556a0b2f3f
Un pot de crédits global (BASIC 100 / PRO 1 000 / BUSINESS 4 000) remplace les seaux par fonction côté débit. Packs one-shot S/M/L, admin sans seaux legacy, usage multi-features, hints de coût, anti-flash thème et hydratation admin. Inclut aussi le pipeline slides renforcé et la page publique polishée.
104 lines
3.7 KiB
TypeScript
104 lines
3.7 KiB
TypeScript
/**
|
||
* Server-side extraction of usable assets from a note BEFORE any slide LLM call.
|
||
* Inspired by PPTAgent (retrieve relevant content per slide) and DeckForge (data_needs).
|
||
* Critical for STEM notes: formulas must be harvested, not hoped for from the model.
|
||
*/
|
||
|
||
export interface SourceAssets {
|
||
formulas: string[]
|
||
numbers: Array<{ label: string; value: number; raw: string }>
|
||
keySentences: string[]
|
||
hasMath: boolean
|
||
hasNumbers: boolean
|
||
wordCount: number
|
||
}
|
||
|
||
/** Extract LaTeX / equation-like fragments from note plain text or HTML. */
|
||
export function extractFormulas(raw: string): string[] {
|
||
if (!raw) return []
|
||
const found: string[] = []
|
||
const push = (s: string) => {
|
||
const t = s.replace(/\s+/g, ' ').trim()
|
||
if (t.length >= 2 && t.length <= 280 && !found.includes(t)) found.push(t)
|
||
}
|
||
|
||
// $$ ... $$
|
||
for (const m of raw.matchAll(/\$\$([\s\S]+?)\$\$/g)) push(m[1] || '')
|
||
// \[ ... \]
|
||
for (const m of raw.matchAll(/\\\[([\s\S]+?)\\\]/g)) push(m[1] || '')
|
||
// \( ... \)
|
||
for (const m of raw.matchAll(/\\\(([\s\S]+?)\\\)/g)) push(m[1] || '')
|
||
// $ ... $ (single line)
|
||
for (const m of raw.matchAll(/(?<![\\$])\$([^$\n]{2,120})\$(?!\$)/g)) push(m[1] || '')
|
||
// Common latex commands in free text
|
||
for (const m of raw.matchAll(
|
||
/(?:^|[\s(])((?:\\frac\{[^}]+\}\{[^}]+\}|\\partial|\\nabla|\\sum|\\int|\\frac|\\cdot|\\times|\\left|\\right|y'|y''|d[xy]\/d[xy]|∂)[^\n]{0,100})/g,
|
||
)) {
|
||
push(m[1] || '')
|
||
}
|
||
// Equation-like: f(x) = ..., dy/dx = ...
|
||
for (const m of raw.matchAll(
|
||
/(?:^|\n)\s*([A-Za-zΑ-ω∂∇][A-Za-z0-9_'"′]*\s*(?:\([^)]*\))?\s*[=≈≡]\s*[^\n]{3,100})/g,
|
||
)) {
|
||
const line = (m[1] || '').trim()
|
||
if (/[=≈]/.test(line) && /[a-zA-Z0-9]/.test(line)) push(line)
|
||
}
|
||
|
||
return found.slice(0, 24)
|
||
}
|
||
|
||
export function extractNumbers(raw: string): Array<{ label: string; value: number; raw: string }> {
|
||
if (!raw) return []
|
||
const out: Array<{ label: string; value: number; raw: string }> = []
|
||
const re =
|
||
/(?:^|[^\d])((?:≈|~)?\s*-?\d+(?:[.,]\d+)?)\s*(%|€|\$|k|m|bn|ms|s|°C)?\b/gi
|
||
let m: RegExpExecArray | null
|
||
while ((m = re.exec(raw)) !== null && out.length < 20) {
|
||
const numStr = (m[1] || '').replace(/[≈~\s]/g, '').replace(',', '.')
|
||
const value = parseFloat(numStr)
|
||
if (!Number.isFinite(value)) continue
|
||
const unit = m[2] || ''
|
||
// crude label: 40 chars before
|
||
const start = Math.max(0, m.index - 40)
|
||
const ctx = raw.slice(start, m.index).replace(/\s+/g, ' ').trim()
|
||
const label = ctx.split(/[.;:!?\n]/).pop()?.trim().slice(-32) || `n${out.length + 1}`
|
||
out.push({ label, value: unit === '%' ? value : value, raw: `${numStr}${unit}` })
|
||
}
|
||
return out
|
||
}
|
||
|
||
export function extractKeySentences(raw: string, max = 16): string[] {
|
||
if (!raw) return []
|
||
const plain = raw
|
||
.replace(/<[^>]+>/g, ' ')
|
||
.replace(/\$\$[\s\S]+?\$\$/g, ' ')
|
||
.replace(/\\\[[\s\S]+?\\\]/g, ' ')
|
||
.replace(/\s+/g, ' ')
|
||
.trim()
|
||
const parts = plain
|
||
.split(/(?<=[.!?])\s+/)
|
||
.map((s) => s.trim())
|
||
.filter((s) => s.length >= 40 && s.length <= 220)
|
||
const out: string[] = []
|
||
for (const p of parts) {
|
||
if (out.length >= max) break
|
||
if (!out.some((x) => x.slice(0, 40) === p.slice(0, 40))) out.push(p)
|
||
}
|
||
return out
|
||
}
|
||
|
||
export function extractSourceAssets(raw: string): SourceAssets {
|
||
const formulas = extractFormulas(raw)
|
||
const numbers = extractNumbers(raw)
|
||
const keySentences = extractKeySentences(raw)
|
||
const wordCount = raw.replace(/<[^>]+>/g, ' ').split(/\s+/).filter(Boolean).length
|
||
return {
|
||
formulas,
|
||
numbers,
|
||
keySentences,
|
||
hasMath: formulas.length > 0 || /équat|equat|différen|differen|dériv|deriv|intégr|integr|EDO|ODE|PDE|latex/i.test(raw),
|
||
hasNumbers: numbers.length >= 2,
|
||
wordCount,
|
||
}
|
||
}
|