Fixed Xapian search: issue #19 #34

Open
rskthomas wants to merge 3 commits from issue_19 into main
4 changed files with 27 additions and 10 deletions

View File

@ -70,14 +70,8 @@ class SearchView(InventaryMixin):
return self.search_hids(query, offset, limit)
devices = []
dev_id = []
for x in matches:
# devices.append(self.get_annotations(x))
dev = self.get_annotations(x)
if dev.id not in dev_id:
devices.append(dev)
dev_id.append(dev.id)
devices.append(self.get_annotations(x))
count = matches.size()
# TODO fix of pagination, the count is not correct

View File

@ -49,8 +49,10 @@ class Build:
self.create_annotations()
def index(self):
timestamp = self.json['timestamp']
snap = json.dumps(self.json)
index(self.user.institution, self.uuid, snap)
index(self.user.institution, self.get_hid(self.json) , self.uuid, timestamp, snap)
def generate_chids(self):
self.algorithms = {

View File

@ -1,4 +1,5 @@
import xapian
from datetime import datetime
# database = xapian.WritableDatabase("db", xapian.DB_CREATE_OR_OPEN)
@ -28,11 +29,17 @@ def search(institution, qs, offset=0, limit=10):
)
enquire = xapian.Enquire(database)
enquire.set_query(final_query)
enquire.set_sort_by_value_then_relevance(0, True)
#colapse key is device_id
enquire.set_collapse_key(1)
matches = enquire.get_mset(offset, limit)
return matches
def index(institution, uuid, snap):
def index(institution, device_id, uuid, timestamp, snap):
uuid = 'uuid:"{}"'.format(uuid)
matches = search(institution, uuid, limit=1)
if matches and matches.size() > 0:
@ -50,6 +57,18 @@ def index(institution, uuid, snap):
indexer.index_text(snap)
indexer.index_text(uuid, 10, "uuid")
# indexer.index_text(snap, 1, "snapshot")
# store device_id, uuid and timestamp
doc.add_value(1, device_id)
doc.add_value(2, str(uuid))
try:
timestamp_dt = datetime.strptime(timestamp, '%Y-%m-%d %H:%M:%S.%f')
timestamp_unix = timestamp_dt.timestamp()
doc.add_value(0, xapian.sortable_serialise(timestamp_unix))
except ValueError as e:
print(f"Error parsing timestamp: {e}")
institution_term = "U{}".format(institution.id)
doc.add_term(institution_term)

View File

@ -101,6 +101,8 @@ def create_index(doc, user):
if not doc or not doc.get('uuid'):
return []
_timestamp = doc['endTime']
_uuid = doc['uuid']
_device_id = doc['CUSTOMER_ID']
ev = json.dumps(doc)
index(user.institution, _uuid, ev)
index(user.institution, _device_id, _uuid, _timestamp, ev)