Regression test for #2552. Deletes the background of project 35 (owned by
testuser6) and then fetches the project to confirm the title is still
'Test35 with background'.
Fixes#2552. RemoveProjectBackground was passing a minimal Project struct
(only ID set) through UpdateProject, which always includes 'title' in its
Cols() list. This caused XORM to write the zero-value empty title to the
DB, wiping the real project title. Now uses ClearProjectBackground which
only updates background_file_id and background_blur_hash.
Register CachedAvatar and Photo with encoding/gob so Redis can properly
deserialize them. Migrate both to use RememberValue[T] which calls
GetWithValue() internally, fixing the broken type assertion when Redis
is the keyvalue backend.
Also removes the recursion-depth fallback in upload.go since
RememberValue eliminates the type mismatch failure mode entirely.
Instead of hardcoding the workbox version string in the service worker,
read it from workbox-precaching/package.json via Vite's define option.
This ensures the service worker always references the correct workbox
version that is actually installed.
Resolves#2549
Allow users to skip the first N data rows when importing CSV files.
This is useful when the CSV contains metadata rows before the actual
task data begins. Adds skip_rows to ImportConfig (backend) and a
number input in the parsing options UI (frontend).
Add a new CSV migration module that allows users to import tasks from
any CSV file with custom column mapping and parsing options.
Backend changes:
- New CSV migrator module with detection, preview, and import endpoints
- Auto-detection of delimiter, quote character, and date format
- Suggested column mappings based on column name patterns
- Transactional import using InsertFromStructure
Frontend changes:
- New CSV migration UI with two-step flow (upload -> mapping -> import)
- Column mapping selectors for all task attributes
- Live preview showing first 5 tasks with current mapping
- Parsing option controls for delimiter and date format
The CSV migrator creates a parent "Imported from CSV" project with
child projects based on the project column if provided, or a default
"Tasks" project for tasks without a specified project.
The docker/metadata-action uses github.sha for the SHA tag, which for
pull_request_target events is the base branch commit, not the PR head.
The comment step was independently constructing SHA tags from the PR
head SHA, causing preview URLs that didn't match any actual docker
image tag. Now reads the actual tags from docker meta output instead.
Syncs the sort choice to a ?sort=field:order URL parameter so it
survives page refreshes and can be shared. The default position sort
is omitted from the URL to keep links clean.
Previously only the "To-Do" default bucket was deleted, leaving "Doing"
and "Done" as duplicates alongside migration-provided buckets. Now all
default-created buckets are removed when migration data already provides
bucket assignments for all tasks.
Add comprehensive tests for the WeKan conversion function including
edge cases (empty board, orphan cards, color mapping, multiple
checklists, unsupported fields) and a realistic JSON fixture file.
Add a file-based migration importer that reads WeKan board JSON exports
and creates Vikunja projects with kanban buckets, tasks, labels,
checklists, and comments.
WeKan lists become kanban buckets. Checklists are converted to HTML
task lists in the description. Card descriptions and comments are
converted from markdown to HTML using goldmark. Label colors are
mapped from WeKan's CSS color names to their actual hex values.
Resolves#2490. Users with team access on a parent project were not seeing
subtask relations for tasks in child projects because getUserProjectsStatement
does not walk the project hierarchy. The fix wraps the base query in a
recursive CTE that traverses child projects via parent_project_id.
When closing a task modal opened from the Gantt view, the date range
query parameters were lost because closeModal() reconstructed the
route with only projectId and viewId. Now preserves query parameters
from the backdrop view.
Replace static initialDateRange snapshot with reactive filters.value
references inside the computed flatPickerConfig. This ensures the
Flatpickr defaultDate always reflects the current user-selected range
instead of the mount-time values.
Ref #2462
kolaente
dependabot[bot]
renovate[bot]
Frederick [Bot]
Claude
surfingbytes