Momento/memento-note/lib/ai/services/interactive-page-llm.service.ts

/**
 * LLM generation for interactive pages (spec §7) — split into short calls:
 * 1. generatePagePlan   → hero + overview + section list (one fast call)
 * 2. generatePageSection → blocks of ONE section, incl. a content-matched demo
 *
 * Splitting keeps every LLM round-trip well under the client abort (~75 s),
 * unlike the original single-shot page generation that timed out (~150 s).
 * Validation failures are fed back verbatim (max 2 attempts per call).
 */

import { generateText } from 'ai'
import { z } from 'zod'
import type { AIProvider } from '@/lib/ai/types'
import { cleanAIJsonResponse } from '@/lib/ai/utils/clean-ai-response'
import { extractSourceAssets } from '@/lib/ai/services/slide-source-assets'
import {
  INTERACTIVE_DEMO_SCHEMA_VERSION,
  PATTERN_IDS,
  PANEL_TYPES,
  INTENT_IDS,
  SPEAK_WRITING_RULE,
} from '@/lib/interactive-demo'
import {
  CALLOUT_KINDS,
  INTERACTIVE_PAGE_CAPS,
  validateInteractivePage,
  normalizeInteractivePageCandidate,
  type PageSection,
  type PageValidationIssue,
} from '@/lib/interactive-page'
import { catalogForPrompt } from '@/lib/simulators'
import thermoFixture from '@/lib/interactive-page/fixtures/thermo-page.json'

const MAX_ATTEMPTS = 2

// ── Shared helpers ───────────────────────────────────────────────────────────

function stripToPlain(html: string): string {
  return html
    .replace(/<[^>]+>/g, ' ')
    .replace(/&nbsp;/g, ' ')
    .replace(/&amp;/g, '&')
    .replace(/&lt;/g, '<')
    .replace(/&gt;/g, '>')
    .replace(/&quot;/g, '"')
    .replace(/&#39;/g, "'")
    .replace(/\s+/g, ' ')
    .trim()
}

function extractJsonObject(raw: string): unknown | null {
  if (!raw) return null
  const cleaned = cleanAIJsonResponse(raw)
  const tryParse = (s: string): unknown | null => {
    try {
      return JSON.parse(s)
    } catch {
      return null
    }
  }
  const stripTrailingCommas = (s: string) => s.replace(/,\s*([}\]])/g, '$1')

  let parsed = tryParse(cleaned) ?? tryParse(stripTrailingCommas(cleaned))
  if (parsed) return parsed

  const fence = cleaned.match(/```(?:json)?\s*([\s\S]*?)```/i)
  const candidate = fence?.[1]?.trim() ?? cleaned
  parsed = tryParse(candidate) ?? tryParse(stripTrailingCommas(candidate))
  if (parsed) return parsed

  const start = candidate.indexOf('{')
  const end = candidate.lastIndexOf('}')
  if (start >= 0 && end > start) {
    const slice = candidate.slice(start, end + 1)
    parsed = tryParse(slice) ?? tryParse(stripTrailingCommas(slice))
    if (parsed) return parsed
  }
  return null
}

function assetsBlock(assets: ReturnType<typeof extractSourceAssets>): string {
  let msg = ''
  if (assets.formulas.length) {
    msg += `\nFORMULES_EXTRAITES (réutilise telles quelles en $...$ — KaTeX inline):\n`
    msg += assets.formulas
      .slice(0, 14)
      .map((f, i) => `${i + 1}. ${f}`)
      .join('\n')
    msg += '\n'
  }
  if (assets.keySentences.length) {
    msg += `\nPHRASES_CLES:\n`
    msg += assets.keySentences
      .slice(0, 8)
      .map((s) => `- ${s}`)
      .join('\n')
    msg += '\n'
  }
  if (assets.numbers.length) {
    msg += `\nDONNEES_NUMERIQUES (source des blocs "stats"/"table" — ne pas inventer d'autres chiffres):\n${JSON.stringify(assets.numbers.slice(0, 10))}\n`
  }
  return msg
}

// ── Golden demo examples (from the hand-crafted thermo fixture) ─────────────

type FixtureDemoBlock = { type: string; demo: unknown }

function fixtureDemo(panelType: string): string | null {
  for (const section of thermoFixture.sections) {
    for (const block of section.blocks as FixtureDemoBlock[]) {
      if (block.type !== 'demo') continue
      const demo = block.demo as {
        scene?: { panels?: { type: string }[] }
      }
      const panels = demo?.scene?.panels ?? []
      if (panels.some((p) => p.type === panelType)) {
        return JSON.stringify(block.demo)
      }
    }
  }
  return null
}

// ── 1. Page plan ─────────────────────────────────────────────────────────────

const DEMO_KINDS = ['svg-scene', 'chart', 'heatmap-matrix', 'simulation', 'none'] as const
export type PagePlanDemoKind = (typeof DEMO_KINDS)[number]

const planSectionSchema = z.object({
  title: z.string().min(1),
  goal: z.string().min(1),
  demoKind: z.enum(DEMO_KINDS),
  demoGoal: z.string().optional(),
})

const pagePlanSchema = z.object({
  heroTitle: z.string().min(1),
  heroSubtitle: z.string().optional(),
  overviewLead: z.string().min(1),
  overviewCards: z
    .array(
      z.object({
        badge: z.string().min(1),
        title: z.string().min(1),
        body: z.string().min(1),
        intent: z.enum(INTENT_IDS).optional(),
      })
    )
    .min(INTERACTIVE_PAGE_CAPS.minOverviewCards)
    .max(INTERACTIVE_PAGE_CAPS.maxOverviewCards),
  sections: z.array(planSectionSchema).min(2).max(5),
})

export type PagePlanSection = z.infer<typeof planSectionSchema>
export type PagePlan = z.infer<typeof pagePlanSchema>

export type GeneratePagePlanResult =
  | { ok: true; plan: PagePlan; attempts: number }
  | { ok: false; error: string; reason?: string; attempts: number }

function buildPlanSystemPrompt(lang: string): string {
  return `You plan an interactive pedagogical page (PageSpecV1) for a Memento note, in the style of the Kimi "Attention Residuals" explainer.
You output ONLY the PLAN as one JSON object — sections content is generated later, one call per section.

OUTPUT: a single JSON object only. No markdown fences. No commentary. No <think> blocks.

SCHEMA:
{
  "heroTitle": string,          // the SUBJECT of the note, never a generic title
  "heroSubtitle": string,       // one sentence, autoportant
  "overviewLead": string,       // the central idea in ONE self-contained paragraph (may use $KaTeX$ inline)
  "overviewCards": [ { "badge": string, "title": string, "body": string, "intent?": ${JSON.stringify(INTENT_IDS)} } ],  // 2–4 key concepts, small-caps badges (ex. PROBLEM / APPROACH / RESULT)
  "sections": [ { "title": string, "goal": string, "demoKind": ${JSON.stringify(DEMO_KINDS)}, "demoGoal?": string } ]  // 2–5
}

COUVERTURE (règle n°1):
- The page covers the CORE of the source: its 2–5 major ideas, one section each.
- NEVER build the page on the note's simplest example.
- First section = the problem / context; last section = synthesis / results.
- If the content does not benefit from an interactive page, REFUSE: return { "error": "unsuitable_content", "reason": "…" } instead.

MATCHING CONTENU → DÉMO (choose demoKind per section, "none" when no visual helps):
- Catégories × propriétés (comparatif, matrice, échanges) → "heatmap-matrix"
- Loi / relation / courbe / évolution chiffrée → "chart"
- Processus / flux / cycle / architecture → "svg-scene"
- Avant/après, comparaison d'états → "svg-scene" (2 groupes de nœuds) et précise-le dans demoGoal
- Loi chiffrée avec paramètres manipulables (η = 1 − Tc/Th, COP, loi physique, modèle) → "simulation" (l'apprenant manipule des curseurs)
- Une démo seulement si elle ENSEIGNE mieux que le texte. Max 3 sections avec démo/simulation.

RULES:
- Language of ALL strings = "${lang}" (the note's language).
- demoGoal: one sentence stating what the demo must show (used by the next LLM call).
- No colors anywhere. Intents only.`
}

export async function generatePagePlan(input: {
  content: string
  lang?: string
  provider: AIProvider
}): Promise<GeneratePagePlanResult> {
  const lang = input.lang || 'fr'
  const assets = extractSourceAssets(input.content)
  const plain = stripToPlain(input.content).slice(0, 6000)
  const system = buildPlanSystemPrompt(lang)
  const model = input.provider.getModel()

  let lastError = 'unknown'
  for (let attempt = 1; attempt <= MAX_ATTEMPTS; attempt++) {
    let user = `Plan the interactive page for THIS note.

EXCERPT_START
${plain}
EXCERPT_END
${assetsBlock(assets)}`
    if (attempt > 1) {
      user += `\nPREVIOUS_ANSWER_INVALID: ${lastError}\nReturn a corrected FULL JSON object matching the schema exactly.`
    }

    const { text: raw } = await generateText({
      model,
      system,
      prompt: user,
      temperature: attempt === 1 ? 0.3 : 0.1,
    })

    const parsed = extractJsonObject(raw)
    if (!parsed) {
      lastError = 'Model did not return parseable JSON'
      console.warn('[interactive-page/plan] unparseable JSON', raw.slice(0, 300))
      continue
    }

    const refusal = parsed as { error?: string; reason?: string }
    if (refusal?.error === 'unsuitable_content') {
      return {
        ok: false,
        error: 'unsuitable_content',
        reason: refusal.reason || 'Contenu inadapté à une page interactive',
        attempts: attempt,
      }
    }

    const result = pagePlanSchema.safeParse(parsed)
    if (result.success) {
      return { ok: true, plan: result.data, attempts: attempt }
    }
    lastError = result.error.issues
      .slice(0, 6)
      .map((i) => `${i.path.join('.')}: ${i.message}`)
      .join(' | ')
    console.warn('[interactive-page/plan] schema failed', lastError)
  }

  return { ok: false, error: 'plan_failed', reason: lastError, attempts: MAX_ATTEMPTS }
}

// ── 2. Section generation ────────────────────────────────────────────────────

export type GeneratePageSectionInput = {
  content: string
  lang?: string
  provider: AIProvider
  pageTitle: string
  sectionId: string
  section: PagePlanSection
}

export type GeneratePageSectionResult =
  | { ok: true; section: PageSection; attempts: number }
  | { ok: false; issues?: PageValidationIssue[]; error?: string; attempts: number }

const SECTION_NARRATION_RULES = `NARRATION des démos (champ "speak"):
- 1–2 phrases, 25 mots max, une seule idée par étape, gras sur le concept clé, KaTeX inline ($...$) pour les formules.
- Voix off de prof, jamais de description mécanique.
- Jamais annoter ce qu'on pointTo dans la même étape.
- Chaque élément d'une scène est désigné au moins une fois dans l'acte.
- Scènes denses : svg-scene ≥ 5 nœuds ; heatmap avec valeurs réalistes et variées (intensité ∝ valeur, valeurs affichées).
- Étape finale d'acte : reveal ["*"] + pattern overview.
- Valeurs illustratives → "disclaimer" obligatoire.`

function buildSectionSystemPrompt(lang: string, hasMath: boolean): string {
  return `You generate ONE section of an interactive pedagogical page (PageSpecV1) for a Memento note, in the style of the Kimi "Attention Residuals" explainer.

OUTPUT: a single JSON object only. No markdown fences. No commentary. No <think> blocks.

SCHEMA:
{
  "title": string,
  "blocks": [ Block, ... ]   // 2–6 blocks
}
Block types (discriminated by "type"):
- { "type": "prose", "md": string }                      // light markdown + $KaTeX$ inline
- { "type": "formula", "tex": string, "caption?": string }   // KaTeX block for key relations
- { "type": "callout", "kind": ${JSON.stringify(CALLOUT_KINDS)}, "title": string, "md": string }
- { "type": "chart", "payload": { "chartType": "line"|"bar"|"area", "series": [{ "id": string, "label?": string, "values": number[], "intent?": IntentId }] }, "caption?": string }
- { "type": "stats", "items": [{ "value": string, "label": string, "intent?": IntentId }] }   // 2–5, REAL figures from the note only
- { "type": "table", "columns": string[], "rows": string[][], "caption?": string }   // every row.length === columns.length
- { "type": "demo", "demo": InteractiveDemoV1, "caption?": string }
- { "type": "sim", "sim": SimRef, "caption?": string }   // interactive simulation with sliders
IntentId: ${JSON.stringify(INTENT_IDS)}

SimRef — TWO forms:
(A) CATALOG simulator (PREFERRED when the section matches one): { "simId": "<id from catalog>", "title?": string, "preset?": { "<paramId>": number }, "disclaimer?": string }
    → You ONLY pick the simId and preset values (from the note's real numbers within the allowed ranges). The app runs the simulation.
(B) GENERIC formula simulator: { "simId": "generic-formula", "title": string, "params": [{ "id", "symbol", "label", "min", "max", "step", "defaultValue", "unit?", "intent?" }] (1–4), "computed": [{ "id", "symbol", "label", "expr", "unit?", "intent?" }] (1–6), "visual": { "kind": "gauges" } | { "kind": "bars" } | { "kind": "curve", "xParamId", "expr" }, "disclaimer?": string }
    → expr = plain math over param ids: + - * / ^ % parentheses, functions sqrt abs exp ln log round min max, constants pi e. NO other identifiers.
SIMULATOR CATALOG (pick from this, else generic-formula):
${catalogForPrompt(lang)}

InteractiveDemoV1 (schemaVersion ${INTERACTIVE_DEMO_SCHEMA_VERSION}):
{
  "schemaVersion": 1, "id": "demo.<slug>", "lang": "${lang}", "disclaimer?": string,
  "scene": { "id": string, "panels": [Panel] },
  "acts": [ { "id": "a1", "title": string, "pattern?": Pattern, "steps": [Step] } ]
}
Panel types: ${PANEL_TYPES.join(', ')} (max 2 panels; heatmap cells ids: r{row}.c{col}, triangular:"lower" when appropriate)
Patterns: ${PATTERN_IDS.join(', ')}
Step: { "id": "a1.s1", "speak": string, "pattern?", "pointTo?": string[], "reveal?": [{"ids": string[], "scope": "transient|act|scene"}], "annotate?": [...] }
svg-scene: nodes[{id,label?,intent?}] (labels may hold $KaTeX$), edges[{id,from,to,style?,weight?,intent?}]

${SECTION_NARRATION_RULES}

MÉCANIQUE (the validator rejects otherwise):
- Unique semantic ids; NO colors (hex/rgb) anywhere — intents only.
- Blocks per section ≤ ${INTERACTIVE_PAGE_CAPS.maxBlocksPerSection}.
- Every pointTo/reveal/annotate reference must be an id of the active scene.
- "stats"/"table" ONLY with figures actually present in the source.
- All human-facing strings in "${lang}".
RÈGLE D'OR — AUCUN VISUEL DÉCORATIF (checked after generation, violations are rejected):
- Every visual block must TEACH something the text alone cannot. Ask: "what does the learner understand after, that they didn't before?" No answer → no visual block.
- "heatmap-matrix" ONLY when the note contains genuinely matrix-shaped data (table of values, correlations, confusion matrix). Otherwise FORBIDDEN.
- "chart" ONLY with numeric series actually present in the source note. Otherwise FORBIDDEN.
- svg-scene: ≥ 5 nodes, dense, real vocabulary of the note — never 3 generic boxes.
- When in doubt: prose/formula/callout only.
${hasMath ? '- STEM: reuse formulas from FORMULES_EXTRAITES verbatim in formula blocks, prose and demo speak ($...$).' : ''}`
}

function buildSectionUserPrompt(
  input: GeneratePageSectionInput,
  assets: ReturnType<typeof extractSourceAssets>,
  opts?: { issues?: PageValidationIssue[]; previousJson?: string }
): string {
  const plain = stripToPlain(input.content).slice(0, 6000)
  const { section } = input
  let msg = `Generate the section "${section.title}" of the interactive page "${input.pageTitle}".

SECTION_GOAL: ${section.goal}
${
  section.demoKind === 'simulation'
    ? `SIMULATION_REQUIRED: include ONE "sim" block. Prefer a catalog simulator if the section matches one (bind the note's real values into "preset"); otherwise "generic-formula" with the section's key relation.
SIM_GOAL: ${section.demoGoal || section.goal}`
    : section.demoKind !== 'none'
      ? `DEMO_REQUIRED: include ONE "demo" block of kind "${section.demoKind}".
DEMO_GOAL: ${section.demoGoal || section.goal}`
      : `NO demo/sim block for this section — rich prose/formula/callout/stats/table only.`
}

EXCERPT_START
${plain}
EXCERPT_END
${assetsBlock(assets)}`

  if (section.demoKind === 'heatmap-matrix' || section.demoKind === 'svg-scene') {
    const golden = fixtureDemo(section.demoKind)
    if (golden) {
      msg += `\nEXEMPLE_D_OR (imite sa densité et sa qualité de narration — jamais moins ; adapte au contenu de CETTE note):\n${golden}\n`
    }
  }

  msg += `\nCONTRAINTES: section autoportante, fidèle au contenu réel de la note (jamais un exemple jouet). 2–6 blocks.`

  if (opts?.issues?.length) {
    msg += `\n\nPREVIOUS_JSON_FAILED_VALIDATION. Fix ALL issues and return a corrected FULL JSON.\n`
    msg += opts.issues
      .slice(0, 12)
      .map((i) => `- [${i.code}] ${i.path}: ${i.message}`)
      .join('\n')
  }
  if (opts?.previousJson) {
    msg += `\n\nPREVIOUS_JSON_START\n${opts.previousJson.slice(0, 10000)}\nPREVIOUS_JSON_END`
  }
  return msg
}

/** Wrap a section candidate in a minimal page so the shared validator runs (incl. demo delegation). */
function validateSectionCandidate(
  candidate: unknown,
  sectionId: string,
  lang: string
): { ok: true; section: PageSection } | { ok: false; issues: PageValidationIssue[] } {
  const wrapped = {
    schemaVersion: 1,
    id: 'page.section-check',
    lang,
    hero: { kicker: 'CHECK', title: 'Section check' },
    sections: [
      typeof candidate === 'object' && candidate !== null
        ? { id: sectionId, ...(candidate as Record<string, unknown>) }
        : candidate,
    ],
  }
  const normalized = normalizeInteractivePageCandidate(wrapped, lang)
  if (!normalized) {
    return {
      ok: false,
      issues: [
        { code: 'normalize_failed', path: '', message: 'Section not normalizable' },
      ],
    }
  }
  const result = validateInteractivePage(normalized)
  if (!result.ok) return { ok: false, issues: result.issues }
  const section = result.page.sections.find((s) => s.id === sectionId)
  if (!section) {
    return {
      ok: false,
      issues: [{ code: 'section_missing', path: 'sections', message: 'Section lost in normalization' }],
    }
  }
  return { ok: true, section }
}

/**
 * Quality gate (règle d'or): reject decorative visuals AFTER schema validation —
 * heatmap/chart whose values don't come from the note, svg-scene too sparse.
 * Issues are fed back to the LLM for a repair attempt.
 */
function qualityGateSection(
  section: PageSection,
  assets: ReturnType<typeof extractSourceAssets>
): PageValidationIssue[] {
  const out: PageValidationIssue[] = []
  const sourceValues = assets.numbers.map((n) => n.value)
  const valueInSource = (v: number) =>
    sourceValues.some((sv) => sv !== 0 && Math.abs(sv - v) / Math.max(1, Math.abs(sv)) < 0.06)

  for (const [bi, block] of section.blocks.entries()) {
    const path = `blocks[${bi}]`
    if (block.type === 'chart') {
      const values = block.payload.series.flatMap((s) => s.values)
      const grounded = values.filter(valueInSource).length
      if (values.length > 0 && grounded / values.length < 0.5) {
        out.push({
          code: 'decorative_data',
          path,
          message:
            'Chart values are not from the note (decorative data). Use real series from the source or remove the chart.',
        })
      }
    }
    if (block.type === 'demo') {
      for (const panel of block.demo.scene.panels) {
        if (panel.type === 'heatmap-matrix') {
          const values = panel.payload.values.flat()
          const grounded = values.filter(valueInSource).length
          if (values.length > 0 && grounded / values.length < 0.5) {
            out.push({
              code: 'decorative_data',
              path,
              message:
                'Heatmap values are not from the note (decorative data). Only use heatmap-matrix for real matrix-shaped source data.',
            })
          }
        }
        if (panel.type === 'svg-scene' && panel.payload.nodes.length < 5) {
          out.push({
            code: 'scene_too_poor',
            path,
            message:
              'svg-scene has fewer than 5 nodes (too poor pedagogically). Densify: ≥5 nodes with real vocabulary and formulas from the note.',
          })
        }
      }
    }
  }
  return out
}

export async function generatePageSection(
  input: GeneratePageSectionInput
): Promise<GeneratePageSectionResult> {
  const lang = input.lang || 'fr'
  const assets = extractSourceAssets(input.content)
  const system = buildSectionSystemPrompt(
    lang,
    assets.hasMath || assets.formulas.length > 0
  )
  const model = input.provider.getModel()

  let lastIssues: PageValidationIssue[] = []
  let lastRaw = ''
  let lastNormalizedJson = ''

  for (let attempt = 1; attempt <= MAX_ATTEMPTS; attempt++) {
    const user = buildSectionUserPrompt(input, assets, {
      issues: attempt > 1 ? lastIssues : undefined,
      previousJson: attempt > 1 ? lastNormalizedJson || lastRaw : undefined,
    })

    const { text: raw } = await generateText({
      model,
      system,
      prompt: user,
      temperature: attempt === 1 ? 0.3 : 0.1,
    })
    lastRaw = raw

    const parsed = extractJsonObject(raw)
    if (!parsed) {
      lastIssues = [
        {
          code: 'invalid_json',
          path: '',
          message: 'Model did not return parseable JSON',
        },
      ]
      console.warn(
        `[interactive-page/section ${input.sectionId}] attempt ${attempt}: unparseable JSON`,
        raw.slice(0, 300)
      )
      continue
    }
    lastNormalizedJson = JSON.stringify(parsed)

    const result = validateSectionCandidate(parsed, input.sectionId, lang)
    if (result.ok) {
      const qualityIssues = qualityGateSection(result.section, assets)
      if (!qualityIssues.length) {
        return { ok: true, section: result.section, attempts: attempt }
      }
      lastIssues = qualityIssues
      console.warn(
        `[interactive-page/section ${input.sectionId}] attempt ${attempt}: quality gate`,
        qualityIssues.slice(0, 6)
      )
      continue
    }
    lastIssues = result.issues
    console.warn(
      `[interactive-page/section ${input.sectionId}] attempt ${attempt}: validation failed`,
      result.issues.slice(0, 8)
    )
  }

  return { ok: false, issues: lastIssues, attempts: MAX_ATTEMPTS }
}