Filtering by Vocabulary-Specific Keywords

The ProjectFilter class supports creating multiple keyword filters, each tied to a specific vocabulary. This allows users to filter projects by different types of controlled keywords (e.g., science keywords, platforms, instruments) rather than having a single generic “Keywords” filter.

Autocomplete for Large Vocabularies

For extremely large vocabularies (e.g., GCMD Science Keywords with thousands of concepts), FairDM uses django-autocomplete-light to efficiently load concepts on demand rather than loading entire querysets into the browser.

Concept Autocomplete View

The ConceptAutocomplete view at /autocomplete/concept/ provides filtered concept lookups:

# URL: /autocomplete/concept/?vocabulary=gcmd-science-keywords&q=ocean
# Returns: Concepts from the specified vocabulary matching the search term

Features:

• Filters by vocabulary name via URL parameter or forwarded field

• Searches concept labels and names

• Only authenticated users can access

• Returns results ordered by label

Usage in custom widgets:

from dal import autocomplete
from django import forms

field = forms.ModelMultipleChoiceField(
    queryset=Concept.objects.all(),
    widget=autocomplete.ModelSelect2Multiple(
        url='concept-autocomplete?vocabulary=gcmd-science-keywords',
        attrs={
            'data-placeholder': 'Select keywords...',
            'data-minimum-input-length': 2,
        }
    )
)

Keyword Form Configuration

How It Works

The ProjectFilter base class has a KEYWORD_VOCABULARIES configuration attribute that accepts a list of vocabulary specifications. For each vocabulary specified, the filter automatically creates a separate filter field that:

1. Only shows concepts from that specific vocabulary

2. Only shows concepts that are actually used by projects

3. Filters the keywords relationship on the Project model

Basic Usage

To add vocabulary-specific filters, create a custom filter class that extends ProjectFilter and specify the vocabularies you want:

from django.utils.translation import gettext_lazy as _
from fairdm.core.project.filters import ProjectFilter


class MyProjectFilter(ProjectFilter):
    """Custom project filter with vocabulary-specific keyword filters."""
    
    KEYWORD_VOCABULARIES = [
        ("gcmd-science-keywords", _("Science Keywords")),
        ("gcmd-platforms", _("Platforms")),
        ("gcmd-instruments", _("Instruments")),
    ]

This will automatically create three new filter fields:

• keywords_gcmd_science_keywords - labeled “Science Keywords”

• keywords_gcmd_platforms - labeled “Platforms”

• keywords_gcmd_instruments - labeled “Instruments”

How Filter Names Are Generated

The filter field name is generated by:

1. Prefixing with keywords_

2. Converting the vocabulary name to a valid Python identifier (replacing - with _)

For example:

• "gcmd-science-keywords" → keywords_gcmd_science_keywords

• "my-custom-vocab" → keywords_my_custom_vocab

Finding Vocabulary Names

Vocabulary names in the database correspond to the name field on vocabulary objects. To find available vocabularies:

from research_vocabs.models import Vocabulary

# List all vocabularies
for vocab in Vocabulary.objects.all():
    print(f"{vocab.name}: {vocab.title}")

Common GCMD vocabularies include:

• gcmd-science-keywords

• gcmd-platforms

• gcmd-instruments

• gcmd-locations

• gcmd-projects

Widget Customization

By default, vocabulary filters use CheckboxSelectMultiple widgets for multi-select functionality. The filters use OR logic, meaning projects matching ANY selected keyword will be returned.

If you need different widget behavior, you can override the filter creation in __init__:

class MyProjectFilter(ProjectFilter):
    KEYWORD_VOCABULARIES = [
        ("gcmd-science-keywords", _("Science Keywords")),
    ]
    
    def __init__(self, *args, **kwargs):
        super().__init__(*args, **kwargs)
        
        # Customize the widget for a specific vocabulary filter
        self.filters["keywords_gcmd_science_keywords"].widget = forms.SelectMultiple(
            attrs={"class": "form-select", "size": "10"}
        )

Empty State Handling

The filters automatically exclude concepts that aren’t used by any projects. This means if no projects use concepts from a vocabulary, that filter will have an empty queryset and won’t show any options.

Complete Example

# myapp/filters.py
from django.utils.translation import gettext_lazy as _
from fairdm.core.project.filters import ProjectFilter


class GeoscienceProjectFilter(ProjectFilter):
    """Project filter configured for geoscience research portals."""
    
    KEYWORD_VOCABULARIES = [
        ("gcmd-science-keywords", _("Science Keywords")),
        ("gcmd-platforms", _("Observation Platforms")),
        ("gcmd-instruments", _("Instruments")),
        ("gcmd-locations", _("Locations")),
    ]

Then use it in your view configuration or registry:

# myapp/config.py
from .filters import GeoscienceProjectFilter

PROJECT_FILTER = GeoscienceProjectFilter

Performance Considerations

The filter queries are optimized to:

• Only load concepts from specified vocabularies

• Only load concepts actually used by projects (via projects__isnull=False)

• Use distinct() to avoid duplicate results

For large vocabularies, consider adding database indexes on the Concept model’s vocabulary_id field if not already present.