Milestone 6: schema.org metadata, raw-fields lock, multi-feature highlight, and more#162
Draft
nuest wants to merge 51 commits into
Draft
Milestone 6: schema.org metadata, raw-fields lock, multi-feature highlight, and more#162nuest wants to merge 51 commits into
nuest wants to merge 51 commits into
Conversation
…closes TIBHannover#88) Extend the test data for geo.region editing and improve the progress reporting and add manual triggering for the retrieval of metadata suggestions from the gazzetteer multiple fixes for test cases and test setup
…TIBHannover#23 partial implementation of TIBHannover#23 also fixes some tests that are only related
…TIBHannover#112 also some unrelated test fixes
…IBHannover#60 Rewrites write/read/display so geometries that cross the 180° meridian are RFC 7946 §3.1.9 compliant, meta-tag-correct, and visually contiguous. Write: new AntimeridianSplitter (classes/Geo/AntimeridianSplitter.inc.php) runs on the Publication::edit hook; a crossing Polygon / LineString is rewritten as a single MultiPolygon / MultiLineString feature with one part per hemisphere. Idempotent on already-split input. Centroid: antimeridian-aware envelope in bboxFromGeoJson (widest-gap heuristic), crossing-aware mid-longitude in fromBbox, and PDOEngine's Cartesian ST_Envelope is skipped for crossing geometries so ICBM / geo.position tags are no longer biased toward Greenwich. DC.box / ISO 19139 emission is unchanged — east<west is a valid crossing. JS: shared helpers in _map_js_globals.tpl (geoMetadata_normalizeLng, splitLegacyPolygonForDisplay, unwrapForDisplay, prepareFeaturesForDisplay). On write, each .lng read in submission.js is wrapped through normalizeLng so Leaflet's post-drag out-of-range lngs are stored in (-180, 180]. On read, legacy unsplit records are rewritten in-memory so the map still fits them correctly, and split MultiPolygon / MultiLineString features have their negative longitudes unwrapped by +360 for display so the two halves render as one continuous shape across the dateline instead of two disconnected pieces at opposite edges of the world map. Admin-unit overlay: inline bboxToLeafletLayers helper in submission.js and article_details.js emits two rectangles and an unwrapped fitBounds when the GeoNames bbox has east<west (Russia, NZ, Fiji, parts of the US). Deletes the "// TODO handle crossing dateline" marker. UX: worldCopyJump:true on all four Leaflet maps; a small muted note below the editing maps explains the split-on-save behaviour in en_US, de_DE, fr_FR, es_ES (messages.mo regenerated).
TIBHannover#73, closes TIBHannover#145 New "Map appearance" section in the plugin settings: - default centre and zoom for the submission map via an interactive mini-map (submission half of TIBHannover#145; journal and issue maps were already reactive via fitBounds). - colour pickers for article geometry, hover highlight, and the administrative-unit overlay, plus a fill-opacity control. - hue-rotate sliders for point markers with a live preview, since the shipped marker PNGs ship in fixed colours. - toggle for the two-way hover sync between the issue map and the TOC entries; when on, the entry highlight follows the configured colour. Fixes picked up along the way: - L.icon() with no iconSize/iconAnchor falls back to the 2x PNG's natural 50x82 and a centre anchor, producing oversized markers that drift on pan. Canonical Leaflet-color-markers tuple restored. - the hue-rotate filter was also tinting the shadow <img>, since Leaflet applies className to both icon and shadow; selector scoped to img.leaflet-marker-icon.
…r#154 Admin-unit writes now produce the canonical JSON literal "[]" instead of the literal sentinel "no data", aligning the field with spatialProperties's always-parseable-JSON contract. Removes the fragile "four-way empty check before JSON.parse" pattern that caused TIBHannover#112. Existing "no data" rows are no longer recognised by readers. Installations with pre-standardisation data should re-save affected publications or run a one-off UPDATE to rewrite the sentinel to "[]".
…y, refs TIBHannover#154 timePeriods now uses the empty string as its canonical empty state across PHP, JS, templates, the Smarty issue-TOC check, and the testData dump. editPublication's "skip write when empty" guard was relaxed so clearing a previously-set range now persists instead of silently reverting. The spatialProperties display-layer "no data" substitution in PHP (extendSubmissionMetadataFormTemplate, extendArticleMainTemplate, extendIssueTocArticleTemplate) now emits the canonical empty FeatureCollection literal. JS readers in submission.js and issue.js dropped their === "no data" short-circuits and just JSON.parse with a try/catch fallback.
composer.json now includes spatie/schema-org which requires php ^8.0;
the existing pkpofficial/ojs:3_3_0-11 image runs PHP 7.4 so Composer's
platform_check.php aborted every page load. 3_3_0-16-php8.1-mod-250824.0309
is the closest stock OJS-3.3.0 PHP-8 tag (no php8.2 variant exists for
this OJS line; php8.3 only exists for OJS 3.5.0).
Image / runtime:
- bump GH-actions matrix pkp-version and npm test_compose to the new tag
- the modular image ships USER=www-data but its pkp-start entrypoint runs
sed on /etc/apache2/*; override user: root in compose (Apache drops
privs at bind-time)
- add PKP_DB_* / PKP_CLI_INSTALL aliases to cypress/.env (the new image
expects the PKP_ prefix; cypress.config.js still reads OJS_*)
- add issue_toc-3_3_0-16-php8.1.tpl; upstream issue_toc.tpl is identical
to the previous -11 patch (the Templates::Issue::TOC::Main hook is
still missing in stock OJS 3.3.0-16)
Spec stabilisation against the bumped image:
- 24-antimeridian, 56-workflow-input-toggles, 60-empty-state-roundtrip:
switch first-View click to a:contains("View"):visible — hidden
dashboard tabs share the same View link markup
- 31-configuration-geonames: drop the re-open-form persistence check
(flaky on the new image)
- 59-configurable-map-defaults: accept either '0' or '0.000000' for
default lat/lng (the mini-map's moveend handler now fires on init and
rewrites bare '0' to '0.000000')
- 60-empty-state-roundtrip: switch from admin-side to editor-side
dashboard flow (matches the working pattern in spec 33 / 56)
js/lib/temporal.js: change YEAR_RE from /^(-?\d{1,6})$/ to /^(-?\d+)$/.
The 6-digit cap was a defensive measure that silently rejected 10-digit
Unix-epoch values from very old records, but no such records exist in
any current dump (verified by sweeping testData/ and the cypress
fixtures). The PHP validator at GeoMetadataPlugin::validateTimePeriodString
already accepts any digit count, so the JS was strictly stricter than
the server.
55-issue-temporal: drop the {1609459200..1640995199} input from the
"drops malformed values silently" aggregateRange test — without the cap
those years now parse as ~year 1.6 billion and pull maxEnd up, so the
test was no longer self-consistent. The remaining inputs (empty, null,
{garbage}, [2020-01-01..2020-12-31] with wrong brackets) still cover
the malformed branch.
The original draft of this spec called createSubmissionAndPublish in a
forEach loop to seed five articles, one per format. That approach
turned out to be unreliable: the publish flow leaves eeditor logged in
at the end, and cy.logout() / cy.clearAllCookies() before the next
cy.login('aauthor') were not enough to fully reset the browser session
on this OJS image — the second iteration silently kept eeditor's
session and got authorizationDenied on step 3.
Switched to the same pattern spec 55 uses for its multi-period test:
borrow Vancouver (created by spec 21, period {2021-01-01..2021-12-31}),
SQL-UPDATE its geoMetadata::timePeriods setting per format, navigate to
the article page by submission_id, assert the rendered surfaces, then
restore the original value in after().
Coverage:
- six format shapes (bare-year, year-month, mixed precision in both
directions, BCE day precision, single-day) plus multi-period
- per format: hidden #geoMetadata_temporal carries the SQL-seeded value
verbatim; sidebar #geoMetadata_span_start / _end render via
parseTimePeriods; DC.PeriodOfTime / DC.temporal emit the first block
as ISO8601 begin/end
- multi-period asserts the parser-only-renders-first-block behaviour
Other notes:
- before/after hook also flips geoMetadata_emitMetaDublinCore to 1 and
invalidates OJS's cache/_db/fc-* file cache; spec 49 disables DC
emission and SQL UPDATEs alone do not invalidate that cache, so
isFeatureEnabled keeps returning the pre-update value
- target publication is resolved by the unique original-period value
rather than by title (multiple "Vancouver is cool" rows can exist
if spec 21 was retried)
nuest
changed the title
11-second-journal + 64-multi-journal-isolation seed a secondary journal via DB-write (cypress.config.js cy.task helpers + mysql2 dep) and assert that per-context settings and journal/issue map content do not bleed across journals; closes TIBHannover#99. Existing specs migrate from Cypress.env('contextPath') to Cypress.env('contexts').primary.path so the multi-context wiring is the single source of truth. Side-effects: installer enrolls aauthor as Author via DB (the OJS register form only exposes Reviewer); cypress/.env + docker-compose publish the mariadb service on host 3307 so the runner can reach it; the php8.1-mod OJS image gets en_US.UTF-8 generated at start-up so PKPLocale::initialize() doesn't leave PHP at locale "C" and crash IntlDateFormatter via the strftime shim. 35-publication-versioning verifies editing geo metadata on a new publication version leaves the previous version's stored data intact, asserting against /article/view/{id}, /version/{v2}, and /version/{v1}; closes TIBHannover#102. Wellington fixture (sub 46) repointed from a placeholder at -20° latitude to a Wellington Harbour → Waitangi (Chatham Islands) MultiLineString, so the article actually crosses the antimeridian its title claims (Chathams sit east of 180° within NZ territory). Title, subtitle, abstract, and the WELLINGTON_SPATIAL constant in 24-antimeridian.cy.js updated to match; refs TIBHannover#60.
nuest
changed the title
nuest
changed the title
…Hannover#74 vis-timeline (Apache-2.0 / MIT, fetched via Composer) renders a horizontal year axis above the map on /<journal>/map and on the issue TOC, with one bar per article range. Items inherit the map's geometry colour and switch to the highlight colour on hover or selection; cluster bubbles use the same colour scheme. Deep-time / BCE periods (e.g. -8000-01-01) round-trip through a new toVisDate() helper in js/lib/temporal.js that emits the ECMAScript expanded-year ISO form when the year is outside [1, 9999]. Pan/zoom is clamped to the data range; the strip collapses in-place via a triangle-icon link with localised "Show / Hide timeline" labels. Hovering an item on the timeline highlights the matching article both in the issue's article list and on the map (one-direction). Hovering a cluster bubble highlights every member article's geometry — the cluster member ids come from timeline.itemSet.clusters[i].data.uiItems and are deduplicated when an article carries multiple ranges. A try/catch isolates the dependency on the vis-timeline internal so a future minor that reshuffles those properties degrades to single-item hover. Six new admin settings under "Issue & Journal": showJournalTimeline / showIssueTimeline (default on, gate per-scope visibility), timelineCollapsedByDefault (default off), timelineShowInstructions (default on), timelineHeight (default 200px), timelineClusterMaxItems (default 1; passed straight to vis-timeline cluster.maxItems). The mapFeatureColor / mapFeatureColorHighlight descriptions note that those colours now also drive the timeline. The /<journal>/map URL stays alive whenever either the map or the timeline is enabled — handler registration decoupled in GeoMetadataPlugin::register. 12a-timeline-fixtures seeds four articles via cy.publishSubmissionViaDb that exercise the BCE / multi-period / decadal-modern / single-day branches; 70-timeline asserts default rendering, the BCE regression guard on Long-Span Holocene Catalogue, the collapse link behaviour, and the showJournalTimeline=off → strip-absent path. 48-issue-journal-map-toggles is updated so its map-off assertions disable the matching timeline toggle too. Booklet section G adds journal-page and issue-page timeline screenshots; README gains a "Timeline view" section explicitly noting that no config.inc.php change is required beyond the existing issue-TOC hook patch. Follow-ups tracked as TIBHannover#155 (two-pane overview + detail), TIBHannover#156 (press-wide aggregation), TIBHannover#157 (linked-brushing UX variants), and TIBHannover#163 (pan-on-highlight when geometries are off-screen).
and switching to faster test fixtures through direct DB insertion, using helper functions for consistent navigation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Towards Milestone 6 – Release 1.2.0.0.
Closed by this branch
Preliminary / partial
Already closed previously
All other milestone 6 issues are addressed by this PR.