tchewik/isanlp_corefhd repository overview

from isanlp import PipelineCommon
from isanlp.processor_remote import ProcessorRemote
from isanlp.processor_razdel import ProcessorRazdel
from processor_corefhd import ProcessorCorefHD

syntax_address = ('localhost', 3334)   # docker run --rm -d -p 3334:3333 --name spacy_ru tchewik/isanlp_spacy:ru
coref_address = ('localhost', 3336)  # docker run --rm -d -p 3336:3333 --name coref_ru tchewik/isanlp_corefhd:base

ppl_base = PipelineCommon([
    (ProcessorRazdel(), ['text'],
     {'tokens': 'tokens',
      'sentences': 'sentences'}),
    (ProcessorRemote(syntax_address[0], syntax_address[1], '0'),
     ['tokens', 'sentences'],
     {'lemma': 'lemma',
      'postag': 'postag',
      'morph': 'morph',
      'syntax_dep_tree': 'syntax_dep_tree',
      'entities': 'entities'}),
    (ProcessorRemote(coref_address[0], coref_address[1], '0'),
     ['text', 'tokens', 'sentences',
      'lemma', 'postag', 'syntax_dep_tree', 'entities'],
     {0: 'entity_clusters'})
])

from isanlp import PipelineCommon
from isanlp.processor_remote import ProcessorRemote
from isanlp.processor_razdel import ProcessorRazdel
from processor_corefhd import ProcessorCorefHD

syntax_address = ('localhost', 3334)   # docker run --rm -d -p 3334:3333 --name spacy_ru tchewik/isanlp_spacy:ru
rst_address = ('localhost', 3335)   # docker run --rm -d -p 3335:3333 --name rst_ru tchewik/isanlp_rst:2.1-rstreebank
coref_address = ('localhost', 3336)  # docker run --rm -d -p 3336:3333 --name coref_ru tchewik/isanlp_corefhd:rh

ppl_rst = PipelineCommon([
    (ProcessorRazdel(), ['text'],
     {'tokens': 'tokens',
      'sentences': 'sentences'}),
    (ProcessorRemote(syntax_address[0], syntax_address[1], '0'),
     ['tokens', 'sentences'],
     {'lemma': 'lemma',
      'postag': 'postag',
      'morph': 'morph',
      'syntax_dep_tree': 'syntax_dep_tree',
      'entities': 'entities'}),
    (ProcessorRemote(rst_parser_address[0], rst_parser_address[1], 'default'),
     ['text', 'tokens', 'sentences', 'postag', 'morph', 'lemma', 'syntax_dep_tree'],
     {'rst': 'rst'}),
    (ProcessorRemote(coref_address[0], coref_address[1], '0'),
     ['text', 'tokens', 'sentences',
      'lemma', 'postag', 'syntax_dep_tree', 'entities', 'rst'],
     {'entity_clusters': 'entity_clusters'})
])