def get_int_from_locus(self, x):
# returns integer from a locus,
# e.g. 2v => 2 * 2 + 1
# returns 0 if not a number (e.g. unumbered)
n = utils.get_int(re.sub(ur'^(\d+).*$', ur'\1', x), 0) * 2
if len(x) and 'v' in x[-1]:
n += 1
return n
def segment_units(self):
self.segunits = []
self.init_variant_patterns()
t0 = datetime.now()
# TODO: derive the info from the faceted_search settings.py or from a new
# settings variable.
# arguments
args = self.options
# unpack hilite
args['hilite'] = args.get('hilite', '').split(',')
args['hilite_groups'] = [self.get_group_key_from_pattern_key(self.get_pattern_from_id(pid)['key']) for pid in args['hilite'] if pid and self.get_pattern_from_id(pid)]
args['ignore'] = args.get('ignore', '')
args['exclude'] = args.get('exclude', '')
args['ulimit'] = dputils.get_int(args.get('ulimit', 10), 10)
args['urange'] = args.get('urange', '')
# Get the text units
hand_filters.chrono('units:')
self.stats = stats = {'duration_segmentation': 0, 'range_size': 0, 'patterns': {}, 'groups': {}}
for pattern in self.get_patterns():
group = re.sub(ur'-\d+$', '', 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 replace(match):
# !!! ASSUME pb is not in <p> or anything else
number = re.sub(ur'^.*"([^"]+)".*$', ur'\1', match.group(1))
if len(number) == len(match.group(1)):
number = self.rep_option
ret = u'<p><span data-dpt="location" data-dpt-loctype="locus">%s</span></p>' % number
self.rep_option = get_int(number, default=self.rep_option) + 1
return ret
def replace(match):
# !!! ASSUME pb is not in <p> or anything else
number = re.sub(ur'^.*"([^"]+)".*$', ur'\1', match.group(1))
if len(number) == len(match.group(1)):
number = self.rep_option
ret = u'<p><span data-dpt="location" data-dpt-loctype="locus">%s</span></p>' % number
self.rep_option = get_int(number, default=self.rep_option) + 1
return ret
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()
