Skip to content

Rebuild cata and legacy_cata area data from a 4.3.4 dump - #821

Merged
Xian55 merged 4 commits into
devfrom
feature/cata-area-data
Aug 2, 2026
Merged

Xian55 merged 4 commits into
devfrom
feature/cata-area-data

Conversation

@Xian55

@Xian55 Xian55 commented Aug 2, 2026

Copy link
Copy Markdown
Owner

Brings the Cataclysm clients onto the dump pipeline the Mists clients already use, and fixes two script bugs found doing it. No addon changes.

The data

cata and legacy_cata were still on Wowhead-era files predating the pipeline. Rebuilt from TrinityCore 4.3.4 (TDB_full_world_434.22011) — the era those clients actually run — rather than the SkyFire 5.4.8 dump the four clients share for creatures.json.

That choice matters more than it looks. SkyFire's old-world Kalimdor is thin:

SkyFire 5.4.8 TDB 4.3.4
spawns inside Feralas 1575 4721
flagged NPCs there 13 74

Thin enough that Feralas, Felwood and Searing Gorge produced no file at all — a failure mode with no error, just a zone that silently has no vendors.

cata 88 zones, legacy_cata 87. Ironforge and Undercity gain files of their own: the area grid projects outdoor terrain only, so their NPCs used to convert against the enclosing zone's bounds and land inside the mountain — Ironforge's Gryphon Master at [67.2, 26.9] instead of [55.5, 47.7]. Vashj'ir, Gilneas City and both Stranglethorn halves gain files too.

Zone-hood is per era

A legacy_ client's WorldMapArea is not always from its own era. legacy_cata is built from 8.1.0.27826 — deliberately, since 4.4.2 shifts the UiMapId space the addon's Lua table maps against — and that build reports Cataclysm's starting valleys as zones owning their own map.

A 4.3.4 client does not. Standing in Coldridge Valley it reports Dun Morogh. Splitting those NPCs into 6176.json created a file nothing ever opens and stripped them out of the file the client does read: Dun Morogh fell from 90 entries to 21.

The non-legacy sibling is era-correct and matches what each client reports — 4.4.2 has no Coldridge row at all, 5.5.4 has one — so it now decides what counts as a zone, while bounds stay with the client's own file. Drops 201 areas for legacy_cata (the valleys, Emberstone Mine, New Tinkertown) and keeps Ironforge, Undercity, Gilneas City, Vashj'ir and the Stranglethorn halves. Provably a no-op for legacy_mop, whose mapped set is identical to mop's.

Column casing

TrinityCore writes gameobject_template.Data0, SkyFire data0. rows() compared exactly and aborted on a dump that had the column all along. Now folds case, exact spelling still wins.

Testing

Reference values hold on all four clients: Innkeeper Farley in Elwynn [43.77, 65.8] (the documented validation value) and Ironforge's Gryth Thurden [55.5, 47.74]. Coldridge Valley correctly has a file on the Mists clients and none on the Cataclysm ones — matching what each client reports in game. Re-running reproduces the committed output.

scripts/CLAUDE.md records which dump each era uses, the zone-hood rule and the casing trap; none of the three were discoverable from the code.

Known gap, not addressed here

legacy_mop/mop still have no data for Feralas, Felwood, Searing Gorge, Un'Goro, Mount Hyjal or Silithus, since SkyFire is the right source for those clients otherwise (it carries Pandaria). A --fill-from second-dump option looked like the answer but currently yields gathering nodes without their NPCs, so it is deliberately left out of this PR.

🤖 Generated with Claude Code

Xian55 and others added 4 commits August 2, 2026 20:49
TrinityCore writes gameobject_template.Data0, SkyFire writes data0. rows()
compared exactly, so the area extractor aborted on a 4.3.4 dump that carried
the column all along:

  `gameobject_template` has no column(s) ['data0']

Exact spelling still wins, so a table holding both spellings is unambiguous.
Cores are known to drift on names as well as case - faction_A, npc_rank - and
those still need explicit candidates; this only removes the casing trap.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
A legacy_ client's WorldMapArea is not always from its own era. legacy_cata is
built from 8.1.0.27826 - deliberately, because 4.4.2 shifts the UiMapId space
the addon's Lua table maps against - and that build calls Cataclysm's starting
valleys zones with their own maps.

A 4.3.4 client does not. Standing in Coldridge Valley it reports Dun Morogh, so
splitting those NPCs into 6176.json produced a file nothing ever opens while
1.json lost the NPCs that are actually reachable there. Dun Morogh fell from 90
entries to 21.

The non-legacy sibling is era-correct and says exactly what each client reports:
4.4.2 carries no Coldridge row at all, 5.5.4 carries one. So the sibling decides
what counts as a zone; bounds still come from the client's own file, because that
is what AreaDB and the Leaflet page convert with.

Drops 201 areas for legacy_cata - the valleys, Emberstone Mine, New Tinkertown -
while keeping Ironforge, Undercity, Gilneas City, Vashj'ir and both Stranglethorn
halves. Provably a no-op for legacy_mop, whose mapped set is identical to mop's.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Both clients were on Wowhead-era files predating the dump pipeline. Rebuilt from
TrinityCore 4.3.4 (TDB_full_world_434.22011), which is the era they actually run,
rather than the SkyFire 5.4.8 dump the four clients share for creatures.json:
SkyFire's old-world Kalimdor is thin enough that Feralas, Felwood and Searing
Gorge came out with no file at all.

cata 88 zones, legacy_cata 87. Ironforge and Undercity gain files of their own -
the grid projects outdoor terrain only, so their NPCs used to convert against the
enclosing zone's bounds and land inside the mountain. Vashj'ir, Gilneas City and
both Stranglethorn halves gain files too. The starting valleys correctly do not:
a 4.3.4 client reports Dun Morogh standing in Coldridge Valley.

Verified against the documented reference and across clients: Innkeeper Farley in
Elwynn reads [43.77, 65.8], and Ironforge's Gryph Master reads [55.5, 47.74] on
all four clients.

scripts/CLAUDE.md records which dump each era uses, the zone-hood rule and the
column-casing trap, since none of the three were discoverable from the code.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Cataclysm split 5339 into Northern Stranglethorn (33) and The Cape of
Stranglethorn (5287), so a 4.3.4 client never reports 5339 and nothing opens
5339.json. Both halves have their own file, rebuilt from the 4.3.4 dump.

The 128 entries dropped here were Wowhead-era data plotted against 5339's bounds,
which span both halves - the exact miscalibration extract-area-sqldump.py was
written to avoid.

Left 5146.json alone: 4.4.2 does map Vashj'ir, so a client may still report it
even though the dump files its content under the three sub-areas.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
@Xian55 Xian55 added bugfix This pull request fixes an issue. ai Artificial Intelligence tool has been contributed. labels Aug 2, 2026
@Xian55
Xian55 merged commit 6d4100f into dev Aug 2, 2026
1 check passed
@Xian55
Xian55 deleted the feature/cata-area-data branch August 7, 2026 10:30
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

ai Artificial Intelligence tool has been contributed. bugfix This pull request fixes an issue.

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant