pybis.py

#!/usr/bin/env python
# -*- coding: utf-8 -*-

"""
pybis.py

Work with openBIS from Python.

"""

from __future__ import print_function
import os
import random

import requests
from requests.packages.urllib3.exceptions import InsecureRequestWarning

requests.packages.urllib3.disable_warnings(InsecureRequestWarning)

import copy
import time
import json
import re
from urllib.parse import urlparse, urljoin, quote
import zlib
import base64
from collections import namedtuple
from texttable import Texttable
from tabulate import tabulate

from pybis.utils import parse_jackson, check_datatype, split_identifier, format_timestamp, is_identifier, is_permid, nvl
from pybis.property import PropertyHolder, PropertyAssignments
from pybis.masterdata import Vocabulary
from . import data_set as pbds

import pandas as pd
from pandas import DataFrame, Series

import threading
from threading import Thread
from queue import Queue

from datetime import datetime

PYBIS_PLUGIN = "dataset-uploader-api"


def _definitions(entity):
    entities = {
        "Space": {
            "attrs_new": "code description".split(),
            "attrs_up": "description".split(),
            "attrs": "code permId description registrator registrationDate modificationDate".split(),
            "multi": "".split(),
            "identifier": "spaceId",
        },
        "Project": {
            "attrs_new": "code description space attachments".split(),
            "attrs_up": "description space attachments".split(),
            "attrs": "code description permId identifier space leader registrator registrationDate modifier modificationDate attachments".split(),
            "multi": "".split(),
            "identifier": "projectId",
        },
        "Experiment": {
            "attrs_new": "code type project tags attachments".split(),
            "attrs_up": "project tags attachments".split(),
            "attrs": "code permId identifier type project tags attachments".split(),
            "multi": "tags attachments".split(),
            "identifier": "experimentId",
        },
        "Sample": {
            "attrs_new": "code type parents children space experiment tags attachments".split(),
            "attrs_up": "parents children space experiment tags attachments".split(),
            "attrs": "code permId identifier type parents children components space experiment tags attachments".split(),
            "ids2type": {
                'parentIds': {'permId': {'@type': 'as.dto.sample.id.SamplePermId'}},
                'childIds': {'permId': {'@type': 'as.dto.sample.id.SamplePermId'}},
                'componentIds': {'permId': {'@type': 'as.dto.sample.id.SamplePermId'}},
            },
            "identifier": "sampleId",
            "cre_type": "as.dto.sample.create.SampleCreation",
            "multi": "parents children components tags attachments".split(),
        },
        "SemanticAnnotation": {
            "attrs_new": "permId entityType propertyType predicateOntologyId predicateOntologyVersion predicateAccessionId descriptorOntologyId descriptorOntologyVersion descriptorAccessionId".split(),
            "attrs_up": "entityType propertyType predicateOntologyId predicateOntologyVersion predicateAccessionId descriptorOntologyId descriptorOntologyVersion descriptorAccessionId ".split(),
            "attrs": "permId entityType propertyType predicateOntologyId predicateOntologyVersion predicateAccessionId descriptorOntologyId descriptorOntologyVersion descriptorAccessionId creationDate".split(),
            "ids2type": {
                "propertyTypeId": { 
                    "permId": "as.dto.property.id.PropertyTypePermId"
                },
                "entityTypeId": { 
                    "permId": "as.dto.entity.id.EntityTypePermId"
                },
            },
            "identifier": "permId",
            "cre_type": "as.dto.sample.create.SampleCreation",
            "multi": "parents children components tags attachments".split(),
        },
        "DataSet": {
            "attrs_new": "type experiment sample parents children components tags".split(),
            "attrs_up": "parents children experiment sample components tags".split(),
            "attrs": "code permId type experiment sample parents children components tags accessDate dataProducer dataProductionDate registrator registrationDate modifier modificationDate dataStore measured".split(),

            "ids2type": {
                'parentIds': {'permId': {'@type': 'as.dto.dataset.id.DataSetPermId'}},
                'childIds': {'permId': {'@type': 'as.dto.dataset.id.DataSetPermId'}},
                'componentIds': {'permId': {'@type': 'as.dto.dataset.id.DataSetPermId'}},
                'containerIds': {'permId': {'@type': 'as.dto.dataset.id.DataSetPermId'}},
            },
            "multi": "parents children container".split(),
            "identifier": "dataSetId",
        },
        "Material": {
            "attrs_new": "code description type creation tags".split(),
            "attrs": "code description type creation registrator tags".split()
        },
        "Tag": {
            "attrs_new": "code description experiments samples dataSets materials".split(),
            "attrs": "code description experiments samples dataSets materials registrationDate".split(),
        },
        "Person": {
            "attrs_new": "userId space".split(),
            "attrs_up": "space".split(),
            "attrs": "permId userId firstName lastName email roleAssignments space registrationDate ".split(),
            "multi": "".split(),
            "identifier": "userId",
        },
        "AuthorizationGroup" : {
            "attrs": "code description users roleAssignments registrator registrationDate modificationDate".split(),
            "attrs_new": "code description userIds".split(),
            "multi": "users".split()

        },
        "attr2ids": {
            "space": "spaceId",
            "project": "projectId",
            "sample": "sampleId",
            "samples": "sampleIds",
            "dataSet": "dataSetId",
            "dataSets": "dataSetIds",
            "experiment": "experimentId",
            "experiments": "experimentIds",
            "material": "materialId",
            "materials": "materialIds",
            "container": "containerId",
            "component": "componentId",
            "components": "componentIds",
            "parents": "parentIds",
            "children": "childIds",
            "tags": "tagIds",
            "userId": "userId",
            "users": "userIds",
        },
        "ids2type": {
            'spaceId': {'permId': {'@type': 'as.dto.space.id.SpacePermId'}},
            'projectId': {'permId': {'@type': 'as.dto.project.id.ProjectPermId'}},
            'experimentId': {'permId': {'@type': 'as.dto.experiment.id.ExperimentPermId'}},
            'tagIds': {'code': {'@type': 'as.dto.tag.id.TagCode'}},
        },
    }
    return entities[entity]


def get_search_type_for_entity(entity):
    """ Returns a dictionary containing the correct search criteria type
    for a given entity.

    Example::
        get_search_type_for_entity('space')
        # returns:
        {'@type': 'as.dto.space.search.SpaceSearchCriteria'}
    """
    search_criteria = {
        "space": "as.dto.space.search.SpaceSearchCriteria",
        "userId": "as.dto.person.search.UserIdSearchCriteria",
        "email": "as.dto.person.search.EmailSearchCriteria",
        "firstName": "as.dto.person.search.FirstNameSearchCriteria",
        "lastName": "as.dto.person.search.LastNameSearchCriteria",
        "project": "as.dto.project.search.ProjectSearchCriteria",
        "experiment": "as.dto.experiment.search.ExperimentSearchCriteria",
        "experiment_type": "as.dto.experiment.search.ExperimentTypeSearchCriteria",
        "sample": "as.dto.sample.search.SampleSearchCriteria",
        "sample_type": "as.dto.sample.search.SampleTypeSearchCriteria",
        "dataset": "as.dto.dataset.search.DataSetSearchCriteria",
        "dataset_type": "as.dto.dataset.search.DataSetTypeSearchCriteria",
        "external_dms": "as.dto.externaldms.search.ExternalDmsSearchCriteria",
        "material": "as.dto.material.search.MaterialSearchCriteria",
        "material_type": "as.dto.material.search.MaterialTypeSearchCriteria",
        "vocabulary_term": "as.dto.vocabulary.search.VocabularyTermSearchCriteria",
        "tag": "as.dto.tag.search.TagSearchCriteria",
        "authorizationGroup": "as.dto.authorizationgroup.search.AuthorizationGroupSearchCriteria",
        "role_assignment": "as.dto.roleassignment.search.RoleAssignmentSearchCriteria",
        "person": "as.dto.person.search.PersonSearchCriteria",
        "code": "as.dto.common.search.CodeSearchCriteria",
        "sample_type": "as.dto.sample.search.SampleTypeSearchCriteria",
        "global": "as.dto.global.GlobalSearchObject",
    }
    return {
        "@type": search_criteria[entity]