1 |
38675
|
sandro.lab
|
import urllib2
|
2 |
|
|
from flask import json
|
3 |
|
|
from DLIObject import DLIObject
|
4 |
|
|
import pysolr
|
5 |
|
|
|
6 |
|
|
|
7 |
|
|
class QueryResolver(object):
|
8 |
|
|
def __init__(self, solr_url):
|
9 |
|
|
self.solr_url = solr_url
|
10 |
|
|
self.solr_server = pysolr.Solr(self.solr_url)
|
11 |
|
|
self.facet_renaming = {'entitytype': "Typology", 'provenance': "Content Provider",
|
12 |
|
|
"Content Provider": 'provenance', "Typology": 'entitytype'}
|
13 |
|
|
|
14 |
|
|
def get_facet(self, facet, query=None):
|
15 |
|
|
data = []
|
16 |
|
|
|
17 |
|
|
f = 'entitytype'
|
18 |
|
|
if query is not None:
|
19 |
|
|
data.append({'facet': self.facet_renaming[f], 'query_name':f, 'values': [
|
20 |
|
|
{'key': facet[f][i], 'value': facet[f][i + 1]} for i in range(0, len(facet[f]), 2)]})
|
21 |
|
|
else:
|
22 |
|
|
data.append({'facet': self.facet_renaming[f], 'query_name':f,
|
23 |
|
|
'values': [{'key': facet[f][i], 'value': facet[f][i + 1]} for i in
|
24 |
|
|
range(0, len(facet[f]), 2)]})
|
25 |
|
|
f = 'provenance'
|
26 |
|
|
data.append({'facet': self.facet_renaming[f], 'query_name':f,
|
27 |
|
|
'values': [{'key': facet[f][i], 'value': facet[f][i + 1]} for i in range(0, len(facet[f]), 2)]})
|
28 |
|
|
|
29 |
|
|
return data
|
30 |
|
|
|
31 |
|
|
|
32 |
|
|
"""
|
33 |
|
|
def get_pointing_to(self, query, pointing_type):
|
34 |
|
|
|
35 |
|
|
result = self.solr_server.search(q=query, rows=10, **{
|
36 |
|
|
'facet': 'true',
|
37 |
|
|
'facet.field': ['relatedobjtype'],
|
38 |
|
|
'fq': 'entitytype:"%s"' % pointing_type
|
39 |
|
|
})
|
40 |
|
|
fields = result.facets['facet_fields']
|
41 |
|
|
return [{'key': fields['relatedobjtype'][i], 'value': fields['relatedobjtype'][i + 1]} for i in
|
42 |
|
|
range(0, len(fields['relatedobjtype']), 2)]
|
43 |
|
|
"""
|
44 |
|
|
|
45 |
|
|
def get_item(self, persistent_identifier):
|
46 |
|
|
query = 'objidentifier:"%s"' % persistent_identifier
|
47 |
|
|
response = self.solr_server.search(q=query)
|
48 |
|
|
query_response = []
|
49 |
|
|
for hit in response:
|
50 |
|
|
result = hit['__result'][0]
|
51 |
|
|
query_response.append(DLIObject(result.encode('utf8')))
|
52 |
|
|
|
53 |
|
|
return {'objects': query_response}
|
54 |
|
|
|
55 |
|
|
def get_stats_detail(self):
|
56 |
|
|
|
57 |
|
|
data =urllib2.urlopen("http://node0.d.dli.research-infrastructures.eu:8080/dli/mvc/ui/stats/getLastStats")
|
58 |
|
|
return json.loads(data.read())
|
59 |
|
|
|
60 |
|
|
"""
|
61 |
|
|
|
62 |
|
|
response =self.solr_server.search(q='*:*', rows=0, **{
|
63 |
|
|
'facet':'true',
|
64 |
|
|
'facet.field':['provenance', 'relationprovenance']
|
65 |
|
|
})
|
66 |
|
|
res = response.facets['facet_fields']
|
67 |
|
|
data= res['provenance']
|
68 |
|
|
repo_name = [data[i] for i in range(len(data)) if (i % 2) == 0]
|
69 |
|
|
stat_dict = {}
|
70 |
|
|
total_item = [(data[i], data[i+1]) for i in range(len(data)) if (i % 2) == 0]
|
71 |
|
|
for repo in repo_name:
|
72 |
|
|
stat_dict[repo] ={'totalObject': 0, 'totalPublication': 0, 'totalDatasets': 0, 'totalRelations': 0}
|
73 |
|
|
for item in total_item:
|
74 |
|
|
stat_dict[item[0]]['totalObject'] =item[1]
|
75 |
|
|
data= res['relationprovenance']
|
76 |
|
|
total_rels = [(data[i], data[i+1]) for i in range(len(data)) if (i % 2) == 0]
|
77 |
|
|
for item in total_rels:
|
78 |
|
|
if item[0] not in stat_dict:
|
79 |
|
|
stat_dict[item[0]] ={'totalObject': 0, 'totalPublication': 0, 'totalDatasets': 0, 'totalRelations': 0}
|
80 |
|
|
stat_dict[item[0]]['totalRelations'] =item[1]
|
81 |
|
|
response =self.solr_server.search(q='*:*', rows=0, **{
|
82 |
|
|
'facet':'true',
|
83 |
|
|
'facet.field':['provenance'],
|
84 |
|
|
'fq': ['entitytype:"publication"']
|
85 |
|
|
})
|
86 |
|
|
res = response.facets['facet_fields']
|
87 |
|
|
data= res['provenance']
|
88 |
|
|
total_pubs = [(data[i], data[i+1]) for i in range(len(data)) if (i % 2) == 0]
|
89 |
|
|
for item in total_pubs:
|
90 |
|
|
if item[0] not in stat_dict:
|
91 |
|
|
stat_dict[item[0]] = {'totalObject': 0, 'totalPublication': 0, 'totalDatasets': 0, 'totalRelations': 0}
|
92 |
|
|
stat_dict[item[0]]['totalPublication'] =item[1]
|
93 |
|
|
|
94 |
|
|
response =self.solr_server.search(q='*:*', rows=0, **{
|
95 |
|
|
'facet':'true',
|
96 |
|
|
'facet.field':['provenance'],
|
97 |
|
|
'fq': ['entitytype:"dataset"'] })
|
98 |
|
|
res = response.facets['facet_fields']
|
99 |
|
|
data= res['provenance']
|
100 |
|
|
total_datasets = [(data[i], data[i+1]) for i in range(len(data)) if (i % 2) == 0]
|
101 |
|
|
for item in total_datasets:
|
102 |
|
|
if item[0] not in stat_dict:
|
103 |
|
|
stat_dict[item[0]] ={}
|
104 |
|
|
stat_dict[item[0]]['totalDatasets'] =item[1]
|
105 |
|
|
|
106 |
|
|
stats =[]
|
107 |
|
|
|
108 |
|
|
for item in stat_dict:
|
109 |
|
|
data = stat_dict[item]
|
110 |
|
|
data['name'] = item
|
111 |
|
|
stats.append(data)
|
112 |
|
|
|
113 |
|
|
return stats
|
114 |
|
|
"""
|
115 |
|
|
|
116 |
|
|
|
117 |
|
|
def get_stats(self):
|
118 |
|
|
|
119 |
|
|
data =urllib2.urlopen("http://node0.d.dli.research-infrastructures.eu:8080/dli/mvc/ui/stats/getAllStats")
|
120 |
|
|
return json.loads(data.read())
|
121 |
|
|
|
122 |
|
|
|
123 |
|
|
|
124 |
|
|
|
125 |
|
|
def query(self, q, browse_fields, start=None, page_size=10, only_result=False, filter_query=None):
|
126 |
|
|
params = {}
|
127 |
|
|
if browse_fields is not None:
|
128 |
|
|
params['facet'] = 'true',
|
129 |
|
|
params['facet.field'] = browse_fields
|
130 |
|
|
params['facet.mincount'] = 1
|
131 |
|
|
|
132 |
|
|
if filter_query is not None:
|
133 |
|
|
params['fq'] = filter_query
|
134 |
|
|
if start is not None:
|
135 |
|
|
params['start'] = start
|
136 |
|
|
response = self.solr_server.search(q, rows=page_size, **params)
|
137 |
|
|
query_response = []
|
138 |
|
|
for hit in response:
|
139 |
|
|
result = hit['__result'][0]
|
140 |
|
|
query_response.append(DLIObject(result.encode('utf8')))
|
141 |
|
|
if not only_result:
|
142 |
|
|
facet = self.get_facet(response.facets['facet_fields'], q)
|
143 |
|
|
else:
|
144 |
|
|
facet = self.get_facet(response.facets['facet_fields'])
|
145 |
|
|
return {'objects': query_response, 'facet': facet, 'total': response.hits}
|