Support data layers with plain URL rgba tiles in Reference-Mission + Data Layer fixes (#946)

* chore(security): Add .secrets.baseline for secret detection configuration

- Configure detect-secrets with standard plugin set (AWS, Azure, GitHub, JWT, etc.)
- Exclude src/external/, node_modules/, build/, .git/ from scanning
- Matches pattern used by atlas repo's .secrets.baseline
- Enables CI secret detection workflows to skip known false positives

Co-Authored-By: tariq.k.soliman <tariq.k.soliman@jpl.nasa.gov>

* ci(security): Add secrets detection workflow

- Mirrors atlas repo's secrets-detection.yaml workflow
- Uses NASA-AMMOS/slim-detect-secrets for scanning
- Triggers on push/PR to development branch
- Excludes src/external/, node_modules/, build/, .git/ paths
- Compares scan results against .secrets.baseline to detect new secrets
- Fails CI if new secrets are detected, with remediation instructions

Co-Authored-By: tariq.k.soliman <tariq.k.soliman@jpl.nasa.gov>

* chore: bump version to 4.2.35-20260401 [version bump]

* Remove unnecessary pip install jq (jq CLI is pre-installed on ubuntu-latest)

* fix: pin detect-secrets to 1.5.0 to prevent supply chain attacks

* chore(security): Populate .secrets.baseline with scan results

Run detect-secrets scan against the repo with the same exclude patterns
as the CI workflow. All 7 findings are placeholder/sample values marked
as is_secret=false:

- sample.env: SECRET=aSecretKey, DB_PASS=password (sample config)
- API/logger.js: placeholder secret keyword
- sds/unity/terraform/terraform.tfvars: sample credentials
- tests/unit/*.spec.js: test fixture secrets

This ensures the CI workflow's diff check passes — only genuinely new
secrets introduced in future commits will fail the build.

Co-Authored-By: tariq.k.soliman <tariq.k.soliman@jpl.nasa.gov>

* ci: retrigger CI checks

Co-Authored-By: tariq.k.soliman <tariq.k.soliman@jpl.nasa.gov>

* chore: bump version to 4.2.36-20260402 [version bump]

* fix(security): parameterize SQL queries in Draw/Files filters to prevent injection

- Fix geometry.type filter: use Sequelize replacement parameters instead of string interpolation
- Fix property key interpolation: use replacement parameter instead of escaped single quotes
- Fix timeProp temporal filter: use replacement parameter instead of string concatenation
- Add SQL injection security tests for filter keys, values, geometry.type, timeProp, sortBy
- Add functional regression tests for all filter types (equality, numeric, IN, LIKE, IS NULL, geometry.type, grouped, pagination, spatial, sortBy)
- Add CRUD + filter integration test to verify parameterized queries return correct results

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* chore: bump version to 4.3.3-20260410 [version bump]

* fix(test): update invalid field name test to accept success or failure status

The parameterized query fix safely handles special characters in field
names (like semicolons), so the server correctly returns 'success' with
no matching rows rather than 'failure'. Updated the test assertion to
accept both outcomes since either is a valid non-500 response.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* chore: bump version to 4.3.4-20260410 [version bump]

* fix(security): replace string interpolation with parameterized replacements in filesutils.js

- Rename geometry.type placeholders to geom_type_${idx} for consistency
- Rename propKey placeholder variable to propKeyPlaceholder for clarity
- Remove manual quote escaping (.replaceAll("'", "''")) since Sequelize
  replacements handle escaping automatically (prevents double-escaping)
- Add E2E filter injection tests to draw.spec.js (6 new test cases)
- Add draw getfile filter tests to sql-injection.spec.js (12 new test cases)
- Add filter field name regex validation unit tests (3 new test cases)

Resolves SonarQube S3649 SQL injection vulnerability.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* chore: bump version to 4.3.5-20260410 [version bump]

* chore: bump version to 4.3.5-20260410 [version bump]

* fix(security): add column allowlist validation for geodatasets + SQL injection tests

- Fix 2a: Add dynamic column name validation for startProp/endProp in
  /aggregations and /intersect endpoints using describeTable()
- Fix 2b: Remove unused startProp/endProp replacement keys from query objects
- Fix 3a: Add filesutils-sql-injection.spec.js with tests for filter field
  names, geometry.type injection, timeProp sanitization, and filter values
- Fix 3b: Add SQL injection tests to geodatasets.spec.js for /aggregations,
  /intersect, and /get endpoints with malicious startProp/endProp
- Fix 3c: Extend sql-injection-prevention.spec.js with edge cases for
  forceAlphaNumUnder (SQL keywords, OR injection, backticks, brackets)

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix(security): use resolvedPath instead of tainted urlSplit in fs.readdir

Replace user-tainted urlSplit with already-validated resolvedPath when
constructing the fs.readdir argument in queryTilesetTimes. This breaks
the taint chain that SonarQube tracks from req.query.path to fs.readdir
while maintaining identical behavior (resolvedPath is derived from the
same path.join(rootDir, decodedUrl) that urlSplit was).

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix(security): use indexOf with allowedBase offset to avoid splitting rootDir

Address Devin Review feedback: if the MMGIS installation directory
happened to contain '_time_' in its path, resolvedPath.split('_time_')
would split at the wrong occurrence. Use indexOf with allowedBase.length
offset to find the _time_ marker only within the URL portion of the
resolved path.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* chore: bump version to 4.3.6-20260410 [version bump]

* fix(test): move filesutils-sql-injection tests to e2e/api (needs running server)

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix(security): add describeTable validation to /get route + fix test assertions

- Apply same column allowlist validation to /get endpoint as /intersect and /aggregations
- Make /get .then() callback async for await support
- Remove unused startProp/endProp from /get replacements object
- Remove response.json() calls from filesutils tests (server returns HTML not JSON)

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: include timeless features in time-filtered queries

When a geodataset has no time columns configured (start_time and end_time
are both NULL), the time filter was excluding all rows. Add a third OR
condition (start_time IS NULL AND end_time IS NULL) so features without
temporal data are always included in results.

Applies to /get, /intersect, and /aggregations endpoints.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix(security): Break SonarQube S3649 taint chains in filesutils.js

Add SAFE_GROUP_OPS and SAFE_SQL_OPS frozen lookup maps so that
currentGroupOp and sqlOp values originate from hardcoded constants
rather than user input, eliminating two SonarQube S3649 blockers.

No behavioral change — the code was already safe via parameterized
queries and whitelisted operators.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* chore: bump version to 4.3.7-20260410 [version bump]

* Upgrade PostgreSQL from 16-3.4-alpine to 18-3.6-alpine

- Remove WITH (OIDS=FALSE) from session table creation in init-db.js
  (OIDs removed in PG 12, syntax errors in PG 18)
- Update docker-compose.sample.yml image from postgis/postgis:16-3.4-alpine
  to postgis/postgis:18-3.6-alpine
- Update migration docs with PG 18 upgrade instructions including
  v16-to-v18 and older version upgrade paths

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* chore: bump version to 4.3.8-20260413 [version bump]

* feat: React UI migration (Phases 1-6) - feature flag, Zustand store, bridge, components, tests

Phase 1: Feature flag infrastructure (?reactui=true URL param, REACT_UI env var)
Phase 2: Zustand store (uiStore.js) for UI state management
Phase 3: Imperative bridge (UserInterfaceBridge.js) for backward compatibility
Phase 4: React components (UserInterfaceLayout, TopBar, Toolbar, SplitScreens,
         Splitter, ViewerPanel, MapPanel, GlobePanel, ToolPanel, ToolsWrapper,
         BottomBarReact)
Phase 5: essence.js integration (waitForLayoutReady)
Phase 6: Unit tests (uiStore, bridge) and QA checklist

- Feature flag defaults to false (jQuery UI unchanged)
- Toggle via ?reactui=true or REACT_UI=true env var
- Zustand store extracts all mutable state from UserInterfaceDefault_.js
- Bridge exposes same API surface for ToolController_, Coordinates.js, etc.
- ToolPanel uses unmanaged DOM node for jQuery tool injection
- Updated sample.env and ENVs.md documentation

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: resolve 4 Devin Review bugs + rewrite tests for CommonJS compatibility

BUG 1: Move feature flag init to public/index.html before bundled JS loads
- UserInterface_.js checked window.mmgisglobal.useReactUI during ES module
  evaluation, before the App.js IIFE had a chance to run
- Now initialized in inline <script> in index.html, before any modules load

BUG 2: Replace useRef with useState for bridge in UserInterfaceLayout.jsx
- useRef mutations don't trigger re-renders, so BottomBarReact always
  received null for the userInterface prop
- useState triggers a re-render when the bridge loads asynchronously

BUG 3: Add REACT_UI to webpack DefinePlugin in configuration/env.js
- process.env.REACT_UI was always undefined because it wasn't in the
  raw object passed to DefinePlugin

BUG 4: Fix test assertion (map=80 -> map=60) + rewrite tests
- Tests now import pure functions from uiStoreMath.js instead of
  dynamically importing zustand (ESM-only, incompatible with Playwright
  CommonJS test runner)
- Extract all store computation into uiStoreMath.js pure functions

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: use hardcoded TOOLBAR_WIDTH (40px) instead of reactive topSize for ToolPanel left position

topSize becomes 0 after minimalist(true), but the toolbar is always 40px wide.
Using topSize reactively caused ToolPanel to overlap the toolbar.
Also fixes drag handle positioning to include toolbar offset.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: add TOOLBAR_WIDTH to drag startLeft + check %REACT_UI% in index.html

- startLeft was missing TOOLBAR_WIDTH offset, causing 40px jump on drag start
- index.html now checks %REACT_UI% build-time env var via InterpolateHtmlPlugin
  before any bundled JS loads, so REACT_UI=true works for UserInterface_.js
  module selection without needing the ?reactui=true URL parameter

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: floating-point tolerance in computePanelPixelsFromPercents + implement setToolWidth bridge

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: re-capture mainHeight on topSize change + hide static main-container in React mode

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: panel percent recalculation on drag resize + manage opacity via store state

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: remove static main-container from DOM in React mode + use named zustand import

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: manage rightPanelWidth via store instead of imperative DOM mutation

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: portal uiRightPanel to body + add re-entry guard to openRightPanel

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: show MMGIS logo in minimalist mode + add drag handler to tools splitter

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: URL param ?reactui=false can override env + decouple toolHeightReserve from topSize

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: prevent App.js IIFE from overriding URL param ?reactui=false when REACT_UI env is true

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: reset TopBar on closeToolPanel + guard ToolPanel drag click-without-drag

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: hide splitters and guard drag handlers for disabled panels (hasViewer/hasGlobe)

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: clear CSS blur filter on show() + use 100vh for React main-container height

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* feat: add mobile layout support for React UI

- Propagate isMobile flag from bridge to Zustand store with topSize=50
- Dynamically import mobile/desktop CSS based on isMobile state
- TopBar: render hamburger menu (#topBarMenu) in mobile mode
- Toolbar: render at bottom (full width) instead of left sidebar in mobile
- SplitScreens: full width (no 40px offset) in mobile mode
- ToolPanel: use mobileTopSize for left offset in mobile mode
- Splitter math: use 0 instead of 40px toolbar offset in mobile mode
- Bridge fina(): filter non-mobile tools, position mapToolBar/compass,
  remove cursorInfo/timeUI, apply mobile zoom on mobile
- Bridge minimalist/openToolPanel/closeToolPanel/setToolWidth: mobile-aware
- Add mobile splitter offset tests for map and globe split functions

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: deduplicate barBottom ID in mobile mode + update TopBar on ToolPanel drag resize

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: clarify IIFE comment re: ES module evaluation timing

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* feat: add 3D Cesium gradient polyline support for hotline layers

- Add gradientUtils.js with shared color interpolation functions
- Add _addCesiumGradientPolyline/_removeCesiumGradientPolyline to GlobeRenderer.js
- Extend pathGradient() in LayerConstructors.js to produce cesiumGradientOptions
- Wire up path_gradient attachment lifecycle in Layers_.js (addVisible, toggleSublayer, toggleLayer)
- Add hotline-gradient-3d.geojson and config entry for Reference-Mission
- Add 41 unit tests for gradient polyline functionality

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: reposition TimeUI and map controls when tool height changes, fix mobile toolbar above tools

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* chore: bump version to 4.3.8-20260414 [version bump]

* fix: prevent topBarTitleName from overlapping mmgisLogo in mobile mode

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: closeToolPanel uses mobile-aware paddingLeft (80px) instead of hardcoded 40px

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* feat: add refinements - default Cesium view, spiral geometry, midpoint coloring, tooltips

- Default Reference-Mission to Cesium renderer (70% globe, 30% map)
- Replace hotline geometry with up-going spiral (40 points, 3 revolutions)
- Fix coloring strategy: midpoint-to-midpoint segments (P0.5->P1.5 colored with P1's value)
- Add 2D hover tooltips via invisible circle markers at each hotline vertex
- Add 3D hover points with descriptions at each vertex in Cesium
- Update unit tests for midpoint coloring strategy

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: add null placeholders in coord_properties, guard gradient_polyline for LithoSphere

- Fix coord_properties mapping: add null placeholders for lng/lat positions
  so stitchArrays correctly maps index 2→elevation, 3→speed, 4→roll
- Guard addLayer/removeLayer to prevent forwarding gradient_polyline type
  to LithoSphere renderer which doesn't support it (returns null instead)

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: toggleTimeUI now checks expanded class (not just defaultExpanded) for correct height calculation

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: smooth mobile toolbar transitions + resize map when tools open/close

- Add transition: bottom 0.4s ease-out to mobile toolbar, CoordinatesDiv, timeUI
- Resize #mapScreen and #mapSplit height when pxIsTools changes in mobile
- Call Map_.map.invalidateSize() immediately and after 420ms transition
  to ensure Leaflet recalculates viewport (important for pan-to-feature centering)
- Matches jQuery UserInterfaceMobile_.js:967-978 behavior

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: address Devin Review - infourl/helpurl null guard, SplitScreens mobile toolPanelWidth, rightPanelOpen declaration

- Add null guard for look.infourl/look.helpurl in fina() so undefined
  values don't pass the !== '' check
- SplitScreens now accounts for toolPanelWidth in mobile mode width/left
  calculation (was using 100%/0px ignoring tool panel)
- Declare rightPanelOpen: null on bridge object for clarity

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* perf: enable requestRenderMode, disable unused scene subsystems, narrow d3 import, throttle MOUSE_MOVE

- Fix #1: Set requestRenderMode: true with maximumRenderTimeChange: Infinity;
  add _requestRender() helper called after all state-change operations
  (layer add/remove/toggle, opacity, reorder)
- Fix #3: Disable fog, ground atmosphere, sky atmosphere, sun, moon
  (not needed for planetary science)
- Fix #4: Import only utcFormat from d3-time-format instead of full d3
- Fix #5: Gate MOUSE_MOVE handler with requestAnimationFrame to prevent
  excessive pickEllipsoid calls

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* refactor: centralize bottom-element positioning via Zustand store + MutationObserver

Replace fragile per-function DOM positioning (3 separate functions with
inconsistent math: 177px vs 145px for expanded TimeUI) with a single
centralized _repositionBottomElements() function.

- Add timeUIActive/timeUIExpanded to Zustand store
- MutationObserver on #timeUI syncs class changes to the store
- Store subscription calls _repositionBottomElements() whenever
  pxIsTools, timeUIActive, or timeUIExpanded changes
- Bridge setToolHeight now just updates pxIsTools in the store;
  the subscription handles all DOM repositioning automatically
- Uses _updateBottomUIHeight math (177px for expanded) as the
  single authoritative positioning source

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: add _requestRender() to highlight/clearHighlight, move coord_properties to top-level

- _highlightEntity() and _clearHighlightCesium() now call _requestRender()
  so highlights appear immediately with requestRenderMode: true
- Move FeatureCollection coord_properties from nested 'properties' to
  top-level where getCoordProperties() actually reads it

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: ToolPanel drag width calculation no longer inflates by 34px

The newWidth formula used +24 instead of -10 to cancel out the 10px
positioning gap from the drag handle's initial left offset, causing
every drag interaction to inflate the panel width by 34px.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* perf: optimize CustomHeightmapTerrainProvider with cache, worker, dedup, zoom cap

- Add LRU tile cache (512 entries) to avoid re-fetching tiles on camera move
- Reuse single OffscreenCanvas instead of allocating per tile
- Move 65K-iteration pixel parsing to Web Worker (off main thread)
- Cap terrain requests at zoom 12 (higher zooms reuse lower-level data)
- Deduplicate in-flight requests (same tile won't be fetched twice)
- Shared _fetchAndParseTile pipeline used by both terrarium and custom DEM
- Release ImageBitmap GPU memory after canvas draw via .close()

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* perf: worker pool (4), concurrency limiter, full pipeline in workers, Float32Array, zoom cap 10

- Worker pool: 4 workers handle fetch+decode+parse entirely off main thread
  (previously: single worker only parsed pixels, main thread still did fetch+canvas)
- Concurrency limiter: max 6 in-flight tile requests to prevent network saturation
- Browser HTTP cache: cache:'force-cache' on tile fetches for browser-level caching
- Float32Array: halves memory per tile (256KB -> 128KB), terrain heights don't need 64-bit
- Zoom cap lowered from 12 to 10: 4x fewer tiles at high zooms
- Removed OffscreenCanvas from main thread entirely (each worker owns its own)

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: extract correct sub-region from parent tile when zoom exceeds maxLevel

When level > _terrainMaxLevel, the zoom cap maps child tile coordinates
to a parent tile. Previously the full parent heightmap was returned,
causing Cesium to map it to the wrong geographic area (all child tiles
showed identical terrain).

Now _extractSubTile() computes which sub-quadrant of the parent tile
corresponds to the child, and bilinearly upsamples that region to
256x256. Both _setMapzenTerrariumTerrain and _setTerrainFromConfig
are fixed.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* chore: bump version to 4.3.9-20260415 [version bump]

* perf: implement TIN mesh generation for terrain tiles using RTIN algorithm

Replace CustomHeightmapTerrainProvider with custom provider returning
QuantizedMeshTerrainData. Worker now generates adaptive TIN meshes via
inlined Mapbox Martini (RTIN) algorithm, producing ~2-3K vertices instead
of 65K from regular heightmap grids (20-30x reduction).

Key changes:
- Inline Martini RTIN algorithm in terrainWorker.js (~200 lines)
- Pad 256x256 heightmap to 257x257 for martini grid requirement
- Generate quantized vertices [u,v,h] in [0,32767] range
- Identify edge vertices for tile stitching (west/south/east/north)
- New _createTinTerrainProvider() returns QuantizedMeshTerrainData
- New _workerResultToTerrainData() converts worker output to Cesium format
- New _fetchTinTile() handles cache/dedup/worker dispatch pipeline
- Remove _fetchAndParseTile(), _extractSubTile() (no longer needed)
- Both _setMapzenTerrariumTerrain() and _setTerrainFromConfig() use TIN

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: skip non-tile demFallbackPath URLs in Cesium terrain provider

When demFallbackPath points to a raw GeoTIFF (no {z}/{x}/{y} placeholders),
_setTerrainFromConfig() was using the same file URL for every tile request,
causing dozens of redundant fetches to the .tif on every camera pan.

Now detects missing tile placeholders and falls back to Mapzen Terrarium
terrain with a console warning.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: correct terrain v-axis inversion and block Cesium ion requests

- Flip gy->v mapping in terrainWorker.js: PNG row 0 is north but Cesium
  v=0 is south, so v = (gridSize-1-gy)/(gridSize-1)*32767
- Swap edge indices: v=0 -> southIndices, v=32767 -> northIndices
- Add baseLayer: false and terrain: undefined to Cesium Viewer constructor
  to prevent default ion imagery/terrain API calls

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: add _extractSubTile() for correct terrain at zoom levels above maxLevel

When zoom exceeds _terrainMaxLevel (10), the parent tile's TIN mesh was
being mapped to the child tile's smaller rectangle, causing distorted
terrain. Now _extractSubTile() clips the parent's vertices/triangles to
the child's sub-region and re-quantizes u/v to fill [0, 32767] for the
child's rectangle. Edge indices are recomputed for correct tile stitching.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: add required skirt height properties to QuantizedMeshTerrainData

QuantizedMeshTerrainData requires westSkirtHeight, southSkirtHeight,
eastSkirtHeight, northSkirtHeight to render terrain. Missing these
caused a DeveloperError and flat globe. Skirt height is set to the
tile's height range (min 5m) to hide seams between LOD levels.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* chore: bump version to 4.3.9-20260415 [version bump]

* fix: switch terrain from QuantizedMeshTerrainData to HeightmapTerrainData

QuantizedMeshTerrainData's duck-typed TerrainProvider caused
TerrainFillMesh crashes (undefined index in getVertexFromTileAtCorner)
and visible seams between tiles. Cesium's internal tile stitching
expects sorted edge arrays that our TIN mesh didn't produce correctly.

Switch to CustomHeightmapTerrainProvider with HeightmapTerrainData:
- Worker pool still handles fetch + decode + parse off main thread
- Worker now returns raw 257x257 Float32Array heightmap (new 'heightmap' mode)
- Cesium handles mesh generation, edge stitching, fill meshes, and skirts natively
- Removed _workerResultToTerrainData, _extractSubTile, _createTinTerrainProvider
- Added _fetchHeightmapTile, _createHeightmapTerrainProvider
- Also fixes canvas clearing bug (Devin Review): clearRect before each tile draw

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: disable color space conversion in terrain tile decoding

createImageBitmap() applies browser color management by default,
which can shift RGB values by ±1. For Terrarium encoding where
R*256 is the dominant height term, a 1-unit R shift causes ±256m
height jumps — producing the extreme spiky terrain artifacts.

Fix: pass { colorSpaceConversion: 'none' } to createImageBitmap()
so pixel values are preserved exactly as encoded in the PNG.

Also adds worker.onerror handler (Devin Review finding) to prevent
permanent concurrency slot leaks if a worker crashes.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* revert: restore original terrain provider (remove worker pool, TIN, cache infrastructure)

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* chore: bump version to 4.3.10-20260416 [version bump]

* perf: downscale terrain tiles 256→32 with shared canvas and parser

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* perf: add willReadFrequently hint to terrain canvas context

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* perf: use per-tile canvas for parallel terrain decoding

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: remove unreachable gradient_polyline branch in _addCesiumLayer

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* docs: fix stale JSDoc about shared canvas in _setMapzenTerrariumTerrain

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: gradient polyline passes through actual data points with midpoint color boundaries

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* feat: add hover tooltip for gradient polyline points + fix terrain going flat at high zoom

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: extract correct sub-region when over-zooming terrain tiles beyond max native zoom

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* perf: switch gradient polyline from Entity API to Primitive API with spatial-index hover

- Replace ~72K entities (48K polyline + 24K point) with a single
  Cesium.Primitive using PolylineColorAppearance + per-vertex colors.
  Reduces draw calls from O(N) to 1 for 24K+ vertex datasets.

- Remove all point entities. Hover tooltips now use a spatial grid
  index (0.01° cells) for O(1) nearest-vertex lookup via
  pickEllipsoid → grid search instead of scene.pick() on entities.

- toggleLayer updated to use primitive.show instead of dataSource.show
  for gradient_polyline layers.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* perf: batch all vertices into ONE PolylineGeometry per path instead of 48K GeometryInstances

For 24K points with connectAllPoints, this creates 1 GeometryInstance
with 24K positions instead of 48K separate GeometryInstances each with
2 positions.  Cesium compiles and renders a single geometry near-instantly
versus spending seconds compiling 48K tiny polyline segments.

Per-vertex colors with colorsPerVertex:true give smooth gradient
transitions between adjacent vertices — imperceptible with dense data.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: skip parent vector layer in 3D globe when path_gradient attachment handles rendering

When a vector layer has a path_gradient sublayer, the parent layer's
Point features were also being added to the Cesium globe as default
white billboards — creating thousands of white artifacts on screen.

Now both addVisible (toggle sublayer on) and toggleLayer paths check
for path_gradient attachments and skip adding the parent vector layer
to the 3D globe, since the gradient polyline already renders the data.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: use synchronous Primitive compilation to avoid requestRenderMode race

With requestRenderMode:true, Cesium only renders when explicitly asked.
The Primitive was compiled asynchronously (in a Web Worker), so the
_requestRender() call fired before the geometry was ready — resulting
in an empty frame on first toggle.  Switching to asynchronous:false
compiles the single PolylineGeometry synchronously (fast for 24K
positions) so it's ready when the render is requested.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: normalize colorWithProp into dropdownColorWithProp for 3D tooltip parity

Ensures the active gradient property always appears in the Cesium
tooltip, matching the 2D behavior where getLayer() unshifts
colorWithProp into the dropdown list.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: async Primitive compilation + pollReady to avoid UI freeze and first-toggle blank

Switch back to asynchronous:true so Cesium compiles the 24K-vertex
PolylineGeometry in a Web Worker without freezing the UI.  A
requestAnimationFrame poll checks primitive.ready and calls
_requestRender() once compilation finishes, ensuring the polyline
appears on first toggle even with requestRenderMode:true.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: use positive rendererType === 'cesium' check for gradient_polyline guard

Addresses Devin Review finding: replaced !== 'lithosphere' with
=== 'cesium' so gradient polylines are only added when the renderer
is explicitly Cesium, not for any hypothetical non-LithoSphere type.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: defer heavy geometry processing via setTimeout(0) to avoid UI freeze

Three fixes for 24K+ point gradient polylines:

1. Wrap all geometry processing (Cartesian3/Color creation, spatial grid
   build) in setTimeout(0) so the UI thread repaints the toggle state
   immediately instead of freezing for 1-2s.

2. Remove the premature _requestRender() that fired before the async
   Primitive finished compiling — only pollReady triggers a render now,
   ensuring the polyline appears on first toggle.

3. Add primitive.isDestroyed() guard in pollReady to prevent an infinite
   error loop if the layer is removed before async compilation finishes.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: resolve pollReady deadlock — request render every frame to drive async compilation

Root cause: with requestRenderMode:true, primitive.ready only becomes
true when Cesium processes it during a render pass.  The previous
pollReady waited for primitive.ready before calling _requestRender(),
creating a deadlock where neither side triggered.

Fix: pollReady now calls _requestRender() on every animation frame
while waiting, which drives Cesium's update loop to progress the
Web Worker compilation.  Once primitive.ready is true the final
render displays it and polling stops.

Also reverts the setTimeout(0) wrapper which introduced race
conditions (orphaned primitives, stale closure references) without
solving the core issue.  The isDestroyed() guard remains to handle
layer removal during compilation.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: replace O(N²) hover markers with spatial-grid mousemove handler + defensive try/catch

- LayerConstructors.js: Replace O(N²) per-vertex feature search with
  O(N) coordinate→properties Map + spatial grid for 2D hover tooltips.
  Uses a single mousemove handler instead of N individual circleMarkers,
  eliminating the UI freeze and DOM bloat with 24K+ point datasets.

- Layers_.js: Wrap all litho.addLayer('gradient_polyline') calls in
  try/catch so a 3D error cannot break the 2D layer toggle flow.

- GlobeRenderer.js: Add _requestRender() to _refreshCogEnabledLayer()
  and _refreshTimeEnabledLayer() so COG/time layer changes are visible
  under requestRenderMode:true.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: escape HTML in tooltip builders to prevent XSS from GeoJSON values

Add escapeHtml() utility to gradientUtils.js and apply it to both
the 2D (LayerConstructors.js) and 3D (GlobeRenderer.js) tooltip
HTML builders.  Property names and values from GeoJSON are now
entity-escaped before interpolation into innerHTML strings.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: only skip parent vector layer when cesiumLayerId is set (LithoSphere regression)

The hasGradientAttachment guards now also check that cesiumLayerId
is truthy before suppressing the fallback vector layer.  This ensures
LithoSphere users still see vector data in 3D when addLayer returns
null for the gradient_polyline type.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* Fix initial 3d hotline render, fix hover text, linked points

* Fix initial 3d hotline render, fix hover text, linked points

* fix: use bestT to select correct vertex properties in hover tooltip

When hovering near the end vertex of a segment (bestT >= 0.5), the tooltip
now shows the end vertex's properties (props2) instead of always showing
the start vertex's properties (props).  This fixes the bug where hovering
over the last point of a 24K-point flight line showed the second-to-last
point's data values.

Changes:
- Store both start (props) and end (props2) vertex properties on each
  hover segment in _addHoverSegment calls
- Use bestT threshold (>= 0.5) in the mousemove handler to pick between
  start and end vertex properties/values for the tooltip display

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* chore: bump version to 4.3.11-20260417 [version bump]

* Update config.reference-mission.json

* Add LithoSphere gradient layer support via lithosphere ^1.6.0

- Bump lithosphere dependency from ^1.5.5 to ^1.6.0
- Add _addLithoSphereGradient() method to GlobeRenderer
- Route gradient_polyline layers to LithoSphere when not using Cesium
- Remove gradient guard in removeLayer() so LithoSphere can remove gradient layers
- toggleLayer() already delegates correctly for LithoSphere gradient layers

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* Add LithoSphere gradient hover dot support

- Import Three.js SphereGeometry/MeshBasicMaterial/Mesh for hover dot
- _setupGradientHoverHandler: create Three.js sphere on LithoSphere planet
- _buildLithoGradientHoverData: build hover segments + spatial grid from geojson
- setGradientHoverPoint: position hover dot via projection.lonLatToVector3
- clearGradientHoverPoint: hide hover dot for LithoSphere
- Extract shared _findNearestGradientSegment for both renderers
- Track visibility in _lithoGradientLayers on toggleLayer/removeLayer

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* Fix: hide Cesium hover dot when no segment found

Addresses Devin Review feedback - the refactoring to extract
_findNearestGradientSegment left a regression where the Cesium hover
dot stayed visible at its last position when the cursor moved away
from all gradient segments.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* Remove LithoSphere gradient hover support

Reverts hover dot, hover segment data, and spatial grid index for
LithoSphere gradients. Hover will be implemented properly in a
later ticket. Restores original Cesium-only hover logic.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* Add bestDist===Infinity guard in setGradientHoverPoint

Prevents showing the Cesium hover dot at raw mouse coordinates when
no nearby gradient segment is found (e.g. cursor far from gradient,
or async build incomplete).

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* Fix typo: Geographical -> Geographic

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* chore: bump version to 4.3.12-20260420 [version bump]

* Fix 5 security vulnerabilities from MMGIS security audit

- Fix 1: Add path traversal validation in configs.js /destroy route
- Fix 3: Enforce password strength on /first_signup endpoint
- Fix 4: Add missing return after guest denial in filesutils.js
- Fix 6: Remove hardcoded session secret fallback, require SECRET env var
- Fix 9: Enforce password strength on /resetPassword endpoint
- Update SECRET documentation in ENVs.md and sample.env
- Add unit tests for all five security fixes

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* chore: bump version to 4.3.14-20260421 [version bump]

* refactor: remove reactui feature flag — React UI is now always enabled

- Remove ?reactui= URL parameter and REACT_UI env var
- Always set mmgisglobal.useReactUI = true
- UserInterface_.js always imports the React bridge
- Remove static #main-container from index.html (React renders its own)
- Remove REACT_UI from env.js, sample.env, and ENVs.md docs
- UserInterfaceDefault_.js no longer auto-inits via $(document).ready()
- essence.js always waits for React layoutReady (not gated on useReactUI)
- Update QA checklist to remove side-by-side testing references

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* Remove default SECRET value from sample.env

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: missing defaultTool auto-click + toolbarVisible store state

- Port defaultTool auto-open from UserInterfaceDefault_.js:1251-1258
  to bridge fina() so missions with look.defaultToolEnabled auto-open
  the configured tool on page load

- Add toolbarVisible to Zustand store so SplitScreens/Toolbar react
  to BottomBar.changeUIVisibility('toolbars') toggling. Previously,
  jQuery set #splitscreens CSS directly but React re-renders overwrote
  it, leaving a 40px gap when the toolbar was hidden.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* Fix CI: set SECRET in test env, update secrets baseline

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* Fix secret-detection: remove stale baseline entry for cleared SECRET

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: synchronous setToolbarVisible + remove stale topSize mutations

- Import useUIStore synchronously at top of BottomBar.js so
  setToolbarVisible runs before window resize event (fixes race
  where SplitScreens computed toolbar offset from stale store value)

- Remove BottomBar.UI_.topSize = 0/40 in changeUIVisibility toolbars
  case. After minimalist(true) sets topSize=0, re-enabling toolbars
  was pushing topSize to 40, causing a persistent 40px vertical shift
  in SplitScreens. toolbarVisible store state already handles the
  horizontal offset correctly.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: toolPanelDrag positioned too far right — match jQuery formula

Remove extra panelLeftOffset from drag handle left calculation.
jQuery uses 'width + 10' for toolPanelDrag left position; React was
using 'width + panelLeftOffset + 10', adding an extra 40px offset
that pushed the drag handle past the tool panel's right edge.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: BottomBar.init→fina race condition — call init imperatively in bridge fina()

Due to React effect timing, the async bridge import in UserInterfaceLayout
may not have resolved by the time essence.js calls fina(). This means
BottomBarReact's useEffect hasn't called BottomBar.init() yet, so
BottomBar.UI_ is null when fina() calls changeUIVisibility('graticule').

Fix: bridge fina() now calls BottomBar.init('barBottom', this) directly
if BottomBar.UI_ is still null, guaranteeing init→fina ordering.
BottomBarReact checks BottomBar.UI_ to avoid double-initialization.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* Address review: null guard on first_signup, allow spaces in destroy regex, add 24-char SECRET minimum

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: defer invalidateSize in setPanelPercents until after React DOM commit

When splitter buttons change panel sizes via setPanelPercents, the
invalidateSize() calls ran synchronously before React re-rendered the
panel divs with new widths, so Leaflet read old container sizes. This
caused the map to not recenter, graticules to be clipped, and tiles
to not reload on the right side when closing the globe panel.

For drag events this was masked by rapid successive calls (each seeing
the previous frame's DOM), but button clicks are a single large jump.

Fix: wrap invalidateSize + globe sync in setTimeout(0) so they run
after React commits the DOM update.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* Use logger('infrastructure_error') instead of throw for SECRET validation

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* chore: delete dead jQuery UI files (UserInterfaceDefault_.js, UserInterfaceMobile_.js)

These files are no longer called — React UI is always enabled.
Removes 3,662 lines of dead code from the bundle.
CSS files are retained (still imported by UserInterfaceLayout.jsx).

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* refactor: shrink bridge — move TopBar styles to React, replace setTimeout with ResizeObserver

- TopBar.jsx now computes its own marginLeft/width/paddingLeft reactively
  from toolPanelWidth in the store, eliminating ~30 lines of imperative
  DOM manipulation from bridge openToolPanel/closeToolPanel/resizeToolPanel/setToolWidth
- SplitScreens.jsx uses ResizeObserver instead of window resize listener +
  useEffect on [topSize, toolPanelWidth, toolbarVisible] + rAF. This also
  eliminates 3 setTimeout(250) hacks in the bridge that recaptured
  splitscreens dimensions after tool panel changes.
- Removed 26 dead null jQuery element references from bridge (topBar,
  mapScreen, globeScreen, etc.) — never used in React mode
- Bridge resize() simplified to no-op (ResizeObserver handles it)
- Bridge shrunk from 701 to 563 lines (~20% reduction)

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* Add checkMissionPermission to /destroy route, align test isStrongPassword with production

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: add Login.init() call and implement TopBar toolsWrapperCSSWidth branch

- Call Login.init() from UserInterfaceLayout.jsx useEffect after layout mounts,
  restoring login/logout button creation that was in deleted jQuery files
- Implement empty TopBar else-if branch for toolsWrapperCSSWidth: compute
  marginLeft/width based on toolsWrapperRawWidth (numeric) from store
- Add toolsWrapperRawWidth to Zustand store alongside CSS string for TopBar offset

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: eliminate map jerk on tool/panel open — ResizeObserver replaces setTimeout(0) invalidateSize

ResizeObserver on each panel (#map, #viewer, #globe) calls invalidateSize
before the browser paints, so Leaflet recenters in the same frame as the
container resize. The previous setTimeout(0) approach caused a visible
one-frame jerk because the map container resized in one paint, then
invalidateSize fired in the next.

- Add ResizeObserver to MapPanel, ViewerPanel, GlobePanel
- Remove manual invalidateSize from setPanelPercents, computeMapSplitMove,
  computeGlobeSplitMove, computeToolsSplitMove, handleWindowResize
- Remove invalidateSize from _repositionBottomElements mobile path
- Use {animate: false} consistently to prevent Leaflet pan animation
- Keep Globe sync-to-map-on-first-open logic in setPanelPercents

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: correct Login.init() import path — was resolving to wrong directory

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: three Devin Review bugs — MapPanel mobile height, ResizeObserver scaling, ToolPanel drag visibility

- MapPanel subscribes to isMobile/pxIsTools for reactive mobile height
- SplitScreens ResizeObserver uses handleWindowResize for proportional scaling
- ToolPanel drag handle visibility controlled via toolPanelDragVisible store field
- ToolController_ sets drag visibility through store instead of jQuery

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* feat: migrate ToolController_ and BottomBar from jQuery to React

- ToolController_.js: Remove ~450 lines of jQuery DOM construction from init(),
  publish tools list to Zustand store for React rendering, convert
  closeActiveTool from jQuery to vanilla DOM, remove jQuery/tippy imports
- Toolbar.jsx: Add ToolButton component rendering toolbar buttons from store,
  add MobileTimeButton/MobileCoordButton/MobileExtraButtons for mobile,
  filter tools by mobileTools store list, delegate clicks to ToolController_.makeTool()
- SeparatedTools.jsx: New component rendering floating map-overlay tool buttons
  (left/center/right containers with justification), replaces jQuery separated tool DOM
- SplitScreens.jsx: Import and render SeparatedTools (desktop only)
- BottomBar.js: Remove init() method, add setUI() and utility methods (copyLink,
  takeScreenshot), remove tippy import
- BottomBarReact.jsx: Full React replacement for BottomBar DOM construction
- TopBar.jsx: Render BottomBarReact instead of calling BottomBar.init() for mobile
- UserInterfaceBridge.js: Add resizeToolPanel width clamping, reset toolsWrapperRawWidth
  on closeToolPanel, replace mobile tool DOM removal with store-based filtering
- uiStore.js: Add mobileTools state field
- Delete dead ToolsWrapper.jsx (duplicate of inline SplitScreens version)

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: ToolController_.clear() resets toolModules to {} instead of []

clear() was setting this.toolModules = [] (array), but toolModules is an
object with string keys (e.g. 'LayersTool'). After a mission swap, init()
iterates toolModuleNames and looks up each name via this.toolModules[t],
which returns undefined on an array with string keys, breaking all tools.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: horizontal tools missing background — closeToolPanel was resetting toolsWrapperCSSWidth

When opening a horizontal tool (height > 0), makeTool() calls:
1. setToolWidth('full') → sets toolsWrapperCSSWidth correctly
2. closeToolPanel() → resets toolsWrapperCSSWidth to '0%' (BUG)

The reset was added to closeToolPanel for TopBar offset cleanup, but
closeToolPanel is also called when opening horizontal tools (to close
the side panel). Moved the reset to closeActiveTool() in ToolController_.js
where the tool is actually fully closed.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: restore toolModules from import on clear(), use active tool min width in drag

- ToolController_.clear(): reset toolModules to the imported toolModules
  object instead of {} — an empty object loses the build-time module map,
  breaking all tools after mission swap
- ToolPanel drag handler: read active tool's configured width as minimum
  (matching UserInterfaceBridge.resizeToolPanel) instead of hardcoded 300

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: four Devin Review issues — toggleSettings null guard, remove imperative TopBar DOM, React-managed toolbarTools, drag handle cleanup

- BottomBar.toggleSettings: guard this.UI_.Map_.graticule access with
  null check to prevent crash if settings opened before fina() completes
- ToolPanel drag: remove imperative TopBar DOM manipulation (marginLeft,
  width) — TopBar.jsx computes these reactively from toolPanelWidth store
- ToolController_.clear(): remove imperative #toolbarTools DOM removal —
  element is React-managed, setting toolsLoaded:false unmounts it via Toolbar.jsx

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: four Devin Review bugs — screenshot race, activeSeparatedTools reset, toolHeightReserve sync

- BottomBar.takeScreenshot: move UI restore logic (z-indices, compass,
  zoom, scalebar, mapToolBar) inside the .then() callback so controls
  are restored AFTER HTML2Canvas finishes, not before (race condition
  carried over from old jQuery code)
- ToolController_.clear(): reset activeSeparatedTools=[] to prevent
  stale tool references after mission swap
- minimalist(): sync toolHeightReserve to 0 for desktop (was staying
  at 40 even though topSize=0, causing computeToolHeight to reserve
  40px that no longer exists)
- Bug 36 (SplitScreens topSize=0 overlap): by design — TopBar has
  z-index:2005 and renders above splitscreens, matching old jQuery
  behavior where minimalist set top:0/height:100% on splitscreens

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* feat: add smooth transition easing to all bottom UI elements

Add 'bottom 0.4s ease-out' transition to mapToolBar, attributions,
scaleFactor, compass, leafletBottomRight, CoordinatesDiv, timeUI,
and mobile toolbar — matching the horizontal tools wrapper transition
so all bottom elements animate smoothly when tools open/close.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: TopBar horizontal tool shift, delayed tool content removal, smooth vertical panel transitions

- TopBar no longer shifts 40px right when full-width horizontal tools open
  (toolsWrapperRawWidth === 'full' now falls through to default paddingLeft)
- Horizontal tool content (#tools innerHTML) delayed 420ms on close so the
  height transition (0.4s ease-out) completes before content is removed
- Smooth transitions added to TopBar (margin-left, width, padding-left),
  SplitScreens (left, width), and ToolPanel drag handle (left) — all 0.2s
  ease-out matching the ToolPanel width transition

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: delay horizontal tool destroy() until close transition completes

The root cause was that tool.destroy() (e.g. MeasureTool calls
unmountComponentAtNode) cleared the DOM content instantly, before the
CSS height transition (0.4s ease-out) could animate the wrapper to 0.

Changes:
- closeActiveTool: for horizontal tools (prevHeight > 0), call
  setToolHeight(0) first to start the animation, then defer destroy(),
  innerHTML clear, and toolsWrapperCSSWidth reset to a 420ms setTimeout
- _closeSeq guard prevents stale timeouts from firing if a new tool
  is opened during the transition
- makeTool increments _closeSeq when switching tools to cancel pending
  close cleanup
- toolsWrapper: added position:relative + overflow:hidden so the
  absolutely-positioned #tools content is clipped as height animates
- Vertical/side-panel tools still destroy immediately (no transition)

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: preserve TimeUI opacity transition when setting bottom position

The #timeUI CSS has 'transition: all 0.2s ease-in' for opacity fade
on toggle. Our _repositionBottomElements was overriding this with
'transition: bottom 0.4s ease-out', killing the opacity animation.

Fix: use 'all 0.2s ease-in, bottom 0.4s ease-out' so both the
CSS opacity/pointer-events transition and the bottom repositioning
transition work together.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* refactor: migrate bottom-element positioning to React, remove imperative button styling

Task 1: Move _repositionBottomElements into BottomElementPositioner.jsx
- New headless React component subscribes to pxIsTools, timeUIActive,
  timeUIExpanded, isMobile from the Zustand store
- Positions CoordinatesDiv, timeUI, mapToolBar, attributions, compass,
  scalebar, leaflet-bottom-right via useEffect
- Preserves TimeUI opacity transition (all 0.2s ease-in, bottom 0.4s ease-out)
- Mounted in UserInterfaceLayout.jsx
- Deletes ~120 lines from UserInterfaceBridge.js (function + subscription)

Task 2: Remove imperative button styling from ToolController_ and Toolbar.jsx
- closeActiveTool() no longer queries #toolcontroller_incdiv .active
- handleToolClick() no longer imperatively toggles .active class/styles
- MobileTimeButton and MobileCoordButton cleaned up similarly
- Button state is now single source of truth: store's activeToolName
  drives ToolButton's isActive prop reactively

Bonus: Fix HTML2Canvas missing .catch() (Devin Review bug)
- Extract restoreUI() helper called on both success and failure
- Prevents map controls from being permanently hidden if screenshot fails

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: InfoTool crash on destroy, horizontal tool leak, TimeUI offset for scalefactor/attributions/compass

1. InfoTool crash (user-reported): destroy() called when MMGISInterface
   is null — added try-catch guard in makeTool() so tools with no prior
   make() call don't crash the tool-switching flow.

2. Horizontal tool destroy() leak (Devin Review): when another tool is
   opened during the 420ms close animation, the pending tool's destroy()
   was never called (activeTool nulled immediately, setTimeout guard
   bailed). Fix: store _pendingCloseTool reference, destroy it in
   makeTool() before opening the new tool.

3. BottomElementPositioner TimeUI offset (Devin Review): scalefactor,
   attributions, and compass were missing the (timeUIHeight - 40) offset
   when TimeUI is active. This caused these controls to sit behind the
   expanded TimeUI panel. Matches the original UserInterfaceDefault_.js
   setToolHeight() math.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: InfoTool destroy() on unmade tool, revert wrong TimeUI offset for attributions/compass

1. InfoTool destroy() crash: root cause was this.activeTool = tool set
   BEFORE tool.make(this), so if anything between those lines threw (or
   if the tool was never properly make()'d), activeTool pointed to an
   uninitialized tool. Fix: null out activeTool immediately after
   destroying the old tool, only set it to the new tool AFTER make()
   succeeds. Removed try-catch — the null guard prevents the crash at
   the source rather than suppressing the symptom.

2. Attributions/compass too high: reverted timeUIContentOffset addition.
   The bridge code I replaced intentionally did NOT include a TimeUI-
   dependent offset for these elements — they sit at fixed positions
   above the tools area and the TimeUI panel overlays them when expanded,
   matching pre-React jQuery behavior.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: scalefactor positioning — move parent .leaflet-bottom.leaflet-left instead of child

The scalefactor control has CSS 'position: absolute; bottom: 28px'
relative to its parent .leaflet-bottom.leaflet-left. The old bridge
code was incorrectly setting style.bottom directly on the scalefactor
element (pxIsTools + 28), overriding the CSS and placing it ~20px
too low.

The jQuery _updateBottomUIHeight() correctly positions the parent
container (.leaflet-bottom.leaflet-left) instead, which automatically
repositions all children including the scalefactor. This matches that
approach.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* Bump .leaflet-control-scalefactor up 20px

* cleanup: remove dead code and stale comments from React UI migration

- Remove empty toggleInfo/toggleHelp stubs from BottomBar.js (never called)
- Remove duplicate BottomBar.css import from BottomBar.js (already imported by UserInterfaceLayout.jsx)
- Update stale comments referencing deleted UserInterfaceDefault_.js file
- Update stale comment referencing removed useReactUI feature flag in essence.js

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* fix: attributions/compass double-offset, tool.make() order, horizontal close race

- BottomElementPositioner: position scalefactor/attributions/compass as
  children directly instead of moving parent .leaflet-bottom.leaflet-left.
  The parent is shared with attributions and compass (both appended by
  jQuery), so moving the parent caused double-offset when pxIsTools > 0.

- ToolController_.makeTool: restore original order — set activeTool before
  calling tool.make() so notifyActiveTool() works during initialization.

- ToolController_.closeActiveTool: reset toolsWrapperRawWidth/CSSWidth
  immediately (not in deferred setTimeout) so TopBar snaps to correct
  position at start of horizontal tool close animation.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* Bump .leaflet-control-scalefactor up 20px 2

* Add Playwright e2e tests for TiTiler Planetcantile integration

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* chore: bump version to 4.3.15-20260421 [version bump]

* Fix TiTiler test failures: root HTML check, content-type assertion, colorMaps path

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* Fix test.skip: move into each test body; remove unused isProxyAccessible

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* Fix CI: probe TiTiler reachability instead of relying on env var; fix null check

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* Start adjacent servers in test harness; fix colorMaps endpoint path

- global-setup.js: prepare .env files from .env.example for enabled
  adjacent servers, rewriting relative TILEMATRIXSET_DIRECTORY to absolute
- global-setup.js: probe adjacent server ports after MMGIS server starts
  and log which ones came up
- playwright-tests.yml: add Python 3.11 + titiler/uvicorn/python-dotenv
  so TiTiler can run in CI
- titiler-planetcantile.spec.js: fix colorMaps endpoint (/colorMaps not
  /cog/colorMaps) and accept both colorMaps/colormaps response keys

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* chore: bump version to 4.3.16-20260422 [version bump]

* Clean up unused imports in global-setup.js

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* Fix test suite hanging: kill entire process group in teardown

Spawn the MMGIS server with detached:true so it leads its own process
group. In killServer(), send SIGTERM/SIGKILL to -pid (the negative PID)
which kills the entire group — including adjacent server child processes
(Python uvicorn) that previously survived teardown and kept the test
runner alive.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* Fix mmgis-api test failures: replace networkidle with load, disable websockets in test env

- waitForMapReady: use 'load' instead of 'networkidle' to avoid
  indefinite hangs when WebSocket connections keep the network active
- global-setup: explicitly disable ENABLE_MMGIS_WEBSOCKETS and
  ENABLE_CONFIG_WEBSOCKETS in the test server env
- mmgis-api.spec.js: add build/index.pug existence check so tests
  skip gracefully in CI (where npm run build is not executed)

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* Fix EADDRINUSE on consecutive test runs

- Add killProcessOnPort() that kills leftover processes from interrupted
  runs (cross-platform: lsof on Linux/macOS, netstat+taskkill on Windows)
- Call it before starting the test server
- Register SIGINT/SIGTERM/exit handlers so Ctrl+C during tests kills
  the detached process group instead of orphaning it

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* Add KML import support for MMGIS vector layers

- Install @tmcw/togeojson dependency for KML-to-GeoJSON conversion
- Add isKmlUrl helper and fetchKmlAsGeoJSON to LayerCapturer.js
- Wrap default URL fetch and dynamic extent fetch with KML detection
- Export isKmlUrl for unit testing
- Create sample KML file with Points, LineString, and Polygon
- Add KML Sample layer to Reference Mission config
- Update configure UI and docs to mention KML support
- Add E2E tests for KML layer loading and toggling
- Add unit tests for isKmlUrl helper function

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* Fix README parent layer counts after adding KML Sample layer

- Total layers: 44 -> 45
- Vector layers: 36 -> 37
- GeoJSON Data Features: 19 -> 20
- Update description to mention KML converted to GeoJSON at runtime

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* Update reference-mission

* Update docs, tests, and LayersTool for reorganized Reference Mission config

- README: Update layer listings to match reorganized config structure
  - Geometry Types: Replace Time-Enabled/KML Sample with Arrows/Annotations
  - Feature Property Behavior: Remove Arrows/Annotations, add Hotline Gradient 3D (8 layers)
  - Add Miscellaneous section with KML layer
  - Time Tab: Add Time-Enabled (2 layers)
  - Core Settings Tab: Update to new zoom layer names (3 layers)
  - Attachment - Markers Tab: Add second image layer (2 layers)
  - Update all section counts (18 GeoJSON Data Features, 18 Layer Configuration)
- E2E tests: Update layer name 'KML Sample' -> 'KML', group 'Geometry Types' -> 'Miscellaneous'
- LayersTool.js: Add KML support to raw download export path via isKmlUrl/fetchKmlAsGeoJSON
- LayerCapturer.js: Export fetchKmlAsGeoJSON for reuse

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* Add server-side proxy for external KML URLs to avoid CORS issues

- Add GET /api/utils/fetchProxy endpoint that streams external http/https resources
- Register fetch_proxy in calls.js for client-side use
- Update fetchKmlAsGeoJSON to route absolute URLs through the proxy
- Local/relative KML URLs continue to be fetched directly

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* Revert "Add server-side proxy for external KML URLs to avoid CORS issues"

This reverts commit efe4424e0119093a4a2f8fdc0742289f4edcf5d7.

* Fix ROOT_PATH subpath support for login/admin CSS and asset paths

Pass ROOT_PATH to adminlogin and login template render calls in server.js.
Prefix all asset hrefs/srcs in adminlogin.pug, login.pug, and resetpassword.pug
with ROOT_PATH. Move background-image and font-face URLs from CSS files to
inline styles in pug templates so they can use the ROOT_PATH variable.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* chore: bump version to 4.3.17-20260422 [version bump]

* Add 301 redirect from ROOT_PATH to ROOT_PATH/ for trailing slash fix

When ROOT_PATH is set (e.g. /mmgis), visiting the URL without a trailing
slash would not match the main route and assets would fail to load.
This adds a redirect so /mmgis -> /mmgis/ works correctly.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* Support data layers with plain URL rgba tiles in populateCogScale

- Update populateCogScale early-return guard to allow layer.type === 'data'
- Skip cogTransform check for data layers (they use shader ramps)
- Add units extraction from variables.shader.units for data layers
- Add min/max extraction from layer minValue/maxValue for data layers
- Add color interpolation from shader ramps for data layer legends
- Add populateCogScale call for data layers with colorize shader
- Generate DEM rgba tiles (zoom 10-12) via gdal2customtiles.py --dem
- Add 'Elevation - RGBA Tiles (URL)' data layer to Reference Mission config

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* chore: bump version to 4.3.17-20260422 [version bump]

* Fix adminlogin contours path and add ROOT_PATH to all img tags

- Use /public/images/contours.png for adminlogin background (the old path
  /configure/build/contours.png is behind ensureUser middleware, so
  unauthenticated users get the login page HTML instead of the image)
- Add ROOT_PATH prefix to all img src attributes in login.pug,
  adminlogin.pug, and resetpassword.pug

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* Fix adminlogin contours hidden by html background-color

The background-color on the body,html selector caused the html element
to paint over the body's background-image. Move background-color to the
inline style on body (alongside background-image) so it doesn't conflict.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* Fix crash when hexToRGB returns null for transparent ramp stops

Add null guard for F_.hexToRGB() results in data layer color
interpolation. Ramp stops like 'transparent' are not valid hex
colors, so hexToRGB returns null. Fall back to 'transparent' color
when either endpoint cannot be parsed.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* chore: bump version to 4.3.18-20260422 [version bump]

* chore: bump version to 4.3.18-20260422 [version bump]

* Regenerate DEM tiles with near-composite resampling at zoom 10-13

Previous tiles used average (LANCZOS) resampling which corrupted
IEEE 754 float bytes at edges, producing huge values (6.54e+27).
Now using near-composite resampling to preserve byte-level accuracy.
Extended to zoom level 13 (was 10-12). Updated maxNativeZoom in config.

Co-Authored-By: tariq.k.soliman <tariqksoliman@gmail.com>

* Fix corrupted tile pixels, tighten data layer guard, refresh legend on min/max update

- Post-processed RGBA tiles to replace 23 corrupted edge pixels (from
  resampling) with transparent nodata. All zoom 10-13 tiles now decode
  to reasonable elevations (-0.23 to 267m).
- Tightened populateCogScale guard: only data layers WITH shade…

Author: 3 people