Companion to --validate-{wom,wob,woc,whm,all}: now the .glb files
we emit (and any third-party .glbs we receive) get the same deep-
check treatment.
wowee_editor --info-glb model.glb # show metadata, exit 0
wowee_editor --validate-glb model.glb # show + exit 1 on errors
Shared parser, different verdict policy. Checks:
- 12-byte header: magic = 'glTF' (0x46546C67), version = 2,
totalLength matches actual file size
- JSON chunk: type = 'JSON' (0x4E4F534A), length within file
- BIN chunk: type = 'BIN\0' (0x004E4942), length within file
- asset.version is '2.0' (per-spec required string)
- accessor.bufferView indices in range
- bufferView.byteOffset+byteLength fits within BIN chunk
Reports counts: meshes, primitives, accessors, bufferViews, buffers,
plus chunk sizes. JSON mode emits structured output for CI scripts.
Verified on a real --export-whm-glb output (256-primitive terrain,
929KB total): all checks pass clean. Synthesized a broken file
with corrupted magic bytes: --validate-glb correctly reports
"magic is not 'glTF'" and exits 1.
Visualizing quest chains in plain text rapidly becomes unreadable
past ~10 quests. This emits a DOT file that renders to a labeled
DAG you can paste into a wiki or PR description:
wowee_editor --export-quest-graph custom_zones/MyZone
dot -Tpng custom_zones/MyZone/quests.dot -o quests.png
Node coloring conveys completion-readiness at a glance:
- lightgreen: has objectives + reward (will complete in-game)
- lightyellow: has objectives but no reward (uncommon)
- lightgray: no objectives (won't complete — common authoring bug)
- mistyrose dashed: synthetic node for a quest ID referenced by
nextQuestId but missing from quests.json (broken chain)
Edges:
- solid black: valid chain link
- dashed red labeled 'missing': dangling nextQuestId
Labels include quest ID, title (DOT-escaped for safety), required
level, and XP reward — enough context that the graph stands alone
without needing to cross-reference quests.json.
Verified on a 3-quest zone with chain 1->2->3->999 (last broken):
DOT output has 3 colored nodes (lightgreen for the 2 quests with
objectives, lightgray for the third), 2 solid edges, 1 dashed-red
edge to a synthetic mistyrose <missing> node.
Catches dangling references that --validate (which only checks
open-format file presence) doesn't:
wowee_editor --check-zone-refs custom_zones/MyZone
Zone refs: custom_zones/MyZone
objects checked : 12 (2 missing)
quests checked : 5 (1 bad NPC refs)
FAILED — 3 issue(s):
- object[3] missing: World/Doodad/Tree.m2
- object[7] missing: World/Building/Inn.wmo
- quest[1] 'Hunt' giver 999 not in creatures.json
Two checks:
1. Every objects.json model path resolves on disk in any of the
conventional roots (zone-local, output/, custom_zones/, Data/),
trying both the open (.wom/.wob) and proprietary (.m2/.wmo)
variants, with a case-fold fallback for case-sensitive Linux
filesystems where extracted assets are usually lowercased.
2. Every quest's questGiverNpcId / turnInNpcId references a
creature in creatures.json (when the zone has any). Filter:
only flag IDs < 100000 since production wires upstream NPCs
with 6-digit IDs that legitimately reference outside content.
Errors capped at 30 listed (with full counts in the summary) so
a misconfigured zone with 500 broken refs doesn't drown the
output. Exit 1 on any issue so CI can gate.
Verified: zone with 1 creature, 2 dead model refs, 1 bad NPC
giver — all 3 issues reported with precise indices and exit 1.
Enumerates every external model path a zone references — both directly
placed (objects.json) and indirectly via WOB doodad placements. Useful
when packaging a content pack to confirm every asset will ship:
wowee_editor --list-zone-deps custom_zones/MyZone
Zone deps: custom_zones/MyZone
WOBs scanned : 3
unique paths total : 12
Direct M2 placements (5 unique):
World/Generic/Lamp.m2
World/Generic/Tree.m2 ×8
...
Direct WMO placements (2 unique):
World/StaticObject/Stormwind/HumanInn.wmo
WOB doodad M2 refs (5 unique):
World/Furniture/Chair.wom
World/Furniture/Table.wom
...
Aggregation:
- Counts duplicates so frequently-used assets show '×N' instead of
cluttering the list with N identical lines.
- Walks WOBs in the zone dir recursively (catches buildings/ subdir).
- Also recurses into WOBs referenced by direct WMO placements so
transitive doodad deps surface — matches the runtime's actual load
chain.
- Sorted output (std::map) so diffs across versions are stable.
JSON mode emits per-category arrays for programmatic consumption
(e.g. piping into a script that checks each path exists in Data/).
Verified on a synthesized zone with 3 M2 placements (one duplicated)
+ 1 WMO placement + no WOBs: 3 unique total, Tree.m2 correctly
deduped with ×2 count, JSON arrays well-formed.
Two more drill-down inspectors that complement --info-batches:
wowee_editor --info-textures HumanMale.wom
WOM textures: HumanMale.wom (3 textures)
idx blp png path
0 y y Character/Human/Male/HumanMaleHair.blp
1 y - Character/Human/Male/HumanMaleBody.blp
2 - - Character/Human/Male/HumanMaleEye.blp
wowee_editor --info-doodads Stormwind.wob
WOB doodads: Stormwind.wob (3 placements)
idx scale pos (x, y, z) rot (x, y, z) model
0 1.50 ( 1.0, 0.0, 1.0) ( 0.0, 90.0, 0.0) Furniture/Chair.wom
...
--info-textures: cross-checks each referenced texture against both
the proprietary BLP and the open PNG sidecar on disk (each lookup
tries the literal path AND falls back to Data/<lowercased path>
since WoW assets are case-sensitive on Linux but designers usually
type Mixed Case). The y/- columns let you spot which textures are
missing before --pack-wcp would fail at runtime.
--info-doodads: model path + position + rotation + scale per
instance. Companion to --info-textures (GPU resources) — this tracks
scene composition.
Both have --json mode that emits per-entry records with index for
programmatic consumption. Verified on synthesized model with 3
texture refs + building with 3 doodads; all fields render correctly.
Third and final glTF 2.0 binary exporter — terrain heightmaps now
ship to the modern web 3D ecosystem alongside WOM models and WOB
buildings:
wowee_editor --export-whm-glb custom_zones/MyZone/MyZone_30_30
Mesh layout matches --export-whm-obj exactly (9x9 outer vertex grid
per chunk, 8x8 quads -> 2 tris each, hole bits respected) so the
.glb and .obj of the same source align spatially when overlaid.
Per-chunk primitive scheme:
- One global vertex pool (positions + normals interleaved by section).
- One global index pool packed sequentially.
- Each loaded chunk gets its own primitive sliced from the shared
index bufferView via byteOffset.
- Designers can hide chunks individually in three.js for area-by-area
inspection of large terrains.
v1 simplifications:
- Normals synthesized as +Z (terrain Z-up). Real per-vertex normals
with cross-chunk smoothing is a follow-up; viewers can compute
their own from positions in the meantime.
- No UVs (no per-chunk material yet — that depends on splat alpha
layers which are a separate format problem).
Verified on a 256-chunk scaffolded zone: 20736 verts, 32768 tris,
256 primitives. First chunk's primitive has 384 indices (128 tris
× 3). Position accessor bounds match WoW coord system for tile
(30, 30): min ~(533, 533), max ~(1067, 1067). Total .glb 929KB,
all spec-compliant.
Pairs naturally with --copy-zone for the 'duplicate then re-populate'
templating workflow:
wowee_editor --copy-zone custom_zones/Base 'Variant'
wowee_editor --clear-zone-content custom_zones/Variant --all
# ... fresh content can now be added without seeing the source's
# creatures/objects/quests bleed through.
Selective wipe via flags so you can keep some content:
--clear-zone-content ZONE --creatures # wipe NPCs only
--clear-zone-content ZONE --quests --objects # quests + props
--clear-zone-content ZONE --all # nuke everything
Implementation:
- Deletes (not blank-writes) so subsequent --info-* commands report
'no content' rather than 'total: 0' which falsely implies the
file existed. Missing files are the canonical 'no content' state.
- Refuses to run with no flags (avoids ambiguous 'do you mean
everything?' silent foot-gun).
- Resets ZoneManifest::hasCreatures when wiping creatures so server
module gen doesn't expect an NPC table that's no longer there.
- Reports per-file action (removed / skipped) plus total count.
Verified end-to-end: scaffolded zone, populated all 3 content
files, --creatures wipes one (others remain), --all wipes the
rest while reporting the already-removed one as 'skipped'.
--info shows aggregate batch count; this drills into each one:
wowee_editor --info-batches HumanMale
WOM batches: HumanMale.wom (v3, 2 batches)
idx iStart iCount tris blend flags texture
0 0 3 1 opaque - Body.blp
1 3 3 1 alpha two-sided no-zwrite Hair.blp
Useful for debugging:
- 'why is this submesh transparent?' -> check blendMode column
- 'why is the back face missing?' -> check flags for two-sided
- 'why is depth wrong?' -> check no-zwrite flag
- 'which batch has the bad UV?' -> indexStart/indexCount range
- 'why does this material show wrong texture?' -> textureIndex + path
Decodes:
- Blend modes: 0=opaque, 1=alpha-test, 2=alpha, 3=add
- Flags: 0x01=unlit, 0x02=two-sided, 0x04=no-zwrite
Falls back to a 'no batches (WOM1/WOM2 single-material model)' note
when called on an older-version model that doesn't have a batch
table.
Verified on a synthesized 2-batch model (opaque body + alpha hair
with two-sided + no-zwrite flags): all fields decoded correctly,
table aligns, JSON output enumerable.
Symmetric to --clone-creature and --clone-quest. Common workflow:
place one tree just right (rotation, scale dialed in), then clone N
copies along a path:
for x in 100 110 120 130 140; do
wowee_editor --clone-object $Z 0 $x 0 0
done
Defaults match --clone-creature: 5-yard X offset prevents z-fighting.
Rotation and scale carry over verbatim (so a tilted barrel stays
tilted). uniqueId is reset so the new placement doesn't collide with
the source's identifier in any downstream system that dedups by it.
CRUD-clone surface is now complete:
--clone-quest <zone> <questIdx> [newTitle]
--clone-creature <zone> <idx> [newName] [dx dy dz]
--clone-object <zone> <idx> [dx dy dz]
Verified: scaffolded zone, added a tree at (100, 200, 50, scale=1.5).
Cloned twice (default offset, custom offset). Result: 3 trees with
correctly-shifted positions and preserved scale=1.50.
Mirrors --export-glb (WOM -> .glb) for the WOB format. Buildings now
also reach the modern web 3D viewer ecosystem with zero conversion:
wowee_editor --export-wob-glb House # -> House.glb
wowee_editor --export-wob-glb House out.glb
Mapping for multi-group buildings:
- Per-group vertex arrays merged into a single global pool packed
into the BIN chunk (positions, normals, UVs interleaved by section).
- Each group becomes one primitive in a single mesh.
- Per-group indices offset by the group's vertex base so the merged
pool indexing still resolves to the right vertices.
- Per-group indices accessor sliced from a shared bufferView via
byteOffset (no buffer duplication).
- mode=4 (TRIANGLES), uint32 indices, vec3 float positions/normals,
vec2 float UVs — same layout as --export-glb.
Verified on a 2-group building (4-vert floor + 3-vert wall, 9
indices total): output .glb has 7 verts, 2 primitives with the
right per-group index counts (6 floor, 3 wall) sliced from the
shared 36-byte index bufferView. BIN = 7*32 + 9*4 = 260 bytes.
OBJ is universal but ancient (1992) — it can't carry skinning,
animations, or PBR materials. glTF 2.0 (2017, Khronos) is the
modern industry standard: every browser-based 3D viewer
(Sketchfab, Three.js, Babylon.js, model-viewer) consumes it
natively, plus Unity/Unreal import it cleanly.
wowee_editor --export-glb Tree # -> Tree.glb
wowee_editor --export-glb Tree out.glb
Shipping WOM through .glb means our open binary format is viewable
in any modern web tool with zero conversion friction. Big win for
the open-format ecosystem reach.
Implementation (single-file binary .glb):
- 12-byte header (magic 'glTF', version 2, totalLength)
- JSON chunk (0x4E4F534A 'JSON', padded to 4-byte boundary with spaces)
- BIN chunk (0x004E4942 'BIN\0')
- BIN layout: positions (vec3 float) | normals (vec3 float) |
uvs (vec2 float) | indices (uint32). 32 bytes/vert keeps the
index region naturally 4-byte aligned for free.
- Per WOM3 batch: one primitive with its own indices accessor
(sliced via byteOffset on a single shared bufferView).
- Position accessor includes min/max bounds for viewer auto-framing.
v1 limitations (deliberate):
- Bones / animations not yet emitted. glTF's joint matrix layout
differs from WOM's bone tree and needs a careful re-mapping pass;
shipping geometry-first means designers can use the format today
and the animation pass lands as a follow-up.
- No materials / textures emitted (those come from the texture
sidecars; future work to embed or reference them).
Verified: WOM(3 verts, 1 tri) -> .glb(108-byte BIN, 856-byte JSON,
1116-byte total). JSON is spec-compliant glTF 2.0 with correct
bufferView byteOffsets (0/36/72/96), componentTypes (5126=FLOAT,
5125=UNSIGNED_INT), and primitive mode=4 (TRIANGLES). Will open in
any glTF viewer without modification.
Shell-script-equivalent of looping over every zone in a project dir,
in a single editor invocation. Substitutes '{}' in the command with
each zone's path (find -exec convention):
wowee_editor --for-each-zone custom_zones -- \
wowee_editor --validate-all {}
wowee_editor --for-each-zone custom_zones -- \
wowee_editor --info-creatures {}/creatures.json
Implementation:
- Walks projectDir for child dirs containing zone.json (canonical
'is this a zone?' test the rest of the editor uses).
- Sorts results so output is deterministic across runs.
- Each token after '--' is shell-escaped via single-quote wrapping
with embedded ' replaced by '\'' — safe against names with spaces
and quotes.
- {} replacement done per-token so paths with spaces don't get split.
- fflush(stdout) before std::system so the [zone] header lands above
the child output instead of after (parent stdout buffered, child
unbuffered to terminal).
- Returns failed-run count as exit code (capped at 255 for POSIX).
Verified: scaffolded 2 zones, ran --info-creatures across both via
for-each-zone. Output ordered correctly with [zone] headers + child
output, exit 0 when all succeed.
Designer workflow: archetype one 'patrol guard' with stats + faction
+ behavior + equipment dialed in, then stamp it across spawn points
around a town:
for x in 100 110 120 130; do
wowee_editor --clone-creature $Z 0 "Guard $x" $x 200 50
done
Defaults give safe behavior:
- Position offset by (5, 0, 0) yards so the copy doesn't z-fight
the original when no offset is passed.
- Name appended with ' (copy)' if no newName given.
- id reset (auto-assigned by NpcSpawner).
- Patrol path NOT offset — patrol points are world-space waypoints,
not relative to the spawn; cloning that vector unchanged matches
designer intent. Re-author the path on the clone if needed.
Verified: scaffolded zone, added Guard, cloned twice (default
offset, custom name + 3D offset). list-creatures shows 3 spawns
with correctly differentiated positions and names.
Completes the proprietary-format inspector lineup. Pairs naturally
with --info-wot / --info-whm (the open WOT/WHM equivalents) so users
can verify the conversion preserves chunk counts, doodad placements,
and WMO references:
wowee_editor --info-adt World/Maps/Azeroth/Azeroth_32_48.adt
ADT: ...
version : 18
file bytes : 450192
coord : (0, 0)
chunks loaded : 256/256
height range : [-0.00, 0.00]
hole chunks : 0 (with cave/gap masks)
water chunks : 0
textures : 0
doodad names : 0 (0 placements)
wmo names : 1 (1 placements)
Reports loaded chunk count (out of fixed 256), height min/max across
all loaded chunks (with NaN guard so corrupted heights don't poison
the range), hole chunks (cave/gap masks), water chunks, texture and
doodad/WMO name table sizes, and placement counts.
Verified on a real WoW ADT (ahnqirajtemple_29_46.adt, 450KB):
correctly reports 256/256 chunks loaded, 1 WMO name + 1 placement.
JSON mode emits structured output for CI scripts.
The format-inspector lineup is now complete:
Proprietary: BLP / DBC / M2 / WMO / ADT
Open: PNG / JSON DBC / WOM / WOB / WOC / WOT/WHM / WCP
Every format on both sides of the open-format bridge has an inspector.
Common designer workflow: build a base quest with objectives + rewards
once, then make N variants ('Slay Wolves' / 'Slay Bears' / 'Slay
Tigers') with the same shape. Doing this through individual --add-*
commands means re-typing all objectives + items each time. --clone-quest
deep-copies an existing quest in one shot:
wowee_editor --clone-quest $Z 0 # 'Foo' -> 'Foo (copy)'
wowee_editor --clone-quest $Z 0 'Slay Bears' # custom title
Carries:
- All objectives (deep copy via vector value-copy)
- All item rewards
- XP / coin reward fields
- requiredLevel, giver, turnIn
Resets:
- id (set to 0; addQuest auto-assigns a fresh one)
- nextQuestId (cleared — chaining the clone to the same next quest
would corrupt chain semantics; user can re-set it explicitly)
Verified: scaffolded zone, added quest with 2 objectives + 1 item +
xp=250. Cloned twice (default name, custom name). Result: 3 quests
in list, both clones have full objective list and reward intact.
Round out the format-inspector lineup. The wowee open formats had
inspectors (--info-wom, --info-wob); these are the proprietary
counterparts that pair with --convert-m2 / --convert-wmo so users
can verify what the conversion preserves vs drops:
wowee_editor --info-m2 Character/Human/Male/HumanMale.m2
wowee_editor --info-wmo World/wmo/Stormwind/Stormwind.wmo
--info-m2 reports verts/tris, bones, sequences (animations),
batches, textures, materials, attachments, particles, ribbons,
collision tris, and bound radius. Auto-merges <base>00.skin if
present (WotLK+ M2s store geometry there) so vertex/index counts
match what gets rendered.
--info-wmo reports group count + portals + lights + doodads +
materials + textures + total verts/tris across loaded groups.
Auto-merges matching <base>_NNN.wmo group files; pre-resizes the
groups vector so loadGroup populates the right slots.
Verified against real WoW assets:
nexusraid_skya.m2: v264, 20917 verts, 22940 tris, 44 bones,
1 sequence, 44 batches, 28 textures, 42 materials.
ed_zd_ziggurat.wmo: v17, 1 group (1 loaded), 8 materials, 7
textures, 4609 verts, 3650 tris from the group file.
Bug caught during testing: initial snprintf used an 8-byte buffer
for '_NNN.wmo' (which is 8 chars + NUL = 9), silently truncating
to '_000.wm' and failing every group lookup. Bumped to 16 bytes
with a comment so the trap doesn't get re-stepped.
The --info-* suite covers wowee open formats and their JSON/PNG
sidecars. BLP itself was the gap — the proprietary source format
that asset_extract converts FROM. Useful for verifying source
files before --convert-blp-png:
wowee_editor --info-blp Texture.blp
BLP: Texture.blp (BLP2)
size : 128 x 128
channels : 4
format : BLP2
compression: DXT5
mip levels : 16
file bytes : 23044
decoded RGBA bytes: 65536
Magic-checks the first 4 bytes (BLP1 or BLP2) before invoking the
loader so feeding a non-BLP gets a clear 'not a BLP1/BLP2 file'
message instead of a confusing 'invalid' from the decoder. Reports
both file size (compressed-on-disk) and decoded RGBA byte count
(width*height*4) so users can see the compression ratio at a glance.
JSON mode emits the same fields for CI/scripting.
Verified on a real WoW BLP (pvp-banner-emblem-1.blp): correctly
identifies as BLP2 / DXT5 / 128x128 / 16 mips. Total format-
inspector lineup is now complete (BLP/PNG/DBC-bin/DBC-json plus
all 5 wowee binary formats).
--list-quests shows the quest roster but doesn't drill into objectives
or item rewards (--info-quests just counts). These complete the
detail picture and unblock --remove-quest-objective by exposing the
0-based objIdx the user needs:
wowee_editor --list-quest-objectives custom_zones/Z/quests.json 0
Quest 0 ('Hunt'): 3 objective(s)
idx type count target description
0 kill 5 Wolf Slay 5 Wolf
1 collect 3 Pelt Collect 3 Pelt
2 talk 1 Mayor Talk to Mayor
wowee_editor --list-quest-rewards custom_zones/Z/quests.json 0
Quest 0 ('Hunt') rewards:
xp : 999
coin : 5g 30s 99c
items : 2
[0] Item:Sword
[1] Item:Shield
Both have --json mode for CI/scripting. Range-check questIdx and
exit 1 on out-of-range with a precise message.
Verified: scaffolded zone, added quest with 3 objectives + 2 items
+ full coin reward; both listings show the right data, JSON output
is well-formed and enumerable.
Round out the tile lifecycle CLI. --add-tile shipped earlier; these
are the symmetric counterparts:
wowee_editor --list-tiles custom_zones/MyZone
wowee_editor --list-tiles custom_zones/MyZone --json
wowee_editor --remove-tile custom_zones/MyZone 31 30
--list-tiles: shows tile coords AND on-disk presence of .whm/.wot/.woc
per tile, so you can spot manifest-vs-disk drift before pack-wcp would
complain. JSON mode emits per-tile records for programmatic checks.
--remove-tile: drops the manifest entry AND deletes the .whm/.wot/.woc
files for that tile so the zone stays consistent (no orphan sidecars).
Refuses to remove the last tile — server module gen and pack-wcp both
expect at least one, so a zero-tile zone would just be a footgun. Use
'rm -rf custom_zones/X/' for total removal.
Verified end-to-end: scaffolded zone, added 2 more tiles, built WOC
on one. --list-tiles shows 3 tiles with correct file presence (y/y/y
on the built one, y/y/- on the others). Removed (31, 30) — 2 files
deleted, 2 tiles remaining. Tried removing last tile after dropping
to 1: correctly refused with exit 1.
Companion to --convert-dbc-json and --convert-json-dbc. Refresh one
PNG sidecar without re-running asset_extract --emit-open across the
whole tree:
wowee_editor --convert-blp-png Texture.blp # -> Texture.png
wowee_editor --convert-blp-png Texture.blp out.png # custom path
Same code path as the asset_extract emitter (BLPLoader::load +
stbi_write_png), with the same dimension/buffer guards (rejects
0x0 or >8192x8192, rejects images where data is shorter than
width*height*4).
Verified on a real WoW BLP (Data/interface/pvp-banner-emblem-1.blp,
128x128 RGBA): converted cleanly, --info-png on the result reports
'128 x 128 / 8-bit / rgba (4 channels) / 24136 file bytes' (PNG
zlib compression on a sparse banner texture).
Symmetric counterpart to --add-quest-objective. Quest design is
iterative — when an objective gets reworked or trimmed, you need
to remove the wrong one without nuking the entire quest.
wowee_editor --remove-quest-objective <zoneDir> <questIdx> <objIdx>
Both indices are 0-based and reported by --info-quests / --list-quests.
Range-checks both: questIdx against quest count, objIdx against the
selected quest's objective count. Each errors with a precise out-of-
range message.
Verified: scaffolded zone, added quest with 3 objectives (kill /
collect / talk), removed objective at index 1 (the collect),
--info-quests confirms '1 kill, 0 collect, 1 talk' afterward.
Bad quest and bad objective indices both rejected with exit 1.
Quest-authoring CLI is now fully symmetric:
--add-quest --remove-quest
--add-quest-objective --remove-quest-objective
--add-quest-reward-item (additive only — items are dedup'd)
--set-quest-reward (idempotent field-level update)
--scaffold-zone creates a zone with one tile; some zones (continent
fragments, large dungeons) span multiple ADT tiles. This extends an
existing zone:
wowee_editor --add-tile custom_zones/MyZone 29 30 # default baseHeight=100
wowee_editor --add-tile custom_zones/MyZone 28 31 250.5 # custom baseHeight
What it does:
- Generates a fresh blank-flat WHM/WOT pair via the same factory
--scaffold-zone uses, so output is consistent.
- Appends (tx, ty) to ZoneManifest::tiles. Save() rebuilds the
files-block from tiles, so the new adt_TX_TY entry appears
automatically in zone.json.
Safety:
- Tile coord must be in WoW grid [0, 64) per axis; rejects 99,99.
- Refuses if the tile is already in the manifest (catches typos).
- Refuses if the .whm/.wot files exist on disk but aren't in the
manifest (catches manifest-out-of-sync drift from hand edits).
- Optional baseHeight allows seeding flat terrain at a non-default
elevation.
Verified end-to-end: scaffolded 1-tile zone, added 2 more tiles
(one with custom height). Result: 3 tiles in manifest, 6 files on
disk, files-block has all 3 adt_TX_TY entries. Duplicate and
out-of-range cases both rejected with exit 1.
Standalone DBC <-> JSON converters. asset_extract --emit-open already
does this in bulk during extraction, but designers often need to
refresh ONE sidecar after a hand-edit, or push an edited JSON back to
binary DBC for private-server compat:
wowee_editor --convert-dbc-json Spell.dbc # -> Spell.json
wowee_editor --convert-json-dbc Spell.json # -> Spell.dbc
wowee_editor --convert-dbc-json Spell.dbc out.json # custom output
DBC -> JSON: uses the wowee-dbc-json-1.0 schema (matches asset_extract
output exactly so the editor's runtime overlay loader picks it up).
Same string/float/uint heuristic the existing emitter uses.
JSON -> DBC: builds a Blizzard-format-compatible binary DBC with:
- 20-byte WDBC header (magic + recordCount + fieldCount + recordSize +
stringBlockSize)
- Records as little-endian uint32 fields (always LE per spec)
- Deduped string block with leading NUL so offset=0 = empty string
- Tolerates JSON numeric types (string/float/int/bool/null) and
derives fieldCount from row size if header omits it
Verified with a hand-built 3-record/2-field DBC: DBC->JSON->DBC->JSON
round-trips with identical records[] arrays, no data loss. Output
is byte-loadable by the existing DBC parser.
Completes the open-format -> universal-text bridge for the last
binary geometry format. WHM was the missing one; designers now have
OBJ exports for all four (WOM models, WOB buildings, WOC collision,
WHM terrain).
wowee_editor --export-whm-obj custom_zones/MyZone/MyZone_30_30
Mesh layout:
- 9x9 outer vertex grid per chunk (skips the 8x8 inner verts the
engine uses for 4-tri fans). That's 81 verts and 128 tris per
chunk; full ADT = 20736 verts + 32768 tris.
- One OBJ 'g chunk_X_Y' per MapChunk so designers can hide chunks
individually in Blender (e.g. to inspect a single problem area).
- Hole bits respected — cave-entrance quads correctly disappear.
- Coords match WoweeCollisionBuilder's outer-grid layout exactly,
so an --export-whm-obj and --export-woc-obj of the same source
align spatially when overlaid in Blender. (Verified: first vertex
of both is (1066.67, 1066.67, ~98.5) for a tile (30, 30) export.)
- UVs are simply row/8, col/8 in [0,1] per chunk so a checker
texture renders at the canonical scale for size reference.
Verified: scaffolded zone -> WHM/WOT auto-built -> --export-whm-obj
produces 256 chunks loaded, 20736 verts, 32768 faces, 256 'g'
blocks. Counts exactly match the chunk × outer-grid math.
--copy-zone duplicates and renames; --rename-zone does it in place
without doubling disk usage. Useful when fixing typos or rebranding
a zone without touching its data:
wowee_editor --rename-zone custom_zones/Old 'Brand New Name'
What changes:
- zone.json mapName: Old -> Brand_New_Name
- zone.json displayName: 'Old' -> 'Brand New Name'
- zone.json files block (adt_NN_NN, wdt): regenerated from new slug
- slug-prefixed files: Old_28_30.whm -> Brand_New_Name_28_30.whm
- the directory itself: custom_zones/Old/ -> custom_zones/Brand_New_Name/
Order of operations matters for crash safety: slug-prefixed files
get renamed first (atomic per-file via fs::rename), then the
manifest is rewritten in the still-named source dir, then the
directory is moved last. If the dir-move fails we surface the
manifest-already-updated state so the user can recover.
Refuses to run if target dir already exists (avoids silent merge),
or if both slug AND displayName already match the target (no-op).
Verified end-to-end: scaffolded 'Old' with 1 creature, renamed to
'Brand New Name'. Result: dir renamed, .whm/.wot files renamed,
zone.json fully updated, creatures.json preserved with 1 entry.
WOC is the open collision format used for movement queries. When the
player gets stuck or walks somewhere they shouldn't, you need to SEE
the mesh — eyeballing flag bits in --info-woc only goes so far.
wowee_editor --export-woc-obj custom_zones/Z/Z_30_30.woc
Each flag class gets its own OBJ 'g' block so designers can hide
categories independently in Blender to debug:
- walkable / steep / water / indoor / nonwalkable
- and combinations like 'walkable_water' for shallow swimable areas
Implementation notes:
- Triangle-soup topology preserved (no vertex dedupe). Adjacent
triangles often have different flags; deduping would merge
categories and lose the boundary.
- Bucket by exact flag value (uint8) so combination flags become
their own group rather than counting twice.
- Index math: triangle t vertex k -> OBJ index (t*3 + k + 1).
- Header comment preserves source path, triangle counts, walkable/
steep counts, and tile (x, y) for provenance.
Verified on a freshly-scaffolded zone's auto-built WOC (32768
triangles, all walkable on flat terrain): export reports
'1 flag class(es)', single 'g walkable' block in the OBJ, last
face index correctly = 32768*3 = 98304.
Closes the WOB <-> universal-format round trip, mirroring what
--import-obj does for WOM. Workflow now works end-to-end for
buildings:
asset_extract # WMO -> WOB (open binary)
--export-wob-obj # WOB -> OBJ (universal text)
... edit in Blender / MeshLab / Maya ...
--import-wob-obj # OBJ -> WOB (back to engine format)
--validate-wob # confirm consistency
Mapping handles:
- Each OBJ 'g name' starts a new WOB group; faces under it become
that group's indices. Default 'imported' group catches faces
before any 'g' directive (raw OBJ files from non-DCC sources).
- Per-group dedupe table on (pos, uv, normal) triples — each WOB
group has its own local vertex array, so the global OBJ index
pool gets remapped per group.
- '_outdoor' suffix stripped + isOutdoor flag set, mirroring
--export-wob-obj's naming convention.
- Doodad placements recovered from the # doodad ... comment lines
--export-wob-obj writes; round-trip preserves them via re-parse.
- UV V flipped back (1.0 - v) so the round trip is exact.
- Per-group bounds + global bound radius computed from positions.
Verified: WOB(2 groups, 7 verts, 3 tris, 1 doodad) -> OBJ ->
WOB(2 groups, 7 verts, 3 tris, 1 doodad). validate-wob clean,
info-wob shows the same counts. Materials and portals don't
survive the OBJ trip (no semantic equivalent) — that's documented.
Closes the quest-authoring CLI: --add-quest creates a quest, then
--add-quest-objective adds completion conditions, and now these add
the rewards.
wowee_editor --add-quest-reward-item $Z 0 'Item:Sword' 'Item:Shield' 'Item:Potion'
wowee_editor --set-quest-reward $Z 0 --xp 999 --gold 5 --silver 30
--add-quest-reward-item greedy-consumes any trailing positional args
that don't start with '-', so a whole loot table can be added in one
invocation rather than N round-trips through load+save.
--set-quest-reward updates only the fields explicitly passed (--xp /
--gold / --silver / --copper) — avoids the round-trip-and-clobber
footgun of a 'replace whole reward' command. Field flags are
order-independent. At least one flag is required (else error).
Both validate questIdx against quest count and bail with exit 1 on
out-of-range / missing file / bad numeric value.
Verified: scaffolded zone, added quest with default xp=100, batch-
added 3 item rewards in one shot, then set xp=999/gold=5/silver=30
in one shot. --info-quests reports total XP=999, with-items=1.
PNG (BLP sidecar) and JSON DBC (DBC sidecar) didn't have inspectors —
the existing --info-* suite only covered the binary native formats
(WOM/WOB/WOC/WOT/WHM/WCP). These fill the gap so debugging the
asset_extract --emit-* output doesn't need GIMP / a JSON pretty-printer:
wowee_editor --info-png Textures/Sky01.png
PNG: Textures/Sky01.png
size : 256 x 256
bit depth : 8
color : rgba (4 channels)
file bytes: 142336
wowee_editor --info-jsondbc db/Spell.json
JSON DBC: db/Spell.json
format : wowee-jsondbc-1
source : Spell.dbc
records : 47882 (header) / 47882 (actual)
fields : 234
PNG inspector reads only the IHDR chunk (24 bytes total) — no pixel
decode — so it works instantly on huge files. Validates the
8-byte signature, parses big-endian width/height, derives channel
count from PNG color type (grayscale/rgb/palette/grayscale+alpha/
rgba per spec table 11.1).
JSON DBC inspector parses with nlohmann::json, reports the schema
fields (format/source/recordCount/fieldCount), and cross-checks
recordCount against actual records[] array length. Exits 1 on
count mismatch — catches truncated extracts where the header lies
about how much data follows. Verified with hand-rolled 2x3 RGBA
PNG (correct dims) and JSON DBC files (one matched, one mismatched
99 vs 1).
WOM (the open M2 replacement) already round-trips through OBJ; this
extends the universal-format bridge to WOB (the open WMO replacement)
so buildings can also be edited in Blender / MeshLab / Maya / etc.
wowee_editor --export-wob-obj House # writes House.obj
wowee_editor --export-wob-obj House out.obj # custom path
Mapping decisions:
- Each WOB group becomes one OBJ 'g' block (named after the group;
outdoor groups get an '_outdoor' suffix). Preserves the room/floor
structure for downstream selection and per-area editing.
- Single global vertex pool with per-group offsets (OBJ requires v
indices to be globally 1-based; we track a running vertOffset).
- UV V flipped (1.0 - v) so texturing matches Blender bottom-left
convention, same as --export-obj for WOM.
- Doodad placements written as # comment lines at the end. OBJ has
no native concept for instanced models, but emitting them as
structured comments keeps the placement data recoverable for
tools that want to re-instance them.
- Portals and material flags drop on the floor — OBJ has no
semantics for either. The native WOB always remains canonical.
Verified on a synthesized 2-group house (4-vert floor + 3-vert wall,
1 doodad): output OBJ has 7 verts / 7 vt / 7 vn entries, 2 'g'
blocks with proper index offsetting, doodad comment line preserved.
--add-quest creates a quest shell but quests with no objectives never
complete in-game. This fills the gap:
wowee_editor --add-quest-objective <zoneDir> <questIdx> \
<kill|collect|talk|explore|escort|use> <targetName> [count]
Workflow:
Z=custom_zones/MyZone
wowee_editor --add-quest $Z 'Hunt Wolves' 100 100 250 5
wowee_editor --add-quest-objective $Z 0 kill 'Wolf' 5
wowee_editor --add-quest-objective $Z 0 collect 'Wolf Pelt' 3
Auto-generates a description from type+count+name so addons and
tooltips show something sensible ('Slay 5 Wolf', 'Collect 3 Wolf
Pelt', 'Talk to Mayor'). Designers can hand-edit quests.json after
the fact for bespoke prose; the auto-text is the floor, not the
ceiling.
Quest index matches --list-quests output (0-based). Out-of-range
index, unknown type, or missing quests.json all error with clear
messages and exit 1.
Verified: scaffolded zone, added 2 quests + 3 objectives across
them; --info-quests reports '1 kill, 1 collect, 1 talk'. Bad type
and bad index both rejected.
The natural counterpart to --diff-wcp (which compares two .wcp
archives) — operates on the unpacked side of the workflow. Shows
exactly what changed across zone.json fields, creature roster,
object placements, and quest list:
wowee_editor --diff-zone custom_zones/Base custom_zones/Variant
Diff: custom_zones/Base vs custom_zones/Variant
manifest : 2 field diff(s)
~ mapName: 'Base' -> 'Variant'
~ displayName: 'Base' -> 'Variant'
creatures : 2 vs 2
- Bear
+ Tiger
quests : 1 vs 2
+ Defeat the Tiger
Pairs naturally with --copy-zone: template a base zone, fork a
variant, then diff to see exactly what was customized. Useful for
PR review when a designer modifies a zone — diff against the
upstream version to scope the change.
Comparison strategy: sorted set diff on stable identifying fields
(creature.name, object.path, quest.title). This intentionally hides
position/orientation changes since those are continuous and would
flag every pixel-perfect tweak as a diff — content-level changes
are the signal here.
Exit 0 if identical, 1 otherwise (so CI can gate). JSON mode emits
per-category onlyA/onlyB arrays + manifestDiffs list + totalDiffs
count for programmatic consumption.
Verified: diffed two zones forked from a common base (one with
creature swap + new quest); reported 5 diffs across manifest +
creatures + quests with exit 1. Same zone vs itself reports
IDENTICAL with exit 0.
Closes the open-format authoring loop:
asset_extract # M2 -> WOM (open binary)
--export-obj # WOM -> OBJ (universal text)
... edit in Blender / MeshLab / ZBrush / Maya ...
--import-obj # OBJ -> WOM (back to engine format)
The same WOM file ships in custom zones, gets validated by
--validate-wom, and renders identically through the existing pipeline
— no proprietary M2 ever needs to touch the authoring path.
Parser handles:
- v / vt / vn pools, deduped on (pos, uv, normal) triples so the
resulting WOM vertex buffer stays compact
- 1-based AND negative (relative) face indices
- f tokens in v, v/t, v//n, and v/t/n forms
- Triangles, quads, and convex n-gons (fan-triangulated)
- CRLF line endings
- Reverses --export-obj's V flip (1.0 - v) so UVs round-trip exactly
- Auto-computes boundMin/Max/Radius from positions (renderer culls
by these — wrong values make the model disappear)
- Output WOM is WOM1 (static); bones/anims/material flags don't
exist in OBJ and stay empty by design
Verified end-to-end: WOM -> OBJ -> WOM -> validate-wom yields a
5-vert, 6-tri pyramid back identical to the input. Bounds, vertex
count, index count, and name all preserved.
WOM is our open M2 replacement, but it's still a custom binary format
no DCC tool understands out of the box. OBJ is the universally
supported text format that opens directly in Blender, MeshLab,
ZBrush, Maya — basically every 3D tool ever made:
wowee_editor --export-obj Tree # writes Tree.obj
wowee_editor --export-obj Tree out.obj # custom output path
This closes the loop for content authors:
asset_extract -> WOM (open binary) -> OBJ (universal text) ->
edit in Blender -> back to WOM via a future --import-obj.
Layout details that matter for downstream tools:
- 1-based face indices (OBJ standard)
- UV V flipped (1.0 - v) so texturing matches between Vulkan
top-left and Blender bottom-left conventions
- Per-batch groups when WOM3 batches exist, named with the
texture basename so material assignment carries through
- Single 'mesh' group for WOM1/WOM2 models
- Header comment preserves provenance (source, version, counts)
Verified on a synthesized 5-vert pyramid (4 base + apex, 6 tris):
output OBJ has 5 v / 5 vt / 5 vn entries, 6 f lines, opens cleanly
in MeshLab. Build green, ctest 31/31.
Verbose enumeration of every entry, complementing the existing
--info-* (summary counts) and unblocking --remove-* (which takes a
0-based index).
wowee_editor --list-creatures custom_zones/Z/creatures.json
creatures.json: custom_zones/Z/creatures.json (2 total)
idx name lvl display pos (x, y, z)
0 Wolf 7 11430 (100.0, 200.0, 50.0)
1 Black Bear 12 11445 (110.0, 210.0, 55.0)
The 'idx' column is exactly what --remove-creature/object/quest
takes, closing the workflow loop:
wowee_editor --list-creatures $Z/creatures.json |
grep -i 'wolf' | awk '{print $1}' |
xargs -I{} wowee_editor --remove-creature $Z {}
Each command has matching --json mode that emits per-entry records
with index, key fields, and position. Verified on a synthesized
zone with 2 creatures + 2 objects + 2 quests; tables align, indices
match, JSON is well-formed and round-trips through jq cleanly.
Symmetric counterparts to --add-{creature,object,quest} from earlier
batches:
wowee_editor --remove-creature <zoneDir> <index>
wowee_editor --remove-object <zoneDir> <index>
wowee_editor --remove-quest <zoneDir> <index>
Index is 0-based and is reported in the corresponding --info-* output;
nothing identifies entries reliably across reloads, so name-based
removal would silently delete the wrong row when duplicates exist.
Pair them in scripts:
idx=$(wowee_editor --info-creatures $Z/creatures.json --json |
jq '.spawns | map(.name) | index("Wolf")')
wowee_editor --remove-creature $Z $idx
Each prints the removed entry's name/path and the new total so
scripts can verify the right row went away. Out-of-range indices
exit 1 with a clear message.
Verified end-to-end: scaffolded zone, added 3 creatures + 2 objects
+ 2 quests, removed one of each by index, totals correctly went
3->2/2->1/2->1. Bad index 99 properly errors out.
--info-extract reports sidecar coverage % per format, but doesn't say
which files are missing. This makes it actionable:
wowee_editor --list-missing-sidecars Data/
Output: one path per line, prefixed with the missing extension so
shell tools can filter:
png Data/Textures/Skybox/Sky01.blp
json Data/DBFilesClient/SoundEntries.dbc
wom Data/Character/Human/Male/HumanMale.m2
Pipe into xargs to drive a targeted re-extract:
wowee_editor --list-missing-sidecars Data/ |
awk '/^png/ {print $2}' |
xargs asset_extract --emit-png-only
Skips WMO group files (Foo_NNN.wmo) since only the parent file gets
a .wob sidecar — they would otherwise inflate the missing list with
hundreds of false positives per WMO.
Exit 1 when anything is missing (so CI can gate). JSON mode emits
arrays per format type for programmatic consumption. Verified
against synthetic dir with 5 files (4 lacking sidecars + 1 with
.png present): all 4 reported, the one with sidecar correctly
omitted.
Round out the per-format validator suite. Open-format zone validation
now covers all four binary formats:
--validate-wom Tree
--validate-wob House
--validate-woc terrain.woc
--validate-whm Zone_28_30
--validate-all custom_zones/Zone1 # runs everything
WOC checks: finite vertex coords on every triangle, no degenerate
triangles (two verts identical), known flag bits only (0x0F mask),
tile coords within WoW grid (< 64), bounds.min <= bounds.max.
WHM/WOT checks: finite heights across all 145 verts/chunk, finite
chunk position vectors, tile coord in [0, 64), reasonable height
envelope ([-10000, 10000] is a generous outer bound — beyond that
suggests units confusion), placements have finite positions and
nameId within doodadNames/wmoNames table size.
validate-all now reports all four format counts (WOM/WOB/WOC/WHM)
and aggregates errors. Verified end-to-end: a fresh scaffolded zone
with --build-woc yields 256/256 chunks loaded, 32768 walkable
triangles, validate-all PASSED. Synthesized WOC with 0xFF flags
correctly fails with 'unknown flag bits 0xFF' and exit 1.
Walks a directory recursively and runs the deep validators on every
.wom and .wob it finds. Single CI gate for an entire zone tree:
wowee_editor --validate-all custom_zones/MyZone --json
Reports per-file failures (capped at first 20 to keep output bounded)
plus aggregate counts so you know which file to drill into with
--validate-wom or --validate-wob individually.
Refactor: pulled the validation bodies out of --validate-wom and
--validate-wob into static helpers (validateWomErrors / validateWobErrors)
returning vector<string>. The per-file commands now share the same
logic as --validate-all — fix one, fix all three. ~200 lines of
duplicate validation code consolidated.
Verified end-to-end: seeded /tmp dir with 2 WOMs (1 with DAG bone
violation) + 1 valid WOB, --validate-all reports 'WOM: 2 total, 1
failed' / 'WOB: 1 total, 0 failed' with the bad file's full path and
error printed below. JSON mode emits per-file failure list for CI.
Companion to --validate-wom; same load-is-lenient story for buildings:
wowee_editor --validate-wob House [--json]
Cross-references checked per group:
- indices.size() divisible by 3
- every index < vertices.size()
- material textures non-empty
- group boundMin <= boundMax per axis
Building-level:
- portal groupA/groupB references real groups (or -1 = exterior)
- portal polygon has >= 3 vertices
- doodad modelPath non-empty
- doodad scale finite and > 0
- boundRadius >= 0
Errors batched (caps to 3 listed plus '... and N more') so a thousand
bad indices in one group don't drown the report. JSON mode emits the
same error list for CI consumption. Exit 1 on any failure so shell
scripts can gate on it.
Verified against a synthesized 2-group building with intentional
flaws across every category — caught all 4 (group indices count,
empty material texture, short portal polygon, empty doodad model)
and exited 1.
The WOM loader is intentionally lenient — it clamps out-of-range
indices to 0, resets bad bone parents to -1, and skips invalid
batches. That keeps broken files from crashing the renderer, but
also hides corruption that authoring scripts should catch BEFORE
the file ships.
wowee_editor --validate-wom Tree [--json]
Cross-references checked (none auto-fixed by load()):
- bone DAG order: parent must be strictly less than self index
- animation boneKeyframes count == bone count when both nonzero
- batch indexStart+Count <= total indexCount
- batch indexCount divisible by 3
- batch textureIndex < texturePaths size
- boundMin <= boundMax per axis, boundRadius >= 0
- header version in [1,3], indices count divisible by 3
Verified against a synthesized 3-bone model with parent=self+1
(invalid DAG order) — load() preserves it as written, validator
reports 'bone 1 parent=2 not strictly less (DAG order)' and
exits 1. JSON mode emits errorCount + errors[] + passed boolean
for CI scripts to gate on.
Duplicate an existing zone to a new slug:
wowee_editor --copy-zone custom_zones/Original "My New Zone"
Workflow this enables: scaffold one base zone, populate it with
creatures/objects/quests, then copy-zone N times to create variants
without re-scaffolding each. Designers can template a 'forest base'
zone and stamp it into Dark Forest, Frozen Forest, etc.
What it does:
- Recursive copy preserves any subdirs (e.g. data/ for DBC sidecars)
- Reads source slug from zone.json (not the dir name) to know what
prefix to rewrite — handles users who renamed dirs without
touching the manifest
- Renames slug-prefixed files (Original_28_30.whm -> NewSlug_28_30.whm,
matches both _-suffixed and .-suffixed forms)
- Saves a fresh zone.json via ZoneManifest::save which rebuilds the
files-block from mapName, so the manifest references the renamed
files correctly
Verified end-to-end: scaffolded Original, added creature + quest,
copied to 'My New Zone'. Result: 2 files renamed, zone.json
mapName/displayName/files all updated, creatures.json + quests.json
copied verbatim.
Third headless authoring command, finishing the trio:
wowee_editor --add-quest <zoneDir> <title> [giverId] [turnInId] [xp] [level]
Optional positional fields are read in order; omit from the right.
Bare 'wowee_editor --add-quest zone Title' produces a valid quest
with default values, matching the editor GUI's New Quest behavior.
Verified: appended 3 quests (250+100+0 XP) to a scaffolded zone,
--info-quests --json reports total=3, totalXp=450, withReward=3.
Mirrors --add-creature for the object placer:
wowee_editor --add-object <zoneDir> <m2|wmo> <gamePath> <x> <y> <z> [scale]
Lets shell scripts populate objects/buildings without the GUI:
for tree in 100,200 150,250 200,300; do
x=${tree%,*}; y=${tree#*,}
wowee_editor --add-object "$zone" m2 'World/Doodad/Tree.m2' $x $y 0
done
Loads existing objects.json first then appends, so multiple
invocations build up. Optional scale slot lets the caller set
non-default size (clamped to >0 by the load-time guard).
Verified end-to-end: scaffolded zone → added m2 + wmo → info-objects
reports total=2 with the correct scale range.
Appends a single creature spawn to a zone's creatures.json. First
real authoring tool that doesn't need the GUI placement system —
useful for batch-populating zones via shell script:
for npc in goblin spider wolf; do
wowee_editor --add-creature "$zone" "$npc" 100 200 50
done
Args: <zoneDir> <name> <x> <y> <z> [displayId] [level]
- displayId 0 → SQL exporter substitutes 11707 (generic humanoid)
- level defaults to 1
- Coordinates are render-space (renderX=wowY, renderY=wowX)
Loads any existing creatures.json first then appends, so multiple
invocations build up the file. The standard NPC spawner caps
(50k creatures) protect against runaway scripts.
Schema:
asset_extract --purge-proprietary Data --json
{
"dir": "Data",
"confirmed": false,
"candidates": 21570,
"removed": 0,
"totalBytes": 17175328768
}
"candidates" = files where a sidecar exists and is at least as
new (would be deleted on --confirm-purge). "removed" reflects the
actual delete count when --confirm-purge is set; 0 in dry-run.
Lets CI scripts gate on the dry-run candidate count before actually
purging:
count=$(asset_extract --purge-proprietary Data --json | jq .candidates)
[ "$count" -gt 1000 ] && asset_extract --purge-proprietary Data --confirm-purge
The asset_extract --json flag now applies to both upgrade and purge
modes.
Mirrors the wowee_editor --info-extract --json schema so CI scripts
can use a single jq filter for both pre-check and post-upgrade:
asset_extract --upgrade-extract Data --json
{
"dir": "Data",
"elapsedSeconds": 47.2,
"skipped": 0,
"png": { "ok": 12340, "failed": 0 },
"jsonDbc": { "ok": 240, "failed": 0 },
"wom": { "ok": 8500, "failed": 0 },
"wob": { "ok": 730, "failed": 0 },
"terrain": { "ok": 5660, "failed": 0 }
}
Now both wowee_editor and asset_extract have JSON-mode summaries
covering every command CI is likely to gate on. The extraction
pipeline is fully scriptable end-to-end.
