test_tool_extended.py

#!/bin/python
# Tests for the tool using pytest
# Adapted from check-sfs
# A. Schlemmer, 06/2021

from newcrawler import Crawler
from newcrawler.structure_elements import File, DictTextElement, DictListElement
from newcrawler.identifiable_adapters import IdentifiableAdapter, LocalStorageIdentifiableAdapter
from functools import partial
from copy import deepcopy
from unittest.mock import MagicMock, Mock
from os.path import join, dirname, basename
import yaml
import caosdb as db
from caosdb.apiutils import compare_entities

import pytest
from pytest import raises


def rfp(*pathcomponents):
    """
    Return full path.
    Shorthand convenience function.
    """
    return join(dirname(__file__), *pathcomponents)


def dircheckstr(*pathcomponents, structure_element_type="Directory"):
    """
    Return the debug tree identifier for a given path.
    """
    return ("newcrawler.structure_elements." + structure_element_type + ": " +
            basename(join(*pathcomponents)) + ", " +
            rfp("test_directories", "examples_article", *pathcomponents))


@pytest.fixture
def crawler():
    crawler = Crawler(debug=True)
    crawler.crawl_directory(rfp("test_directories", "examples_article"),
                            rfp("scifolder_extended.yml"))
    return crawler


# @pytest.fixture
# def ident(crawler):
#     ident = LocalStorageIdentifiableAdapter()
#     crawler.identifiableAdapter = ident
    
#     ident.restore_state(rfp("records.xml"))

#     ident.register_identifiable(
#         "Person", db.RecordType()
#         .add_parent(name="Person")
#         .add_property(name="first_name")
#         .add_property(name="last_name"))
#     ident.register_identifiable(
#         "Measurement", db.RecordType()
#         .add_parent(name="Measurement")
#         .add_property(name="identifier")
#         .add_property(name="date")
#         .add_property(name="project"))
#     ident.register_identifiable(
#         "Project", db.RecordType()
#         .add_parent(name="Project")
#         .add_property(name="date")
#         .add_property(name="identifier"))
#     return ident


def test_file_structure_generation(crawler):
    sd = crawler.debug_tree[dircheckstr("SimulationData",
                                        "2020_climate-model-predict", "2020-02-01",
                                        "README.md", structure_element_type="File")]
    assert sd[1]["ReadmeFile"].role == "File"
    assert len(sd[1]["ReadmeFile"].path) > 0
    assert len(sd[1]["ReadmeFile"].file) > 0