""" invenio_tools.factory

"""
import requests

from base import (is_conference,
                  is_institute,
                  is_thesis,
                  MSG_INV_CONF,
                  MSG_INV_CONF_KEY,
                  MSG_NO_CONF,
                  MSG_NO_CONF_ID_KEY,
                  REG_CONF,
                  REG_OAI)

from exception import CdsException
from inveniostore import InvenioStore
from recordconf import RecordConf
from recordinst import RecordInst
from recordpubli import RecordPubli
from recordthesis import RecordThesis


def add_affiliation_keys(recjson):
    """A the affiliation keys to the record describing an institute:

        * The XML record contains the affiliation keys used by inspirehep.net.
          They are located in the field 110__u and 110__t (future).

        * The JSON record does not contains this information.

        * This tool add the affiliation keys to the JSON record.
          They are located:

             +----------------+------------------------------------+
             | field (limbra) | subfield                           |
             +----------------+------------------------------------+
             | corporate_note | identifier, futur_identifier, name |
             +----------------+------------------------------------+

    Args
        recjson (dict): record data (MarcJSON)

    """
    url = "https://inspirehep.net/record/%i" % recjson[u"recid"]
    rep = requests.get(url, params={"ot": "110", "of": "txt"})

    # decode the string: '000recid 110__ $$aXXX$$bYYY$$tZZZ\n'
    txt = rep.content.replace("\n", "")
    li = txt[txt.find("$"):].split("$$")

    di = {}
    for el in li:
        if len(el) == 0:
            continue
        di[el[0:1]] = el[1:]

    recjson[u"corporate_note"] = {u"identifier": di["u"],
                                  u"future_identifier": di["t"]}


def add_conference_data(recjson):
    """Add the conference data to the recjson.

    It adds the following field and subfield::
         +---------------+-----------------------------------------------+
         | field         | subfield                                      |
         +---------------+-----------------------------------------------+
         | meeting_name  | closing_date, coference_code, country, date,  |
         |               | location, opening_date, year                  |
         | meeting_note  | recid, url                                    |
         +---------------+-----------------------------------------------+

    Args:
        recjson (dict): record data (MarcJSON)

    Note:
        * Fields are not added when there is no conference identifier and
          no conference key in the recjson.
        * The method CheckAndFix.is_conference will identify that case.

    """
    # ........................................................................
    #
    # Retrieve conference identifier and the host
    #     - the algorithm depend on the store
    #     - for cds use aleph_linking_page
    #     - for inspire use  publication_info.cnum
    #
    conf_id, conf_key, host = None, None, None

    if u"aleph_linking_page" in recjson:
        di = recjson[u"aleph_linking_page"]
        conf_id = di[u"sysno"]
        conf_key = di[u"up_link"]
        host = "cds.cern.ch"

    elif u"publication_info" in recjson:
        data = recjson[u"publication_info"]
        data = (data if isinstance(data, list) else [data])

        for di in data:
            if u"cnum" in di:
                conf_key = di[u"cnum"]
                host = "inspirehep.net"
                break

    if conf_id is None and conf_key is None:
        return

    # ........................................................................
    #
    # Get conference data
    #

    # get the data
    if conf_id is not None:
        conf_id = (conf_id if isinstance(conf_id, int) else int(conf_id))
        confjson = get_conference_data(host, conf_id=conf_id)

    else:
        confjson = get_conference_data(host, key=conf_key)

    #
    # extract the conference url
    #     * information is in confjson[url]
    #     * in most of the case it is a dictionary
    #     * it happen that it is a list. The first entry is for the conference
    #       home page while the second one is for the proceeding (cds 2270940)
    #     - in other case the url is not defined (cds 2258914)

    confurl = u""
    if u"url" in confjson:
        obj = confjson[u"url"]
        confurl = (obj[u"url"] if isinstance(obj, dict) else obj[0][u"url"])

    # ........................................................................
    #
    # Add conference data to the recjson
    #
    recjson[u"meeting_name"] = confjson[u"meeting_name"]
    recjson[u"meeting_note"] = {u"recid": confjson[u"recid"], u"url": confurl}


def build_record(recjson):
    """Transform a JSON object into a record

    Args:
        recjson (dict):
            record data in a JSON format.

    Return
        Record:
            either RecordConf, RecordInst, RecodPubli or RecordThesis

    Raises:

    """
    if is_conference(recjson):
        add_conference_data(recjson)
        upcast_record = RecordConf(recjson)

    elif is_institute(recjson):
        add_affiliation_keys(recjson)
        upcast_record = RecordInst(recjson)

    elif is_thesis(recjson):
        upcast_record = RecordThesis(recjson)

    else:
        upcast_record = RecordPubli(recjson)

    return upcast_record


def get_conference_data(host, conf_id=None, key=None):
    """Get the conference data identified by its id or key.

    Args:
        host (unicode):
            possible values are ``cds.cern.ch`` or ``inspirehep.net``.

        conf_id (int):
            the conference identifier in the store.
            This is the preferred way.

        key (unicode): the conference key in the store.

    Returns:
        dict:
            The conference data (MarcJSON).

    Raises:
        CdsException:
            - conference record with a wrong identifier
            - conference not found

    """
    cds = InvenioStore(host)

    # ........................................................................
    #
    # search by id
    #
    if conf_id is not None:
        recjson = cds.get_record(conf_id)
        if recjson["recid"] != conf_id:
            raise CdsException(MSG_INV_CONF)
        return recjson

    # ........................................................................
    #
    # search by key in cds.cern.ch
    #
    if key is not None and host == "cds.cern.ch":
        ids = cds.get_ids(p=key)

        for conf_id in ids:
            recjson = cds.get_record(conf_id)

            if match_conference_key(recjson, key):
                return recjson

        raise CdsException(MSG_NO_CONF)

    # ........................................................................
    #
    # search by key in inspirehep.net
    #
    if key is not None and host == "inspirehep.net":
        key = key.replace("/", "-")

        if not REG_CONF.match(key):
            raise CdsException(MSG_INV_CONF_KEY)

        ids = cds.get_ids(cc="Conferences", p="111__g:%s" % key)

        for conf_id in ids:
            recjson = cds.get_record(conf_id)

            if match_conference_key(recjson, key):
                return recjson

        raise CdsException(MSG_NO_CONF)


def match_conference_key(recjson, conf_key):
    """Return ``True`` when the record corresponds to a conference identified
    by its key.

    Args:
        recjson (dict):
            record formatted MarcJSON.

        conf_key (unicode):
            conference key

    Returns
        bool:

    """
    if u"meeting_name" in recjson:
        for di in recjson[u"meeting_name"]:

            subfield = u"coference_code"
            if subfield in di and di[subfield] == conf_key:
                return True

    return False