[anantjain341]

Backend for "Pagination and filter implementation" (https://github.com/CivicDataLab/DataSpace/issues/277): use cases (https://github.com/CivicDataLab/DataSpace/issues/278), collaboratives (https://github.com/CivicDataLab/DataSpace/issues/279), AI models (https://github.com/CivicDataLab/DataSpace/issues/280) and publications (https://github.com/CivicDataLab/DataSpace/issues/287).

Adds four new queries for the dashboard tables, built the same way as datasetsTable (#215):
useCasesTable, collaborativesTable, aiModelsTable, publicationsTable.
The existing useCases, collaboratives, aiModels and publications queries are untouched, so nothing that runs today changes.

What you get in one call

A page of rows, the total count, and the per-status numbers for the tabs. All four look the same:

useCasesTable(filters, sortOptions, limit, offset, includePublic) {
  totalItemsCount                     # total after filters, for "Showing 1–20 of 57"
  statusCounts { status count }       # for the tab labels: Drafts (3), Published (6)
  data { ...any use case field... }
}

data rows are the same types the old queries return (TypeUseCase, TypeCollaborative, TypeAIModel, TypePublication), so select whatever the table shows.

Mapping the current table UI

The four dashboards have the same layout: two tabs and a table.

On screen	Send / read
Tabs Drafts / Published	a status filter: DRAFT / PUBLISHED. For AI models: REGISTERED (Draft tab) / ACTIVE (Published tab). Numbers for the tab labels come from statusCounts
Title column (use cases, collaboratives, publications)	title
Display Name / Type / Provider (AI models)	displayName, modelType, provider
Date Created / Date Modified	created, modified (AI models: createdAt, updatedAt)
Delete / Unpublish buttons	need id (and slug where the link uses it), unchanged
Sorting a column (today fixed to modified, newest first)	sortOptions: [{field: "modified", direction: "desc"}] (AI models: updated_at)
A search box, if added	filter title with icontains (AI models: display_name)
Showing x–y of z, Previous / Next, if added	limit, offset, totalItemsCount

Publications today fetch 100 rows and split the tabs in the browser, so anything past 100 never shows. With publicationsTable the status filter, paging and tab counts all come from the server.

Filters

A list of filters, each: which field, how to compare, what value.
All filters are combined with AND. The value is always a string.

{"field": "title",          "condition": "icontains", "value": "health"}
{"field": "status",         "condition": "in",        "value": "DRAFT,PUBLISHED"}
{"field": "sectors__slug",  "condition": "in",        "value": "health,education"}
{"field": "created",        "condition": "gte",       "value": "2026-01-01"}

Fields you can filter on:

useCasesTable

• title, status, running_status, id
• organization, organization__slug
• sectors__id, sectors__slug, sectors__name
• geographies__id, geographies__name, geographies__code
• sdgs__code, sdgs__slug
• tags__value
• started_on, completed_on, created, modified (ISO date, e.g. 2026-01-01)

collaborativesTable: the same as use cases, without running_status.

aiModelsTable

• display_name, name, status, model_type, provider, domain, id
• organization, organization__slug
• sectors__id, sectors__slug, sectors__name
• geographies__id, geographies__name, geographies__code
• tags__value
• created_at, updated_at (ISO date)

publicationsTable

• title, status, license, id
• resource_type (id), resource_type__slug
• organization, organization__slug
• sectors__id, sectors__slug, sectors__name
• geographies__id, geographies__name, geographies__code
• publication_date, created, modified (ISO date)

Field names are the model's names in snake_case (display_name, created_at), as in datasetsTable. Filtering on several sectors, geographies or tags never returns a row twice.

Ways to compare:

• exact (default), iexact — equals
• contains, icontains — text search (use icontains for a search box)
• startswith, endswith and their i versions
• gt, gte, lt, lte — for dates and numbers
• in — any of a comma-separated list
• isnull — "true" or "false"

Anything outside this list returns a GraphQL error that names the allowed fields.
Need another column in the filter panel? It is a one-line change on the backend; ask.

Sorting

"sortOptions": [{"field": "modified", "direction": "desc"}]

Fields:

• useCasesTable: title, created, modified, status, running_status, started_on, completed_on
• collaborativesTable: title, created, modified, status, started_on, completed_on
• aiModelsTable: display_name, name, created_at, updated_at, status, model_type, provider
• publicationsTable: title, created, modified, status, publication_date, download_count

Several keys are allowed. Default when omitted: newest first (created, AI models created_at). The dashboards today sort by last modified, so send that sort to keep the same order.

Pagination

• limit = page size, 1 to 100 (default 50)
• offset = rows to skip
• pages = ceil(totalItemsCount / limit), offset = (page − 1) × limit

statusCounts is computed with every filter applied except the status filter, so the tab numbers do not change when switching tabs.

Who sees what

Same rules as each module's current list query. Send the usual headers: Authorization: Bearer <token> and, inside an organisation, organization: <slug>.

• In an organisation: that organisation's rows.
• Logged in, no organisation: your own rows (superuser sees all). For AI models and publications: your own rows that have no organisation.
• Not logged in: published rows (AI models: public and active ones). This differs from datasetsTable, which returns nothing here; it follows what useCases, collaboratives, aiModels and publications already do.
• includePublic: true adds every published row (AI models: public and active).

Notes on statuses:

• Use cases and collaboratives can also be ARCHIVED. The current design has no tab for them, so they show on neither tab, the same as today.
• AI models have more statuses (DRAFT, VALIDATING, AUDITING, APPROVED, FLAGGED, DEPRECATED). statusCounts lists every status that has rows, so a later Approved tab only needs the filter.

Full example

query UseCasesTable($filters: [FilterSpec!], $sort: [SortSpec!], $limit: Int, $offset: Int) {
  useCasesTable(filters: $filters, sortOptions: $sort, limit: $limit, offset: $offset) {
    totalItemsCount
    statusCounts { status count }
    data { id title slug status created modified }
  }
}

{
  "filters": [{"field": "status", "condition": "exact", "value": "DRAFT"}],
  "sort": [{"field": "modified", "direction": "desc"}],
  "limit": 20,
  "offset": 0
}

AI models, Published tab:

aiModelsTable(
  filters: [{field: "status", value: "ACTIVE"}]
  sortOptions: [{field: "updated_at", direction: "desc"}]
  limit: 20
) {
  totalItemsCount
  statusCounts { status count }
  data { id displayName modelType provider createdAt updatedAt }
}

Errors

Ordinary GraphQL errors, for example:

• Cannot filter on 'xyz'. Allowed fields: …
• limit (500) must be between 1 and 100
• offset (-1) must be 0 or more
• A value of the wrong type (text for an id or a date) returns Django's validation message, the same as datasetsTable today.

No migration, no new dependency. Helpers live in api/utils/qs_utils.py. The status counting is now one shared helper there (status_counts), used by these four tables and by datasetsTable (second commit; it keeps its DatasetStatusCount type, and the GraphQL schema is unchanged).

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!