()
]
data = {
'url_title': author.url_title,
'url': reverse('wikiprox-api-author', args=([author.url_title]), request=request),
'absolute_url': reverse('wikiprox-author', args=([author.url_title]), request=request),
'title': author.title,
'title_sort': author.title_sort,
'body': author.body,
'modified': author.modified,
'articles': articles,
}
return Response(data)
def article(request, url_title="index", printed=False, template_name="wikiprox/page.html"):
"""
"""
alt_title = url_title.replace("_", " ")
try:
page = Page.get(url_title)
page.scrub()
except NotFoundError:
page = None
if not page:
try:
page = Page.get(alt_title)
page.scrub()
except NotFoundError:
page = None
if not page:
# might be an author
author_titles = [author.title for author in Author.authors()]
if url_title in author_titles:
return HttpResponseRedirect(reverse("wikiprox-author", args=[url_title]))
elif alt_title in author_titles:
return HttpResponseRedirect(reverse("wikiprox-author", args=[alt_title]))
raise Http404
if (not page.published) and (not settings.MEDIAWIKI_SHOW_UNPUBLISHED):
template_name = "wikiprox/unpublished.html"
elif printed:
template_name = "wikiprox/article-print.html"
else:
template_name = "wikiprox/article.html"
# DDR objects
# show small number of objects, distributed among topics
TOTAL_OBJECTS = 10
PAGE_OBJECTS = 8
try:
terms_objects = page.ddr_terms_objects(size=TOTAL_OBJECTS)
ddr_error = None
except requests.exceptions.ConnectionError:
terms_objects = []
ddr_error = "ConnectionError"
except requests.exceptions.Timeout:
terms_objects = []
ddr_error = "Timeout"
ddr_objects = ddr.distribute_list(terms_objects, PAGE_OBJECTS)
ddr_objects_width = 280
ddr_img_width = ddr_objects_width / (PAGE_OBJECTS / 2)
return render_to_response(
template_name,
{
"page": page,
"ddr_error": ddr_error,
"ddr_objects": ddr_objects,
"ddr_objects_width": ddr_objects_width,
"ddr_img_width": ddr_img_width,
},
context_instance=RequestContext(request),
)
def authors(request, format=None):
"""DOCUMENTATION GOES HERE.
"""
data = [
{
'title': author.title,
'title_sort': author.title_sort,
'url': reverse('wikiprox-api-author', args=([author.url_title]), request=request),
}
for author in Author.authors()
]
return Response(data)
def authors(request, template_name="wikiprox/authors.html"):
return render_to_response(
template_name, {"authors": Author.authors(num_columns=4)}, context_instance=RequestContext(request)
)