* fix: fallback to english challenges
All challenges will use the english version if a translated file is not
available. SHOW_NEW_CURRICULUM still gates what's shown in the client.
* refactor: use closures to simplify createChallenge
* refactor: remove messy destructure
* refactor: add meta via helper
* fix: fallback to [] for meta.required
* fix: repair challenge.block
* refactor: use CONST_CASE for meta + challenge dirs
* fix: catch empty superblocks immediately
* fix: clean up path.resolves
* fix: invalid syntax in JS project steps
* fix: default to english comments and relax tests
Instead of always throwing errors when a comment is not translated, the
tests now warn while SHOW_UPCOMING_CHANGES is true, so that tests will
pass while we're developing and allow translators time to work.
They still throw when SHOW_UPCOMING_CHANGES is false to catch issues
in production
* test: update createCommentMap test
* refactor: delete stale comment
* refactor: clarify validate with explanatory consts
* feat: throw if audited cert falls back to english
* fix: stop testing upcoming localized curriculum
* feat: add function to map from path to superblock
* fix: use superblocks as keys in curriculum.json
* fix: handle new superblock names
* fix: keep deleting relational-database
* refactor: DRY up certification and ProjectModal
* fix: use sensible keys for SolutionViewer
* refactor: handle legacy solutions like new ones
* refactor: correct CompletedChallenge type
* fix: store challengeType for multifile projects
* fix: use challengeType to set display type
* feat: use dropdown to display project + code
* refactor: isOpen -> showCode to avoid a clash
We need to be able both show the code and show the completed project
* refactor: remove redundant parts of projectPreview
* refactor: fix project preview types
* feat: wip, using existing modal to show project
* feat: show projects on timeline
* feat: display projects on time-line
* chore: use consistent case for GitHub
* fix(a11y): translate show solution/view
* refactor: rename showFilesSolution
* refactor: use self-closing tag
* fix: remove hardcoding (certification + timeline)
* fix: remove hardcoding (settings)
* test: supply store and mock ga
* fix: include challengeType for projects
Co-authored-by: Tom <20648924+moT01@users.noreply.github.com>
* refactor: remove space
Co-authored-by: Tom <20648924+moT01@users.noreply.github.com>
* fix: key -> filekey on challenge submission
* fix: handle submissions without files
Co-authored-by: Tom <20648924+moT01@users.noreply.github.com>
* refactor: light tweaks for readability
* refactor: simplify metadata functions
* fix: most tests
* test: fix utils tests
* test: simplify mocks
* WIP: update get-last-step-file-content
* feat: finish create-next-step
* fix: type error
* test: provide mock meta.json for test
* refactor: get meta path from project path
* refactor: get project name from path
* refactor: simplify getProjectMetaPath further
Also removes some excessive mocking
* refactor: remove more mocks, always clear .env
* feat: update create-next-step
* feat: update create-empty steps
Also refactors slightly, so it's easier to insert steps into the meta
* docs: update challenge-helper-script docs
* feat: create-step-between
* refactor: allow metadata parse errors to propagate
* fix: convert reorderSteps to renameSteps
* refactor: create-step-between -> insert-step
* feat: update delete-step
* refactor: consolidate commands into commands.ts
* refactor: clean up and consolidation
* refactor: more cleanup
* fix: make cli args consistent
Everything accepts a single integer and nothing else
* refactor: renameSteps -> updateStepTitles
* docs: update with the names and args
* feat: add step validating meta + files are synced
* chore: remove /refetch-user-completed-challenges
* chore: remove /update-my-theme
Theme changes are handled by /update-user-flag
I left 'update-my-theme' in the blocklist to allow for future use.
* chore: remove /certificate/verify-can-claim-cert
The client has not been using this for several weeks
* feat: return 410 for redundant endpoints
* chore: order endpoints
It's important to not pass frame to transform when using
transformWithFrame. Instead we need to pass either frame.contentDocument
or frame.contentDocument.documentElement
* refactor: re-organise show-project-links
* refactor: update ChallengeFile's declared shape
* fix: handle missing challenge solution
* refactor: use display function for Certification
* refactor: use display function for TimeLine
* refactor: use common component for timeline + cert
* fix: handle legacy solutions
* refactor: use widget for certifications
* refactor: reorganise ShowDisplayWidget
* refactor: remove unused ids
* test: pass dataCy, not projectTitle, to widget
* chore: kebabify
* revert: add id back for dropdown
Co-authored-by: Shaun Hamilton <shauhami020@gmail.com>
* revert: add the ids back
Co-authored-by: Shaun Hamilton <shauhami020@gmail.com>
* feat: make all sources available in tests
The original contents of all files (index.html, script.js etc.) will be
accessible in tests. For example __sources['index.html'] gets the
original html file and nothing else.
* feat: use __file to get user code
* fix: scroll to certification section
* test: refactor and update show-cert
Pulls out the flaky code from the before hook (where a single failure
kills the suite) into the test. Also, the visit to settings to view the
certification only needed to happen once.
* feat: allow more 1000 steps to be created at once
* refactor: start migrating to typescript
* refactor: delete-step to ts
* refactor: migrated some helpers
* refactor: migrate create-empty-steps
* refactor: migrate create-step-between
* refactor: finish migrating to TS
* refactor: migrate tests
* fix: ensure mock.restore is done after each test
* fix: prevent double-tscing
* fix: repair the tests
* chore: use ts-node for scripts
We don't need the performance boost of incremental compilation and
ts-node is easier to work with
* refactor: consolidate tsconfigs
* refactor: replace gulp
* fix: use ts-node for build-curriculum
* fix: allow ts compilation of config
* feat: create and use create:config script
* fix: add /config to eslint projects
* fix: remove gulp script
* fix: handle the sorting of transformed files
We can't use the ext property, since that is transformed, but the
history is maintained.
* test: make solutions polyVinyls for sorting
* refactor: inline prepareChallenge
* refactor: make solutions polys in getChallenges
* fix: convert each solution to a poly
* fix: skip empty solutions
* fix: get challenge file history in client
* refactor: stop calling isPoly from the client
* fix: remove debug console.log
* refactor: remove unnecessary sorting step
* fix: use an iframe to preserve head and body
* fix: remove unnecessary parsing of html
The contents gets inserted into the DOM during transformHtml, which
is always part of the build pipeline
* fix: pipe contents through iframe
* refactor: use the same code for both transforms
* fix: try to handle test errors better
Co-authored-by: moT01 <20648924+moT01@users.noreply.github.com>
* fix: keep challengeFile order on update
* refactor: name variable challengeFile, not file
Co-authored-by: Shaun Hamilton <shauhami020@gmail.com>
* fix: prettify
Co-authored-by: Shaun Hamilton <shauhami020@gmail.com>
* fix: ensures jquery exists for those challenges
For tests frame-runner.js imports jquery, but jquery challenges always
need it
* Revert "Revert "fix: only include frame-runner when running tests (#44337)" (#44419)"
This reverts commit c19bf504b8.
* fix: only render profile urls
* feat: warn user when submitting data url
* fix: prevent dataurls being saved to db
* fix: tests and imports
Not sure why jest didn't like the es imports, but they aren't necessary
so I dropped them.
* fix: check for url protocol
* fix: handle missing solutions correctly
Rather than creating an [[]] the parser now creates [] which isEmpty().
This makes the test suite check the next challenge for a solution.
In addition, the logic for testing solutions was fixed.
* chore: update snapshots
* test: build new superblock in node.js-tests CI
* test: allow forward slash in superblock slug
* fix: borked tests oops
* test: ignore duplicated projects
* fix: i did not break these shaun did :)
* fix: idIndex is index of id not id
Co-authored-by: Nicholas Carrigan <nhcarrigan@gmail.com>
* refactor: use enum
* feat: create links between legacy and new
* refactor: pull out legacy-links component
* fix: import order
* fix: provide icon for new superblock
* feat: only show the copy on the new superblock
* fix: use the proper i18n keys correctly
* feat: put the new superblock at the end of the map
* feat: update the new superblock title
* test: update for soft launch order
* fix: show original project links on /settings
Soft launch only.
* test: use SHOW_NEW_CURRICULUM in tests
* fix: rename the remaining rwd-22s
* feat: make the messge prettier
* fix: message copy
* refactor: make comment make sense
Co-authored-by: Ahmad Abdolsaheb <ahmad.abdolsaheb@gmail.com>
Co-authored-by: Nicholas Carrigan (he/him) <nhcarrigan@gmail.com>
* chore: update superblock properties
* refactor: remove isAudited from certs
It was always defaulting to English, anyway.
* refactor: get superblock from metas, not dirs
* fix: rename responsive-web-design-22
This converts the slugs to /2022/responsive-web-design
* refactor: remove unused export
* test: update test with new curriculum config
* refactor: use key not [key]
* refactor: remove directory reference from test
* fix: store completed js projects in redux
These were missed when we migrated from files to challengeFiles
* test: put the data-cy on the correct button
Rather than the fallback. The tests were only passing because the
fallback was kicking in and, since that was fixed, the tests started to
fail!
* refactor: stop spreading challenge over the node
Instead of creating a Gatsby node with id of challenge.id, we create a
single node field that has all the challenge data.
While this makes the GraphQL queries more verbose, it means we're free
to create multiple nodes with the same challenge.id.
* test: update time-line test for new GQL schema
* test: update mocks with new GQL schema
* refactor: create superOrder from superblock
* chore: drop superOrders from meta
* fix: return null for 'certifications'
* fix: only set superOrder if it is needed
* fix(tests): remove filter from order tests
* test: use getSuperOrder, since superOrder has gone
* test: ignore certificates
* fix: getSuperOrder throw for 'certifications'
* test: remove certificate special case
It wasn't necessary, anyway (it had never been working because I put
certificate not certification)
Co-authored-by: Nicholas Carrigan <nhcarrigan@gmail.com>
* test: login more directly
* test: separate login from other visits
In a single test, Cypress can only visit within a single domain, hence
the separation.
* chore: fail slowly for all strategies
* test: user certified user for showing cert
* test: fix and cleanup certifications
* fix: replace index with script/styles as needed
* fix: remove redundant fileKey
It's overwritten by createPoly, so the parser does not need to create it
* fix: curriculum test suite
* Update client/src/templates/Challenges/classic/MultifileEditor.js
Co-authored-by: Shaun Hamilton <shauhami020@gmail.com>
Co-authored-by: Shaun Hamilton <shauhami020@gmail.com>
react-monaco-editor needed rolling back due to an incompatibility with
react 16
monaco-editor then needed rolling back and the editor config, too.
Some @types needed to be added, some lint deps needed updating.
* feat: add data for preview to challengeMeta
* feat: allow creation of project preview frames
* feat: make project preview data available for frame
* refactor: simplify reducer
* feat: show project preview for first challenge
* feat: show project preview on MultiFile challenges
* test: check for presence/absence of preview modal
* fix: simplify previewProject saga
* test: uncomment project preview test
* fix: increase modal size + change modal title
* modal-footer
* feat: adjust preview size
* fix: remove margin, padding, and line-height for preview of finished projects
* Revert "fix: remove margin, padding, and line-height for preview of finished projects"
This reverts commit 0db11a0819.
* fix: remove margin on all previews
* refactor: use closeModal('projectPreview') for clarity
Co-authored-by: Shaun Hamilton <shauhami020@gmail.com>
* fix: get started -> start coding!
* fix: update closeModal type
Co-authored-by: moT01 <20648924+moT01@users.noreply.github.com>
Co-authored-by: Ahmad Abdolsaheb <ahmad.abdolsaheb@gmail.com>
Co-authored-by: Shaun Hamilton <shauhami020@gmail.com>
* feat: give editable region own background
* refactor: remove unused prop
* fix: remove debugging log
Co-authored-by: Shaun Hamilton <shauhami020@gmail.com>
Co-authored-by: Shaun Hamilton <shauhami020@gmail.com>
* fix: only reset hints and test button on restart
Previously it would be triggered by any change to the editor contents
* refactor: move output node reset into own function
* refactor: reorganise init/update editable region args
* fix: control when line decorations update
* refactor: clean up updateEditableRegion
* refactor: MultifileEditor to functional component.
* fix: make editor acquire focus once on mount
Now the editors can leave the dom (e.g. if a tab is clicked), but an
editor will only call for focus if the MultifileEditor itself remounts
* refactor: remove forbidden regions
We aren't using them and they're implemented badly. To recreate them, we
should simply track the editable region and update them when it's Range
changes.
* refactor: simplify the refs considerably
There's no point keeping an object with properties for all possible
editors when only one is created at once.
Instead of relying on heuristics like "does the current content
differ from the initial content?" this just resets if the reset button
has been pressed.
For legacy challenges, there's only one editor and it must be focused.
For project steps, only the editor with the editable region should call
for focus.
* fix: stop showing completion modal on steps
* feat: submit steps with ctrl+enter
* fix: handle ctrl+enter when not focussing editor
* fix: reset tests when user types
* refactor: pass showCompletionModal as an option
Otherwise we have to write executeChallenge(true) which does not mean
what you might reasonably expect.
* fix: always executeChallenge when not on step
* fix: update frontend project show
* fix: handle missing payload
* refactor: isProjectStep -> hasEditableRegion
* refactor: more renaming
* fix: make meta.json control multifile editor use
* fix: update the challengeSchema correctly
* Update client/src/templates/Challenges/classic/editor.tsx
Co-authored-by: Shaun Hamilton <shauhami020@gmail.com>
* fix: remove logging
Co-authored-by: Shaun Hamilton <shauhami020@gmail.com>
Co-authored-by: Shaun Hamilton <shauhami020@gmail.com>
