def matchall():
""" Re-do all matching. """
from datawire.store import load_frame
from datawire.model import Frame
from datawire.processing.matching import match
for frame_ref in Frame.all():
frame = load_frame(frame_ref.urn)
if frame is None:
continue
match(frame)
def get(urn):
# TODO: authz checks.
data = load_frame(urn)
if data is None:
raise NotFound('Frame: %s' % urn)
headers = {
'X-Backend-Location': frame_url(urn),
'ETag': data['hash'],
'Cache-Control': 'public; max-age: 8460000'
}
return jsonify(data, headers=headers)
def backsearch(entity, step=5000):
# TODO: Check if the string is already tracked, use existing results.
found_count = 0
pattern = entity.pattern
q = Frame.all().order_by(Frame.submitted_at.desc())
for offset in range(0, BACKSEARCH_LIMIT, step):
log.info("Backsearch [%s] at %s (found: %s)", entity.text, offset, found_count)
if 0 == q.limit(step).offset(offset).count():
return
for frame_obj in q.limit(step).offset(offset):
frame = load_frame(frame_obj.urn)
matches = match(frame, pattern, [entity.id])
found_count += len(matches)
if len(matches):
db.session.commit()
if found_count >= BACKSEARCH_FIND:
return
def backsearch(entity, step=5000):
# TODO: Check if the string is already tracked, use existing results.
found_count = 0
pattern = entity.pattern
q = Frame.all().order_by(Frame.submitted_at.desc())
for offset in range(0, BACKSEARCH_LIMIT, step):
log.info("Backsearch [%s] at %s (found: %s)", entity.text, offset,
found_count)
if 0 == q.limit(step).offset(offset).count():
return
for frame_obj in q.limit(step).offset(offset):
frame = load_frame(frame_obj.urn)
matches = match(frame, pattern, [entity.id])
found_count += len(matches)
if len(matches):
db.session.commit()
if found_count >= BACKSEARCH_FIND:
return
def match(urn):
""" Test entity matching. """
from datawire.store import load_frame
from datawire.processing.matching import match
frame = load_frame(urn)
match(frame)
