mirror of
https://gitlab.ub.uni-bielefeld.de/sfb1288inf/nopaque.git
synced 2024-12-24 10:34:17 +00:00
cqi package update and changes to match the new functions
This commit is contained in:
parent
90ef828e61
commit
83a607728d
@ -32,12 +32,14 @@ class CQiClient:
|
||||
|
||||
def connect(self, username='anonymous', password=''):
|
||||
status = self.api.ctrl_connect(username, password)
|
||||
self.corpora = CorpusCollection(self)
|
||||
return status
|
||||
|
||||
def disconnect(self):
|
||||
del self.corpora
|
||||
return self.api.ctrl_bye()
|
||||
|
||||
def ping(self):
|
||||
return self.api.ctrl_ping()
|
||||
|
||||
@property
|
||||
def corpora(self):
|
||||
return CorpusCollection(client=self)
|
||||
|
@ -1,139 +1,119 @@
|
||||
class AttributeCollection:
|
||||
def __init__(self, client, corpus):
|
||||
self.client = client
|
||||
self.corpus = corpus
|
||||
self.alignment = AlignmentAttributeCollection(client, corpus)
|
||||
self.positional = PositionalAttributeCollection(client, corpus)
|
||||
self.structural = StructuralAttributeCollection(client, corpus)
|
||||
from .ressource import Collection, Model
|
||||
|
||||
|
||||
class Attribute:
|
||||
class Attribute(Model):
|
||||
"""
|
||||
This is a class representing an attribute. Attributes denote the general
|
||||
category of information. A specific occurence is identified by an Id.
|
||||
|
||||
Attributes:
|
||||
client (CQiClient): A connected client pointing at the server that this
|
||||
object is on.
|
||||
corpus (Corpus): The corpus, this attribute belongs to.
|
||||
name (str): The name of the Attribute.
|
||||
size (int): The number of occurences of this attribute within the corpus.
|
||||
"""
|
||||
|
||||
def __init__(self, client, corpus, name):
|
||||
self.client = client
|
||||
self.corpus = corpus
|
||||
self.name = name
|
||||
self._name = '{}.{}'.format(corpus.name, name)
|
||||
self.size = client.api.cl_attribute_size(self._name)
|
||||
id_attribute = 'api_name'
|
||||
|
||||
@staticmethod
|
||||
def _attrs(client, corpus, name):
|
||||
api_name = '{}.{}'.format(corpus.attrs['api_name'], name)
|
||||
return {'api_name': api_name,
|
||||
'name': name,
|
||||
'size': client.api.cl_attribute_size(api_name)}
|
||||
|
||||
def drop(self):
|
||||
return self.client.api.cl_drop_attribute(self._name)
|
||||
return self.client.api.cl_drop_attribute(self.attrs['api_name'])
|
||||
|
||||
|
||||
class AlignmentAttributeCollection:
|
||||
def __init__(self, client, corpus):
|
||||
self.client = client
|
||||
class AttributeCollection(Collection):
|
||||
model = Attribute
|
||||
|
||||
def __init__(self, client=None, corpus=None):
|
||||
super(AttributeCollection, self).__init__(client=client)
|
||||
self.corpus = corpus
|
||||
|
||||
def get(self, name):
|
||||
return AlignmentAttribute(self.client, self.corpus, name)
|
||||
def get(self, attribute_name):
|
||||
return self.prepare_model(self.model._attrs(self.client, self.corpus,
|
||||
attribute_name))
|
||||
|
||||
def list(self):
|
||||
return [AlignmentAttribute(self.client, self.corpus, attr) for attr in
|
||||
self.client.api.corpus_alignment_attributes(self.corpus.name)]
|
||||
raise NotImplementedError
|
||||
|
||||
|
||||
class AlignmentAttribute(Attribute):
|
||||
def cpos_by_ids(self, id_list):
|
||||
return self.client.api.cl_alg2cpos(self._name, id_list)
|
||||
return self.client.api.cl_alg2cpos(self.attrs['api_name'], id_list)
|
||||
|
||||
def ids_by_cpos(self, cpos_list):
|
||||
return self.client.api.cl_cpos2alg(self._name, cpos_list)
|
||||
return self.client.api.cl_cpos2alg(self.attrs['api_name'], cpos_list)
|
||||
|
||||
|
||||
class PositionalAttributeCollection:
|
||||
def __init__(self, client, corpus):
|
||||
self.client = client
|
||||
self.corpus = corpus
|
||||
|
||||
def get(self, name):
|
||||
return PositionalAttribute(self.client, self.corpus, name)
|
||||
class AlignmentAttributeCollection(AttributeCollection):
|
||||
model = AlignmentAttribute
|
||||
|
||||
def list(self):
|
||||
return [PositionalAttribute(self.client, self.corpus, attr) for attr in
|
||||
self.client.api.corpus_positional_attributes(self.corpus.name)]
|
||||
return [self.prepare_model(self.model._attrs(self.client, self.corpus, attr)) # noqa
|
||||
for attr in self.client.api.corpus_alignment_attributes(self.corpus.attrs['api_name'])] # noqa
|
||||
|
||||
|
||||
class PositionalAttribute(Attribute):
|
||||
def __init__(self, client, corpus, name):
|
||||
super(PositionalAttribute, self).__init__(client, corpus, name)
|
||||
self.lexicon_size = client.api.cl_lexicon_size(self._name)
|
||||
@staticmethod
|
||||
def _attrs(client, corpus, name):
|
||||
attrs = super(PositionalAttribute, PositionalAttribute)._attrs(client, corpus, name) # noqa
|
||||
attrs['lexicon_size'] = client.api.cl_lexicon_size(attrs['api_name'])
|
||||
return attrs
|
||||
|
||||
def cpos_by_id(self, id):
|
||||
return self.client.api.cl_id2cpos(self._name, id)
|
||||
return self.client.api.cl_id2cpos(self.attrs['api_name'], id)
|
||||
|
||||
def cpos_by_ids(self, id_list):
|
||||
return self.client.api.cl_idlist2cpos(self._name, id_list)
|
||||
return self.client.api.cl_idlist2cpos(self.attrs['api_name'], id_list)
|
||||
|
||||
def freqs_by_ids(self, id_list):
|
||||
return self.client.api.cl_id2freq(self._name, id_list)
|
||||
return self.client.api.cl_id2freq(self.attrs['api_name'], id_list)
|
||||
|
||||
def ids_by_cpos(self, cpos_list):
|
||||
return self.client.api.cl_cpos2id(self._name, cpos_list)
|
||||
return self.client.api.cl_cpos2id(self.attrs['api_name'], cpos_list)
|
||||
|
||||
def ids_by_regex(self, regex):
|
||||
return self.client.api.cl_regex2id(self._name, regex)
|
||||
return self.client.api.cl_regex2id(self.attrs['api_name'], regex)
|
||||
|
||||
def ids_by_values(self, value_list):
|
||||
return self.client.api.cl_str2id(self._name, value_list)
|
||||
return self.client.api.cl_str2id(self.attrs['api_name'], value_list)
|
||||
|
||||
def values_by_cpos(self, cpos_list):
|
||||
return self.client.api.cl_cpos2str(self._name, cpos_list)
|
||||
return self.client.api.cl_cpos2str(self.attrs['api_name'], cpos_list)
|
||||
|
||||
def values_by_ids(self, id_list):
|
||||
return self.client.api.cl_id2str(self._name, id_list)
|
||||
return self.client.api.cl_id2str(self.attrs['api_name'], id_list)
|
||||
|
||||
|
||||
class StructuralAttributeCollection:
|
||||
def __init__(self, client, corpus):
|
||||
self.client = client
|
||||
self.corpus = corpus
|
||||
class PositionalAttributeCollection(AttributeCollection):
|
||||
model = PositionalAttribute
|
||||
|
||||
def get(self, name):
|
||||
return StructuralAttribute(self.client, self.corpus, name)
|
||||
|
||||
def list(self, filters={}):
|
||||
attrs = [StructuralAttribute(self.client, self.corpus, attr) for attr
|
||||
in self.client.api.corpus_structural_attributes(
|
||||
self.corpus.name)]
|
||||
for k, v in filters.items():
|
||||
if k == 'part_of':
|
||||
attrs = list(filter(lambda x: x.name.startswith(v.name + '_'),
|
||||
attrs))
|
||||
return attrs
|
||||
def list(self):
|
||||
return [self.prepare_model(self.model._attrs(self.client, self.corpus, attr)) # noqa
|
||||
for attr in self.client.api.corpus_positional_attributes(self.corpus.attrs['api_name'])] # noqa
|
||||
|
||||
|
||||
class StructuralAttribute(Attribute):
|
||||
def __init__(self, client, corpus, name):
|
||||
super(StructuralAttribute, self).__init__(client, corpus, name)
|
||||
self.has_values = \
|
||||
client.api.corpus_structural_attribute_has_values(self._name)
|
||||
@staticmethod
|
||||
def _attrs(client, corpus, name):
|
||||
attrs = super(StructuralAttribute, StructuralAttribute)._attrs(client, corpus, name) # noqa
|
||||
attrs['has_values'] = client.api.corpus_structural_attribute_has_values(attrs['api_name']) # noqa
|
||||
return attrs
|
||||
|
||||
def cpos_by_id(self, id):
|
||||
return self.client.api.cl_struc2cpos(self._name, id)
|
||||
return self.client.api.cl_struc2cpos(self.attrs['api_name'], id)
|
||||
|
||||
def ids_by_cpos(self, cpos_list):
|
||||
return self.client.api.cl_cpos2struc(self._name, cpos_list)
|
||||
return self.client.api.cl_cpos2struc(self.attrs['api_name'], cpos_list)
|
||||
|
||||
def lbound_by_cpos(self, cpos_list):
|
||||
return self.client.api.cl_cpos2lbound(self._name, cpos_list)
|
||||
return self.client.api.cl_cpos2lbound(self.attrs['api_name'],
|
||||
cpos_list)
|
||||
|
||||
def rbound_by_cpos(self, cpos_list):
|
||||
return self.client.api.cl_cpos2rbound(self._name, cpos_list)
|
||||
return self.client.api.cl_cpos2rbound(self.attrs['api_name'],
|
||||
cpos_list)
|
||||
|
||||
def values_by_ids(self, id_list):
|
||||
return self.client.api.cl_struc2str(self._name, id_list)
|
||||
return self.client.api.cl_struc2str(self.attrs['api_name'], id_list)
|
||||
|
||||
def export(self, first_cpos, last_cpos, context=0, expand_lists=False):
|
||||
first_id, last_id = self.ids_by_cpos([first_cpos, last_cpos])
|
||||
@ -150,9 +130,10 @@ class StructuralAttribute(Attribute):
|
||||
else:
|
||||
cpos_list_lbound = first_cpos
|
||||
rc_rbound = \
|
||||
self.cpos_by_id(min((last_id + context), (self.size - 1)))[1]
|
||||
self.cpos_by_id(min((last_id + context), (self.attrs['size'] - 1)))[1]
|
||||
if rc_rbound != last_cpos:
|
||||
rc_lbound = min((last_cpos + 1), (self.corpus.size - 1))
|
||||
rc_lbound = min((last_cpos + 1),
|
||||
(self.collection.corpus.attrs['size'] - 1))
|
||||
rc = (rc_lbound, rc_rbound)
|
||||
cpos_list_rbound = rc_rbound
|
||||
else:
|
||||
@ -164,5 +145,17 @@ class StructuralAttribute(Attribute):
|
||||
'rc': list(range(rc[0], (rc[1] + 1)))}
|
||||
else:
|
||||
match = {'lc': lc, 'c': c, 'rc': rc}
|
||||
lookups = self.corpus.lookups_by_cpos(cpos_list)
|
||||
lookups = self.collection.corpus.lookups_by_cpos(cpos_list)
|
||||
return {'match': match, **lookups}
|
||||
|
||||
|
||||
class StructuralAttributeCollection(AttributeCollection):
|
||||
model = StructuralAttribute
|
||||
|
||||
def list(self, filters={}):
|
||||
attrs = [self.prepare_model(self.model._attrs(self.client, self.corpus, attr)) # noqa
|
||||
for attr in self.client.api.corpus_structural_attributes(self.corpus.attrs['api_name'])] # noqa
|
||||
for k, v in filters.items():
|
||||
if k == 'part_of':
|
||||
attrs = list(filter(lambda x: x.attrs['name'].startswith(v.attrs['name'] + '_'), attrs)) # noqa
|
||||
return attrs
|
||||
|
@ -1,30 +1,24 @@
|
||||
from .attributes import AttributeCollection
|
||||
from .attributes import (AlignmentAttributeCollection,
|
||||
PositionalAttributeCollection,
|
||||
StructuralAttributeCollection)
|
||||
from .ressource import Collection, Model
|
||||
from .subcorpora import SubcorpusCollection
|
||||
|
||||
|
||||
class CorpusCollection:
|
||||
def __init__(self, client):
|
||||
self.client = client
|
||||
class Corpus(Model):
|
||||
id_attribute = 'api_name'
|
||||
|
||||
def get(self, name):
|
||||
return Corpus(self.client, name)
|
||||
|
||||
def list(self):
|
||||
return [Corpus(self.client, corpus) for corpus in
|
||||
self.client.api.corpus_list_coprora()]
|
||||
|
||||
|
||||
class Corpus:
|
||||
def __init__(self, client, name):
|
||||
self.client = client
|
||||
self.name = name
|
||||
self.size = client.api.cl_attribute_size('{}.word'.format(name))
|
||||
# self.info = client.api.corpus_info(name)
|
||||
self.charset = client.api.corpus_charset(name)
|
||||
# self.full_name = client.api.corpus_full_name(name)
|
||||
self.properties = client.api.corpus_properties(name)
|
||||
self.attributes = AttributeCollection(client, self)
|
||||
self.subcorpora = SubcorpusCollection(client, self)
|
||||
@staticmethod
|
||||
def _attrs(client, name):
|
||||
api_name = name
|
||||
return {'api_name': api_name,
|
||||
'name': name,
|
||||
'size': client.api.cl_attribute_size(
|
||||
'{}.word'.format(api_name)),
|
||||
# 'info': client.api.corpus_info(name),
|
||||
'charset': client.api.corpus_charset(api_name),
|
||||
# 'full_name' = client.api.corpus_full_name(name),
|
||||
'properties': client.api.corpus_properties(api_name)}
|
||||
|
||||
def lookups_by_cpos(self, cpos_list):
|
||||
cpos_list = list(set(cpos_list))
|
||||
@ -33,35 +27,64 @@ class Corpus:
|
||||
lookups['cpos_lookup'] = {}
|
||||
for cpos in cpos_list:
|
||||
lookups['cpos_lookup'][cpos] = {}
|
||||
for attr in self.attributes.positional.list():
|
||||
for attr in self.positional_attributes.list():
|
||||
cpos_attr_values = attr.values_by_cpos(cpos_list)
|
||||
for i, cpos in enumerate(cpos_list):
|
||||
lookups['cpos_lookup'][cpos][attr.name] = cpos_attr_values[i]
|
||||
for attr in self.attributes.structural.list():
|
||||
if attr.has_values:
|
||||
lookups['cpos_lookup'][cpos][attr.attrs['name']] = cpos_attr_values[i]
|
||||
for attr in self.structural_attributes.list():
|
||||
if attr.attrs['has_values']:
|
||||
continue
|
||||
cpos_attr_ids = attr.ids_by_cpos(cpos_list)
|
||||
for i, cpos in enumerate(cpos_list):
|
||||
if cpos_attr_ids[i] != -1:
|
||||
lookups['cpos_lookup'][cpos][attr.name] = cpos_attr_ids[i]
|
||||
lookups['cpos_lookup'][cpos][attr.attrs['name']] = cpos_attr_ids[i]
|
||||
occured_attr_ids = list(filter(lambda x: x != -1,
|
||||
set(cpos_attr_ids)))
|
||||
if not occured_attr_ids:
|
||||
continue
|
||||
subattrs = \
|
||||
self.attributes.structural.list(filters={'part_of': attr})
|
||||
self.structural_attributes.list(filters={'part_of': attr})
|
||||
if not subattrs:
|
||||
continue
|
||||
lookup_name = '{}_lookup'.format(attr.name)
|
||||
lookup_name = '{}_lookup'.format(attr.attrs['name'])
|
||||
lookups[lookup_name] = {}
|
||||
for attr_id in occured_attr_ids:
|
||||
lookups[lookup_name][attr_id] = {}
|
||||
for subattr in subattrs:
|
||||
subattr_values = subattr.values_by_ids(occured_attr_ids)
|
||||
for i, subattr_value in enumerate(subattr_values):
|
||||
lookups[lookup_name][occured_attr_ids[i]][subattr.name] = \
|
||||
subattr_name = subattr.attrs['name'][(len(attr.attrs['name']) + 1):]
|
||||
lookups[lookup_name][occured_attr_ids[i]][subattr_name] = \
|
||||
subattr_value
|
||||
return lookups
|
||||
|
||||
def query(self, query, subcorpus_name='Results'):
|
||||
return self.client.api.cqp_query(self.name, subcorpus_name, query)
|
||||
return self.client.api.cqp_query(self.attrs['api_name'],
|
||||
subcorpus_name, query)
|
||||
|
||||
@property
|
||||
def alignment_attributes(self):
|
||||
return AlignmentAttributeCollection(client=self.client, corpus=self)
|
||||
|
||||
@property
|
||||
def positional_attributes(self):
|
||||
return PositionalAttributeCollection(client=self.client, corpus=self)
|
||||
|
||||
@property
|
||||
def structural_attributes(self):
|
||||
return StructuralAttributeCollection(client=self.client, corpus=self)
|
||||
|
||||
@property
|
||||
def subcorpora(self):
|
||||
return SubcorpusCollection(client=self.client, corpus=self)
|
||||
|
||||
|
||||
class CorpusCollection(Collection):
|
||||
model = Corpus
|
||||
|
||||
def get(self, corpus_name):
|
||||
return self.prepare_model(self.model._attrs(self.client, corpus_name))
|
||||
|
||||
def list(self):
|
||||
return [self.prepare_model(self.model._attrs(self.client, corpus))
|
||||
for corpus in self.client.api.corpus_list_coprora()]
|
||||
|
78
app/corpora/cqi/models/ressource.py
Normal file
78
app/corpora/cqi/models/ressource.py
Normal file
@ -0,0 +1,78 @@
|
||||
class Model:
|
||||
"""
|
||||
A base class for representing a single object on the server.
|
||||
"""
|
||||
id_attribute = 'Id'
|
||||
|
||||
def __init__(self, attrs=None, client=None, collection=None):
|
||||
#: A client pointing at the server that this object is on.
|
||||
self.client = client
|
||||
|
||||
#: The collection that this model is part of.
|
||||
self.collection = collection
|
||||
|
||||
#: The raw representation of this object from the API
|
||||
self.attrs = attrs or {}
|
||||
|
||||
@staticmethod
|
||||
def _attrs(client, key):
|
||||
raise NotImplementedError
|
||||
|
||||
def __repr__(self):
|
||||
return "<{}: {}>".format(self.__class__.__name__, self.id)
|
||||
|
||||
def __eq__(self, other):
|
||||
return isinstance(other, self.__class__) and self.id == other.id
|
||||
|
||||
def __hash__(self):
|
||||
return hash("{}:{}".format(self.__class__.__name__, self.id))
|
||||
|
||||
@property
|
||||
def id(self):
|
||||
"""
|
||||
The ID of the object.
|
||||
"""
|
||||
return self.attrs.get(self.id_attribute)
|
||||
|
||||
def reload(self):
|
||||
"""
|
||||
Load this object from the server again and update ``attrs`` with the
|
||||
new data.
|
||||
"""
|
||||
new_model = self.collection.get(self.id)
|
||||
self.attrs = new_model.attrs
|
||||
|
||||
|
||||
class Collection:
|
||||
"""
|
||||
A base class for representing all objects of a particular type on the
|
||||
server.
|
||||
"""
|
||||
|
||||
#: The type of object this collection represents, set by subclasses
|
||||
model = None
|
||||
|
||||
def __init__(self, client=None):
|
||||
#: The client pointing at the server that this collection of objects
|
||||
#: is on.
|
||||
self.client = client
|
||||
|
||||
def list(self):
|
||||
raise NotImplementedError
|
||||
|
||||
def get(self, key):
|
||||
raise NotImplementedError
|
||||
|
||||
def prepare_model(self, attrs):
|
||||
"""
|
||||
Create a model from a set of attributes.
|
||||
"""
|
||||
if isinstance(attrs, Model):
|
||||
attrs.client = self.client
|
||||
attrs.collection = self
|
||||
return attrs
|
||||
elif isinstance(attrs, dict):
|
||||
return self.model(attrs=attrs, client=self.client, collection=self)
|
||||
else:
|
||||
raise Exception("Can't create {} from {}".format(
|
||||
self.model.__name__, attrs))
|
@ -1,54 +1,45 @@
|
||||
from .ressource import Collection, Model
|
||||
from ..api.specification import (CONST_FIELD_KEYWORD, CONST_FIELD_MATCH,
|
||||
CONST_FIELD_MATCHEND, CONST_FIELD_TARGET)
|
||||
|
||||
|
||||
class SubcorpusCollection:
|
||||
def __init__(self, client, corpus):
|
||||
self.client = client
|
||||
self.corpus = corpus
|
||||
class Subcorpus(Model):
|
||||
id_attribute = 'api_name'
|
||||
|
||||
def get(self, name):
|
||||
return Subcorpus(self.client, self.corpus, name)
|
||||
|
||||
def list(self):
|
||||
return [Subcorpus(self.client, self.corpus, subcorpus) for subcorpus in
|
||||
self.client.api.cqp_list_subcorpora(self.corpus.name)]
|
||||
|
||||
|
||||
class Subcorpus:
|
||||
def __init__(self, client, corpus, name):
|
||||
self.client = client
|
||||
self.corpus = corpus
|
||||
self.name = name
|
||||
self._name = '{}:{}'.format(corpus.name, name)
|
||||
self.fields = {}
|
||||
if client.api.cqp_subcorpus_has_field(self._name, CONST_FIELD_MATCH):
|
||||
self.fields['match'] = CONST_FIELD_MATCH
|
||||
if client.api.cqp_subcorpus_has_field(self._name,
|
||||
@staticmethod
|
||||
def _attrs(client, corpus, name):
|
||||
api_name = '{}:{}'.format(corpus.attrs['api_name'], name)
|
||||
fields = {}
|
||||
if client.api.cqp_subcorpus_has_field(api_name, CONST_FIELD_MATCH):
|
||||
fields['match'] = CONST_FIELD_MATCH
|
||||
if client.api.cqp_subcorpus_has_field(api_name,
|
||||
CONST_FIELD_MATCHEND):
|
||||
self.fields['matchend'] = CONST_FIELD_MATCHEND
|
||||
if client.api.cqp_subcorpus_has_field(self._name, CONST_FIELD_TARGET):
|
||||
self.fields['target'] = CONST_FIELD_TARGET
|
||||
if client.api.cqp_subcorpus_has_field(self._name, CONST_FIELD_KEYWORD):
|
||||
self.fields['keyword'] = CONST_FIELD_KEYWORD
|
||||
self.size = client.api.cqp_subcorpus_size(self._name)
|
||||
fields['matchend'] = CONST_FIELD_MATCHEND
|
||||
if client.api.cqp_subcorpus_has_field(api_name, CONST_FIELD_TARGET):
|
||||
fields['target'] = CONST_FIELD_TARGET
|
||||
if client.api.cqp_subcorpus_has_field(api_name, CONST_FIELD_KEYWORD):
|
||||
fields['keyword'] = CONST_FIELD_KEYWORD
|
||||
return {'api_name': api_name,
|
||||
'name': name,
|
||||
'fields': fields,
|
||||
'size': client.api.cqp_subcorpus_size(api_name)}
|
||||
|
||||
def drop(self):
|
||||
return self.client.api.cqp_drop_subcorpus(self._name)
|
||||
return self.client.api.cqp_drop_subcorpus(self.attrs['api_name'])
|
||||
|
||||
def dump(self, field, first, last):
|
||||
return self.client.api.cqp_dump_subcorpus(self._name, field, first,
|
||||
last)
|
||||
return self.client.api.cqp_dump_subcorpus(self.attrs['api_name'],
|
||||
field, first, last)
|
||||
|
||||
def export(self, context=25, cutoff=float('inf'), expand_lists=False,
|
||||
offset=0):
|
||||
if self.size == 0:
|
||||
if self.attrs['size'] == 0:
|
||||
return {"matches": []}
|
||||
first_match = max(0, offset)
|
||||
last_match = min((offset + cutoff - 1), (self.size - 1))
|
||||
last_match = min((offset + cutoff - 1), (self.attrs['size'] - 1))
|
||||
match_boundaries = \
|
||||
zip(self.dump(self.fields['match'], first_match, last_match),
|
||||
self.dump(self.fields['matchend'], first_match, last_match))
|
||||
zip(self.dump(self.attrs['fields']['match'], first_match, last_match), # noqa
|
||||
self.dump(self.attrs['fields']['matchend'], first_match, last_match)) # noqa
|
||||
cpos_list = []
|
||||
matches = []
|
||||
for match_start, match_end in match_boundaries:
|
||||
@ -64,10 +55,11 @@ class Subcorpus:
|
||||
cpos_list_lbound = lc_lbound
|
||||
else:
|
||||
cpos_list_lbound = match_start
|
||||
rc_lbound = min((match_end + 1), (self.corpus.size - 1))
|
||||
rc_lbound = min((match_end + 1),
|
||||
(self.collection.corpus.attrs['size'] - 1))
|
||||
if rc_lbound != match_end:
|
||||
rc_rbound = min((match_end + 1 + context),
|
||||
(self.corpus.size - 1))
|
||||
(self.collection.corpus.attrs['size'] - 1))
|
||||
rc = (rc_lbound, rc_rbound)
|
||||
cpos_list_rbound = rc_rbound
|
||||
else:
|
||||
@ -81,14 +73,30 @@ class Subcorpus:
|
||||
else:
|
||||
match = {'lc': lc, 'c': c, 'rc': rc}
|
||||
matches.append(match)
|
||||
lookups = self.corpus.lookups_by_cpos(cpos_list)
|
||||
lookups = self.collection.corpus.lookups_by_cpos(cpos_list)
|
||||
return {'matches': matches, **lookups}
|
||||
|
||||
def fdist_1(self, cutoff, field, attribute):
|
||||
return self.client.api.cqp_fdist_1(self._name, cutoff,
|
||||
return self.client.api.cqp_fdist_1(self.attrs['api_name'], cutoff,
|
||||
field, attribute._name)
|
||||
|
||||
def fdist_2(self, cutoff, field_1, attribute_1, field_2, attribute_2):
|
||||
return self.client.api.cqp_fdist_2(self._name, cutoff,
|
||||
return self.client.api.cqp_fdist_2(self.attrs['api_name'], cutoff,
|
||||
field_1, attribute_1._name,
|
||||
field_2, attribute_2._name)
|
||||
|
||||
|
||||
class SubcorpusCollection(Collection):
|
||||
model = Subcorpus
|
||||
|
||||
def __init__(self, client=None, corpus=None):
|
||||
super(SubcorpusCollection, self).__init__(client=client)
|
||||
self.corpus = corpus
|
||||
|
||||
def get(self, subcorpus_name):
|
||||
return self.prepare_model(self.model._attrs(self.client, self.corpus,
|
||||
subcorpus_name))
|
||||
|
||||
def list(self):
|
||||
return [self.prepare_model(self.model._attrs(self.client, self.corpus, subcorpus)) # noqa
|
||||
for subcorpus in self.client.api.cqp_list_subcorpora(self.corpus.attrs['api_name'])] # noqa
|
||||
|
@ -67,23 +67,23 @@ def corpus_analysis_query(query):
|
||||
socketio.emit('corpus_analysis_query', response, room=request.sid)
|
||||
return
|
||||
response = {'code': 200, 'desc': None, 'msg': 'OK',
|
||||
'payload': {**query_status, 'match_count': results.size}}
|
||||
'payload': {**query_status, 'match_count': results.attrs['size']}}
|
||||
socketio.emit('corpus_analysis_query', response, room=request.sid)
|
||||
chunk_size = 100
|
||||
chunk_start = 0
|
||||
context = 100
|
||||
progress = 0
|
||||
client.status = 'running'
|
||||
while chunk_start <= results.size:
|
||||
while chunk_start <= results.attrs['size']:
|
||||
if client.status == 'abort':
|
||||
break
|
||||
chunk = results.export(context=context, cutoff=chunk_size,
|
||||
expand_lists=False, offset=chunk_start)
|
||||
chunk['cpos_ranges'] = True
|
||||
if (results.size == 0):
|
||||
if (results.attrs['size'] == 0):
|
||||
progress = 100
|
||||
else:
|
||||
progress = ((chunk_start + chunk_size) / results.size) * 100
|
||||
progress = ((chunk_start + chunk_size) / results.attrs['size']) * 100
|
||||
progress = min(100, int(math.ceil(progress)))
|
||||
response = {'code': 200, 'desc': None, 'msg': 'OK',
|
||||
'payload': {'chunk': chunk, 'progress': progress}}
|
||||
@ -104,12 +104,12 @@ def corpus_analysis_inspect_match(payload):
|
||||
return
|
||||
# Get more context for given match CPOS
|
||||
corpus = client.corpora.get('CORPUS')
|
||||
s = corpus.attributes.structural.get('s')
|
||||
s = corpus.structural_attributes.get('s')
|
||||
match_context = s.export(payload['first_cpos'], payload['last_cpos'],
|
||||
context=3, expand_lists=False)
|
||||
match_context['cpos_ranges'] = True
|
||||
socketio.emit('match_context',
|
||||
{'payload': match_context}, room=request.sid)
|
||||
socketio.emit('match_context', {'payload': match_context},
|
||||
room=request.sid)
|
||||
|
||||
|
||||
def corpus_analysis_session_handler(app, corpus_id, user_id, session_id):
|
||||
|
@ -381,7 +381,7 @@ class ResultsList extends List {
|
||||
token = chunk["cpos_lookup"][cpos];
|
||||
hitCellElement.insertAdjacentHTML("beforeend", `<span class="token" data-cpos="${cpos}">${token["word"]} </span>`);
|
||||
// get text titles of every hit cpos token
|
||||
textTitles.add(chunk["text_lookup"][token["text"]]["text_title"]);
|
||||
textTitles.add(chunk["text_lookup"][token["text"]]["title"]);
|
||||
// add button to trigger more context to every match td
|
||||
var inspectBtn = document.createElement("a");
|
||||
inspectBtn.setAttribute("class", "btn-floating btn-flat waves-effect waves-light grey right inspect disabled");
|
||||
|
Loading…
Reference in New Issue
Block a user