# -*- coding: utf-8 -*-
"""test_all_harvester

"""
import pytest

from gluon import current

from harvest_tools.articles import (
    MSG_NO_EDITOR,
    MSG_TRANSFORM_PREPRINT)

from harvest_tools.base import (
    MSG_FIX_ORIGIN,
    MSG_IN_DB,
    MSG_LOAD,
    MSG_NO_ENTRY,
    MSG_TOOMANY_SYNONYM)

from harvest_tools.checkandfix import (
    MSG_NO_AUTHOR,
    MSG_NO_CONF_DATE,
    MSG_NO_DATE,
    MSG_NO_MY_AUTHOR,
    MSG_NO_REF,
    MSG_NO_YEAR,
    MSG_TEMPORARY_RECORD,
    MSG_TO_MANY_DATE,
    MSG_TO_MANY_FAUTHOR,
    MSG_TO_MANY_YEAR,
    MSG_WELL_FORMED_CONF_DATES,
    MSG_WELL_FORMED_DATE,
    MSG_WELL_FORMED_EDITOR)

from harvest_tools.factory import build_harvester_tool

from harvest_tools.preprints import (
    MSG_PREPRINT_IS_PAPER,
    MSG_PREPRINT_IS_CONFERENCE,
    MSG_PREPRINT_IS_THESIS,
    MSG_PREPRINT_NO_NUMBER)

from harvest_tools.reports import MSG_REPORT_NO_NUMBER
from harvest_tools.thesis import MSG_NO_THESIS
from invenio_tools.base import MSG_NO_CONF, MSG_NO_PUBLISHER


@pytest.fixture(scope="module")
def messages():
    T = current.T

    set_msgs = {
        T(MSG_NO_EDITOR),
        T(MSG_TRANSFORM_PREPRINT),
        T(MSG_FIX_ORIGIN),
        T(MSG_IN_DB),
        T(MSG_LOAD),
        T(MSG_NO_ENTRY % "collaborations"),
        T(MSG_NO_ENTRY % "countries"),
        T(MSG_NO_ENTRY % "publishers"),
        T(MSG_TOOMANY_SYNONYM),
        T(MSG_NO_AUTHOR),
        T(MSG_NO_CONF),
        T(MSG_NO_CONF_DATE),
        T(MSG_NO_DATE),
        T(MSG_NO_MY_AUTHOR),
        T(MSG_NO_PUBLISHER),
        T(MSG_NO_REF),
        T(MSG_NO_THESIS),
        T(MSG_NO_YEAR),
        T(MSG_PREPRINT_IS_PAPER),
        T(MSG_PREPRINT_IS_CONFERENCE),
        T(MSG_PREPRINT_IS_THESIS),
        T(MSG_PREPRINT_NO_NUMBER),
        T(MSG_REPORT_NO_NUMBER),
        T(MSG_TEMPORARY_RECORD),
        T(MSG_TO_MANY_DATE),
        T(MSG_TO_MANY_FAUTHOR),
        T(MSG_TO_MANY_YEAR),
        T(MSG_WELL_FORMED_CONF_DATES),
        T(MSG_WELL_FORMED_DATE),
        T(MSG_WELL_FORMED_EDITOR)}

    return set_msgs


def test_astro_gamma(messages):
    """Test all harvesters for the astro-gamma team.

    """
    # Parameters
    # Select the current year in order to test different case
    db = current.db
    id_team = 2
    year = current.request.now.year

    # get the list of harvester
    query = db.harvesters.id_teams == id_team

    # process
    for harvester in db(query).iterselect():

        tool = build_harvester_tool(
            db,
            harvester.id_teams,
            harvester.id_projects,
            harvester.controller,
            harvester.id_categories,
            year_start=str(year),
            year_end="",
            dry_run=True,
            debug=False)

        tool.process_url(harvester.host, harvester.collections)

        # analyse the log
        # Number of article cannot be check since it evolve within a year
        # Only test that there are no unexpected messages

        msgs = set([el.txt for el in tool.logs])
        assert msgs.issubset(messages)


def test_atlas_all(messages):
    """Test all harvesters for the Atlas team.

    """
    # Parameters
    # Select the current year in order to test different case
    db = current.db
    id_team = 3
    year = current.request.now.year

    # get the list of harvester
    query = db.harvesters.id_teams == id_team

    # process
    for harvester in db(query).iterselect():

        tool = build_harvester_tool(
            db,
            harvester.id_teams,
            harvester.id_projects,
            harvester.controller,
            harvester.id_categories,
            year_start=str(year),
            year_end="",
            dry_run=True,
            debug=False)

        tool.process_url(harvester.host, harvester.collections)

        # analyse the log
        # Number of article cannot be check since it evolve within a year
        # Only test that there are no unexpected messages

        msgs = set([el.txt for el in tool.logs])
        assert msgs.issubset(messages)


def test_lhcb_all(messages):
    """Test all harvesters for the LHCb team.

    """
    # Parameters
    # Select the current year in order to test different case
    db = current.db
    id_team = 7
    year = current.request.now.year

    # get the list of harvester
    query = db.harvesters.id_teams == id_team

    # process
    for harvester in db(query).iterselect():

        tool = build_harvester_tool(
            db,
            harvester.id_teams,
            harvester.id_projects,
            harvester.controller,
            harvester.id_categories,
            year_start=str(year),
            year_end="",
            dry_run=True,
            debug=False)

        tool.process_url(harvester.host, harvester.collections)

        # analyse the log
        # Number of article cannot be check since it evolve within a year
        # Only test that there are no unexpected messages

        msgs = set([el.txt for el in tool.logs])
        assert msgs.issubset(messages)