/* ============================================================================ 8LovePatterns — Adaptive scoring engine (MVP, calibration-ready) ---------------------------------------------------------------------------- Pure logic. The item bank lives in items.jsx (window.LP_RAW_ITEMS), so weights stay easy to read and edit there. This file implements: • weighted multidimensional scoring (8 archetypes on 9 latent dimensions) • Anchor as a TRANSVERSAL regulation factor (never a 9th profile) • grip (emprise) of the dominant mechanism, moderated by the Anchor • internal confidence + 3-phase adaptive selection + intelligent stop 40–60 Weights are hand-set priors, meant to be re-calibrated later on real data. 8LovePatterns is a relationship self-reflection tool — not a clinical diagnosis. The engine works in the bank's archetype namespace (French keys), and maps the final dominant to the site's short codes (inc/gue/cam/sau/bas/fug/alc/mir). ========================================================================== */ /* ---- Latent dimensions -------------------------------------------------- */ const LP_DIMENSIONS = [ 'pursuit_intensity','hypervigilance','self_adaptation','rescuing', 'emotional_fortress','withdrawal','intellectualization','external_validation', 'anchor_regulation', ]; /* Archetype keys used inside the bank → primary latent dimension */ const LP_ARCH_DIM = { incendiaire:'pursuit_intensity', guetteur:'hypervigilance', cameleon:'self_adaptation', sauveur:'rescuing', bastion:'emotional_fortress', fugitif:'withdrawal', alchimiste:'intellectualization', miroir:'external_validation', }; const LP_ARCH = Object.keys(LP_ARCH_DIM); /* Bank archetype name → site short code (used by Result / Profile / data.jsx) */ const LP_CODE = { incendiaire:'inc', guetteur:'gue', cameleon:'cam', sauveur:'sau', bastion:'bas', fugitif:'fug', alchimiste:'alc', miroir:'mir', }; /* Grand strategy per archetype (public copy) */ const LP_STRATEGY = { incendiaire:'I pursue', miroir:'I pursue', guetteur:'I control', alchimiste:'I control', cameleon:'I erase myself', sauveur:'I erase myself', bastion:'I protect myself', fugitif:'I flee', }; /* 5-level Likert; converted to -2..+2 (lpVal) */ const LP_LIKERT = [ { v:5, label:'Completely' }, { v:4, label:'Somewhat' }, { v:3, label:'In between' }, { v:2, label:'Not really' }, { v:1, label:'Not at all' }, ]; function lpVal(likert){ return (likert||3) - 3; } // 1..5 → -2..+2 /* Adaptive bounds */ const LP_MIN = 40; const LP_MAX = 60; /* ---- Bank indexes ------------------------------------------------------- */ const LP_ITEMS = (window.LP_RAW_ITEMS || []).slice(); const LP_BY_ID = {}; LP_ITEMS.forEach(it => { LP_BY_ID[it.id] = it; }); /* The archetype an item most strongly raises (max positive weight) */ function lpPrimaryArch(item){ let best=null, bw=0; for (const k in item.archetypeWeights){ const w=item.archetypeWeights[k]; if(w>bw){ bw=w; best=k; } } return best; } /* For an anchor_contextualized item, the mechanism it regulates = the (dampened) archetype key present in its weights (most negative). */ function lpAnchorContext(item){ let ctx=null, mn=0; for (const k in item.archetypeWeights){ const w=item.archetypeWeights[k]; if(w i.phase==='core'); const mech = core.filter(i => i.itemType==='mechanism' || i.itemType==='situation'); const rest = core.filter(i => i.itemType!=='mechanism' && i.itemType!=='situation'); const byArch = {}; LP_ARCH.forEach(a => byArch[a] = mech.filter(i => lpPrimaryArch(i)===a)); const order = []; for (let round=0; round<4; round++){ LP_ARCH.forEach(a => { if(byArch[a][round]) order.push(byArch[a][round].id); }); } // weave the remaining core items (anchors + contrasts) roughly every 5 rest.forEach((it,k)=>{ const pos = Math.min(order.length, 4 + k*5); order.splice(pos, 0, it.id); }); return order; })(); /* ============================================================================ SCORING — recomputed from scratch on every answer (cheap, deterministic). ========================================================================== */ function LP_score(answers){ const archRaw={}, archMax={}, archPos={}, archFacets={}; const dimRaw={}, dimMax={}; let anchorRaw=0, anchorMax=0, anchorItems=0; // anchorItems = dedicated anchor items only const anchorCtxByArch={}; // normalized anchor responses per regulated archetype const contrastPairs={}; // answered contrast items per "a|b" LP_ARCH.forEach(a=>{ archRaw[a]=0; archMax[a]=0; archPos[a]=0; archFacets[a]=new Set(); }); LP_DIMENSIONS.forEach(d=>{ dimRaw[d]=0; dimMax[d]=0; }); for (const id in answers){ const it = LP_BY_ID[id]; if(!it) continue; const val = lpVal(answers[id]); // -2..+2 // 1–3 · archetypes (positive AND negative weights both count) for (const k in it.archetypeWeights){ const w = it.archetypeWeights[k]; if(archRaw[k]===undefined){ archRaw[k]=0; archMax[k]=0; archPos[k]=0; archFacets[k]=new Set(); } archRaw[k] += val*w; archMax[k] += 2*Math.abs(w); if(w>0){ archPos[k] += 1; if(it.facet) archFacets[k].add(it.facet); } } // 4 · latent dimensions (primary full, secondary half) if(it.primaryDimension && dimRaw[it.primaryDimension]!==undefined){ dimRaw[it.primaryDimension]+=val; dimMax[it.primaryDimension]+=2; } if(it.secondaryDimension && dimRaw[it.secondaryDimension]!==undefined){ dimRaw[it.secondaryDimension]+=val*0.5; dimMax[it.secondaryDimension]+=1; } // 5 · Anchor (any item with anchorWeight contributes; only dedicated items are "counted") if(it.anchorWeight){ anchorRaw += val*it.anchorWeight; anchorMax += 2*Math.abs(it.anchorWeight); if(it.itemType==='anchor_direct' || it.itemType==='anchor_contextualized'){ anchorItems++; if(it.itemType==='anchor_contextualized'){ const ctx = lpAnchorContext(it); if(ctx) (anchorCtxByArch[ctx]=anchorCtxByArch[ctx]||[]).push((val+2)/4*100); } } } // 7 · contrast pair bookkeeping if(it.itemType==='contrast' && it.discriminatesBetween && it.discriminatesBetween.length===2){ const key = it.discriminatesBetween.slice().sort().join('|'); contrastPairs[key]=(contrastPairs[key]||0)+1; } } // normalize raw[-max,+max] → 0..100 (50 = neutral / unmeasured) const norm = (raw,max)=> max>0 ? Math.round(((raw+max)/(2*max))*100) : 50; const archScore={}; LP_ARCH.forEach(a=> archScore[a]=norm(archRaw[a],archMax[a])); const dimScore={}; LP_DIMENSIONS.forEach(d=> dimScore[d]=norm(dimRaw[d],dimMax[d])); const anchorScore = norm(anchorRaw, anchorMax); const ranking = LP_ARCH.map(a=>({ code:a, score:archScore[a], items:archPos[a]||0, facets:(archFacets[a]?archFacets[a].size:0) })).sort((x,y)=> y.score-x.score); const top1=ranking[0], top2=ranking[1], top3=ranking[2]; const gap = top1.score - top2.score; return { archScore, dimScore, anchorScore, anchorItems, anchorCtxByArch, contrastPairs, ranking, top1, top2, top3, gap, archFacets, archPos, answeredCount: Object.keys(answers).length }; } /* ---- Grip (emprise) of the dominant mechanism --------------------------- * emprise ≈ activation of the leader + intensity − contextual Anchor for that same leader − general Anchor. High Anchor → the pattern is more "integrated". */ function LP_grip(state){ const top1 = state.top1.code; const activation = state.top1.score; // 0..100 const ctx = state.anchorCtxByArch[top1]; const ctxAnchor = (ctx && ctx.length) ? (ctx.reduce((s,v)=>s+v,0)/ctx.length) : null; let grip = activation - 0.45 * Math.max(0, state.anchorScore - 50) // general regulation - 0.40 * (ctxAnchor!=null ? Math.max(0, ctxAnchor - 50) : 0); // regulation of THIS mechanism grip = Math.round(Math.max(0, Math.min(100, grip))); let level = 'Emerging'; if(grip >= 66) level = 'Very active'; else if(grip < 42) level = 'Integrated'; return { gripScore:grip, gripLevel:level, ctxAnchor }; } /* ---- Internal confidence in the dominant result ------------------------- */ function LP_confidence(state){ const strength = state.top1.score; const sep = Math.min(state.gap*5, 100); const coverage = Math.min((state.top1.facets/4)*100, 100); return Math.round(Math.max(0, Math.min(100, 0.45*strength + 0.40*sep + 0.15*coverage))); } /* ============================================================================ PHASE 3 · INTELLIGENT STOP ========================================================================== */ function LP_shouldStop(state, answeredCount){ if(answeredCount >= LP_MAX) return true; if(answeredCount < LP_MIN) return false; const conf = LP_confidence(state); const top1Items = state.top1.items; // items touching the leader const facets = state.top1.facets; // distinct facets covered const anchorOK = state.anchorItems >= 3; // ≥3 dedicated Anchor items const pairClose = state.gap < 10; const pairKey = [state.top1.code, state.top2.code].sort().join('|'); const contrastForPair = state.contrastPairs[pairKey] || 0; if(top1Items < 4) return false; // ≥4 items on the leader if(facets < 3) return false; // several facets covered if(!anchorOK) return false; // Anchor measured enough if(pairClose && contrastForPair < 2) return false; // close pair needs ≥2 contrasts if(conf >= 70 && state.gap >= 10) return true; // confident & separated → stop return false; } /* ============================================================================ PHASE 2 · ADAPTIVE SELECTION — pick the most informative next item. ========================================================================== */ function LP_selectNext(state, asked){ const askedSet = new Set(asked); const unused = LP_ITEMS.filter(i => !askedSet.has(i.id)); if(!unused.length) return null; const t1 = state.top1.code, t2 = state.top2.code, t3 = state.top3.code; const top = [t1,t2,t3]; const coveredT1 = state.archFacets[t1] || new Set(); // 1 — close leaders → separate them with a contrast item if(state.gap < 12){ let c = unused.filter(i => i.itemType==='contrast' && i.discriminatesBetween.includes(t1) && i.discriminatesBetween.includes(t2)); if(c.length) return c[0].id; c = unused.filter(i => i.itemType==='contrast' && (i.discriminatesBetween.includes(t1) || i.discriminatesBetween.includes(t2))); if(c.length) return c[0].id; } // 2 — leader clear but Anchor under-measured → contextualized Anchor on the leader if(state.anchorItems < 3 || !(state.anchorCtxByArch[t1])){ let c = unused.filter(i => i.itemType==='anchor_contextualized' && lpAnchorContext(i)===t1); if(c.length) return c[0].id; c = unused.filter(i => i.itemType==='anchor_direct' || i.itemType==='anchor_contextualized'); if(c.length) return c[0].id; } // 3 — broaden facet coverage of the leader let c = unused.filter(i => (i.archetypeWeights[t1]||0) > 0 && i.facet && !coveredT1.has(i.facet)); if(c.length) return c[0].id; // 4 — otherwise, the item that informs the most-probable profiles c = unused.filter(i => Object.keys(i.archetypeWeights).some(k => top.includes(k) && i.archetypeWeights[k] > 0)); if(c.length) return c[0].id; // 5 — fallback return unused[0].id; } /* Estimated length label for the progress UI (adaptive, not fixed) */ function LP_estimatedTotal(){ return 'usually 40–60'; } /* ============================================================================ PUBLIC RESULT — one dominant profile is shown; everything else stays internal. ========================================================================== */ function LP_deriveResult(answers){ const state = LP_score(answers); const grip = LP_grip(state); const conf = LP_confidence(state); const a = state.anchorScore; let anchorLabel = 'Developing'; if(a >= 66) anchorLabel = 'Steady'; else if(a < 42) anchorLabel = 'Still forming'; const dims = LP_DIMENSIONS.filter(d=>d!=='anchor_regulation') .map(d=>({ d, s:state.dimScore[d] })).sort((x,y)=>y.s-x.s); const toCode = r => ({ code:LP_CODE[r.code], arch:r.code, score:r.score }); return { dominant: LP_CODE[state.top1.code], // ← the only public field used by Result strategy: LP_STRATEGY[state.top1.code], // ----- internal data (kept for future calibration; not shown publicly) ----- top: [toCode(state.top1), toCode(state.top2), toCode(state.top3)], archScore: state.archScore, dimScore: state.dimScore, anchorScore: state.anchorScore, anchorLabel, gripScore: grip.gripScore, gripLevel: grip.gripLevel, // 'Very active' | 'Emerging' | 'Integrated' confidence: conf, strongDims: dims.slice(0,3).map(x=>x.d), weakDims: dims.slice(-3).map(x=>x.d), askedCount: state.answeredCount, facetsCovered: state.top1.facets, }; } /* ============================================================================ PREMIUM . resultData (saved to localStorage for the premium report page) ---------------------------------------------------------------------------- The premium report (premium-report.html) is a separate page. It does NOT run the engine; it only reads the object we save here. This keeps the funnel simple: the test computes everything once, stores it, the report reads it. We reuse the rich scores the engine already produced (no second scoring system). The 8 measured archetype scores become user-facing "trigger dimensions", plus the Anchor. Everything is normalized 0..100. ========================================================================== */ /* localStorage key used by the premium report page. */ const LP_RESULT_KEY = 'lovepattern_resultData'; /* Bank archetype name → premium trigger-dimension id (1:1, both are measured). */ const LP_DIM_OF_ARCH = { incendiaire:'chaudFroid', guetteur:'silence', alchimiste:'hyperanalyse', sauveur:'sauvetage', bastion:'controle', fugitif:'distance', cameleon:'effacement', miroir:'flouRelationnel', }; /* Anchor level from the 0..100 anchor score, per the product thresholds: 0..39 → Très actif (very active) 40..69 → En progression (in progress) 70..100→ Intégré (integrated) */ function LP_anchorLevel(anchorScore){ if(anchorScore >= 70) return 'Intégré'; if(anchorScore >= 40) return 'En progression'; return 'Très actif'; } /* Build the resultData object from a full engine result (LP_deriveResult). */ function LP_buildResultData(result){ const A = result.archScore || {}; const dimensions = {}; for(const arch in LP_DIM_OF_ARCH){ dimensions[LP_DIM_OF_ARCH[arch]] = (A[arch] != null ? A[arch] : 50); } dimensions.ancre = result.anchorScore != null ? result.anchorScore : 50; // top 3 trigger dimensions (Anchor excluded) by score → declencheurs principaux const declencheursPrincipaux = Object.keys(dimensions) .filter(id => id !== 'ancre') .sort((a,b) => dimensions[b] - dimensions[a]) .slice(0,3); const dominantArch = (result.top && result.top[0] && result.top[0].arch) || 'bastion'; return { profilDominant: dominantArch, // full bank key (e.g. "incendiaire") code: result.dominant, // short code (e.g. "inc") scoreProfil: (result.top && result.top[0] && result.top[0].score) || 50, strategy: result.strategy || '', niveauAncre: LP_anchorLevel(dimensions.ancre), gripLevel: result.gripLevel || 'Emerging', dimensions, declencheursPrincipaux, // array of dimension ids dateTest: new Date().toISOString(), }; } /* Build + persist. Called by App.onResult right after the test finishes. */ function LP_saveResultData(result){ try { const data = LP_buildResultData(result); localStorage.setItem(LP_RESULT_KEY, JSON.stringify(data)); return data; } catch(e){ return null; } } Object.assign(window, { LP_DIMENSIONS, LP_ARCH, LP_ARCH_DIM, LP_CODE, LP_STRATEGY, LP_LIKERT, LP_MIN, LP_MAX, LP_ITEMS, LP_BY_ID, LP_CORE_ORDER, LP_score, LP_grip, LP_confidence, LP_shouldStop, LP_selectNext, LP_estimatedTotal, LP_deriveResult, LP_RESULT_KEY, LP_DIM_OF_ARCH, LP_anchorLevel, LP_buildResultData, LP_saveResultData, });