path: root/ishtar_common/data_importer.py

#!/usr/bin/env python
# -*- coding: utf-8 -*-
# Copyright (C) 2013-2017 Étienne Loks  <etienne.loks_AT_peacefrogsDOTnet>

# This program is free software: you can redistribute it and/or modify
# it under the terms of the GNU Affero General Public License as
# published by the Free Software Foundation, either version 3 of the
# License, or (at your option) any later version.

# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
# GNU Affero General Public License for more details.

# You should have received a copy of the GNU Affero General Public License
# along with this program.  If not, see <http://www.gnu.org/licenses/>.

# See the file COPYING for details.

import copy
import datetime
import os
import logging
import re
import sys
import zipfile

from django.conf import settings
from django.contrib.auth.models import User
from django.contrib.gis.geos.error import GEOSException
from django.db.models.fields import FieldDoesNotExist
from django.core.exceptions import FieldError
from django.core.files import File
from django.db import IntegrityError, DatabaseError, transaction
from django.db.models import Q
from django.template.defaultfilters import slugify
from django.utils.translation import ugettext_lazy as _

from ishtar_common.utils import debug_line_no, get_all_field_names, update_data


NEW_LINE_BREAK = "#####@@@#####"

RE_FILTER_CEDEX = re.compile("(.*) *(?: *CEDEX|cedex|Cedex|Cédex|cédex *\d*)")


def post_importer_action(func):
    def wrapper(self, context, value):
        return func(self, context, value)

    wrapper.importer_trigger = "post"
    return wrapper


def pre_importer_action(func):
    def wrapper(self, context, value):
        return func(self, context, value)

    wrapper.importer_trigger = "pre"
    return wrapper


class ImportFormater(object):
    def __init__(
        self,
        field_name,
        formater=None,
        required=True,
        through=None,
        through_key=None,
        through_dict=None,
        through_unicity_keys=None,
        duplicate_fields=None,
        regexp=None,
        regexp_formater_args=None,
        force_value=None,
        post_processing=False,
        concat=False,
        concat_str=False,
        comment="",
        force_new=None,
        export_field_name=None,
        value_format=None,
        label="",
    ):
        self.field_name = field_name
        if export_field_name:
            self.export_field_name = export_field_name
        else:
            self.export_field_name = field_name
        self.formater = formater
        self.required = required
        self.through = through
        self.through_key = through_key
        self.through_dict = through_dict
        self.through_unicity_keys = through_unicity_keys
        self.duplicate_fields = duplicate_fields or []
        self.regexp = regexp
        self.value_format = value_format
        self.regexp_formater_args = regexp_formater_args or []
        # write this value even if a value exists
        self.force_value = force_value
        # post process after import
        self.post_processing = post_processing
        # concatenate with existing value
        self.concat = concat
        self.concat_str = concat_str
        self.comment = comment
        self.force_new = force_new
        self.label = label

    def reinit_db_target(self, db_target, nb=0, user=None):
        if not self.formater:
            return
        if type(db_target) in (list, tuple):
            db_target = db_target[nb]
        if type(self.formater) not in (list, tuple):
            self.formater.db_target = db_target
            self.formater.init_db_target(user=user)
        else:
            for idx, formater in enumerate(self.formater):
                formater.db_target = db_target
                formater.init_db_target(user=user)

    def init_db_target(self, user=None):
        pass

    def __str__(self):
        return self.field_name

    def report_succes(self, *args):
        return

    def report_error(self, *args):
        return

    def init(
        self, vals, output=None, choose_default=False, import_instance=None, user=None
    ):
        try:
            lst = iter(self.formater)
        except TypeError:
            lst = [self.formater]
        for formater in lst:
            if formater:
                formater.check(
                    vals,
                    output,
                    self.comment,
                    choose_default=choose_default,
                    import_instance=import_instance,
                    user=user,
                )

    def post_process(self, obj, context, value, owner=None):
        raise NotImplemented()


class ImporterError(Exception):
    STANDARD = "S"
    HEADER = "H"

    def __init__(self, message, type="S"):
        self.msg = message
        self.type = type

    def __str__(self):
        return str(self.msg)


class Formater(object):
    def __init__(self, *args, **kwargs):
        self.db_target = kwargs.get("db_target", None)

    def format(self, value):
        return value

    def check(
        self,
        values,
        output=None,
        comment="",
        choose_default=False,
        import_instance=None,
        user=None,
    ):
        return

    def init_db_target(self, user=None):
        pass

    def _base_target_filter(self, user=None):
        # set for all users
        q_or = (
            Q(associated_import__isnull=True)
            & Q(associated_user__isnull=True)
            & Q(associated_group__isnull=True)
        )
        if hasattr(self, "import_instance") and self.import_instance:
            # set for current import
            q_or = q_or | Q(associated_import=self.import_instance)
            if self.import_instance.associated_group:
                # set for associated group
                q_or = q_or | Q(associated_group=self.import_instance.associated_group)
        if user:
            # set for current user
            q_or = q_or | Q(associated_user=user)
        return q_or

    def get_db_target_query(self, user=None):
        if not self.db_target:
            return
        q = self.db_target.keys.filter(is_set=True)
        q_or = self._base_target_filter(user)
        q = q.filter(q_or)
        return q


class ChoiceChecker(object):
    def report_new(self, comment):
        if not self.new_keys:
            return
        msg = 'For "%s" these new associations have been made:\n' % comment
        sys.stderr.write(msg.encode("utf-8"))
        for k in self.new_keys:
            msg = '"%s";"%s"\n' % (k, self.new_keys[k])
            sys.stderr.write(msg.encode("utf-8"))


class UnicodeFormater(Formater):
    def __init__(
        self,
        max_length=None,
        clean=False,
        re_filter=None,
        notnull=False,
        prefix="",
        db_target=None,
        import_instance=None,
        many_split=None,
    ):
        self.max_length = max_length
        self.db_target = db_target
        self.clean = clean
        self.re_filter = re_filter
        self.notnull = notnull
        self.prefix = prefix
        self.import_instance = import_instance
        self.many_split = many_split

    def format(self, value):
        try:
            if type(value) != str:
                value = str(value.strip())
            vals = []
            for v in value.split("\n"):
                v = v.strip()
                if v:
                    vals.append(v)
            value = "\n".join(vals)
            if self.re_filter:
                m = self.re_filter.match(value)
                if m:
                    value = "".join(m.groups())
            if self.clean:
                if value.startswith(","):
                    value = value[1:]
                if value.endswith(","):
                    value = value[:-1]
                value = value.replace(", , ", ", ")
        except UnicodeDecodeError:
            return
        if self.max_length and len(value) > self.max_length:
            raise ValueError(
                _(
                    '"%(value)s" is too long. The max length is %(length)d '
                    "characters."
                )
                % {"value": value, "length": self.max_length}
            )
        if self.notnull and not value:
            return
        if value:
            value = self.prefix + value
        return value


class BooleanFormater(Formater):
    def format(self, value):
        value = value.strip().upper()
        if value in ("1", "OUI", "VRAI", "YES", "TRUE"):
            return True
        if value in ("", "0", "NON", "FAUX", "NO", "FALSE"):
            return False
        raise ValueError(_('"%(value)s" not equal to yes or no') % {"value": value})


class FloatFormater(Formater):
    def format(self, value):
        value = value.strip().replace(",", ".")
        value = value.replace(" ", "")
        if not value:
            return
        try:
            return float(value)
        except ValueError:
            raise ValueError(_('"%(value)s" is not a float') % {"value": value})


class InseeFormater(Formater):
    """
    Formater for "code INSEE" (statistic institute ine France)
    The syntax "CodeINSEE-Year" is accepted (Ishtar trick) in order to manage
    old INSEE (year is the date of creation)
    """

    ERROR = _('"{value}" is not an appropriate INSEE code')

    def format(self, value):
        value = value.strip()
        exp = value.split("-")
        code = exp[0]
        try:
            int(code)
        except ValueError:
            raise ValueError(str(self.ERROR).format(value))
        while len(code) < 5:
            code = "0" + code
        if len(exp) > 2:
            raise ValueError(str(self.ERROR).format(value))
        elif len(exp) == 1:
            return code
        try:
            datetime.datetime.strptime(exp[1], "%Y")
        except ValueError:
            raise ValueError(str(self.ERROR).format(value))
        return code + "-" + exp[1]


class YearFormater(Formater):
    def format(self, value):
        value = value.strip()
        if not value:
            return
        try:
            value = int(value)
            assert value > 0 and value < (datetime.date.today().year + 30)
        except (ValueError, AssertionError):
            raise ValueError(_('"%(value)s" is not a valid date') % {"value": value})
        return value


class YearNoFuturFormater(Formater):
    def format(self, value):
        value = value.strip()
        if not value:
            return
        try:
            value = int(value)
            assert value > 0 and value < datetime.date.today().year
        except (ValueError, AssertionError):
            raise ValueError(_('"%(value)s" is not a valid date') % {"value": value})
        return value


class IntegerFormater(Formater):
    def format(self, value):
        value = value.strip()
        value = value.replace(" ", "")
        if not value:
            return
        try:
            return int(value)
        except ValueError:
            raise ValueError(_('"%(value)s" is not an integer') % {"value": value})


class StrChoiceFormater(Formater, ChoiceChecker):
    def __init__(
        self,
        choices,
        strict=False,
        equiv_dict=None,
        model=None,
        cli=False,
        many_split="",
        db_target=None,
        import_instance=None,
    ):
        if not equiv_dict:
            equiv_dict = {}
        self.choices = list(choices)
        self.strict = strict
        self.equiv_dict = copy.deepcopy(equiv_dict)
        self.cli = cli
        self.model = model
        self.db_target = db_target
        self.create = False
        self.missings = set()
        self.new_keys = {}
        self.match_table = {}
        self.many_split = many_split
        self.import_instance = None
        for key, value in self.choices:
            value = str(value)
            if not self.strict:
                value = slugify(value)
            if value not in self.equiv_dict:
                v = key
                if model and v:
                    v = model.objects.get(pk=v)
                self.equiv_dict[value] = v
        self.init_db_target()

    def init_db_target(self, user=None):
        if not self.db_target:
            return

        q = self.get_db_target_query(user)

        for target_key in list(q.all()):
            key = target_key.key
            if not self.strict:
                key = slugify(key)
            if key in self.equiv_dict:
                continue
            v = target_key.value
            if self.model and v and type(v) in (int, str):
                q = self.model.objects.filter(txt_idx=v)
                if not q.count():
                    q = self.model.objects.filter(pk=v)
                    if not q.count():
                        target_key.delete()
                        continue
                v = q.all()[0]
            self.equiv_dict[key] = v

    def prepare(self, value):
        return str(value).strip()

    def _get_choices(self, comment=""):
        msgstr = comment + " - "
        msgstr += str(
            _('Choice for "%s" is not available. ' "Which one is relevant?\n")
        )
        idx = -1
        for idx, choice in enumerate(self.choices):
            msgstr += "%d. %s\n" % (idx + 1, choice[1])
        idx += 2
        if self.create:
            msgstr += str(_("%d. None of the above - create new")) % idx + "\n"
            idx += 1
        msgstr += str(_("%d. None of the above - skip")) % idx + "\n"
        return msgstr, idx

    def check(
        self,
        values,
        output=None,
        comment="",
        choose_default=False,
        import_instance=None,
        user=None,
    ):
        self.init_db_target(user)

        """
        from ishtar_common.models import TargetKey
        if self.db_target:
            q = {'target': self.db_target,
                 'associated_import': import_instance,
                 'is_set': True
                 }
            for v in self.equiv_dict:
                q['key'] = v
                value = self.equiv_dict[v]
                if hasattr(value, 'pk'):
                    value = value.pk
                q['value'] = value
                with transaction.atomic():
                    try:
                        t, created = TargetKey.objects.get_or_create(**q)
                    except IntegrityError:
                        pass
        """
        if (not output or output == "silent") and not choose_default:
            return
        if self.many_split:
            new_values = []
            r = re.compile(self.many_split)
            for value in values:
                new_values += r.split(value)
            values = new_values
        for value in values:
            base_value = copy.copy(value)
            value = self.prepare(value)
            if value in self.equiv_dict:
                continue
            if output != "cli" and not choose_default:
                self.missings.add(value)
                continue
            msgstr, idx = self._get_choices(comment)
            res = None
            if choose_default:
                res = 1
            while res not in range(1, idx + 1):
                msg = msgstr % value
                sys.stdout.write(msg.encode("utf-8"))
                sys.stdout.write("\n>>> ")
                res = input()
                try:
                    res = int(res)
                except ValueError:
                    pass
            res -= 1
            if res < len(self.choices):
                v = self.choices[res][0]
                if self.model and v:
                    v = self.model.objects.get(pk=v)
                self.equiv_dict[value] = v
                self.add_key(v, value, import_instance)
                self.new_keys[value] = v
            elif self.create and res == len(self.choices):
                self.equiv_dict[value] = self.new(base_value)
                self.choices.append(
                    (self.equiv_dict[value].pk, str(self.equiv_dict[value]))
                )
                self.new_keys[value] = str(self.equiv_dict[value])
            else:
                self.equiv_dict[value] = None
            if self.equiv_dict[value] and self.db_target:
                from ishtar_common.models import TargetKey

                q = {"target": self.db_target, "key": value}
                query = TargetKey.objects.filter(**q)
                query = query.filter(self._base_target_filter(user))
                if query.count():
                    target = query.all()[0]
                    target.value = self.equiv_dict[value]
                    target.is_set = True
                    target.save()
                else:
                    q["associated_import"] = import_instance
                    with transaction.atomic():
                        q["value"] = self.equiv_dict[value]
                        q["is_set"] = True
                        try:
                            TargetKey.objects.create(**q)
                        except IntegrityError:
                            pass
        if output == "db" and self.db_target:
            from ishtar_common.models import TargetKey

            for missing in self.missings:
                q = {"target": self.db_target, "key": missing}
                query = TargetKey.objects.filter(**q)
                query = query.filter(self._base_target_filter(user))
                if query.count():
                    continue
                with transaction.atomic():
                    q["associated_import"] = import_instance
                    try:
                        TargetKey.objects.create(**q)
                    except IntegrityError:
                        pass
        if output == "cli":
            self.report_new(comment)

    def new(self, value):
        return

    def add_key(self, obj, value, importer=None):
        return

    def format(self, value):
        origin_value = value
        value = self.prepare(value)
        if not self.strict:
            value = slugify(value)
        if value in self.equiv_dict:
            self.match_table[origin_value] = self.equiv_dict[value] or ""
            return self.equiv_dict[value]


class TypeFormater(StrChoiceFormater):
    def __init__(
        self,
        model,
        cli=False,
        defaults=None,
        many_split=False,
        db_target=None,
        import_instance=None,
    ):
        if not defaults:
            defaults = {}
        self.create = True
        self.strict = False
        self.model = model
        self.defaults = defaults
        self.many_split = many_split
        self.db_target = db_target
        self.missings = set()
        self.equiv_dict, self.choices = {}, []
        self.match_table = {}
        self.new_keys = {}
        self.import_instance = import_instance
        if self.import_instance:
            for item in model.objects.all():
                self.choices.append((item.pk, str(item)))
                for key in item.get_keys(importer=import_instance):
                    self.equiv_dict[key] = item

    def prepare(self, value):
        return slugify(str(value).strip())

    def add_key(self, obj, value, importer=None):
        obj.add_key(slugify(value), force=True, importer=importer)

    def new(self, value):
        values = copy.copy(self.defaults)
        values["label"] = value
        values["txt_idx"] = slugify(value)
        if "order" in get_all_field_names(self.model):
            order = 1
            q = self.model.objects.values("order").order_by("-order")
            if q.count():
                order = q.all()[0]["order"] or 1
            values["order"] = order
        return self.model.objects.create(**values)


class DateFormater(Formater):
    def __init__(self, date_formats=None, db_target=None, import_instance=None):
        if not date_formats:
            date_formats = ["%d/%m/%Y"]
        self.date_formats = date_formats
        if type(date_formats) not in (list, tuple):
            self.date_formats = [self.date_formats]
        self.db_target = db_target
        self.import_instance = import_instance

    def format(self, value):
        value = value.strip()
        if not value:
            return
        for date_format in self.date_formats:
            try:
                return datetime.datetime.strptime(value, date_format).date()
            except:
                continue
        raise ValueError(_('"%(value)s" is not a valid date') % {"value": value})


class FileFormater(Formater):
    need_archive = True

    def format(self, value, archive):
        value = value.strip()
        if not value:
            return
        zp = zipfile.ZipFile(archive)
        value = value.strip().replace("\\", "/")
        items = value.replace("/", "_").split(".")
        base_dir = settings.MEDIA_ROOT + "imported"
        if not os.path.isdir(base_dir):
            os.mkdir(base_dir)
        filename = base_dir + os.sep + ".".join(items[:-1]) + "." + items[-1]

        try:
            with open(filename, "wb") as f:
                with zp.open(value) as z:
                    f.write(z.read())
            f = open(filename, "rb")
            my_file = File(f)
            # manually set the file size because of an issue with TempFile
            my_file.size = os.stat(filename).st_size
            return my_file
        except KeyError:
            raise ValueError(
                _('"%(value)s" is not a valid path for the ' "given archive")
                % {"value": value}
            )


class StrToBoolean(Formater, ChoiceChecker):
    def __init__(
        self,
        choices=None,
        cli=False,
        strict=False,
        db_target=None,
        import_instance=None,
    ):
        if not choices:
            choices = {}
        self.dct = copy.copy(choices)
        self.cli = cli
        self.strict = strict
        self.db_target = db_target
        self.missings = set()
        self.match_table = {}
        self.new_keys = {}
        self.import_instance = import_instance
        self.init_db_target()

    def init_db_target(self, user=None):
        if not self.db_target:
            return
        q = self.get_db_target_query(user)
        for target_key in q.all():
            key = self.prepare(target_key.key)
            if key in self.dct:
                continue
            v = target_key.format()
            self.dct[key] = v

    def prepare(self, value):
        value = str(value).strip()
        if not self.strict:
            value = slugify(value)
        return value

    def check(
        self,
        values,
        output=None,
        comment="",
        choose_default=False,
        import_instance=None,
        user=None,
    ):
        if (not output or output == "silent") and not choose_default:
            return
        msgstr = comment + " - "
        msgstr += str(
            _('Choice for "%s" is not available. ' "Which one is relevant?\n")
        )
        msgstr += "1. True\n"
        msgstr += "2. False\n"
        msgstr += "3. Empty\n"
        for value in values:
            value = self.prepare(value)
            if value in self.dct:
                continue
            if output != "cli" and not choose_default:
                self.missings.add(value)
                continue
            res = None
            if choose_default:
                res = 1
            while res not in range(1, 4):
                msg = msgstr % value
                sys.stdout.write(msg.encode("utf-8"))
                sys.stdout.write("\n>>> ")
                res = input()
                try:
                    res = int(res)
                except ValueError:
                    pass
            if res == 1:
                self.dct[value] = True
            elif res == 2:
                self.dct[value] = False
            else:
                self.dct[value] = None
            self.new_keys[value] = str(self.dct[value])
        if output == "db" and self.db_target:
            from ishtar_common.models import TargetKey

            for missing in self.missings:
                try:
                    q = {
                        "target": self.db_target,
                        "key": missing,
                        "associated_import": import_instance,
                    }
                    if not TargetKey.objects.filter(**q).count():
                        TargetKey.objects.create(**q)
                except IntegrityError:
                    pass
        if output == "cli":
            self.report_new(comment)

    def format(self, value):
        origin_value = value
        value = self.prepare(value)
        if value in self.dct:
            val = self.dct[value] and "True" or "False"
            self.match_table[origin_value] = _(val)
            return self.dct[value]


logger = logging.getLogger(__name__)


def get_object_from_path(obj, path):
    for k in path.split("__")[:-1]:
        if not hasattr(obj, k):
            return
        obj = getattr(obj, k)
    return obj


class Importer(object):
    SLUG = ""
    NAME = ""
    DESC = ""
    LINE_FORMAT = []
    OBJECT_CLS = None
    UNICITY_KEYS = []
    # if set only models inside this list can be created
    MODEL_CREATION_LIMIT = []
    EXTRA_DEFAULTS = {}
    DEFAULTS = {}
    ERRORS = {
        "header_check": _(
            "The given file is not correct. Check the file "
            "format. If you use a CSV file: check that column separator "
            "and encoding are similar to the ones used by the reference "
            "file."
        ),
        "too_many_cols": _(
            "Too many cols (%(user_col)d) when " "maximum is %(ref_col)d"
        ),
        "no_data": _("No data provided"),
        "value_required": _("Value is required"),
        "not_enough_cols": _("At least %d columns must be filled"),
        "regex_not_match": _("The regexp doesn't match."),
        "improperly_configured": _(
            "Forced creation is set for model {} but this model is not in the "
            "list of models allowed to be created."
        ),
        "does_not_exist_in_db": _(
            "{} with values {} doesn't exist in the "
            "database. Create it first or fix your source file."
        ),
    }

    def _create_models(self, force=False):
        """
        Create a db config from a hardcoded import.
        Not useful anymore?
        """
        from ishtar_common import models

        q = models.ImporterType.objects.filter(slug=self.SLUG)
        if not force and (not self.SLUG or q.count()):
            return
        if force and q.count():
            q.all()[0].delete()
        name = self.NAME if self.NAME else self.SLUG

        model_name = self.OBJECT_CLS.__module__ + "." + self.OBJECT_CLS.__name__
        model_cls, c = models.ImporterModel.object.get_or_create(
            klass=model_name, default={"name": self.OBJECT_CLS.__name__}
        )

        unicity_keys = ""
        if self.UNICITY_KEYS:
            unicity_keys = ";".join(self.UNICITY_KEYS)

        importer = models.ImporterType.objects.create(
            slug=self.SLUG,
            name=name,
            description=self.DESC,
            associated_models=model_cls,
            unicity_keys=unicity_keys,
        )

        for default in self.DEFAULTS:
            values = self.DEFAULTS[default]
            imp_default = models.ImporterDefault.objects.create(
                importer_type=importer, target="__".join(default)
            )
            for key in values:
                if key in ("history_modifier",):
                    continue
                value = values[key]
                if hasattr(value, "txt_idx") and value.txt_idx:
                    value = value.txt_idx
                elif hasattr(value, "pk") and value.pk:
                    value = value.pk
                if callable(value):
                    value = value()
                models.ImporterDefaultValues.objects.create(
                    default_target=imp_default, target=key, value=value
                )
        for idx, line in enumerate(self.line_format):
            idx += 1
            if not line:
                continue
            column = models.ImporterColumn.objects.create(
                importer_type=importer, col_number=idx
            )
            targets = line.field_name
            if type(targets) not in (list, tuple):
                targets = [targets]
            formaters = line.formater
            if type(formaters) not in (list, tuple):
                formaters = [formaters]
            for idx_target, target in enumerate(targets):
                formater = formaters[idx_target]
                formater_name = formater.__class__.__name__
                if formater_name not in models.IMPORTER_TYPES_DCT:
                    formater_name = "UnknowType"
                options = ""
                if formater_name == "TypeFormater":
                    options = formater.model.__module__ + "." + formater.model.__name__
                elif formater_name == "UnicodeFormater":
                    options = str(formater.max_length or "")
                elif formater_name == "DateFormater":
                    options = formater.date_formats[0]
                formater_model, created = models.FormaterType.objects.get_or_create(
                    formater_type=formater_name,
                    options=options.strip(),
                    many_split=getattr(formater, "many_split", None) or "",
                )
                regexp_filter = None
                if getattr(formater, "regexp", None):
                    regexp_filter, created = models.Regexp.objects.get_or_create(
                        regexp=formater.regex, defaults={"name": "Default name"}
                    )
                models.ImportTarget.objects.get_or_create(
                    column=column,
                    target=target,
                    formater_type=formater_model,
                    force_new=getattr(formater, "force_new", False),
                    concat=getattr(formater, "concat", False),
                    concat_str=getattr(formater, "concat_str", ""),
                    regexp_filter=regexp_filter,
                    comment=line.comment,
                )
        return True

    def _get_improperly_conf_error(self, model):
        from ishtar_common.models import ImporterModel

        cls_name = model.__module__ + "." + model.__name__
        q = ImporterModel.objects.filter(klass=cls_name)
        if q.count():
            cls_name = q.all()[0].name
        return ImporterError(str(self.ERRORS["improperly_configured"]).format(cls_name))

    def _get_does_not_exist_in_db_error(self, model, data):
        from ishtar_common.models import ImporterModel

        cls_name = model.__module__ + "." + model.__name__
        q = ImporterModel.objects.filter(klass=cls_name)
        if q.count():
            cls_name = q.all()[0].name
        values = ", ".join(["{}: {}".format(k, data[k]) for k in data])
        raise ImporterError(
            str(self.ERRORS["does_not_exist_in_db"]).format(cls_name, values)
        )

    def __init__(
        self,
        skip_lines=0,
        reference_header=None,
        check_col_num=False,
        test=False,
        history_modifier=None,
        output="silent",
        import_instance=None,
        conservative_import=False,
    ):
        """
        * skip_line must be set if the data provided has got headers lines.
        * a reference_header can be provided to perform a data compliance
          check. It can be useful to warn about bad parsing.
        * test doesn't write in the database
        """
        self.skip_lines = skip_lines
        self.reference_header = reference_header
        self.test = test
        self.errors = []  # list of (line, col, message)
        self.validity = []  # list of (line, col, message)

        # list used for simulation
        self.simulate = False
        self.new_objects, self.updated_objects = [], []
        self.ambiguous_objects, self.not_find_objects = [], []

        self.number_updated = 0
        self.number_created = 0
        self.check_col_num = check_col_num
        self.line_format = copy.copy(self.LINE_FORMAT)
        self.import_instance = import_instance
        self.archive = None
        self.current_csv_line = None
        self.conservative_import = conservative_import
        # for a conservative_import UNICITY_KEYS should be defined
        assert not self.conservative_import or bool(self.UNICITY_KEYS)
        self.DB_TARGETS = {}
        self.match_table = {}
        self.concats = set()
        self.concat_str = {}
        self.to_be_close = []
        if import_instance and import_instance.imported_images:
            self.archive = import_instance.imported_images
        self._defaults = self.DEFAULTS.copy()
        # EXTRA_DEFAULTS are for multiple inheritance
        if self.EXTRA_DEFAULTS:
            for k in self.EXTRA_DEFAULTS:
                if k not in self._defaults:
                    self._defaults[k] = {}
                self._defaults[k].update(self.EXTRA_DEFAULTS[k])
        self.history_modifier = history_modifier
        self.output = output
        if not self.history_modifier:
            if self.import_instance:
                self.history_modifier = self.import_instance.user.user_ptr
            else:
                # import made by the CLI: get the first admin
                self.history_modifier = User.objects.filter(is_superuser=True).order_by(
                    "pk"
                )[0]

    def post_processing(self, idx_line, item):
        # force django based post-processing for the item
        item = item.__class__.objects.get(pk=item.pk)
        item.save()
        if hasattr(item, "RELATED_POST_PROCESS"):
            for related_key in item.RELATED_POST_PROCESS:
                for related in getattr(item, related_key).all():
                    related.save()
        for func, context, value in self._item_post_processing:
            context["import_object"] = self.import_instance
            try:
                getattr(item, func)(context, value)
            except ImporterError as msg:
                self.errors.append((idx_line, None, msg))
        return item

    def initialize(self, table, output="silent", choose_default=False, user=None):
        """
        copy vals in columns and initialize formaters
        * output:
         - silent: no associations
         - cli: output by command line interface and stocked in the database
         - db: output on the database with no interactive association
           (further exploitation by web interface)
         - user: associated user
        """
        assert output in ("silent", "cli", "db")
        vals = []
        for idx_line, line in enumerate(table):
            if self.skip_lines > idx_line:
                continue
            for idx_col, val in enumerate(line):
                if idx_col >= len(self.line_format):
                    break
                if idx_col >= len(vals):
                    vals.append([])
                vals[idx_col].append(val)
        for idx, formater in enumerate(self.line_format):
            if formater and idx < len(vals):
                formater.import_instance = self.import_instance
                if self.DB_TARGETS:
                    field_names = formater.field_name
                    if type(field_names) not in (list, tuple):
                        field_names = [field_names]
                    db_targets = []
                    for field_name in field_names:
                        db_targets.append(
                            self.DB_TARGETS["{}-{}".format(idx + 1, field_name)]
                        )
                    formater.reinit_db_target(db_targets, user=user)

                formater.init(
                    vals[idx],
                    output,
                    choose_default=choose_default,
                    import_instance=self.import_instance,
                    user=user,
                )

    def get_formaters(self):
        return self.line_format

    def importation(
        self,
        table,
        initialize=True,
        choose_default=False,
        user=None,
        line_to_process=None,
        simulate=False,
    ):
        if initialize:
            self.initialize(
                table, self.output, choose_default=choose_default, user=user
            )
        self.simulate = simulate
        return self._importation(table, line_to_process=line_to_process)

    def get_current_values(self, obj):
        return obj

    def _associate_db_target_to_formaters(self):
        if not self.import_instance:
            return
        self.DB_TARGETS = {}
        from ishtar_common.models import ImporterColumn, ImportTarget

        for idx, line in enumerate(self.line_format):
            idx += 1
            if not line:
                continue
            col = ImporterColumn.objects.get(
                importer_type=self.import_instance.importer_type, col_number=idx
            )
            formater = line.formater
            targets = line.field_name
            if type(formater) not in (list, tuple):
                formater = [formater]
                targets = [targets]
            for target in targets:
                tg = target
                if type(target) == list and type(target[0]) == list:
                    tg = target[0]
                self.DB_TARGETS["{}-{}".format(idx, tg)] = ImportTarget.objects.get(
                    column=col, target=tg
                )

    @classmethod
    def _field_name_to_data_dict(
        cls,
        field_name,
        value,
        data,
        force_value=False,
        concat=False,
        concat_str="",
        force_new=False,
    ):
        field_names = field_name
        if type(field_names) not in (list, tuple):
            field_names = [field_name]
        for field_name in field_names:
            keys = field_name.split("__")
            current_data = data
            for idx, key in enumerate(keys):
                if idx == (len(keys) - 1):  # last
                    if concat:
                        if isinstance(value, (list, tuple)):
                            if key not in current_data:
                                current_data[key] = []
                            if not value:
                                continue
                            current_data[key] += value
                        else:
                            if key not in current_data:
                                current_data[key] = ""
                            if not value:
                                continue
                            current_data[key] = (
                                (current_data[key] + (concat_str or ""))
                                if current_data[key]
                                else ""
                            )
                            current_data[key] += value
                    elif force_value and value:
                        if concat_str and key in current_data and current_data[key]:
                            current_data[key] = (
                                str(current_data[key]) + concat_str + str(value)
                            )
                        else:
                            current_data[key] = value
                    elif key not in current_data or not current_data[key]:
                        current_data[key] = value
                    elif concat_str:
                        current_data[key] = (
                            str(current_data[key]) + concat_str + str(value)
                        )
                    if force_new:
                        current_data["__force_new"] = True
                elif key not in current_data:
                    current_data[key] = {}
                current_data = current_data[key]
        return data

    def _importation(self, table, line_to_process=None):
        self.match_table = {}
        table = list(table)
        if not table or not table[0]:
            raise ImporterError(self.ERRORS["no_data"], ImporterError.HEADER)
        if self.check_col_num and len(table[0]) > len(self.line_format):
            raise ImporterError(
                self.ERRORS["too_many_cols"]
                % {"user_col": len(table[0]), "ref_col": len(self.line_format)}
            )
        self.errors = []
        self.validity = []
        self.number_imported = 0
        idx_last_col = 0
        # index of the last required column
        for idx_last_col, formater in enumerate(reversed(self.line_format)):
            if formater and formater.required:
                break
        else:
            idx_last_col += 1
        # min col number to be filled
        self.min_col_number = len(self.line_format) - idx_last_col
        # check the conformity with the reference header
        if (
            self.reference_header
            and self.skip_lines
            and self.reference_header != table[0]
        ):
            raise ImporterError(self.ERRORS["header_check"], type=ImporterError.HEADER)
        self.now = datetime.datetime.now()
        start = datetime.datetime.now()
        total = len(table)
        if self.output == "cli":
            sys.stdout.write("\n")
        results = []
        for idx_line, line in enumerate(table):
            self.idx_line = idx_line
            if line_to_process is not None:
                if line_to_process != idx_line:
                    continue
                if idx_line > line_to_process:
                    return results
            if self.output == "cli":
                left = None
                if idx_line > 10:
                    ellapsed = datetime.datetime.now() - start
                    time_by_item = ellapsed / idx_line
                    if time_by_item:
                        left = ((total - idx_line) * time_by_item).seconds
                txt = "\r* %d/%d" % (idx_line + 1, total)
                if left:
                    txt += " (%d seconds left)" % left
                sys.stdout.write(txt.encode("utf-8"))
                sys.stdout.flush()
            try:
                results.append(self._line_processing(idx_line, line))
            except ImporterError as msg:
                self.errors.append((idx_line, None, msg))
        for item in self.to_be_close:
            item.close()
        return results

    def _line_processing(self, idx_line, line):
        for item in self.to_be_close:
            item.close()
        self.to_be_close = []
        self.idx_line = idx_line
        if self.skip_lines > idx_line:
            self.validity.append(line)
            return
        if not line:
            self.validity.append([])
            return
        if (
            not self.simulate
            and self.import_instance
            and not self.import_instance.has_changes(idx_line)
        ):
            self.validity.append(line)
            return

        self._throughs = []  # list of (formater, value)
        self._post_processing = []  # list of (formater, value)
        self._item_post_processing = []
        data = {}

        self.current_csv_line = None
        # raw line for simulation
        if self.simulate:
            self.current_csv_line = line

        n = datetime.datetime.now()
        logger.debug("%s - Processing line %d" % (str(n - self.now), idx_line))
        self.now = n
        n2 = n
        self.c_errors = False
        c_row = []
        for idx_col, val in enumerate(line):
            try:
                self._row_processing(c_row, idx_col, idx_line, val, data)
            except:
                pass

        self.validity.append(c_row)
        if not self.c_errors and (idx_col + 1) < self.min_col_number:
            self.c_errors = True
            self.errors.append(
                (
                    idx_line + 1,
                    idx_col + 1,
                    self.ERRORS["not_enough_cols"] % self.min_col_number,
                )
            )
        if self.c_errors:
            return
        n = datetime.datetime.now()
        logger.debug("* %s - Cols read" % (str(n - n2)))
        n2 = n
        if self.test:
            return
        # manage unicity of items (mainly for updates)
        if "history_modifier" in get_all_field_names(self.OBJECT_CLS):
            data["history_modifier"] = self.history_modifier

        self.new_objects, self.updated_objects = [], []
        self.ambiguous_objects, self.not_find_objects = [], []

        obj, created = self.get_object(self.OBJECT_CLS, data)
        if self.simulate:
            return data
        # print(data)
        # print(self._defaults)

        if self.import_instance:
            self.import_instance.add_imported_line(self.idx_line)

        if self.import_instance and hasattr(obj, "imports") and created:
            obj.imports.add(self.import_instance)

        if created:
            self.number_created += 1
        else:
            self.number_updated += 1

        if not created and "defaults" in data:
            for k in data["defaults"]:
                setattr(obj, k, data["defaults"][k])
            obj.save()
        n = datetime.datetime.now()
        logger.debug("* %s - Item saved" % (str(n - n2)))
        n2 = n
        for formater, value in self._throughs:
            n = datetime.datetime.now()
            logger.debug(
                "* %s - Processing formater %s" % (str(n - n2), formater.field_name)
            )
            n2 = n
            data = {}
            if formater.through_dict:
                data = formater.through_dict.copy()
            if formater.through_key:
                data[formater.through_key] = obj
            data[formater.field_name] = value
            through_cls = formater.through
            if formater.through_unicity_keys:
                data["defaults"] = {}
                for k in list(data.keys()):
                    if k not in formater.through_unicity_keys and k != "defaults":
                        data["defaults"][k] = data.pop(k)
            created = False
            if "__force_new" in data:
                if (
                    self.MODEL_CREATION_LIMIT
                    and through_cls not in self.MODEL_CREATION_LIMIT
                ):
                    raise self._get_improperly_conf_error(through_cls)
                created = data.pop("__force_new")
                t_obj = through_cls.objects.create(**data)
            else:
                if (
                    not self.MODEL_CREATION_LIMIT
                    or through_cls in self.MODEL_CREATION_LIMIT
                ):
                    t_obj, created = through_cls.objects.get_or_create(**data)
                else:
                    get_data = data.copy()
                    if "defaults" in get_data:
                        get_data.pop("defaults")
                    try:
                        t_obj = through_cls.objects.get(**get_data)
                    except through_cls.DoesNotExist:
                        raise self._get_does_not_exist_in_db_error(
                            through_cls, get_data
                        )
            if not created and "defaults" in data:
                t_obj = t_obj.__class__.objects.get(pk=t_obj.pk)
                for k in data["defaults"]:
                    setattr(t_obj, k, data["defaults"][k])
                t_obj.save()
            if self.import_instance and hasattr(t_obj, "imports") and created:
                t_obj.imports.add(self.import_instance)
        if not obj:
            return data

        for formater, val in self._post_processing:
            formater.post_process(obj, data, val, owner=self.history_modifier)

        self.post_processing(idx_line, obj)
        return data

    def _row_processing(self, c_row, idx_col, idx_line, val, data):
        if idx_col >= len(self.line_format):
            return

        formater = self.line_format[idx_col]

        if formater and formater.post_processing:
            self._post_processing.append((formater, val))

        if not formater or not formater.field_name:
            c_row.append(_("Not imported"))
            return

        if formater.regexp:
            # multiline regexp is a mess...
            val = val.replace("\n", NEW_LINE_BREAK)
            match = formater.regexp.match(val)
            if not match:
                if formater.required:
                    self.errors.append(
                        (idx_line + 1, idx_col + 1, self.ERRORS["value_required"])
                    )
                    self.c_errors = True
                elif not val.strip():
                    c_row.append("")
                    return
                val = val.replace(NEW_LINE_BREAK, "\n")
                self.errors.append(
                    (
                        idx_line + 1,
                        idx_col + 1,
                        str(self.ERRORS["regex_not_match"]) + val,
                    )
                )
                c_row.append("")
                return
            val_group = []
            for g in formater.regexp.findall(val):
                if isinstance(g, (tuple, list)):
                    g = "".join(g)
                val_group.append(g.replace(NEW_LINE_BREAK, "\n") if g else "")
            val = "".join(val_group)

        field_names = formater.field_name
        if not isinstance(field_names, (list, tuple)):
            field_names = [field_names]

        c_values = []
        for idx_fields, field_name in enumerate(field_names):
            func = formater.formater
            if type(func) in (list, tuple):
                func = func[idx_fields]
            if not callable(func) and type(func) == str:
                func = getattr(self, func)

            values = [val]
            many_values = getattr(func, "many_split", None)
            if many_values:
                values = re.split(func.many_split, values[0])
                # filter empty entries on m2m such as "my-value & "
                if len(values) > 1:
                    values = [v for v in values if v.strip()]
                    if not values:
                        # keep an empty value if there is only empty values
                        values = [values[0]]
            formated_values = []
            field_name = formater.field_name
            force_new = formater.force_new
            if type(field_name) in (list, tuple):
                field_name = field_name[idx_fields]
            if type(force_new) in (list, tuple):
                force_new = force_new[idx_fields]
            if formater.concat:
                self.concats.add(field_name)
            concat_str = formater.concat_str
            if type(formater.concat_str) in (list, tuple):
                concat_str = concat_str[idx_fields]
            if concat_str:
                self.concat_str[field_name] = concat_str

            if self.DB_TARGETS:
                formater.import_instance = self.import_instance
                formater.reinit_db_target(
                    self.DB_TARGETS["{}-{}".format(idx_col + 1, field_name)], idx_fields
                )
            for idx, v in enumerate(values):
                try:
                    """
                    # disable regex cut
                    if formater.regexp_formater_args:
                        args = []
                        for idx in formater.regexp_formater_args[idx_fields]:
                            args.append(val_group[idx])
                        value = func.format(*args)
                    else:
                    """
                    if getattr(func, "need_archive", False):
                        value = func.format(v, archive=self.archive)
                    else:
                        value = func.format(v)
                except ValueError as e:
                    if formater.required:
                        self.c_errors = True
                    self.errors.append((idx_line + 1, idx_col + 1, str(e)))
                    c_values.append("")
                    return
                if formater.value_format and value is not None and value != "":
                    value = formater.value_format.format(value)
                if hasattr(value, "close"):
                    # opened file are kept - they need to be listed and closed
                    # later
                    self.to_be_close.append(value)
                formated_values.append(value)
            if hasattr(func, "match_table"):
                if field_name not in self.match_table:
                    self.match_table[field_name] = {}
                self.match_table[field_name].update(func.match_table)

            value = formated_values
            if not many_values:
                value = formated_values[0]

            printed_values = value
            if type(value) not in (list, tuple):
                printed_values = [value]
            try:
                # don't reunicode - unicoded values
                c_values.append(" ; ".join([v for v in printed_values]))
            except TypeError:
                c_values.append(" ; ".join([str(v) for v in printed_values]))
            if value is None and formater.required:
                self.c_errors = True
                self.errors.append(
                    (idx_line + 1, idx_col + 1, self.ERRORS["value_required"])
                )
                return

            field_names = [field_name]
            force_news = [force_new]
            concats = formater.concat
            concat_str = [concat_str]
            if idx_fields == 0:
                # duplicate fields are only for the first occurrence
                for duplicate_field in formater.duplicate_fields:
                    if type(duplicate_field[0]) in (list, tuple):
                        duplicate_field, force_new, concat, conc_str = duplicate_field[
                            0
                        ]
                    else:
                        duplicate_field, force_new, concat, conc_str = duplicate_field
                    field_names += [duplicate_field]
                    force_news += [force_new]
                    concats += [concat]
                    concat_str += [conc_str]

            if formater.through:
                self._throughs.append((formater, value))
            else:
                for idx, f_name in enumerate(field_names):
                    self._field_name_to_data_dict(
                        f_name,
                        value,
                        data,
                        formater.force_value,
                        force_new=force_news[idx],
                        concat=concats[idx],
                        concat_str=concat_str[idx],
                    )
        c_row.append(" ; ".join([v for v in c_values]))

    def _get_field_m2m(self, attribute, data, c_path, new_created, field_object):
        """
        Manage and m2m field from raw data

        :param attribute: attribute name
        :param data: current data dictionary
        :param c_path: attribute path from the main model point of view
        :param new_created: dict of forced newly created items to prevent
        multiple creation
        :param field_object: django field object for this attribute
        :return: None
        """
        m2ms = []

        many_values = data.pop(attribute)
        model = None
        if hasattr(field_object, "rel"):
            model = field_object.rel.to
        elif hasattr(field_object, "related_model"):
            model = field_object.related_model
        elif hasattr(field_object, "to"):
            model = field_object.to
        elif hasattr(field_object, "model"):
            model = field_object.model
        if type(many_values) not in (list, tuple):
            many_values = [many_values]

        for val in many_values:
            if val.__class__ == model:
                # the value is a model instance: it is OK!
                m2ms.append((attribute, val))
                continue
            if type(val) != dict:
                # value is not a dict, we don't know what to do with it...
                continue
            vals = []

            # contruct many dict for each values
            default_dict = {}

            # # get default values
            p = [attribute]
            if c_path:
                p = list(c_path) + p
            p = tuple(p)
            if p in self._defaults:
                for k in self._defaults[p]:
                    default_dict[k] = self._defaults[p][k]

            # # init with simple values that will be duplicated
            for key in val.keys():
                if type(val[key]) not in (list, tuple):
                    default_dict[key] = val[key]
            vals.append(default_dict.copy())

            # # manage multiple values
            for key in list(val.keys()):
                if type(val[key]) in (list, tuple):
                    for idx, v in enumerate(val[key]):
                        if len(vals) <= idx:
                            vals.append(default_dict.copy())
                        vals[idx][key] = v

            # check that m2m are not empty
            notempty = False
            for dct in vals:
                for k in dct:
                    if dct[k] not in ("", None):
                        notempty = True
                        break
            if not notempty:
                continue

            field_names = get_all_field_names(model)
            for v in vals:
                if "history_modifier" in field_names:
                    if "defaults" not in v:
                        v["defaults"] = {}
                    v["defaults"]["history_modifier"] = self.history_modifier
                m2m_m2ms = []
                c_c_path = c_path[:]
                for k in list(v.keys()):
                    if k not in field_names:
                        continue
                    self.get_field(model, k, v, m2m_m2ms, c_c_path, new_created)
                if "__force_new" in v:
                    created = v.pop("__force_new")
                    key = ";".join(["{}-{}".format(k, v[k]) for k in sorted(v.keys())])
                    # only one forced creation
                    if attribute in new_created and key in new_created[attribute]:
                        continue
                    if attribute not in new_created:
                        new_created[attribute] = []
                    new_created[attribute].append(key)
                    has_values = bool([1 for k in v if v[k] and k != "defaults"])
                    if has_values:
                        if (
                            self.MODEL_CREATION_LIMIT
                            and model not in self.MODEL_CREATION_LIMIT
                        ):
                            raise self._get_improperly_conf_error(model)
                        if "defaults" in v:
                            default_values = v.pop("defaults")
                            for k in default_values.keys():
                                if k not in v:
                                    v[k] = default_values[k]
                        v = model.objects.create(**v)
                    else:
                        continue
                else:
                    v["defaults"] = v.get("defaults", {})
                    extra_fields = {}
                    # "File" type is a temp object and can be different
                    # for the same filename - it must be treated
                    # separately
                    for field in model._meta.fields:
                        k = field.name
                        # attr_class is a FileField attribute
                        if hasattr(field, "attr_class") and k in v:
                            extra_fields[k] = v.pop(k)
                    created = False
                    if (
                        not self.MODEL_CREATION_LIMIT
                        or model in self.MODEL_CREATION_LIMIT
                    ):
                        try:
                            v, created = model.objects.get_or_create(**v)
                        except FieldError as e:
                            raise ImporterError(
                                str(_("Importer configuration error: " '"{}".')).format(
                                    e
                                )
                            )
                        except Exception as e:
                            msg = str(_('Import error: {} - "{}".')).format(model, e)
                            raise ImporterError(msg)
                    else:
                        get_v = v.copy()
                        if "defaults" in get_v:
                            get_v.pop("defaults")
                        try:
                            v = model.objects.get(**get_v)
                        except model.DoesNotExist:
                            raise self._get_does_not_exist_in_db_error(model, get_v)
                    changed = False
                    for k in extra_fields.keys():
                        if extra_fields[k]:
                            changed = True
                            setattr(v, k, extra_fields[k])
                    if changed:
                        v.save()
                for att, objs in m2m_m2ms:
                    if type(objs) not in (list, tuple):
                        objs = [objs]
                    for obj in objs:
                        getattr(v, att).add(obj)
                if self.import_instance and hasattr(v, "imports") and created:
                    v.imports.add(self.import_instance)
                m2ms.append((attribute, v))
        return m2ms

    def _set_importer_trigger(self, cls, attribute, data):
        """
        An importer trigger is used. Stock it for later execution and remove
        it from current data dict.

        :param cls: current model
        :param attribute: attribute name
        :param data: current data dictionary
        :return: None
        """
        func = getattr(cls, attribute)
        if func.importer_trigger == "pre":
            func(data, data[attribute])
        elif func.importer_trigger == "post":
            self._item_post_processing.append([attribute, data, data[attribute]])
        else:
            logger.warning(
                "Unknow importer_trigger '{}' for '{}'".format(
                    func.importer_trigger, attribute
                )
            )
        data.pop(attribute)

    def get_field(self, cls, attribute, data, m2ms, c_path, new_created):
        """
        Get field from raw data

        :param cls: current model
        :param attribute: attribute name
        :param data: current data dictionary
        :param m2ms: many to many list of tuple: (m2m key, m2m value)
        :param c_path: attribute path from the main model point of view
        :param new_created: dict of forced newly created items to prevent
        multiple creation
        :return: None
        """
        if hasattr(cls, attribute) and getattr(
            getattr(cls, attribute), "importer_trigger", None
        ):
            # importer trigger
            self._set_importer_trigger(cls, attribute, data)
            return
        if attribute == "data":  # json field
            # no need to do anything
            return
        if attribute == "get_default":
            # force evaluation of default value for this field
            return
        try:
            field_object = cls._meta.get_field(attribute)
        except FieldDoesNotExist:
            raise ImporterError(
                str(
                    _(
                        'Importer configuration error: field "{}" does not exist '
                        "for {}."
                    )
                ).format(attribute, cls._meta.verbose_name)
            )
        if field_object.many_to_many:
            try:
                m2ms += self._get_field_m2m(
                    attribute, data, c_path, new_created, field_object
                )
            except Exception as e:
                self.errors.append((self.idx_line, None, str(e)))
            return
        if not hasattr(field_object, "rel") or not field_object.rel:
            return
        if type(data[attribute]) == list:
            # extract the first item from list
            # be careful if the list has more than one item this is arbitrary
            if len(data[attribute]) > 1:
                logger.warning(
                    "Import {}: {} has many when only one is expected. Get "
                    "the first one but it is not OK!".format(
                        self.import_instance, attribute
                    )
                )
            data[attribute] = data[attribute][0]
            return
        if not isinstance(data[attribute], dict):
            # we treat only dict formated values
            return
        # put history_modifier for every created item
        if "history_modifier" in get_all_field_names(field_object.rel.to):
            data[attribute]["history_modifier"] = self.history_modifier
        try:
            c_path.append(attribute)
            data[attribute], created = self.get_object(
                field_object.rel.to, data[attribute].copy(), c_path
            )
        except ImporterError as msg:
            self.errors.append((self.idx_line, None, msg))
            data[attribute] = None

    def get_object(self, cls, data, path=None):
        if not path:
            path = []
        m2ms = []
        if type(data) != dict:
            # if data is not a dict we don't know what to do
            return data, False

        is_empty = not bool(
            [k for k in data if k not in ("history_modifier", "defaults") and data[k]]
        )
        if is_empty:
            # if no value, no creation
            return None, False

        c_path = path[:]

        # get all related fields
        new_created = {}
        try:
            for attribute in list(data.keys()):
                c_c_path = c_path[:]
                if attribute not in data:  # removed by previous get_field
                    continue
                if not attribute:
                    data.pop(attribute)
                    continue
                if not data[attribute]:
                    if hasattr(cls, attribute) and getattr(
                        getattr(cls, attribute), "importer_trigger", None
                    ):
                        data.pop(attribute)
                        continue
                    field_object = cls._meta.get_field(attribute)
                    if field_object.many_to_many:
                        data.pop(attribute)
                    continue
                if attribute != "__force_new":
                    self.get_field(cls, attribute, data, m2ms, c_c_path, new_created)
        except (ValueError, IntegrityError, FieldDoesNotExist) as e:
            try:
                message = str(e)
            except (UnicodeDecodeError, UnicodeDecodeError):
                message = ""
            try:
                data = str(data)
            except UnicodeDecodeError:
                data = ""
            raise ImporterError(
                "Erreur d'import %s %s, contexte : %s, erreur : %s"
                % (str(cls), str("__".join(path)), str(data), message)
            )

        # image field is not serialized
        image, associated_file = None, None
        if "image" in data and data["image"]:
            image = data.pop("image")
        if "associated_file" in data and data["associated_file"]:
            associated_file = data.pop("associated_file")
        create_dict = copy.deepcopy(data)
        if image:
            data["image"] = image
            create_dict["image"] = image
        if associated_file:
            data["associated_file"] = associated_file
            create_dict["associated_file"] = associated_file

        for k in list(create_dict.keys()):
            # filter unnecessary default values but not the json field
            if isinstance(create_dict[k], dict) and k != "data":
                if self.simulate:
                    create_dict[k] = _("* created *")
                else:
                    create_dict.pop(k)
            # File doesn't like deepcopy
            elif type(create_dict[k]) == File:
                create_dict[k] = copy.copy(data[k])

        # default values
        path = tuple(path)
        defaults = {}
        if hasattr(cls, "get_import_defaults"):
            defaults = cls.get_import_defaults()
        if path in self._defaults:
            for k in self._defaults[path]:
                if (k not in data or not data[k]) and self._defaults[path][k]:
                    defaults[k] = self._defaults[path][k]

        if "history_modifier" in create_dict:
            defaults.update({"history_modifier": create_dict.pop("history_modifier")})

        created = False
        post_save_keys = []
        try:
            try:
                dct = {}
                if hasattr(cls, "get_import_defaults"):
                    dct = cls.get_import_defaults() or {}
                dct.update(create_dict.copy())
                for key in dct:
                    if callable(dct[key]):
                        dct[key] = dct[key]()
                        if getattr(dct[key], "post_save", True):
                            dct.pop(key)
                            post_save_keys.append(key)
                if "__force_new" in dct:
                    created = dct.pop("__force_new")
                    if not [k for k in dct if dct[k] is not None]:
                        return None, created
                    new_dct = defaults.copy()
                    new_dct.update(dct)
                    if (
                        self.MODEL_CREATION_LIMIT
                        and cls not in self.MODEL_CREATION_LIMIT
                    ):
                        raise self._get_improperly_conf_error(cls)
                    if not self.simulate:
                        obj = cls.objects.create(**new_dct)
                    else:
                        self.new_objects.append((path, cls, new_dct))
                else:
                    # manage UNICITY_KEYS - only level 1
                    if not path and self.UNICITY_KEYS:
                        for k in list(dct.keys()):
                            if k not in self.UNICITY_KEYS and k != "defaults":
                                if dct[k]:
                                    defaults[k] = dct.pop(k)
                                else:
                                    dct.pop(k)
                    if "get_default" in dct and dct["get_default"]:
                        dct.pop("get_default")
                        new_dct = defaults.copy()
                        new_dct.update(dct)
                        dct = new_dct

                    obj = None
                    if self.simulate:
                        q = cls.objects.filter(**dct)
                        if not q.count():
                            if (
                                self.MODEL_CREATION_LIMIT
                                and cls not in self.MODEL_CREATION_LIMIT
                            ):
                                self.not_find_objects.append((path, cls, dct))
                                return _("* match not find *"), False
                            dct.update(defaults)
                            self.new_objects.append([path, cls, dct])
                            created = True
                        elif q.count() > 1:
                            self.ambiguous_objects.append((path, list(q.all()), dct))
                            if q.count() > 10:
                                return (
                                    _("* the query match more than 10 " "results*"),
                                    False,
                                )
                            else:
                                return (
                                    str(_(" or ")).join(
                                        [str(item) for item in q.all()]
                                    ),
                                    False,
                                )
                        else:
                            self.updated_objects.append([path, q.all()[0], dct, {}])
                            dct["defaults"] = defaults.copy()
                    else:
                        if not dct and not defaults:
                            obj = None
                        else:
                            if (
                                not self.MODEL_CREATION_LIMIT
                                or cls in self.MODEL_CREATION_LIMIT
                            ):
                                dct["defaults"] = defaults.copy()
                                obj, created = cls.objects.get_or_create(**dct)
                            else:
                                try:
                                    obj = cls.objects.get(**dct)
                                    dct["defaults"] = defaults.copy()
                                except cls.DoesNotExist:
                                    raise self._get_does_not_exist_in_db_error(cls, dct)

                    if not created and not path and self.UNICITY_KEYS:
                        updated_dct = {}
                        if self.conservative_import:
                            for k in dct["defaults"]:
                                new_val = dct["defaults"][k]
                                if new_val is None or new_val == "":
                                    continue
                                val = getattr(obj, k)
                                if val is None or val == "":
                                    updated_dct[k] = new_val
                                elif (
                                    k in self.concats
                                    and type(val) == str
                                    and type(new_val) == str
                                ):
                                    updated_dct[k] = val + "\n" + new_val
                        else:
                            for k in dct["defaults"]:
                                new_val = dct["defaults"][k]
                                if new_val is None or new_val == "":
                                    continue
                                if obj and k == "data":
                                    updated_dct[k] = update_data(obj.data, new_val)
                                else:
                                    updated_dct[k] = new_val
                        if updated_dct:
                            if self.simulate:
                                self.updated_objects[-1][-1] = updated_dct
                            else:
                                for k in updated_dct:
                                    setattr(obj, k, updated_dct[k])
                                obj.save()
                if (
                    not self.simulate
                    and self.import_instance
                    and hasattr(obj, "imports")
                    and created
                ):
                    obj.imports.add(self.import_instance)
            except (ValueError, IntegrityError, DatabaseError, GEOSException) as e:
                raise IntegrityError(str(e))
            except cls.MultipleObjectsReturned as e:
                created = False
                if "defaults" in dct:
                    dct.pop("defaults")
                raise IntegrityError(str(e))
                # obj = cls.objects.filter(**dct).all()[0]
            for key in post_save_keys:
                getattr(obj, key)()
            for attr, value in m2ms:
                values = [value]
                if type(value) in (list, tuple):
                    values = value
                if self.simulate:
                    if created:
                        obj_dct = self.new_objects[-1][-1]
                    else:
                        obj_dct = self.updated_objects[-1][-1]
                    obj_dct[attr] = values
                else:
                    for v in values:
                        related_model = getattr(obj, attr)
                        # an intermediary model is used
                        if (
                            hasattr(related_model, "through")
                            and not related_model.through._meta.auto_created
                        ):
                            # try to create it with default attributes
                            inter_model = related_model.through
                            target_name, item_name = None, None
                            for field in inter_model._meta.get_fields():
                                rel_model = getattr(field, "related_model", None)
                                # assume that the first found is correct...
                                if rel_model == v.__class__:
                                    target_name = field.name
                                elif rel_model == obj.__class__:
                                    item_name = field.name
                            assert target_name is not None
                            assert item_name is not None
                            inter_model.objects.get_or_create(
                                **{item_name: obj, target_name: v}
                            )
                        else:
                            getattr(obj, attr).add(v)
                        # force post save script
                        v = v.__class__.objects.get(pk=v.pk)
                        try:
                            v.save()
                        except DatabaseError as e:
                            raise IntegrityError(e.message)
            if self.simulate:
                # put m2m result in data dict
                current_data = data
                if m2ms:
                    for item in path:
                        if item not in current_data:
                            current_data[item] = {}
                        current_data = current_data[item]
                    for key, value in m2ms:
                        if not isinstance(value, list) and not isinstance(value, tuple):
                            value = [value]
                        current_data[key] = value

                if created:
                    return dct, True
                else:
                    # defaults are not presented as matching data
                    dct.pop("defaults")
                    return self.updated_objects[-1][1], False

            if m2ms:
                # force post save script
                obj.save()
            if hasattr(obj, "fix"):
                # post save/m2m specific fix
                obj.fix()
        except IntegrityError as e:
            try:
                message = str(e)
            except (UnicodeDecodeError, UnicodeDecodeError):
                message = ""
            try:
                data = str(data)
            except UnicodeDecodeError:
                data = ""
            raise ImporterError(
                "Erreur d'import %s %s, contexte : %s, erreur : %s"
                % (str(cls), str("__".join(path)), str(data), message)
            )
        return obj, created

    def _format_csv_line(self, values, empty="-"):
        return (
            '"'
            + '","'.join([(v and str(v).replace('"', '""')) or empty for v in values])
            + '"'
        )

    def _get_csv(self, rows, header=None, empty="-"):
        if not rows:
            return ""
        if not header:
            header = []
        csv_v = []
        if header:
            csv_v.append(self._format_csv_line(header, empty=empty))
        for values in rows:
            csv_v.append(self._format_csv_line(values, empty=empty))
        return "\n".join(csv_v)

    def get_csv_errors(self):
        return self._get_csv(self.errors, header=[_("line"), _("col"), _("error")])

    def get_csv_result(self):
        return self._get_csv(self.validity)

    def get_csv_matches(self):
        header = [_("field"), _("source"), _("result")]
        values = []
        for field in self.match_table:
            for source in self.match_table[field]:
                values.append((field, source, self.match_table[field][source]))
        return self._get_csv(values, header=header)

    @classmethod
    def choices_check(cls, choices):
        def function(value):
            choices_dct = dict(choices)
            value = value.strip()
            if not value:
                return
            if value not in choices_dct.values():
                raise ValueError(
                    _('"%(value)s" not in %(values)s')
                    % {
                        "value": value,
                        "values": ", ".join([val for val in choices_dct.values()]),
                    }
                )
            return value

        return function