def work_remote_id_change_action(**kwargs):
    decisions = kwargs['decisions']
    execution_id = kwargs['execution_id']
    names = {
        'enrich': f'enrichment-{execution_id}',
        'change': f'generate-{execution_id}'
    }
    kwargs['names'] = names
    enrich_signature = _build_enrich_signature(**kwargs)
    change_data_signatures = _build_change_data_signatures(**kwargs)
    if change_data_signatures is None:
        decisions.append(CompleteWork())
        return
    great_chain = chain(enrich_signature,
                        group(*tuple(change_data_signatures)))
    chain_results = great_chain(**kwargs)
    if chain_results is None:
        return
    fungal_leech = _build_fungal_leech(**kwargs)
    fungal_leech(**kwargs)
    if not fungal_leech.is_complete:
        return
    post_process_group = _build_post_process_group(**kwargs)
    post_process_results = post_process_group(**kwargs)
    if post_process_results is None:
        return
    decisions.append(CompleteWork())
Пример #2
0
def _build_group(task_args, **kwargs):
    subtask_name = 'work_remote_id_change_action'
    signatures = []
    execution_id = kwargs['execution_id']
    workflow_args = kwargs['workflow_args']
    category_id = workflow_args['category_id']
    changelog_types = task_args.get_argument_value('changelog_types')
    change_category = changelog_types.categories[category_id]
    change_actions = task_args.get_argument_value('change_actions')
    if isinstance(change_actions, list):
        change_actions = change_actions[0]
    for action_id, change_action in change_category.change_types.items():
        change_action = changelog_types[str(action_id)]
        remote_actions = change_actions.get(change_action.action, {})
        if not remote_actions:
            continue
        new_task_args = task_args.replace_argument_value(
            subtask_name, {'action_id': action_id}, action_id)
        subtask_identifier = f'work_action-{change_action}-{execution_id}'
        change_type_signature = SubtaskSignature(subtask_identifier,
                                                 subtask_name,
                                                 task_args=new_task_args,
                                                 **kwargs)
        signatures.append(change_type_signature)
    tuple_signatures = tuple(signatures)
    return group(*tuple_signatures)
Пример #3
0
def _build_index_group(task_args, **kwargs):
    signatures = []
    names = kwargs['names']
    local_id_values = task_args.get_argument_value('local_id_values')
    remote_id_values = task_args.get_argument_value('remote_id_values')
    local_linked_values = local_id_values['linked']
    new_ids = remote_id_values - local_id_values['all']
    newly_linked_ids = (remote_id_values - local_linked_values) - new_ids
    operations = [
        ('unlink_old_ids', local_linked_values - remote_id_values),
        ('link_new_ids', newly_linked_ids),
        ('put_new_ids', new_ids),
    ]
    for operation in operations:
        operation_name = operation[0]
        id_values = operation[1]
        operation_identifier = names[operation_name]
        new_task_arg = {'id_values': id_values}
        new_task_args = task_args.replace_argument_value(
            operation_name, new_task_arg, operation_identifier)
        operation_signature = ActivitySignature(operation_identifier,
                                                operation_name,
                                                task_args=new_task_args,
                                                **kwargs)
        signatures.append(operation_signature)
    if not signatures:
        return None
    return group(*tuple(signatures))
Пример #4
0
def _build_chain(names, **kwargs):
    get_local_ids = ActivitySignature(names['local'], 'get_local_ids', **kwargs)
    get_remote_ids = ActivitySignature(names['remote'], 'get_remote_ids', **kwargs)
    generate_mapping = ActivitySignature(names['map'], 'build_mapping', **kwargs)
    get_ids_group = group(get_local_ids, get_remote_ids)
    get_change_type = ActivitySignature(names['change_types'], 'pull_change_types', **kwargs)
    get_schema_entry = ActivitySignature(names['schema_entry'], 'pull_schema_entry', **kwargs)
    work_fip_links = SubtaskSignature(names['work_links'], 'work_fip_links', **kwargs)
    great_chain = chain(get_schema_entry, get_ids_group, work_fip_links, get_change_type, generate_mapping)
    return great_chain
Пример #5
0
def _build_group(names, task_args, **kwargs):
    subtask_name = 'work_remote_id'
    work_history = kwargs['work_history']
    execution_id = kwargs['execution_id']
    remote_id_values = work_history.get_result(names['remote'])
    work_remote_ids_signatures = []
    for remote_id_value in remote_id_values['remote_id_values']:
        subtask_identifier = f'work_id-{remote_id_value}-{execution_id}'
        new_task_args = task_args.replace_argument_value(subtask_name, {'id_value': remote_id_value}, remote_id_value)
        work_remote_id_signature = SubtaskSignature(subtask_identifier, subtask_name, task_args=new_task_args, **kwargs)
        work_remote_ids_signatures.append(work_remote_id_signature)
    tuple_signatures = tuple(work_remote_ids_signatures)
    return group(*tuple_signatures)
def _build_query_data_group(task_args, **kwargs):
    execution_id = kwargs['execution_id']
    task_name = 'query_credible_data'
    report_args = task_args.get_argument_value('report_args')
    signatures = []
    for query_name, query_args in report_args['queries'].items():
        activity_id = f'{query_name}-{execution_id}'
        new_task_args = task_args.replace_argument_value(
            task_name, query_args, activity_id)
        signature = ActivitySignature(activity_id,
                                      task_name,
                                      task_args=new_task_args,
                                      **kwargs)
        signatures.append(signature)
    if not signatures:
        return None
    return group(*tuple(signatures))
def _build_post_process_group(execution_id, task_args, **kwargs):
    ext_id_values = task_args.get_argument_value('ext_id_values')

    signatures = []
    encounter_ids = []
    for entry in ext_id_values:
        id_type = entry[1]
        id_value = entry[3]
        if id_type == 'ClientVisit':
            encounter_ids.append(id_value)
    if encounter_ids:
        subtask_name = 'post_process_encounters'
        subtask_identifier = f'post_process_encounters-{execution_id}'
        new_task_args = task_args.replace_argument_value(
            subtask_name, {'encounter_ids': encounter_ids}, subtask_identifier)
        signatures.append(
            SubtaskSignature(subtask_identifier,
                             subtask_name,
                             task_args=new_task_args))
    return group(*tuple(signatures))
Пример #8
0
def work_remote_id(task_args, **kwargs):
    subtask_name = 'work_remote_id_change_type'
    decisions = kwargs['decisions']
    execution_id = kwargs['execution_id']
    changelog_types = task_args.get_argument_value('changelog_types')
    signatures = []
    for category_id, change_category in changelog_types.categories.items():
        new_task_args = task_args.replace_argument_value(
            subtask_name, {'category_id': category_id}, category_id)
        subtask_identifier = f'work_type-{change_category}-{execution_id}'
        category_signature = SubtaskSignature(subtask_identifier,
                                              subtask_name,
                                              task_args=new_task_args,
                                              **kwargs)
        signatures.append(category_signature)
    tuple_signatures = tuple(signatures)
    great_group = group(*tuple_signatures)
    group_results = great_group(task_args=task_args, **kwargs)
    if group_results is None:
        return
    decisions.append(CompleteWork())
Пример #9
0
def fungal_leech(**kwargs):
    decisions = kwargs['decisions']
    execution_id = kwargs['execution_id']
    names = {
        'transform': f'transform-{execution_id}',
        'index': f'index-{execution_id}',
        'graph': f'graph-{execution_id}',
    }
    kwargs['names'] = names
    transform_signature = _build_transform_signature(**kwargs)
    transform_results = transform_signature(**kwargs)
    if transform_results is None:
        return
    assimilate_group = _build_assimilate_group(**kwargs)
    assimilate_results = assimilate_group(**kwargs)
    if assimilate_results is None:
        return
    index_signature = _build_index_signature(**kwargs)
    graph_signature = _build_graph_signature(**kwargs)
    load_group = group(index_signature, graph_signature)
    load_results = load_group(**kwargs)
    if load_results is None:
        return
    decisions.append(CompleteWork())
Пример #10
0
def _build_assimilate_group(task_args, **kwargs):
    batch_size = 100
    signatures = []
    execution_id = kwargs['execution_id']
    task_name = 'assimilate'
    transform_results = task_args.get_argument_value('transform')
    batches = []
    batch = []
    for transform_result in transform_results:
        potentials = transform_result['potentials']
        source_vertex = transform_result['source_vertex']
        extracted_data = transform_result['extracted_data']
        for potential in potentials:
            if len(batch) > batch_size:
                batches.append(batch)
                batch = []
            new_arg = {
                'rule_entry': potential[1],
                'potential_vertex': potential[0],
                'source_vertex': source_vertex,
                'extracted_data': extracted_data
            }
            batch.append(new_arg)
    if batch:
        batches.append(batch)
    for batch in batches:
        task_identifier = f'assimilate-{batches.index(batch)}-{execution_id}'
        new_task_args = task_args.replace_argument_value(
            task_name, {'assimilation_batch': batch}, task_identifier)
        signature = ActivitySignature(task_identifier,
                                      task_name,
                                      task_args=new_task_args,
                                      **kwargs)
        signatures.append(signature)
    tuple_signatures = tuple(signatures)
    return group(*tuple_signatures)