Hi @bmihelac,

Related issues:

• #64
• #176
• #415
• #601

Here's where I've got to so far. I thought it would be good to get some feedback before getting stuck into writing new tests, but the current ones are passing, so at least nothing is broken :)

When enabled, instance.full_clean() is called for each row, and any resulting validation errors surface themselves in the preview table like so:

The errors are hidden visually at first, then become visible when you hover over a row's first column, in such a way that it doesn't hide any of the values for the current row. It's pretty crude, but it's a starting point. Some examples:

Anyone wanting to give this a whirl for themselves should be able to install from git using:

pip install git+https://github.com/ababic/django-import-export.git@modelresource-validation-error-handling

Any/all feedback welcome :)

i try to import some data form csv file, but errors occurred as follows:

Errors Line number: 1 - u'id' Traceback (most recent call last): File "/home/zhijiasun/.virtualenvs/tutorial/local/lib/python2.7/site-packages/import_export/resources.py", line 317, in import_data instance, new = self.get_or_init_instance(instance_loader, row) File "/home/zhijiasun/.virtualenvs/tutorial/local/lib/python2.7/site-packages/import_export/resources.py", line 149, in get_or_init_instance instance = self.get_instance(instance_loader, row) KeyError: u'id'

I just modify the data of the csv file which exported by the app and then try to import it. But error happed.

Sir,thanks a lot to develop this great repository. I have little problem to solve , as below the code:

#in models.py
class Author(models.Model):
    author = models.CharField('作者', max_length=50)
    title = models.CharField('标题', max_length=150)
    qualification = models.ForeignKey(Qualification)
    mark = models.ManyToManyField(Mark)
    blog = models.TextField('博客内容')
    time = models.DateField('写作日期')

    def __unicode__(self):
        return unicode(self.author)

    class Meta:
        ordering = ['time']

#in admin.py
class AuthorAdmin(ImportExportModelAdmin, admin.ModelAdmin):
    search_fields = ('author', 'title', 'mark', 'blog')
    list_display = ('author', 'title', 'time')

When I export the data, but the csv file appear to be the retortion, when I modified the second row author data and import,

it cause the error,

how I it be smoothly modified and import？thanks. Allenlee

This is a continuation of @int-ua's work in #725, addressing #332. I've resolved conflicts, tweaked the documentation and tests, but otherwise kept it identical.

I need this feature too, so I'd like to get it merged sooner rather than later. However, there are other tweaks I'd like to make, including:

• Better documentation, including its own page in the Reference
• Format options:
  • arg to list available formats, calling into base_formats rather than being hardcoded
  • arg to specify format to use rather than guessing
• Detect whether supplied class is a Resource or Model rather than needing to specify
• raise_errors currently defaults to True if dry_run is False - is this the right thing to do?
• Option to supply kwargs to the Resource constructor (which I need for my own project)
• Hey, export would be nice too

... but let's start with this, and if I get time for those tweaks I'll make them in a separate PR.

Created a mixin view to export without admin interface.

I think it would be a really great thing if we have a generic class based views for this application.

I tested this patch with a ListView & django_filters.views.FilterView. The 2nd is really useful to replace the AdminFilters

Do many to many fields work via admin imports? I keep getting something like this: ValueError('"<Accession: >" needs to have a value for field "accession" before this many-to-many relationship can be used.',)

It doesn't seem like it can if it's displaying the data to import before importing, as there's no way to get the primary key for the row since it doesn't have one yet?

Thanks,

Dean

Problem

Overriding the forms for the import process is currently more difficult than it could be, and there's a bit of 'difficult to follow' stuff going on in the current implementation - particularly in how get_form_kwargs() is used to create the confirmation form.

Solution

First of all, new import_form_class and confirm_form_class attributes allow you to easily specify custom form classes for the import process without having to override methods.

The current get_import_form() and get_import_confirm_form() methods have been deprecated in favour of using the more descriptively named get_import_form_class() and get_confirm_form_class() methods.

The single get_form_kwargs() method has been deprecated in favour of adding separate methods for each form:

• get_import_form_kwargs(request, form_class, **kwargs)
• get_confirm_form_kwargs(request, form_class, import_form=None, **kwargs)

And to more clearly separate the logic for customising 'intial data' and 'init kwargs', there are separate dedicated methods for each form:

• get_import_form_initial(request, form_class, **kwargs)
• get_confirm_form_initial(request, form_class, import_form=None, **kwargs)

Last but not least, the entire business of form instantiation has been factored out into two new methods:

• create_import_form(request, **kwargs)
• create_confirm_form(request, import_form=None, **kwargs)

These should make it super convenient for modifying forms on the fly.

Paired with the changes from #1038, this should make for a much more developer-friendly implementation.

To do

• [x] Update documentation to reflect changes
• [x] Write tests to cover new behaviour

Describe the bug Django DateTime Import

To Reproduce Import DateTime data and you'll hit a "localtime() cannot be applied to a naive datetime", I've been using the package for a while now and use the admin import page as well as a custom import page for users. Both work fine for non-datetime data.

I've tried this with adding with tz +00 and without 2020-07-17 7:42:39 2020-07-17 7:42:39+00

I've tried the DateTimeWidget where format="%Y-%m-%d %H:%M:%S" format="%Y-%m-%d %H:%M:%S$z" format=None

I've posted a SOF question regarding it https://stackoverflow.com/questions/62990575/django-import-export-datetime-naive-datetime-error

And had a convo specific to it on SOF https://chat.stackoverflow.com/rooms/218260/django-import-export

The end result seems that the issue may be in the django-import-export package.

Versions (please complete the following information):

• django-import-export==2.2.0
• Python 3.7.4
• Django==3.0.2

settings.py

LANGUAGE_CODE = 'en-us'
TIME_ZONE = 'UTC'
USE_I18N = True
USE_L10N = True
USE_TZ = True

Expected behavior I was expecting the datetime to import the dates in my csv file

Screenshots View 2 links above

Additional context View 2 links above

  File "/Users/syabro/.envs/homeowners/lib/python2.7/site-packages/import_export/resources.py", line 16, in <module>
    from django.db.models.related import RelatedObject
ImportError: No module named related

because

The django.db.models.related module has been removed and the Field.related attribute will be removed in Django 2.0.

https://docs.djangoproject.com/en/dev/releases/1.8/#model-field-related

Running on:

• Django-1.6
• django-import-export-0.1.5
• postgresql.x86_64-9.2.5-1
• tablib-0.9.11
• python-2.7.5-9

I have successfully imported data from a csv file.

When I try to add more data manually, I receive the error:

duplicate key value violates unique constraint "mytable_pkey"
DETAIL:  Key (id)=(1) already exists.

From psql:

select max(id) from mytable;
  75

SELECT nextval('mytable_id_seq');
  2

Does django-import-export support the table's id auto-incrementing?

AFAIU it will help create widgets with autocreation of related ForeignKey instances that require two fields already present in the row. For example:

• FK to Person with first_name, last_name
• FK to Quote that requires a Book (let's suppose it's identifiable just by name) and that book is already in the import fields
• FK to POI that is identifiable with two (or even three) coordinates

Problem

ovveriding get_queryset not working when exporting from admin panel

Solution

How did you solve the problem?

add export_selected_related option to include when making the queryset

usage:

class MyResource(resources.ModelResource):
    export_selected_related = ('RELATED_MODEL1', 'RELATED_MODEL2', ...)

Acceptance Criteria

Have you written tests? Have you included screenshots of your changes if applicable? Did you document your changes?

dry_run is passed into import_row() but it is hit and miss as to whether this param (and other named kwargs) are passed to subsequent methods such as after_import_instance().

It would help users (e.g. this example) if these flags were available to other methods. I would propose that they are passed in kwargs rather than as named params, so as not to clutter the method definitions.

Describe the bug I tried to impliment admin class for Books table along with ImportExportModelAdmin and ImportExportActionModelAdmin. while exporting selected records from admin panel i got the error.

To Reproduce Steps to reproduce the behavior:

1. Go to admin.py and impliment custom model admin along with ImportExportModelAdmin and ImportExportActionModelAdmin
2. Got to respective table in admin panel. Select few records and export them as csv file
3. See error

class Base(models.Model):
    slug = models.SlugField(max_length = 200)

    class Meta:
        abstract = True


class Books(Base):
    name = models.CharField('Book name', max_length=100)
    author = models.ForeignKey(Author, blank=True, null=True)
    author_email = models.EmailField('Author email', max_length=75, blank=True)
    imported = models.BooleanField(default=False)
    published = models.DateField('Published', blank=True, null=True)
    price = models.DecimalField(max_digits=10, decimal_places=2, null=True, blank=True)
    categories = models.ManyToManyField(Category, blank=True)

    def __str__(self):
        return self.name


class BaseAdmin(admin.ModelAdmin):
    list_display = ('slug', )


class BooksResource(ModelResource):
    class Meta:
        model = Books


class BooksAdmin(ImportExportModelAdmin, ImportExportActionModelAdmin, BaseAdmin):
    resource_class = [BooksResource]
    list_display = BaseAdmin.list_display + ('name', 'author',  'published', 'price', 'categories', )


admin.site.register(Books, BooksAdmin)

Versions (please complete the following information):

• Django Import Export: [e.g. 3.0.1]
• Python [e.g. 3.9]
• Django [e.g. 4.1]

Expected behavior A clear and concise description of what you expected to happen.

Screenshots If applicable, add screenshots to help explain your problem.

Additional context Add any other context about the problem here.

The after_import_instance() is a misnomer because it is called after the instances is loaded (or created) and not after it is imported. Existing clients rely on this method so it would have to be deprecated correctly rather than simply renaming.

It would also be a good idea to pass the row in as a kwarg so that that users can use row data if they need to.

If a user needs to create related objects prior to import and those objects are identified by a combination of fields, then there isn't a 'clean' way of doing this. When the relation is identified by a single field, then we can override ForeignKeyWidget for this, or create data in before_import_row(). The import will skip any fields which are not present in the row, hence using an overridden ForeignKeyWidget won't work, but perhaps using ForeignKeyWidget in some way would be the best solution. It would have to be called if the field was not present in the row data.

In this related SO issue, I suggest overriding import_obj(), but it would be cleaner to override before_save_instance(). This could be done if the 'row' data is passed to before_save_instance().

Throughout the code base, the param name row is used to define the incoming row (even though this could be JSON or YAML). In import_obj() it is named data, which is less descriptive. We should standardise variable naming for better readability.