Buhen — Expansion to Nubia

The entire mission — configuration block, variables, and all event handlers — lives in one file, imported by src/scripts/missions.js. Key parts of the mission13 { … } block (verified vs mission1.pak scenario 13, 2026-07-26 dump):

mission13 { // Buhen — Expansion to Nubia
  player_rank : 5
  next_mission : 15          // converges on North Dahshur

  // pak Normal funds=10000 loan=5000 debt_interest=20 → int_dcy row (verified)
  initial_funds [20000, 13300, 10000, 6700, 5300]
  debt_interest [10, 15, 20, 25, 30]

  // Gods fixed to Ptah + Seth per pak (GOD_PTAH:1, GOD_SETH:2) — Osiris/Ra removed
  // pop 3000, culture 25, prosperity 25, monuments 9 (pak raw 6 → formula), kingdom 75, housing 10
  win_criteria {
    population    { enabled : true, goal : 3000 }
    culture       { enabled : true, goal : 25 }
    prosperity    { enabled : true, goal : 25 }
    monuments     { enabled : true, goal : 9 }
    kingdom       { enabled : true, goal : 75 }
    housing_level { enabled : true, goal : 10 }
  }
}

There is no choice[]: Buhen and South Dahshur are the two arms of a fork the player picks at the end of mission 11/12; both set next_mission : 15. Empire id is set to 13 via empire.set_id(13) in mission13_on_start (m_013_buhen.js:629) together with a full hide_pak_* cities/routes/ornaments/texts redefine (six openable trading partners — Men-nefer, Selima Oasis, Abu, Meidum, Enkomi, Serabit Khadim — plus seven display-only cities).

Handlers fire only while mission=mission13 is active. Raids use the JS API city.start_foreign_army_invasion with ENEMY_6_KUSHITE; native EVENT_TYPE_INVASION binds exist (B2), but this mission still uses JS poll until B2-migrate. Every pak event index that isn't skipped junk gets its own event_advance_month handler, gated by a once-flag or a per-year "last fired" guard for recurring roots:

• event_mission_start line 620 Start message; empire.set_id(13); city.set_scenario_enemy_id(ENEMY_6_KUSHITE); advisors up to Diplomacy; wires every dormant chain leaf up front. mission13_on_start
• Shared helper builders lines 537–603 Dormant ONLY_VIA_EVENT chain leaves reused by most requests: the three shared KR leaves (+6 / −8 / −4), the granite gift chain kicked off by the year-7 beer festival, and the troops chain kicked off by the year-4 troops request (message → KR+30 → 400-troop re-arm, or refuse/late → status/KR/price cascade). mission13_ensure_shared_kr_leaves / mission13_ensure_granite_gift_leaves / mission13_ensure_troops_chain_leaves
• Once calendar roots lines 654–853 Limestone ladder y1m6 / y3m2 / y4m10 / y7m2 / y9m11 / y11m7; troops y4m7; granite demand y5m1 / y10m0; oil y6m1; beer festival y7m4 (granite gift chain). mission13_event_i<n>_*
• Recurring calendar roots lines 668 / 943 Beer y2m9+, idle-gated so a fresh request never stacks on an unresolved one; clay pit floods every year from y1m0, hitting up to 8 working pits per year. mission13_event_i3_beer_recurring / mission13_clay_pit_flood_recurring
• Kushite raids + Pharaoh favour lines 758–937 / 969 Eight one-shot ENEMY_6_KUSHITE invasions (size 8→36, years 6–20, attack targets vary); a monthly favour tick drops a size-22 Pharaoh army when kingdom rating bottoms out, chaining to size-21 then size-28 waves — the pak's three-wave by_favour chain, mirrored locally since the shared favour helper only supports a single chained wave. mission13_event_i<n>_kushite_invasion / mission13_pharaoh_favour_invasion

Dump helper: src/js/js_test_mission_pak_dump.cpp (__test_mission_pak_dump / GamestateIO::load_mission_pak_raw; ad-hoc tests/99_tmp_mission13_dump.js, deleted after verification).

Popup dialogs and the opening history panel are defined in src/scripts/game_messages_en.js. To edit text, find the entry by key and change the text: field inside its content { … } block.

Audio: Voice/Mission/213_mission.mp3 and Voice/Mission/213_victory.mp3.

• Monuments: pak raw goal is 6; TEMP goal is 9 — one finished small obelisk (weight 2) already produces rating 9 under the additive formula (2.25×2+4.5) in src/scripts/city/monuments.js. Reverts to 6 once per-type weights are recalibrated against the original point table (pending weight recalibration).
• Invasion points omitted: pak inv_land_count / inv_sea_count are both 0 — the engine falls back to the map exit tile for every Kushite raid (scenario_start_invasion_impl).
• Troops defeat chain: set_defeat_action_tag is wired; fulfilling defers to a linked distant battle whose egyptian strength comes from empire-service bataillons (B8 — not equal-abstract). No / weak army → lose → i=36→47; strong enough → ok chain. Refuse/late still skip the battle and fire i=34. Hardcoded kingdom swings from fight_distant_battle are suppressed when request-linked. Smoke: tests/179_troops_distant_battle.js.
• Troops MESSAGE / CITY_STATUS subtypes: i=33 MESSAGE subtype 0 = city saved (already handled). i=34/36 CITY_STATUS subtype 0 = Egyptian city falls — engine now closes trade, marks the city foreign-occupied, and posts the falls message (was a silent no-op / AUD1).
• Empire routes 8/9/10 skipped: the pak polylines exist but no map object references them — orphan data, the same pattern seen in earlier missions.
• Oil×2000: the year-6 request amount is unusually large but matches the pak value exactly — not a transcription error.
• Invasions and Pharaoh's favour are proxied in JS (city.start_foreign_army_invasion) rather than the native EVENT_TYPE_INVASION — same pattern as every mission before B2-migrate.
• Next mission 15 (North Dahshur) is scripted (m_015_north_dahshur.js); both Buhen and South Dahshur set next_mission : 15.

The map layout, terrain, and the kingdom road entry point come from the original Pharaoh campaign pack:

mission1.pak (scenario offset for id 13 — see GamestateIO::get_campaign_scenario_offset, src/io/gamestate/boilerplate.cpp:120)

The binary map is loaded by the engine before the JS script runs. The scripting system is loaded from src/scripts/missions.js, which imports mission/m_013_buhen.js by path. If you rename or move the file, update the import there.