mirror of
https://gitlab.ub.uni-bielefeld.de/sfb1288inf/nlp.git
synced 2024-12-26 08:14:18 +00:00
Create package for stand-off-data-py
This commit is contained in:
parent
4dea95a108
commit
a4b2fc3a65
@ -38,7 +38,7 @@ RUN apt-get install --no-install-recommends --yes \
|
||||
&& pip3 install "spacy==${SPACY_VERSION}"
|
||||
|
||||
|
||||
# Only models that include the following components are compatibel:
|
||||
# Only models that include the following components are compatibel:
|
||||
# lemmatizer, ner, parser, senter, tagger,
|
||||
ENV SPACY_MODELS="de_core_news_md,en_core_web_md,it_core_news_md,nl_core_news_md,pl_core_news_md,zh_core_web_md"
|
||||
ENV SPACY_MODELS_VERSION=3.0.0
|
||||
@ -51,6 +51,13 @@ RUN apt-get install --no-install-recommends --yes \
|
||||
zip
|
||||
|
||||
|
||||
COPY packages .
|
||||
RUN cd stand-off-data-py \
|
||||
&& python3 setup.py build \
|
||||
&& python3 setup.py install \
|
||||
&& cd -
|
||||
|
||||
|
||||
## Install Pipeline ##
|
||||
COPY nlp spacy-nlp vrt-creator /usr/local/bin/
|
||||
|
||||
|
@ -0,0 +1,14 @@
|
||||
import setuptools
|
||||
|
||||
setuptools.setup(
|
||||
name='stand-off-data',
|
||||
author='Patrick Jentsch',
|
||||
author_email='p.jentsch@uni-bielefeld.de',
|
||||
description='A python library to handle stand off data.',
|
||||
classifiers=[
|
||||
'Programming Language :: Python :: 3',
|
||||
'Operating System :: OS Independent',
|
||||
],
|
||||
packages=setuptools.find_packages(),
|
||||
python_requires='>=3.5'
|
||||
)
|
@ -0,0 +1,2 @@
|
||||
# flake8: noqa
|
||||
from .models import StandOffData
|
@ -1,33 +1,17 @@
|
||||
'''
|
||||
'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
|
||||
self.meta = attrs.get('meta', {})
|
||||
self.lookup = {tag_definition.id: tag_definition for tag_definition in
|
||||
[TagDefinition(x) for x in attrs.get('tags', [])]}
|
||||
self.annotations = [TagAnnotation(x, self.lookup) for x in
|
||||
attrs.get('annotations', [])]
|
||||
|
||||
|
||||
class TagAnnotation:
|
||||
def __init__(self, attrs, tag_lookup):
|
||||
def __init__(self, attrs, lookup):
|
||||
self.lookup = lookup
|
||||
self.tag_id = attrs['tag_id']
|
||||
self.tag_lookup = tag_lookup
|
||||
if self.tag_id not in self.tag_lookup:
|
||||
if self.tag_id not in self.lookup:
|
||||
raise Exception('Unknown tag id: {}'.format(self.tag_id))
|
||||
self.start = attrs['start']
|
||||
self.end = attrs['end']
|
||||
@ -35,16 +19,17 @@ class TagAnnotation:
|
||||
raise Exception('start must be lower then end')
|
||||
self.description = attrs.get('description', '')
|
||||
self.properties = [
|
||||
PropertyAnnotation(x, self.tag_lookup[self.tag_id].properties)
|
||||
PropertyAnnotation({**x, 'tag_id': self.tag_id}, self.lookup)
|
||||
for x in attrs.get('properties', [])
|
||||
]
|
||||
for required_property_id in self.tag_lookup[self.tag_id].required_properties:
|
||||
for required_property_id in self.lookup[self.tag_id].required_properties:
|
||||
if required_property_id not in self.properties:
|
||||
raise Exception('Missing required property: {}'.format(required_property_id))
|
||||
raise Exception(
|
||||
'Missing required property: {}'.format(required_property_id))
|
||||
|
||||
@property
|
||||
def name(self):
|
||||
return self.tag_lookup[self.tag_id].name
|
||||
return self.lookup[self.tag_id].name
|
||||
|
||||
def __lt__(self, other):
|
||||
if self.start == other.start:
|
||||
@ -78,17 +63,18 @@ class TagAnnotation:
|
||||
|
||||
|
||||
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:
|
||||
def __init__(self, attrs, lookup):
|
||||
self.lookup = lookup
|
||||
self.property_id = attrs['property_id']
|
||||
self.tag_id = attrs['tag_id']
|
||||
if self.property_id not in self.lookup[self.tag_id].properties:
|
||||
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
|
||||
return self.lookup[self.tag_id].properties[self.property_id].name
|
||||
|
||||
|
||||
class TagDefinition:
|
||||
|
@ -1,3 +1,6 @@
|
||||
from xml.sax.saxutils import escape
|
||||
|
||||
|
||||
def create_vrt(text, stand_off_data):
|
||||
# Devide annotations into CWB's verticalized text format (.vrt) logic
|
||||
p_attrs = [] # positional attributes
|
||||
@ -42,6 +45,34 @@ def create_vrt(text, stand_off_data):
|
||||
if s_attr_end_buffer[s_attr.end]:
|
||||
s_attr_end_buffer[s_attr.end].append(i)
|
||||
else:
|
||||
s_attr_end_buffer[s_attr.end] = [1]
|
||||
s_attr_end_buffer[s_attr.end] = [i]
|
||||
vrt = ''
|
||||
# TODO do the work!
|
||||
vrt += '<text>\n'
|
||||
for p_attr in p_attrs:
|
||||
# s_attr_starts
|
||||
for k in {k: v for k, v in s_attr_start_buffer.items() if k <= p_attr.start}: # noqa
|
||||
s_attrs = s_attr_start_buffer.pop(k)
|
||||
for s_attr in s_attrs:
|
||||
foo = ''
|
||||
for property in s_attr.properties:
|
||||
foo += ' {}="{}"'.format(escape(property.name),
|
||||
escape(property.value))
|
||||
vrt += '<{}{}>\n'.format(escape(s_attr.name), foo)
|
||||
for k in {k: v for k, v in s_attr_end_buffer.items() if k <= p_attr.start}: # noqa
|
||||
s_attrs = s_attr_end_buffer.pop(k)
|
||||
for s_attr in s_attrs:
|
||||
vrt += '</{}>\n'.format(escape(s_attr.name))
|
||||
# s_attr_ends
|
||||
foo = {'lemma': None, 'ner': None, 'pos': None, 'simple_pos': None, 'word': None} # noqa
|
||||
for property in p_attrs.properties:
|
||||
if property.name == 'lemma':
|
||||
foo['lemma'] = escape(property.value)
|
||||
elif property.name == 'ner':
|
||||
foo['ner'] = escape(property.value)
|
||||
elif property.name == 'pos':
|
||||
foo['pos'] = escape(property.value)
|
||||
elif property.name == 'simple_pos':
|
||||
foo['simple_pos'] = escape(property.value)
|
||||
foo['word'] = escape(text[p_attr.start:p_attr.end])
|
||||
vrt += '{word}\t{pos}\t{lemma}\t{simple_pos}\t{ner}\n'.format(**foo)
|
||||
vrt += '</text>\n'
|
||||
|
Loading…
Reference in New Issue
Block a user