#508 update

arekit/contrib/source/nerelbio/io_utils.py

@@ -0,0 +1,62 @@
+from os import path
+
+from arekit.common.experiment.data_type import DataType
+from arekit.contrib.source.nerel.folding.fixed import create_fixed_folding
+from arekit.contrib.source.nerel.io_utils import NerelIOUtils
+from arekit.contrib.source.nerel.utils import iter_filename_and_splittype
+
+
+class NerelBioIOUtils(NerelIOUtils):
+
+    splits = {
+        DataType.Train: "train",
+        DataType.Dev: "dev",
+        DataType.Test: "test"
+    }
+
+    @staticmethod
+    def get_archive_filepath(version):
+        return path.join(NerelBioIOUtils.get_data_root(), "nerel-bio-{}.zip".format(version))
+
+    @staticmethod
+    def get_annotation_innerpath(folding_data_type, filename):
+        assert(isinstance(filename, str))
+        return path.join(NerelBioIOUtils.splits[folding_data_type], "{}.ann".format(filename))
+
+    @staticmethod
+    def get_news_innerpath(folding_data_type, filename):
+        assert(isinstance(filename, str))
+        return path.join(NerelBioIOUtils.splits[folding_data_type], "{}.txt".format(filename))
+
+    @staticmethod
+    def map_doc_to_fold_type(version):
+
+        it = iter_filename_and_splittype(
+            filenames_it=NerelBioIOUtils.iter_filenames_from_zip(version),
+            splits=NerelBioIOUtils.splits.items())
+
+        d2f = {}
+        for filename, split_type in it:
+            d2f[filename] = split_type
+
+        return d2f
+
+    @staticmethod
+    def read_dataset_split(version, docs_limit=None):
+
+        it = iter_filename_and_splittype(
+            filenames_it=NerelBioIOUtils.iter_filenames_from_zip(version),
+            splits=NerelBioIOUtils.splits.items())
+
+        f2d = {}
+        for filename, split_type in it:
+            if split_type not in f2d:
+                f2d[split_type] = []
+            f2d[split_type].append(filename)
+
+        filenames_by_ids, data_folding = create_fixed_folding(train_filenames=f2d[DataType.Train],
+                                                              test_filenames=f2d[DataType.Test],
+                                                              dev_filenames=f2d[DataType.Dev],
+                                                              limit=docs_limit)
+
+        return filenames_by_ids, data_folding

arekit/contrib/source/nerelbio/labels.py

@@ -0,0 +1,265 @@
+from arekit.common.labels.base import Label
+
+
+class ABBREVIATION(Label):
+    pass
+
+
+class ALTERNATIVE_NAME(Label):
+    pass
+
+
+class KNOWS(Label):
+    pass
+
+
+class AGE_IS(Label):
+    pass
+
+
+class AGE_DIED_AT(Label):
+    pass
+
+
+class AWARDED_WITH(Label):
+    pass
+
+
+class PLACE_OF_BIRTH(Label):
+    pass
+
+
+class DATE_DEFUNCT_IN(Label):
+    pass
+
+
+class DATE_FOUNDED_IN(Label):
+    pass
+
+
+class DATE_OF_BIRTH(Label):
+    pass
+
+
+class DATE_OF_CREATION(Label):
+    pass
+
+
+class DATE_OF_DEATH(Label):
+    pass
+
+
+class POINT_IN_TIME(Label):
+    pass
+
+
+class PLACE_OF_DEATH(Label):
+    pass
+
+
+class FOUNDED_BY(Label):
+    pass
+
+
+class HEADQUARTERED_IN(Label):
+    pass
+
+
+class IDEOLOGY_OF(Label):
+    pass
+
+
+class SPOUSE(Label):
+    pass
+
+
+class MEMBER_OF(Label):
+    pass
+
+
+class ORGANIZES(Label):
+    pass
+
+
+class OWNER_OF(Label):
+    pass
+
+
+class PARENT_OF(Label):
+    pass
+
+
+class PARTICIPANT_IN(Label):
+    pass
+
+
+class PLACE_RESIDES_IN(Label):
+    pass
+
+
+class PRICE_OF(Label):
+    pass
+
+
+class PRODUCES(Label):
+    pass
+
+
+class RELATIVE(Label):
+    pass
+
+
+class RELIGION_OF(Label):
+    pass
+
+
+class SCHOOLS_ATTENDED(Label):
+    pass
+
+
+class SIBLING(Label):
+    pass
+
+
+class SUBEVENT_OF(Label):
+    pass
+
+
+class SUBORDINATE_OF(Label):
+    pass
+
+
+class TAKES_PLACE_IN(Label):
+    pass
+
+
+class WORKPLACE(Label):
+    pass
+
+
+class WORKS_AS(Label):
+    pass
+
+
+class CONVICTED_OF(Label):
+    pass
+
+
+class PENALIZED_AS(Label):
+    pass
+
+
+class START_TIME(Label):
+    pass
+
+
+class END_TIME(Label):
+    pass
+
+
+class EXPENDITURE(Label):
+    pass
+
+
+class AGENT(Label):
+    pass
+
+
+class INANIMATE_INVOLVED(Label):
+    pass
+
+
+class INCOME(Label):
+    pass
+
+
+class SUBCLASS_OF(Label):
+    pass
+
+
+class PART_OF(Label):
+    pass
+
+
+class LOCATED_IN(Label):
+    pass
+
+
+class TREATED_USING(Label):
+    pass
+
+
+class ORIGINS_FROM(Label):
+    pass
+
+
+class TO_DETECT_OR_STUDY(Label):
+    pass
+
+
+class AFFECTS(Label):
+    pass
+
+
+class HAS_CAUSE(Label):
+    pass
+
+
+class APPLIED_TO(Label):
+    pass
+
+
+class USED_IN(Label):
+    pass
+
+
+class ASSOCIATED_WITH(Label):
+    pass
+
+
+class HAS_ADMINISTRATION_ROUTE(Label):
+    pass
+
+
+class HAS_STRENGTH(Label):
+    pass
+
+
+class DURATION_OF(Label):
+    pass
+
+
+class VALUE_IS(Label):
+    pass
+
+
+class PHYSIOLOGY_OF(Label):
+    pass
+
+
+class PROCEDURE_PERFORMED(Label):
+    pass
+
+
+class MENTAL_PROCESS_OF(Label):
+    pass
+
+
+class MEDICAL_CONDITION(Label):
+    pass
+
+
+class DOSE_IS(Label):
+    pass
+
+
+class FINDING_OF(Label):
+    pass
+
+
+class CAUSE_OF_DEATH(Label):
+    pass
+
+
+class CONSUME(Label):
+    pass

arekit/contrib/source/nerelbio/reader.py

@@ -0,0 +1,8 @@
+from arekit.contrib.source.nerel.reader import NerelDocReader
+from arekit.contrib.source.nerelbio.io_utils import NerelBioIOUtils
+
+
+class NerelBioDocReader(NerelDocReader):
+
+    def __init__(self, version):
+        super(NerelBioDocReader, self).__init__(version=version, io_utils=NerelBioIOUtils())

arekit/contrib/source/nerelbio/versions.py

@@ -0,0 +1,8 @@
+import enum
+
+
+class NerelBioVersions(enum.Enum):
+    V1 = "v1_0"
+
+
+DEFAULT_VERSION = NerelBioVersions.V1