mirror of
https://gitlab.ub.uni-bielefeld.de/sfb1288inf/nlp.git
synced 2024-09-21 11:42:52 +00:00
127 lines
4.0 KiB
Python
127 lines
4.0 KiB
Python
|
'''
|
||
|
'generator': {
|
||
|
'name': 'nopaque NLP service',
|
||
|
'version': '1.0.0',
|
||
|
'arguments': {
|
||
|
'check_encoding': args.check_encoding,
|
||
|
'language': args.language
|
||
|
}
|
||
|
},
|
||
|
'file': {
|
||
|
'encoding': encoding,
|
||
|
'md5': text_md5.hexdigest(),
|
||
|
'name': os.path.basename(args.input)
|
||
|
}
|
||
|
'''
|
||
|
|
||
|
|
||
|
class StandOffData:
|
||
|
def __init__(self, attrs):
|
||
|
self.tags = {tag_definition.id: tag_definition for tag_definition in
|
||
|
[TagDefinition(x) for x in attrs.get('tags', [])]}
|
||
|
self.annotations = [TagAnnotation(x, self.tags) for x in
|
||
|
attrs.get('annotations', [])]
|
||
|
|
||
|
|
||
|
class TagAnnotation:
|
||
|
def __init__(self, attrs, tag_lookup):
|
||
|
self.tag_id = attrs['tag_id']
|
||
|
self.tag_lookup = tag_lookup
|
||
|
if self.tag_id not in self.tag_lookup:
|
||
|
raise Exception('Unknown tag id: {}'.format(self.tag_id))
|
||
|
self.start = attrs['start']
|
||
|
self.end = attrs['end']
|
||
|
if self.start >= self.end:
|
||
|
raise Exception('start must be lower then end')
|
||
|
self.description = attrs.get('description', '')
|
||
|
self.properties = [
|
||
|
PropertyAnnotation(x, self.tag_lookup[self.tag_id].properties)
|
||
|
for x in attrs.get('properties', [])
|
||
|
]
|
||
|
for required_property_id in self.tag_lookup[self.tag_id].required_properties:
|
||
|
if required_property_id not in self.properties:
|
||
|
raise Exception('Missing required property: {}'.format(required_property_id))
|
||
|
|
||
|
@property
|
||
|
def name(self):
|
||
|
return self.tag_lookup[self.tag_id].name
|
||
|
|
||
|
def __lt__(self, other):
|
||
|
if self.start == other.start:
|
||
|
return self.name == 'token' and other.name != 'token'
|
||
|
else:
|
||
|
return self.start < other.start
|
||
|
|
||
|
def __le__(self, other):
|
||
|
if self.start == other.start:
|
||
|
return self.name == 'token' or other.name != 'token'
|
||
|
else:
|
||
|
return self.start < other.start
|
||
|
|
||
|
def __eq__(self, other):
|
||
|
return self.start == other.start and self.name == other.name
|
||
|
|
||
|
def __ne__(self, other):
|
||
|
return self.start != other.start and self.name != other.name
|
||
|
|
||
|
def __gt__(self, other):
|
||
|
if self.start == other.start:
|
||
|
return self.name != 'token' and other.name == 'token'
|
||
|
else:
|
||
|
return self.start > other.start
|
||
|
|
||
|
def __ge__(self, other):
|
||
|
if self.start == other.start:
|
||
|
return self.name != 'token' or other.name == 'token'
|
||
|
else:
|
||
|
return self.start > other.start
|
||
|
|
||
|
|
||
|
class PropertyAnnotation:
|
||
|
def __init__(self, attrs, property_lookup):
|
||
|
self.property_id = property['property_id']
|
||
|
self.property_lookup = property_lookup
|
||
|
if self.property_id not in self.property_lookup:
|
||
|
raise Exception('Unknown property id: {}'.format(self.property_id))
|
||
|
self.value = property['value']
|
||
|
# TODO: Process attrs['possibleValues'] as self.labels (no id?)
|
||
|
|
||
|
@property
|
||
|
def name(self):
|
||
|
return self.property_lookup[self.property_id].name
|
||
|
|
||
|
|
||
|
class TagDefinition:
|
||
|
def __init__(self, attrs):
|
||
|
self.id = attrs['id']
|
||
|
self.name = attrs['name']
|
||
|
self.description = attrs.get('description', '')
|
||
|
self.properties = {
|
||
|
property_definition.id: property_definition
|
||
|
for property_definition in [
|
||
|
PropertyDefinition(x) for x in attrs.get('properties', [])
|
||
|
]
|
||
|
}
|
||
|
|
||
|
@property
|
||
|
def required_properties(self):
|
||
|
return {property.id: property for property in self.properties
|
||
|
if property.is_required}
|
||
|
|
||
|
|
||
|
class PropertyDefinition:
|
||
|
def __init__(self, attrs):
|
||
|
self.id = attrs['id']
|
||
|
self.name = attrs['name']
|
||
|
self.description = attrs.get('description', '')
|
||
|
self.flags = attrs.get('flags', [])
|
||
|
self.labels = attrs.get('labels', [])
|
||
|
|
||
|
@property
|
||
|
def is_required(self):
|
||
|
return 'required' in self.flags
|
||
|
|
||
|
@property
|
||
|
def has_multiple_values(self):
|
||
|
return 'multiple' in self.flags
|