def test_journal(self):
field, pattern = search_from_reference(
'[1] J. Mars, oh hello, Nucl.Phys. B76 (1974) 477-482')
self.assertEqual(field, 'journal')
self.assert_('Nucl' in pattern)
self.assert_('B76' in pattern)
self.assert_('477' in pattern)
def answer(self, req, user_info, of, cc, colls_to_search, p, f,
search_units, ln):
"""Answer question given by context.
Return (relevance, html_string) where relevance is integer
from 0 to 100 indicating how relevant to the question the
answer is (see C{CFG_WEBSEARCH_SERVICE_MAX_SERVICE_ANSWER_RELEVANCE}
for details), and html_string being a formatted answer.
"""
from invenio.refextract_api import search_from_reference
_ = gettext_set_language(ln)
if f or not self.seems_a_journal_reference(p):
return (0, "")
(field, pattern) = search_from_reference(p.decode('utf-8'))
if field is not "journal":
return (0, "")
recids = perform_request_search(req=req,
p=pattern,
f=field,
cc=cc,
c=colls_to_search)
if not recids:
return (0, "")
if len(recids) == 1:
recid = recids.pop()
user_info = collect_user_info(req)
return (100, """\
<p><span class="journalhint">%s</span></p>
<table style="padding: 5px; border: 2px solid #ccc; margin: 20px"><tr><td>
%s
</td></tr></table>""" % (escape(_("Were you looking for this paper?")),
print_record(recid, ln=ln, user_info=user_info)))
query = "find rawref \"" + p + "\""
query_link = CFG_SITE_URL + '/search?' + urlencode({'p': query})
return (
80, '<span class="journalhint">%s</span>' %
(_("Were you looking for a journal reference? Try: %(x_href)s") % {
"x_href":
'<a href="{0}">{1}</a>'.format(escape(query_link, True),
escape(query))
}, ))
def answer(self, req, user_info, of, cc,
colls_to_search, p, f, search_units, ln):
"""Answer question given by context.
Return (relevance, html_string) where relevance is integer
from 0 to 100 indicating how relevant to the question the
answer is (see C{CFG_WEBSEARCH_SERVICE_MAX_SERVICE_ANSWER_RELEVANCE}
for details), and html_string being a formatted answer.
"""
from invenio.refextract_api import search_from_reference
_ = gettext_set_language(ln)
if f or not self.seems_a_journal_reference(p):
return (0, "")
(field, pattern) = search_from_reference(p.decode('utf-8'))
if field is not "journal":
return (0, "")
recids = perform_request_search(
req=req, p=pattern, f=field, cc=cc, c=colls_to_search)
if not recids:
return (0, "")
if len(recids) == 1:
recid = recids.pop()
user_info = collect_user_info(req)
return (100, """\
<p><span class="journalhint">%s</span></p>
<table style="padding: 5px; border: 2px solid #ccc; margin: 20px"><tr><td>
%s
</td></tr></table>""" % (escape(_("Were you looking for this paper?")),
print_record(recid, ln=ln, user_info=user_info)))
query = "find rawref \"" + p + "\""
query_link = CFG_SITE_URL + '/search?' + urlencode({'p': query})
return (80, '<span class="journalhint">%s</span>' % (
_("Were you looking for a journal reference? Try: %(x_href)s") %
{"x_href": '<a href="{0}">{1}</a>'.format(
escape(query_link, True), escape(query))}, ))
def check_record(record):
"""
ensure that 999C5s is in standard form "p,v,c"
fix if possible, otherwise flag
"""
for pos, val in record.iterfield('999C5s'):
if val.count(',') < 2:
val = val.decode('utf-8')
field, pubnote = search_from_reference(val)
if pubnote and field == 'journal':
if abs(len(pubnote) - len(val)) < 3:
record.amend_field(pos, pubnote)
else:
record.warn("fix ref 999C5s: '{0}' ==> '{1}'".format(val, pubnote))
elif field == 'report':
record.warn("report number in 999C5s, should move '{0}' ==> '999C5r:{1}'".format(
val, pubnote))
else:
record.warn("999C5s in non-standard form: '{0}'".format(
val))
def test_report(self):
field, pattern = search_from_reference(
'[1] J. Mars, oh hello, [hep-ph/0104088]')
self.assertEqual(field, 'report')
self.assertEqual(pattern, 'hep-ph/0104088')
def test_not_recognized(self):
field, pattern = search_from_reference('[1] J. Mars, oh hello')
self.assertEqual(field, '')
self.assertEqual(pattern, '')
def test_not_recognized_unicode_2(self):
field, pattern = search_from_reference(u'Capellà Pere')
self.assertEqual(field, '')
self.assertEqual(pattern, '')
def test_not_recognized_unicode_1(self):
field, pattern = search_from_reference(u'País Valencià')
self.assertEqual(field, '')
self.assertEqual(pattern, '')
def test_journal(self):
field, pattern = search_from_reference('[1] J. Mars, oh hello, Nucl.Phys. B76 (1974) 477-482')
self.assertEqual(field, 'journal')
self.assert_('Nucl' in pattern)
self.assert_('B76' in pattern)
self.assert_('477' in pattern)
def test_report(self):
field, pattern = search_from_reference('[1] J. Mars, oh hello, [hep-ph/0104088]')
self.assertEqual(field, 'report')
self.assertEqual(pattern, 'hep-ph/0104088')
def test_not_recognized(self):
field, pattern = search_from_reference('[1] J. Mars, oh hello')
self.assertEqual(field, '')
self.assertEqual(pattern, '')