def command_unit(self):
# from digipal_text.models import TextUnit
# rs = TextUnit.objects
from digipal_text.models import TextContentXML
from digipal_text.views.viewer import get_fragment_extent, get_all_units
rid = self.get_arg(1)
fitler = {}
if rid:
fitler = {'id': rid}
ctx = TextContentXML.objects.filter(**fitler).first()
cnt = 0
if ctx:
print ctx
location_type = self.get_arg(2, 'locus')
location = self.get_arg(3, None)
units = get_all_units(ctx.content, location_type)
for unit in units:
if location is None or dputils.is_unit_in_range(unit['unitid'], location):
cnt += 1
print '%-10s %-5s %-10s' % (unit['unitid'], len(unit['content']), repr(unit['content'][:10]))
if location:
print repr(unit['content'])
print '%s units' % cnt
def command_unit(self):
# from digipal_text.models import TextUnit
# rs = TextUnit.objects
from digipal_text.models import TextContentXML
from digipal_text.views.viewer import get_fragment_extent, get_all_units
rid = self.get_arg(1)
fitler = {}
if rid:
fitler = {'id': rid}
ctx = TextContentXML.objects.filter(**fitler).first()
cnt = 0
if ctx:
print ctx
location_type = self.get_arg(2, 'locus')
location = self.get_arg(3, None)
units = get_all_units(ctx.content, location_type)
for unit in units:
if location is None or dputils.is_unit_in_range(
unit['unitid'], location):
cnt += 1
print '%-10s %-5s %-10s' % (unit['unitid'],
len(unit['content']),
repr(unit['content'][:10]))
if location:
print repr(unit['content'])
print '%s units' % cnt
def segment_units(self):
self.segunits = []
self.init_variant_patterns()
t0 = datetime.now()
# TODO: derive the info from the faceted_search settings.py or from a new
# settings variable.
# arguments
args = self.options
# unpack hilite
args['hilite'] = args.get('hilite', '').split(',')
args['hilite_groups'] = [self.get_group_key_from_pattern_key(self.get_pattern_from_id(pid)['key']) for pid in args['hilite'] if pid and self.get_pattern_from_id(pid)]
args['ignore'] = args.get('ignore', '')
args['exclude'] = args.get('exclude', '')
args['ulimit'] = dputils.get_int(args.get('ulimit', 10), 10)
args['urange'] = args.get('urange', '')
# Get the text units
hand_filters.chrono('units:')
self.stats = stats = {'duration_segmentation': 0, 'range_size': 0, 'patterns': {}, 'groups': {}}
for pattern in self.get_patterns():
group = re.sub(ur'-\d+$', '', pattern['key'])
self.stats['groups'][group] = 0
for unit in self.get_unit_model().objects.filter(content_xml__id=4).iterator():
# only fief
types = unit.get_entry_type()
if not types or 'F' not in types: continue
# only selected range
if not dputils.is_unit_in_range(unit.unitid, args['urange']): continue
stats['range_size'] += 1
# segment the unit
self.segment_unit(unit, args)
if unit.match_conditions:
self.segunits.append(unit)
hand_filters.chrono(':units')
self.variants = [{'text': variant, 'hits': self.variants[variant]} for variant in sorted(self.variants.keys())]
# stats
stats['result_size'] = len(self.segunits)
stats['result_size_pc'] = int(100.0 * stats['result_size'] / stats['range_size']) if stats['range_size'] else 'N/A'
# limit size of returned result
if args['ulimit'] > 0:
self.segunits = self.segunits[0:args['ulimit']]
stats['duration_segmentation'] = (datetime.now() - t0).total_seconds()
def segment_units(self):
self.segunits = []
self.init_variant_patterns()
t0 = datetime.now()
# TODO: derive the info from the faceted_search settings.py or from a new
# settings variable.
# arguments
args = self.options
# unpack hilite
args['hilite'] = args.get('hilite', '').split(',')
args['hilite_groups'] = [
self.get_group_key_from_pattern_key(
self.get_pattern_from_id(pid)['key']) for pid in args['hilite']
if pid and self.get_pattern_from_id(pid)
]
args['ignore'] = args.get('ignore', '')
args['exclude'] = args.get('exclude', '')
args['ulimit'] = dputils.get_int(args.get('ulimit', 10), 10)
args['urange'] = args.get('urange', '')
# Get the text units
hand_filters.chrono('units:')
self.stats = stats = {
'duration_segmentation': 0,
'range_size': 0,
'patterns': {},
'groups': {}
}
for pattern in self.get_patterns():
group = re.sub(ur'-\d+$', '', pattern['key'])
self.stats['groups'][group] = 0
for unit in self.get_unit_model().objects.filter(
content_xml__id=4).iterator():
# only fief
types = unit.get_entry_type()
if not types or 'F' not in types:
continue
# only selected range
if not dputils.is_unit_in_range(unit.unitid, args['urange']):
continue
stats['range_size'] += 1
# segment the unit
self.segment_unit(unit, args)
if unit.match_conditions:
self.segunits.append(unit)
hand_filters.chrono(':units')
self.variants = [{
'text': variant,
'hits': self.variants[variant]
} for variant in sorted(self.variants.keys())]
# stats
stats['result_size'] = len(self.segunits)
stats['result_size_pc'] = int(
100.0 * stats['result_size'] /
stats['range_size']) if stats['range_size'] else 'N/A'
# limit size of returned result
if args['ulimit'] > 0:
self.segunits = self.segunits[0:args['ulimit']]
stats['duration_segmentation'] = (datetime.now() - t0).total_seconds()
