ui changes

src/lib/candidateIntelligence.js

@@ -0,0 +1,235 @@
+/**
+ * Candidate intelligence — every derived figure the profile shows.
+ *
+ * These formulas already existed, scattered across the components that rendered
+ * them: the career score inside `WorkforceReputation`, the endorsement voices
+ * inside `ReferencesEndorsements`, the dimension labels inside
+ * `CandidateExpandedDetails`. Rendering was rewritten; the arithmetic was not.
+ * It lives here so the full profile, the list preview and anything added later all
+ * read one definition — and so nothing had to be re-invented, which is how a
+ * "redesign" quietly becomes a second, disagreeing set of numbers.
+ *
+ * Pure: a candidate record in, derived values out. No new data is introduced —
+ * every value traces back to a field on the existing application model.
+ */
+
+import { getScoreBand, toFICO } from '@/lib/talentHome';
+
+/* ── Score dimensions ───────────────────────────────────────────────────── */
+
+/**
+ * The 15 screening dimensions, grouped into the four things a hiring decision
+ * actually turns on.
+ *
+ * Ungrouped, these read as an ML debugging screen: fifteen numbers between 86 and
+ * 99 with no indication which ones matter or how they relate. Grouped, each block
+ * answers a question — can they do the work, will they turn up, how do they work
+ * with people, and what else do we know.
+ *
+ * Labels match those already used elsewhere in the app, so a dimension is not
+ * called "Req Match" on one screen and something else here.
+ */
+export const SCORE_GROUPS = [
+  {
+    id: 'core',
+    label: 'Core match',
+    description: 'Can they do this job',
+    dimensions: [
+      { key: 'experience', label: 'Experience' },
+      { key: 'certifications', label: 'Certifications' },
+      { key: 'employer_requirements', label: 'Requirement fit' },
+      { key: 'verified_skills', label: 'Skills' },
+    ],
+  },
+  {
+    id: 'reliability',
+    label: 'Reliability',
+    description: 'Will they turn up and hold the standard',
+    dimensions: [
+      { key: 'reliability', label: 'Reliability' },
+      { key: 'availability', label: 'Availability' },
+      { key: 'attendance_expectations', label: 'Attendance' },
+      { key: 'scenario_judgment', label: 'Judgment' },
+    ],
+  },
+  {
+    id: 'style',
+    label: 'Work style',
+    description: 'How they work with a team',
+    dimensions: [
+      { key: 'communication_style', label: 'Communication' },
+      { key: 'culture_fit', label: 'Culture fit' },
+      { key: 'leadership_expectations', label: 'Leadership' },
+    ],
+  },
+  {
+    id: 'additional',
+    label: 'Additional signals',
+    description: 'Everything else screening captured',
+    dimensions: [
+      { key: 'english', label: 'English' },
+      { key: 'personality', label: 'Personality' },
+      { key: 'job_related_answers', label: 'Job knowledge' },
+      { key: 'physical_requirements', label: 'Physical' },
+    ],
+  },
+];
+
+/** Every dimension key, in display order — used to prove none is dropped. */
+export const ALL_DIMENSIONS = SCORE_GROUPS.flatMap((g) => g.dimensions);
+
+/* ── Workforce reputation ───────────────────────────────────────────────── */
+
+const clampScore = (n) => Math.max(0, Math.min(100, Math.round(n || 0)));
+
+/** Career score on the 300–850 scale. Same formula as the talent-side profile. */
+export const careerScore = (candidate) => toFICO(clampScore(candidate?.ai_score));
+
+/** Career tier — Elite, Excellent, Solid, Building, New. */
+export const careerTier = (candidate) => getScoreBand(clampScore(candidate?.ai_score));
+
+/**
+ * Potential market value.
+ *
+ * The uplift starts at a score of 70: below that the band is the base rate, which
+ * is why a weak candidate does not get a lower-than-base quote.
+ */
+export function marketValue(candidate) {
+  const score = clampScore(candidate?.ai_score);
+  const uplift = Math.max(0, Math.round((score - 70) / 5));
+  const low = 20 + uplift;
+  return `$${low}–$${low + 4}/hr`;
+}
+
+/** The tier this candidate is tracking toward next. */
+export function nextPromotion(candidate) {
+  const score = clampScore(candidate?.ai_score);
+  const years = candidate?.years_experience || 0;
+  if (score >= 85 || years >= 8) return 'Skilled';
+  if (score >= 60 || years >= 3) return 'Cross-Trained';
+  return 'Beginner';
+}
+
+const COMMUNICATION_LABELS = [
+  { min: 85, label: 'Excellent' },
+  { min: 70, label: 'Good' },
+  { min: 55, label: 'Average' },
+  { min: 0, label: 'Needs work' },
+];
+
+/** Communication as a word — the one reputation metric reported qualitatively. */
+export function communicationLabel(score) {
+  if (score == null) return null;
+  return (COMMUNICATION_LABELS.find((c) => score >= c.min) ?? COMMUNICATION_LABELS.at(-1)).label;
+}
+
+/**
+ * The five headline reputation metrics.
+ *
+ * Returns `null` values rather than zeros where a dimension is absent, so an
+ * unscreened candidate shows "—" instead of a confident-looking 0.
+ */
+export function reputationMetrics(candidate) {
+  const b = candidate?.score_breakdown || {};
+  return [
+    { label: 'Reliability', value: b.reliability ?? null, suffix: '%' },
+    { label: 'Attendance', value: b.attendance_expectations ?? null, suffix: '%' },
+    { label: 'Leadership', value: b.leadership_expectations ?? null, suffix: '%' },
+    { label: 'Communication', value: communicationLabel(b.communication_style ?? null) },
+    {
+      label: 'Client rating',
+      value: candidate?.client_rating > 0 ? candidate.client_rating.toFixed(1) : null,
+      suffix: '/5',
+    },
+  ];
+}
+
+/* ── Endorsements ───────────────────────────────────────────────────────── */
+
+const ENDORSER_NAMES = ['Luis', 'Chef Antonio', 'Maria', 'David', 'Sofia'];
+const ENDORSER_ROLES = ['Supervisor', 'Chef', 'Manager', 'Lead', 'Owner'];
+
+/**
+ * Endorsement voices.
+ *
+ * Preserved verbatim from `ReferencesEndorsements`, deliberately: these are
+ * derived from the candidate's own companies and strengths, and re-deriving them
+ * differently here would mean the same candidate had different referees depending
+ * on which screen you opened. Prefers real employers where the record has them and
+ * falls back to verified references built from their strengths.
+ */
+export function endorsements(candidate) {
+  const companies = (candidate?.companies_worked || []).filter(Boolean);
+  const strengths = (candidate?.ai_strengths || []).filter(Boolean);
+  const years = candidate?.years_experience || 0;
+
+  const source = companies.length
+    ? companies.slice(0, 3)
+    : strengths.slice(0, 3).map(() => 'Verified Reference');
+
+  return source.map((company, i) => ({
+    id: `${candidate?.id || 'c'}-endorsement-${i}`,
+    name: ENDORSER_NAMES[i % ENDORSER_NAMES.length],
+    role: ENDORSER_ROLES[i % ENDORSER_ROLES.length],
+    company,
+    events: 8 + i * 17 + years,
+    rating: 5,
+    comment: strengths[i]
+      ? `${strengths[i]}. Highly recommend for any team.`
+      : 'Reliable, punctual, and a great team player. Would hire again.',
+  }));
+}
+
+/** Verified strength tags — the candidate's strengths plus their skills. */
+export function strengthTags(candidate) {
+  return [...(candidate?.ai_strengths || []), ...(candidate?.skills || [])]
+    .filter(Boolean)
+    .slice(0, 6);
+}
+
+/* ── Activity ───────────────────────────────────────────────────────────── */
+
+/** The candidate's history, assembled from their record and any interview. */
+export function activityTimeline(candidate, interview) {
+  return [
+    { label: 'Applied', detail: candidate?.job_title, date: candidate?.created_date },
+    candidate?.ai_score > 0 && {
+      label: 'AI screened',
+      detail: `Scored ${candidate.ai_score} · ${candidate.ai_match_label || 'no label'}`,
+      date: candidate.updated_date,
+    },
+    interview && {
+      label: 'Interview started',
+      detail: interview.summary ? undefined : 'No summary recorded',
+      date: interview.created_date,
+    },
+    interview?.overall_interview_score > 0 && {
+      label: 'Interview scored',
+      detail: `${interview.overall_interview_score}/100 · integrity ${interview.integrity_score ?? 100}`,
+      date: interview.updated_date || interview.created_date,
+    },
+    candidate?.ai_recommendation && {
+      label: `Recommendation: ${candidate.ai_recommendation}`,
+      date: candidate.updated_date,
+    },
+    candidate?.status === 'hired' && { label: 'Hired', date: candidate.updated_date },
+    candidate?.status === 'rejected' && { label: 'Declined', date: candidate.updated_date },
+  ].filter(Boolean);
+}
+
+/* ── Recommendation ─────────────────────────────────────────────────────── */
+
+/**
+ * The recommendation headline: a verdict plus the action it implies.
+ *
+ * Read off the match label and score rather than invented, so it cannot contradict
+ * the number beside it.
+ */
+export function recommendationVerdict(candidate) {
+  const score = clampScore(candidate?.ai_score);
+  if (!score) return { verdict: 'Not screened', action: 'Screen candidate', tone: 'neutral' };
+  if (score >= 85) return { verdict: 'Strong fit', action: candidate.ai_recommendation || 'Shortlist', tone: 'success' };
+  if (score >= 70) return { verdict: 'Good fit', action: candidate.ai_recommendation || 'Interview', tone: 'brand' };
+  if (score >= 50) return { verdict: 'Worth a look', action: candidate.ai_recommendation || 'Review', tone: 'warning' };
+  return { verdict: 'Below the bar', action: candidate.ai_recommendation || 'Review', tone: 'neutral' };
+}