def __init__(self, dascore):
if not dascore:
raise Exception("dascore needed")
self.kws = KeywordSearch(dascore)
config = das_readconfig()
self.colored_scorebar = \
config['keyword_search'].get('colored_scorebar', False)
def __init__(self, dascore):
if not dascore:
raise Exception("dascore needed")
self.kws = KeywordSearch(dascore)
config = das_readconfig()
self.colored_scorebar = \
config['keyword_search'].get('colored_scorebar', False)
def __init__(cls, name, bases, d):
type.__init__(cls, name, bases, d)
# set up only once
if hasattr(cls, 'global_dbs_inst') and cls.global_dbs_inst:
return
print('setUp in metaclass: getting dbs manager ' \
'(and fetching datasets if needed)')
cls.global_dbs_mngr = initialize_global_dbs_mngr(update_required=False)
cls.global_dbs_inst = get_global_dbs_inst()
cls.kws = KeywordSearch(dascore=DASCore(multitask=False))
dasconfig = das_readconfig()
cls.timeout = dasconfig['keyword_search']['timeout']
class KeywordSearchHandler(object):
"""
handles the keyword search and presentation of its results
"""
def __init__(self, dascore):
if not dascore:
raise Exception("dascore needed")
self.kws = KeywordSearch(dascore)
config = das_readconfig()
self.colored_scorebar = \
config['keyword_search'].get('colored_scorebar', False)
@classmethod
def _get_link_to_query(cls, query, kw_query=''):
"""
returns a link to given query taking into account other parameters
already present in cherry.request (e.g. instance, page size, etc)
"""
return url_extend_params(url='/das/request',
input=query,
kwquery=kw_query)
def _prepare_score_bar(self, q):
"""
prepares the score bar for each query (max_w & w is in pixels)
"""
max_w = 50
min_w = 3
score = max(min(q['score'], 1.0), 0.0)
w = math.floor(
(max_w - min_w) * q['scorebar_normalized_score']) + min_w
color_class = (score < 0.35) and 'low' or \
(score < 0.60) and 'avg' or 'high'
return {
'max_w': max_w,
'w': w,
'style': self.colored_scorebar and color_class or 'no-color',
'score': q['score']
}
@classmethod
def _prepare_trace(cls, q):
"""
used for debugging, is passed as escaped html string
to be later used by javascript (implementation doesn't matter much)
"""
t = list(q['trace'])
#t.sort()
thtml = \
'<ul style="width: 500px;background-color: #fff;"><li>' + \
'</li><li>'.join(str(item) for item in t) + \
'</li></ul>' + \
'score: {0:.2f}'.format(q['score'])
return cgi.escape(thtml, quote=True)
@classmethod
def _get_top_entities(cls, proposed_queries, top_k=5):
""" returns k top-scoring entities """
best_scores = {}
for r in proposed_queries:
best_scores[r['entity']] = max(r['score'],
best_scores.get(r['entity']))
best_scores = sorted(best_scores.items(),
key=lambda g_score: g_score[1],
reverse=True)
hi_score_result_types = [g for (g, _) in best_scores[:top_k]]
if len(hi_score_result_types) < 2:
return []
hi_score_result_types.append('any')
return hi_score_result_types
def handle_search(self,
webm,
query,
dbs_inst,
is_ajax=False,
timeout=5,
show_score=False):
"""
performs the search, and renders the search results
"""
err, proposed_queries = self.kws.search(query,
dbs_inst=dbs_inst,
timeout=timeout)
# get top 5 entity types
hi_score_result_types = self._get_top_entities(proposed_queries) \
if len(proposed_queries) > 6 else False
# process the results
for q in proposed_queries:
q['link'] = self._get_link_to_query(q['result'], kw_query=query)
q['trace'] = self._prepare_trace(q)
q['bar'] = self._prepare_score_bar(q)
# TODO: add user info to logs if available
# (e.g. certificate auth, to filter out queries submitted by developers)
return webm.templatepage('kwdsearch_results',
is_ajax=is_ajax,
proposed_queries=proposed_queries,
hi_score_result_types=hi_score_result_types,
err=err,
show_score=show_score)
class KeywordSearchHandler(object):
"""
handles the keyword search and presentation of its results
"""
def __init__(self, dascore):
if not dascore:
raise Exception("dascore needed")
self.kws = KeywordSearch(dascore)
config = das_readconfig()
self.colored_scorebar = \
config['keyword_search'].get('colored_scorebar', False)
@classmethod
def _get_link_to_query(cls, query, kw_query=''):
"""
returns a link to given query taking into account other parameters
already present in cherry.request (e.g. instance, page size, etc)
"""
params = cherrypy.request.params.copy()
params['input'] = query
# will see in logs which suggestion was selected by user for kw_query
if kw_query:
params['kwquery'] = kw_query
das_url = '/das/request?' + urllib.urlencode(params)
return das_url
def _prepare_score_bar(self, q):
"""
prepares the score bar for each query (max_w & w is in pixels)
"""
max_w = 50
min_w = 3
score = max(min(q['score'], 1.0), 0.0)
w = math.floor((max_w - min_w) * q['scorebar_normalized_score']) + min_w
color_class = (score < 0.35) and 'low' or \
(score < 0.60) and 'avg' or 'high'
return {'max_w': max_w,
'w': w,
'style': self.colored_scorebar and color_class or 'no-color',
'score': q['score']}
@classmethod
def _prepare_trace(cls, q):
"""
used for debugging, is passed as escaped html string
to be later used by javascript (implementation doesn't matter much)
"""
t = list(q['trace'])
#t.sort()
thtml = \
'<ul style="width: 500px;background-color: #fff;"><li>' + \
'</li><li>'.join(str(item) for item in t) + \
'</li></ul>' + \
'score: {0:.2f}'.format(q['score'])
return cgi.escape(thtml, quote=True)
@classmethod
def _get_top_entities(cls, proposed_queries, top_k=5):
""" returns k top-scoring entities """
best_scores = {}
for r in proposed_queries:
best_scores[r['entity']] = max(r['score'],
best_scores.get(r['entity']))
best_scores = sorted(best_scores.items(),
key=lambda (g, score): score, reverse=True)
hi_score_result_types = [g for (g, score) in best_scores[:top_k]]
if len(hi_score_result_types) < 2:
return []
hi_score_result_types.append('any')
return hi_score_result_types
def handle_search(self, webm, query, dbs_inst, is_ajax=False, timeout=5,
show_score=False):
"""
performs the search, and renders the search results
"""
err, proposed_queries = self.kws.search(query, dbs_inst=dbs_inst,
timeout=timeout)
# get top 5 entity types
hi_score_result_types = self._get_top_entities(proposed_queries) \
if len(proposed_queries) > 6 else False
# process the results
for q in proposed_queries:
q['link'] = self._get_link_to_query(q['result'], kw_query=query)
q['trace'] = self._prepare_trace(q)
q['bar'] = self._prepare_score_bar(q)
# TODO: add user info to logs if available
# (e.g. certificate auth, to filter out queries submitted by developers)
return webm.templatepage('kwdsearch_results',
is_ajax=is_ajax,
proposed_queries = proposed_queries,
hi_score_result_types = hi_score_result_types,
err=err,
show_score=show_score)
