{ "$schema": "https://uipkge.dev/schema/registry-item.json", "name": "synthetic-data-generator", "title": "Synthetic Data Generator", "type": "registry:block", "files": [ { "path": "packages/registry-angular/blocks/synthetic-data-generator/synthetic-data-generator.component.ts", "content": "import {\n Component,\n DestroyRef,\n Input,\n type OnInit,\n computed,\n inject,\n signal,\n ChangeDetectionStrategy,\n} from '@angular/core'\nimport {\n Check,\n Code2,\n Cpu,\n Database,\n Download,\n FileSpreadsheet,\n Info,\n Layers,\n Lock,\n LucideAngularModule,\n Plus,\n RefreshCw,\n ShieldCheck,\n Sliders,\n Sparkles,\n Trash2,\n} from 'lucide-angular'\nimport { cn } from '@/lib/utils'\nimport { UiBadgeComponent } from '@/ui/badge/badge.component'\nimport { UiButtonComponent } from '@/ui/button/button.component'\nimport {\n UiCardComponent,\n UiCardContentComponent,\n UiCardDescriptionComponent,\n UiCardHeaderComponent,\n UiCardTitleComponent,\n} from '@/ui/card/card.component'\nimport { UiInputComponent } from '@/ui/input/input.component'\nimport {\n UiSelectComponent,\n UiSelectContentComponent,\n UiSelectItemComponent,\n UiSelectTriggerComponent,\n UiSelectValueComponent,\n} from '@/ui/select/select.component'\nimport { UiSeparatorComponent } from '@/ui/separator/separator.component'\nimport { UiSliderComponent } from '@/ui/slider/slider.component'\nimport { UiSwitchComponent } from '@/ui/switch/switch.component'\nimport type {\n FieldDataType,\n FidelityMetrics,\n GeneratedRecord,\n SchemaField,\n SyntheticDataGeneratorProps,\n TemplatePreset,\n} from './synthetic-data-types'\nimport {\n boxMullerRandom as boxMullerRandomHelper,\n CARD_BRANDS,\n FIRST_NAMES,\n LAST_NAMES,\n TEMPLATES,\n} from './synthetic-data-templates'\nimport { UiSyntheticDataPreviewComponent, type SyntheticDataPreviewTab } from './synthetic-data-preview.component'\n\nexport type {\n FieldDataType,\n FidelityMetrics,\n GeneratedRecord,\n SchemaField,\n SyntheticDataGeneratorProps,\n TemplatePreset,\n}\n\nconst ROW_PRESETS = [100, 1000, 5000, 10000, 50000]\nconst EPSILON_PRESETS = [\n { eps: 0.5, label: 'Healthcare' },\n { eps: 1.2, label: 'Standard' },\n { eps: 3.0, label: 'Analytics' },\n]\nconst FIELD_TYPE_BADGE: Record = {\n gaussian_float: 'border-info/20 bg-info/10 text-info',\n name: 'border-chart-1/20 bg-chart-1/10 text-chart-1',\n email: 'border-chart-2/20 bg-chart-2/10 text-chart-2',\n credit_card: 'border-warning/20 bg-warning/10 text-warning',\n category_weights: 'border-success/20 bg-success/10 text-success',\n date_range: 'border-destructive/20 bg-destructive/10 text-destructive',\n}\nconst FORMAT_BUTTON = 'flex cursor-pointer items-center gap-1.5 rounded-md px-2.5 py-1.5 font-medium transition-colors'\n\nfunction cloneFields(key: string): SchemaField[] {\n const tpl = TEMPLATES[key]\n return tpl?.fields ? (JSON.parse(JSON.stringify(tpl.fields)) as SchemaField[]) : []\n}\n\n/**\n * Synthetic dataset generator & privacy engine (Gretel / Tonic style). Port of the React\n * `SyntheticDataGenerator` 1:1: schema template picker, CSV / JSON Lines format toggle, row\n * volume slider + presets, differential-privacy epsilon budget, a custom schema field builder,\n * PII anonymization switches, and the live preview column (`ui-synthetic-data-preview`).\n * `initialTemplate`, `initialRowCount` and `initialEpsilon` seed the state once.\n */\n@Component({\n changeDetection: ChangeDetectionStrategy.Eager,\n selector: 'ui-synthetic-data-generator, [ui-synthetic-data-generator]',\n standalone: true,\n // React renders the root
itself: the host stays out of layout and `class` goes to the root.\n host: { '[attr.class]': '\"contents\"' },\n imports: [\n LucideAngularModule,\n UiBadgeComponent,\n UiButtonComponent,\n UiCardComponent,\n UiCardContentComponent,\n UiCardDescriptionComponent,\n UiCardHeaderComponent,\n UiCardTitleComponent,\n UiInputComponent,\n UiSelectComponent,\n UiSelectContentComponent,\n UiSelectItemComponent,\n UiSelectTriggerComponent,\n UiSelectValueComponent,\n UiSeparatorComponent,\n UiSliderComponent,\n UiSwitchComponent,\n UiSyntheticDataPreviewComponent,\n ],\n template: `\n
\n \n
\n
\n
\n
\n
\n
\n \n
\n

\n Synthetic Dataset Generator & Privacy Engine\n

\n \n \n Differential Privacy Enabled\n \n
\n

\n Mathematical & LLM-guided synthetic tabular dataset synthesizer with configurable (ε, δ)-DP noise\n injection.\n

\n
\n\n \n
\n \n
\n \n \n
\n\n \n \n \n Download Dataset\n \n\n \n \n @if (isGenerating()) {\n \n \n } @else {\n \n }\n {{ isGenerating() ? 'Synthesizing...' : 'Generate Synthetic Batch' }}\n \n
\n
\n\n \n \n
\n \n Schema Template:\n
\n\n
\n \n span]:truncate [&>svg]:shrink-0\"\n >\n \n \n \n @for (tpl of templates; track tpl[0]) {\n {{ tpl[1].name }}\n }\n \n \n
\n\n \n
\n
\n
\n\n \n @if (downloadToast(); as toast) {\n \n
\n \n {{ toast }}\n
\n \n {{ estimatedFileSize() }}\n \n
\n }\n\n \n
\n \n
\n \n
\n
\n
\n
\n
\n \n
\n

Dataset Generation Volume

\n
\n \n {{ rowCount().toLocaleString() }} rows\n \n
\n
\n\n
\n \n
\n
\n 100 rows\n Estimated: {{ estimatedFileSize() }}\n 50,000 rows\n
\n\n \n
\n\n \n
\n Presets:\n @for (preset of rowPresets; track preset) {\n \n }\n
\n
\n
\n\n \n
\n
\n
\n
\n
\n \n
\n
\n

Differential Privacy Budget

\n

Laplace (ε, δ)-DP noise calibration

\n
\n
\n \n ε = {{ epsilon().toFixed(1) }}\n \n
\n
\n\n
\n \n
\n
\n ε=0.1 (High Privacy)\n ε=5.0 (High Fidelity)\n
\n\n \n
\n\n \n
\n \n \n
\n
{{ fidelityMetrics().privacyTier }}
\n

\n Laplace noise scale:\n σ=±{{ fidelityMetrics().laplaceNoiseStd }}\n • Delta: δ=10⁻⁵ • Zero memorization\n guarantee.\n

\n
\n
\n\n \n
\n @for (p of epsilonPresets; track p.eps) {\n \n }\n
\n
\n
\n\n \n
\n
\n
\n
\n
\n \n
\n
\n

Schema Fields Generator

\n

\n {{ activeFields().length }} active synthesizer distributions\n

\n
\n
\n \n \n {{ showAddField() ? 'Cancel' : 'Add Field' }}\n \n
\n
\n\n
\n \n @if (showAddField()) {\n
\n
Configure New Synthetic Field
\n
\n
\n \n \n
\n
\n \n \n \n \n Float Gaussian Distribution\n Full Name Faker\n Email Faker\n Credit Card Faker\n Category Weights\n Date Range\n \n \n
\n
\n\n
\n \n \n Add to Schema\n \n
\n
\n }\n\n \n
\n @for (field of activeFields(); track field.id) {\n \n
\n
\n {{ field.name }}\n \n {{ field.type.replace('_', ' ') }}\n \n
\n

\n {{ field.description || 'Configured synthesizer parameter' }}\n

\n
\n\n \n \n \n
\n }\n
\n
\n
\n\n \n
\n
\n
\n
\n \n
\n
\n

PII Anonymization Policy

\n

Zero linkage & re-identification guarantees

\n
\n
\n
\n\n
\n \n
\n
\n
K-Anonymity (k = 5 Cohort Enforcement)
\n

\n Generalizes quasi-identifiers so each demographic bucket contains ≥ 5 records.\n

\n
\n \n
\n\n
\n\n \n
\n
\n
Salted HMAC-SHA256 for Primary Keys
\n

\n Deterministically pseudonymizes primary keys with ephemeral cryptographic salt.\n

\n
\n \n
\n\n
\n\n \n
\n
\n
Differential Privacy Gaussian Noise
\n

\n Injects calibrated noise calibrated to global sensitivity (Δf / ε).\n

\n
\n \n
\n
\n
\n
\n\n \n
\n \n
\n \n \n `,\n})\nexport class UiSyntheticDataGeneratorComponent implements OnInit {\n @Input('class') className?: string\n @Input() initialTemplate = 'customer-transactions'\n @Input() initialRowCount = 1000\n @Input() initialEpsilon = 1.2\n\n // --- STATE ---\n readonly selectedTemplateKey = signal('customer-transactions')\n readonly outputFormat = signal<'csv' | 'jsonl'>('csv')\n readonly rowCount = signal(1000)\n readonly epsilon = signal(1.2)\n readonly kAnonymity = signal(true)\n readonly saltedHash = signal(true)\n readonly noiseMechanism = signal(true)\n readonly activeFields = signal(cloneFields('customer-transactions'))\n\n readonly isGenerating = signal(false)\n readonly copied = signal(false)\n readonly downloadToast = signal(null)\n readonly previewTab = signal('table')\n readonly generationSeed = signal(48291)\n\n // Field addition state\n readonly showAddField = signal(false)\n readonly newFieldName = signal('')\n readonly newFieldType = signal('gaussian_float')\n\n protected readonly templates = Object.entries(TEMPLATES)\n protected readonly rowPresets = ROW_PRESETS\n protected readonly epsilonPresets = EPSILON_PRESETS\n\n protected readonly Check = Check\n protected readonly Code2 = Code2\n protected readonly Cpu = Cpu\n protected readonly Database = Database\n protected readonly Download = Download\n protected readonly FileSpreadsheet = FileSpreadsheet\n protected readonly Info = Info\n protected readonly Layers = Layers\n protected readonly Lock = Lock\n protected readonly Plus = Plus\n protected readonly RefreshCw = RefreshCw\n protected readonly ShieldCheck = ShieldCheck\n protected readonly Sliders = Sliders\n protected readonly Sparkles = Sparkles\n protected readonly Trash2 = Trash2\n\n private readonly timers = new Set>()\n\n constructor() {\n inject(DestroyRef).onDestroy(() => this.timers.forEach((t) => clearTimeout(t)))\n }\n\n ngOnInit(): void {\n this.selectedTemplateKey.set(this.initialTemplate)\n this.rowCount.set(this.initialRowCount)\n this.epsilon.set(this.initialEpsilon)\n this.activeFields.set(cloneFields(this.initialTemplate))\n }\n\n get rootClass(): string {\n return cn('mx-auto w-full max-w-7xl space-y-6', this.className)\n }\n\n protected readonly templateCategory = computed(() => TEMPLATES[this.selectedTemplateKey()]?.category)\n\n // Dynamic preview records computation\n readonly previewRecords = computed(() => {\n const generationSeed = this.generationSeed()\n const saltedHash = this.saltedHash()\n const noiseMechanism = this.noiseMechanism()\n const epsilon = this.epsilon()\n const rows: GeneratedRecord[] = []\n const count = 6\n\n // Box-Muller Gaussian random generator with differential privacy noise\n const boxMullerRandom = (mean: number, stdDev: number, seedOffset: number): number =>\n boxMullerRandomHelper(mean, stdDev, seedOffset, generationSeed, noiseMechanism, epsilon)\n\n for (let i = 0; i < count; i++) {\n const fn = FIRST_NAMES[(i * 3 + generationSeed) % FIRST_NAMES.length]!\n const ln = LAST_NAMES[(i * 5 + generationSeed + 2) % LAST_NAMES.length]!\n const fullName = `${fn} ${ln}`\n const baseSlug = `${fn.charAt(0).toLowerCase()}.${ln.toLowerCase().replace(/[^a-z]/g, '')}`\n\n const record: GeneratedRecord = {\n id: saltedHash\n ? `syn_${((generationSeed * 9301 + i * 49297) % 233280).toString(16).padStart(6, '0')}`\n : `usr_${1000 + i}`,\n }\n\n for (const field of this.activeFields()) {\n if (field.name.includes('id') || field.name.includes('uuid') || field.name.includes('token')) {\n record[field.name] = record.id\n continue\n }\n\n switch (field.type) {\n case 'name':\n record[field.name] = fullName\n break\n\n case 'email': {\n const dom = field.params.domain || 'synthetic-vault.io'\n record[field.name] = `${baseSlug}@${dom}`\n break\n }\n\n case 'credit_card': {\n const last4 = (((generationSeed * 7 + i * 1337 + 1000) % 9000) + 1000).toString()\n const brand = CARD_BRANDS[i % CARD_BRANDS.length]\n record[field.name] = `${brand} •••• ${last4}`\n break\n }\n\n case 'gaussian_float': {\n const m = field.params.mean ?? 100\n const s = field.params.stdDev ?? 20\n const rawVal = boxMullerRandom(m, s, i)\n const val = Math.max(m * 0.1, rawVal)\n\n if (field.params.currency) {\n record[field.name] = `${field.params.currency}${val.toFixed(2)}`\n } else if (field.params.unit) {\n record[field.name] = `${val.toFixed(1)} ${field.params.unit}`\n } else {\n record[field.name] = parseFloat(val.toFixed(2))\n }\n break\n }\n\n case 'category_weights': {\n const cats = field.params.categories || ['Standard', 'Premium']\n const catIndex = (i + Math.floor(generationSeed / 100)) % cats.length\n record[field.name] = cats[catIndex]!\n break\n }\n\n case 'date_range': {\n const day = ((10 + i * 4) % 28) + 1\n const month = (i % 8) + 1\n const dayStr = day.toString().padStart(2, '0')\n const monthStr = month.toString().padStart(2, '0')\n const hour = (8 + i * 3) % 24\n const hourStr = hour.toString().padStart(2, '0')\n record[field.name] = `2025-${monthStr}-${dayStr} ${hourStr}:24:18 UTC`\n break\n }\n\n default:\n record[field.name] = `value_${i}`\n }\n }\n\n rows.push(record)\n }\n\n return rows\n })\n\n // Statistical fidelity metrics\n readonly fidelityMetrics = computed(() => {\n const epsilon = this.epsilon()\n const rawSimilarity = Math.min(99.4, Math.max(82.0, 99.2 - ((5.0 - epsilon) / 4.9) * 9.6))\n const similarity = parseFloat(rawSimilarity.toFixed(1))\n const leakageRisk =\n this.kAnonymity() && this.saltedHash() && epsilon <= 3.5 ? 0.0 : Math.max(0, (epsilon - 3.5) * 0.04)\n const correlationRetention = parseFloat(Math.min(99.1, 98.6 - ((5.0 - epsilon) / 4.9) * 6.2).toFixed(1))\n const wassersteinDistance = parseFloat((0.032 + ((5.0 - epsilon) / 4.9) * 0.024).toFixed(3))\n const laplaceNoiseStd = (0.12 / Math.max(0.1, epsilon)).toFixed(3)\n\n let privacyTier = 'Balanced Privacy & Utility'\n let privacyBadgeColor = 'bg-info/10 text-info border-info/30'\n\n if (epsilon <= 0.5) {\n privacyTier = 'Maximum Privacy / High DP Noise'\n privacyBadgeColor = 'bg-success/10 text-success border-success/30'\n } else if (epsilon > 2.5) {\n privacyTier = 'High Statistical Utility / Minimal Noise'\n privacyBadgeColor = 'bg-warning/10 text-warning border-warning/30'\n }\n\n return {\n similarity,\n leakageRisk: leakageRisk.toFixed(2),\n correlationRetention,\n wassersteinDistance,\n laplaceNoiseStd,\n privacyTier,\n privacyBadgeColor,\n }\n })\n\n // Estimated file size calculation\n readonly estimatedFileSize = computed(() => {\n const bytesPerRow = this.outputFormat() === 'csv' ? 180 : 380\n const totalKb = (this.rowCount() * bytesPerRow) / 1024\n if (totalKb < 1024) {\n return `~${totalKb.toFixed(1)} KB`\n }\n return `~${(totalKb / 1024).toFixed(2)} MB`\n })\n\n protected formatButtonClass(format: 'csv' | 'jsonl'): string {\n return cn(\n FORMAT_BUTTON,\n this.outputFormat() === format\n ? 'bg-background text-foreground shadow-xs'\n : 'text-muted-foreground hover:text-foreground',\n )\n }\n\n protected rowPresetClass(preset: number): string {\n return cn(\n 'cursor-pointer rounded-md border px-2.5 py-1 text-xs font-medium transition-colors',\n this.rowCount() === preset\n ? 'bg-primary text-primary-foreground border-primary shadow-xs'\n : 'bg-muted/40 text-muted-foreground border-border hover:bg-muted hover:text-foreground',\n )\n }\n\n protected presetLabel(preset: number): string {\n return preset >= 1000 ? `${preset / 1000}k` : String(preset)\n }\n\n protected epsilonPresetClass(eps: number): string {\n return cn(\n 'cursor-pointer rounded-lg border p-2 text-center text-xs transition-colors',\n this.epsilon() === eps\n ? 'border-primary bg-primary/10 text-primary font-semibold'\n : 'border-border bg-background text-muted-foreground hover:bg-muted/40',\n )\n }\n\n protected privacyBannerClass(): string {\n return cn('flex items-start gap-2.5 rounded-lg border p-3 text-xs', this.fidelityMetrics().privacyBadgeColor)\n }\n\n protected fieldTypeBadgeClass(type: FieldDataType): string {\n return cn('px-1.5 py-0 text-xs font-semibold tracking-wider uppercase', FIELD_TYPE_BADGE[type])\n }\n\n protected onRowCountChange(value: number[]): void {\n if (value[0] !== undefined) this.rowCount.set(value[0])\n }\n\n protected onEpsilonChange(value: number[]): void {\n if (value[0] !== undefined) this.epsilon.set(value[0])\n }\n\n protected setNewFieldType(value: string): void {\n this.newFieldType.set(value as FieldDataType)\n }\n\n // Handle template selection switch\n handleTemplateChange(key: string): void {\n this.selectedTemplateKey.set(key)\n if (TEMPLATES[key]) {\n this.activeFields.set(cloneFields(key))\n this.triggerRegeneration()\n }\n }\n\n // Generation trigger\n triggerRegeneration(): void {\n this.isGenerating.set(true)\n this.generationSeed.set(Math.floor(Math.random() * 90000) + 10000)\n this.later(() => this.isGenerating.set(false), 450)\n }\n\n // Add field handler\n handleAddCustomField(): void {\n const name = this.newFieldName()\n if (!name.trim()) return\n\n const type = this.newFieldType()\n const cleanName = name.trim().toLowerCase().replace(/\\s+/g, '_')\n const newField: SchemaField = {\n id: `f-custom-${Date.now()}`,\n name: cleanName,\n label: name.trim(),\n type,\n description: `Custom ${type} schema column`,\n params:\n type === 'gaussian_float'\n ? { mean: 500, stdDev: 120, currency: '$' }\n : type === 'category_weights'\n ? { categories: ['Alpha', 'Beta', 'Gamma'], weights: [0.5, 0.3, 0.2] }\n : {},\n }\n\n this.activeFields.update((prev) => [...prev, newField])\n this.newFieldName.set('')\n this.showAddField.set(false)\n this.triggerRegeneration()\n }\n\n // Remove field handler\n removeField(fieldId: string): void {\n if (this.activeFields().length <= 2) return\n this.activeFields.update((prev) => prev.filter((f) => f.id !== fieldId))\n this.triggerRegeneration()\n }\n\n // Copy JSON lines\n async copyJsonLines(): Promise {\n try {\n const jsonLinesText = this.previewRecords()\n .map((r) => JSON.stringify(r))\n .join('\\n')\n if (navigator?.clipboard?.writeText) {\n await navigator.clipboard.writeText(jsonLinesText)\n }\n this.copied.set(true)\n this.later(() => this.copied.set(false), 2000)\n } catch (err) {\n console.error('Failed to copy to clipboard', err)\n }\n }\n\n // Download simulation\n handleDownloadDataset(): void {\n const rowCount = this.rowCount()\n const filename = `synthetic_${this.selectedTemplateKey()}_${rowCount}_rows.${this.outputFormat() === 'csv' ? 'csv' : 'jsonl'}`\n this.downloadToast.set(`Downloaded ${rowCount.toLocaleString()} synthetic records (${filename})`)\n this.later(() => this.downloadToast.set(null), 3500)\n }\n\n private later(fn: () => void, ms: number): void {\n const t = setTimeout(() => {\n this.timers.delete(t)\n fn()\n }, ms)\n this.timers.add(t)\n }\n}\n", "type": "registry:block", "target": "~/src/app/components/blocks/synthetic-data-generator/synthetic-data-generator.component.ts" }, { "path": "packages/registry-angular/blocks/synthetic-data-generator/synthetic-data-preview.component.ts", "content": "import { Component, EventEmitter, Input, Output, ChangeDetectionStrategy } from '@angular/core'\nimport {\n BarChart3,\n Binary,\n Check,\n Code2,\n Copy,\n LucideAngularModule,\n Shield,\n ShieldCheck,\n Table,\n Zap,\n} from 'lucide-angular'\nimport { UiBadgeComponent } from '@/ui/badge/badge.component'\nimport { UiButtonComponent } from '@/ui/button/button.component'\nimport {\n UiCardComponent,\n UiCardContentComponent,\n UiCardDescriptionComponent,\n UiCardFooterComponent,\n UiCardHeaderComponent,\n UiCardTitleComponent,\n} from '@/ui/card/card.component'\nimport { UiProgressComponent } from '@/ui/progress/progress.component'\nimport {\n UiTableBodyComponent,\n UiTableCellComponent,\n UiTableComponent,\n UiTableHeadComponent,\n UiTableHeaderComponent,\n UiTableRowComponent,\n} from '@/ui/table/table.component'\nimport { UiTabsComponent, UiTabsListComponent, UiTabsTriggerComponent } from '@/ui/tabs/tabs.component'\nimport type { FidelityMetrics, GeneratedRecord, SchemaField } from './synthetic-data-types'\n\nexport type SyntheticDataPreviewTab = 'table' | 'json' | 'metrics'\n\n/** How a preview cell renders its value (React's nested ternary in SyntheticDataPreview). */\ntype CellKind = 'key' | 'category' | 'float' | 'card' | 'email' | 'text'\n\n/**\n * Right-hand column of the synthetic data generator: the fidelity / privacy gauge banner and\n * the live preview card with Table / JSONL / Marginals views. Port of the React\n * `SyntheticDataPreview` 1:1; the parent owns all state and passes it in.\n */\n@Component({\n changeDetection: ChangeDetectionStrategy.Eager,\n selector: 'ui-synthetic-data-preview, [ui-synthetic-data-preview]',\n standalone: true,\n host: { class: 'contents' },\n imports: [\n LucideAngularModule,\n UiBadgeComponent,\n UiButtonComponent,\n UiCardComponent,\n UiCardContentComponent,\n UiCardDescriptionComponent,\n UiCardFooterComponent,\n UiCardHeaderComponent,\n UiCardTitleComponent,\n UiProgressComponent,\n UiTableComponent,\n UiTableHeaderComponent,\n UiTableBodyComponent,\n UiTableRowComponent,\n UiTableHeadComponent,\n UiTableCellComponent,\n UiTabsComponent,\n UiTabsListComponent,\n UiTabsTriggerComponent,\n ],\n template: `\n
\n \n
\n
\n
\n
\n \n \n {{ fidelityMetrics.similarity }}% Distribution Similarity • {{ fidelityMetrics.leakageRisk }}% PII\n Leakage Risk\n \n
\n \n ε-DP Bound: ≤ {{ epsilon.toFixed(1) }}\n \n
\n
\n\n
\n \n
\n
\n Fidelity Score\n \n
\n
\n {{ fidelityMetrics.similarity }}%\n
\n
\n
\n\n \n
\n
\n Covariance Retention\n \n
\n
\n {{ fidelityMetrics.correlationRetention }}%\n
\n
\n
\n\n \n
\n
\n Wasserstein Distance\n \n
\n
\n {{ fidelityMetrics.wassersteinDistance }}\n
\n
Low error bound (< 0.05)
\n
\n
\n
\n\n \n
\n
\n
\n
\n
\n \n
\n
\n

Live Generated Synthetic Data Preview

\n

\n Real-time sampler showing first {{ previewRecords.length }} synthesized records\n

\n
\n
\n\n \n
\n
\n \n \n \n \n
\n
\n
\n
\n\n
\n \n @if (previewTab === 'table') {\n
\n
\n
\n \n \n \n @for (f of activeFields; track f.id) {\n \n {{ f.name }}\n \n }\n \n \n \n @for (row of previewRecords; track $index) {\n \n @for (f of activeFields; track f.id) {\n \n @switch (cellKind(f)) {\n @case ('key') {\n {{ row[f.name] }}\n }\n @case ('category') {\n \n {{ row[f.name] }}\n \n }\n @case ('float') {\n \n {{ row[f.name] }}\n \n }\n @case ('card') {\n {{ row[f.name] }}\n }\n @case ('email') {\n {{ row[f.name] }}\n }\n @default {\n {{ row[f.name] }}\n }\n }\n \n }\n \n }\n \n \n
\n
\n\n
\n \n \n Showing {{ previewRecords.length }} sample records of {{ rowCount.toLocaleString() }} generated rows\n \n Seed: #{{ generationSeed }}\n
\n
\n }\n\n \n @if (previewTab === 'json') {\n
\n
\n Streaming JSON Lines formatted output:\n \n @if (copied) {\n \n } @else {\n \n }\n {{ copied ? 'Copied JSONL' : 'Copy JSON Lines' }}\n \n
\n\n \n @for (rec of previewRecords; track $index; let idx = $index) {\n
\n {{ idx + 1 }} \n {{ '{ ' }}\n @for (entry of entries(rec); track entry[0]; let last = $last) {\n \"{{ entry[0] }}\"\n : \n {{\n jsonValue(entry[1])\n }}\n @if (!last) {\n , \n }\n }\n {{ ' }' }}\n
\n }\n
\n
\n }\n\n \n @if (previewTab === 'metrics') {\n
\n
\n \n
\n
\n Gaussian Distribution Overlap\n 96.2%\n
\n

\n Kolmogorov-Smirnov distance between synthetic batch and continuous Gaussian ground truth.\n

\n
\n
\n\n \n
\n
\n Categorical Frequency Prior\n 98.4%\n
\n

\n Total variation distance across categorical weight assignments.\n

\n
\n
\n
\n\n \n
\n
\n \n Mathematical Privacy Guarantees • Formal Specs\n
\n
\n
\n \n Differential Privacy (ε = {{ epsilon.toFixed(1) }}, δ = 10⁻⁵)\n
\n
\n \n K-Anonymity (k = 5) Cohort Clamping\n
\n
\n \n Zero Direct Row Memorization\n
\n
\n \n Ephemeral In-Memory Compute\n
\n
\n
\n
\n }\n
\n\n \n \n \n Powered by LLM Semantic Synthesis & Laplace Noise Engine\n \n \n Output: {{ outputFormat.toUpperCase() }} • {{ activeFields.length }} fields •\n {{ rowCount.toLocaleString() }} records\n \n
\n \n \n `,\n})\nexport class UiSyntheticDataPreviewComponent {\n @Input() previewTab: SyntheticDataPreviewTab = 'table'\n @Output() previewTabChange = new EventEmitter()\n @Input() previewRecords: GeneratedRecord[] = []\n @Input() activeFields: SchemaField[] = []\n @Input() fidelityMetrics: FidelityMetrics = {\n similarity: 0,\n leakageRisk: '0.00',\n correlationRetention: 0,\n wassersteinDistance: 0,\n laplaceNoiseStd: '0.000',\n privacyTier: '',\n privacyBadgeColor: '',\n }\n @Input() epsilon = 0\n @Input() rowCount = 0\n @Input() outputFormat: 'csv' | 'jsonl' = 'csv'\n @Input() generationSeed = 0\n @Input() copied = false\n @Output() copyJsonLines = new EventEmitter()\n\n protected readonly BarChart3 = BarChart3\n protected readonly Binary = Binary\n protected readonly Check = Check\n protected readonly Code2 = Code2\n protected readonly Copy = Copy\n protected readonly Shield = Shield\n protected readonly ShieldCheck = ShieldCheck\n protected readonly TableIcon = Table\n protected readonly Zap = Zap\n\n protected onTabChange(value: string): void {\n this.previewTabChange.emit(value as SyntheticDataPreviewTab)\n }\n\n /** Keys (id / uuid / token columns) first, then by field type, exactly like React's ternary chain. */\n protected cellKind(f: SchemaField): CellKind {\n if (f.name === 'id' || f.name.includes('id') || f.name.includes('uuid') || f.name.includes('token')) return 'key'\n if (f.type === 'category_weights') return 'category'\n if (f.type === 'gaussian_float') return 'float'\n if (f.type === 'credit_card') return 'card'\n if (f.type === 'email') return 'email'\n return 'text'\n }\n\n protected entries(rec: GeneratedRecord): [string, string | number][] {\n return Object.entries(rec)\n }\n\n protected isNumber(value: string | number): boolean {\n return typeof value === 'number'\n }\n\n protected jsonValue(value: string | number): string {\n return typeof value === 'number' ? String(value) : `\"${value}\"`\n }\n}\n", "type": "registry:block", "target": "~/src/app/components/blocks/synthetic-data-generator/synthetic-data-preview.component.ts" }, { "path": "packages/registry-angular/blocks/synthetic-data-generator/synthetic-data-types.ts", "content": "export type FieldDataType = 'name' | 'email' | 'credit_card' | 'gaussian_float' | 'date_range' | 'category_weights'\n\nexport interface SchemaField {\n id: string\n name: string\n label: string\n type: FieldDataType\n description?: string\n params: {\n mean?: number\n stdDev?: number\n currency?: string\n unit?: string\n categories?: string[]\n weights?: number[]\n domain?: string\n dateStart?: string\n dateEnd?: string\n }\n}\n\nexport interface TemplatePreset {\n id: string\n name: string\n description: string\n category: string\n fields: SchemaField[]\n}\n\nexport interface GeneratedRecord {\n id: string\n [key: string]: string | number\n}\n\nexport interface FidelityMetrics {\n similarity: number\n leakageRisk: string\n correlationRetention: number\n wassersteinDistance: number\n laplaceNoiseStd: string\n privacyTier: string\n privacyBadgeColor: string\n}\n\n/** Inputs of `` (`class` is merged onto the root). */\nexport interface SyntheticDataGeneratorProps {\n class?: string\n initialTemplate?: string\n initialRowCount?: number\n initialEpsilon?: number\n}\n", "type": "registry:block", "target": "~/src/app/components/blocks/synthetic-data-generator/synthetic-data-types.ts" }, { "path": "packages/registry-angular/blocks/synthetic-data-generator/synthetic-data-templates.ts", "content": "import type { TemplatePreset } from './synthetic-data-types'\n\nexport const TEMPLATES: Record = {\n 'customer-transactions': {\n id: 'customer-transactions',\n name: 'Customer Transactions & Demographics Dataset',\n description:\n 'High-dimensional e-commerce customer cohort, synthetic PII, spending distribution, and country codes.',\n category: 'E-Commerce & Retail',\n fields: [\n {\n id: 'f-id',\n name: 'cust_id',\n label: 'Customer ID',\n type: 'name',\n description: 'Salted pseudonymized primary key',\n params: {},\n },\n {\n id: 'f-name',\n name: 'full_name',\n label: 'Full Name',\n type: 'name',\n description: 'Synthetic person name generator',\n params: {},\n },\n {\n id: 'f-email',\n name: 'email_address',\n label: 'Synthetic Email',\n type: 'email',\n description: 'Faker email with domain isolation',\n params: { domain: 'syn-corp.io' },\n },\n {\n id: 'f-card',\n name: 'masked_card',\n label: 'Card PAN',\n type: 'credit_card',\n description: 'Luhn-compliant masked payment token',\n params: {},\n },\n {\n id: 'f-spend',\n name: 'spending_amount',\n label: 'Spending Amount',\n type: 'gaussian_float',\n description: 'Gaussian spending curve (μ=$342.50, σ=$85.20)',\n params: { mean: 342.5, stdDev: 85.2, currency: '$' },\n },\n {\n id: 'f-country',\n name: 'country_code',\n label: 'Country Region',\n type: 'category_weights',\n description: 'Categorical distribution [US, DE, GB, FR, JP]',\n params: {\n categories: ['US', 'DE', 'GB', 'FR', 'JP'],\n weights: [0.45, 0.2, 0.15, 0.1, 0.1],\n },\n },\n {\n id: 'f-created',\n name: 'created_at',\n label: 'Registration Date',\n type: 'date_range',\n description: 'ISO-8601 timestamp range (2024-2025)',\n params: { dateStart: '2024-01-01', dateEnd: '2025-08-20' },\n },\n ],\n },\n 'healthcare-ehr': {\n id: 'healthcare-ehr',\n name: 'Healthcare Patient Clinical Records (EHR)',\n description: 'HIPAA-compliant patient cohort with biomarker telemetry, blood pressure, and triage severity tiers.',\n category: 'Healthcare & Life Sciences',\n fields: [\n {\n id: 'f-h-id',\n name: 'patient_id',\n label: 'Patient ID',\n type: 'name',\n description: 'Deterministic HIPAA salted identifier',\n params: {},\n },\n {\n id: 'f-h-name',\n name: 'patient_name',\n label: 'Patient Name',\n type: 'name',\n description: 'Synthetic medical record moniker',\n params: {},\n },\n {\n id: 'f-h-email',\n name: 'contact_email',\n label: 'Contact Email',\n type: 'email',\n description: 'Synthetic provider contact',\n params: { domain: 'med-synthetic.org' },\n },\n {\n id: 'f-h-glucose',\n name: 'blood_glucose',\n label: 'Glucose (mg/dL)',\n type: 'gaussian_float',\n description: 'Fast fasting blood glucose (μ=108.4, σ=22.1)',\n params: { mean: 108.4, stdDev: 22.1, unit: 'mg/dL' },\n },\n {\n id: 'f-h-bp',\n name: 'systolic_bp',\n label: 'Systolic BP',\n type: 'gaussian_float',\n description: 'Arterial blood pressure (μ=124.0, σ=15.5)',\n params: { mean: 124.0, stdDev: 15.5, unit: 'mmHg' },\n },\n {\n id: 'f-h-tier',\n name: 'triage_tier',\n label: 'Triage Severity',\n type: 'category_weights',\n description: 'Clinical priority classification',\n params: {\n categories: ['Routine', 'Elevated', 'Urgent', 'Critical'],\n weights: [0.52, 0.28, 0.14, 0.06],\n },\n },\n {\n id: 'f-h-date',\n name: 'admission_date',\n label: 'Admission Date',\n type: 'date_range',\n description: 'Encounter timestamps',\n params: { dateStart: '2024-06-01', dateEnd: '2025-08-20' },\n },\n ],\n },\n 'fintech-fraud': {\n id: 'fintech-fraud',\n name: 'Fintech Fraud Detection & Risk Scoring',\n description: 'Transaction monitoring telemetry for anti-fraud classifiers with skewed anomalous patterns.',\n category: 'Fintech & Risk',\n fields: [\n {\n id: 'f-ff-id',\n name: 'tx_token',\n label: 'Transaction ID',\n type: 'name',\n description: 'Cryptographic ledger token',\n params: {},\n },\n {\n id: 'f-ff-name',\n name: 'merchant_name',\n label: 'Merchant Name',\n type: 'name',\n description: 'Synthetic counterparty vendor',\n params: {},\n },\n {\n id: 'f-ff-email',\n name: 'auth_email',\n label: 'Auth Email',\n type: 'email',\n description: 'Tokenized transaction origin email',\n params: { domain: 'syn-pay.net' },\n },\n {\n id: 'f-ff-card',\n name: 'card_pan',\n label: 'Card PAN',\n type: 'credit_card',\n description: 'Tokenized 16-digit card surrogate',\n params: {},\n },\n {\n id: 'f-ff-amt',\n name: 'settlement_usd',\n label: 'Settlement ($)',\n type: 'gaussian_float',\n description: 'Transaction gross volume (μ=$1420.00, σ=$480.00)',\n params: { mean: 1420.0, stdDev: 480.0, currency: '$' },\n },\n {\n id: 'f-ff-verdict',\n name: 'risk_verdict',\n label: 'Risk Verdict',\n type: 'category_weights',\n description: 'Ground truth classifier target',\n params: {\n categories: ['Legitimate', 'Suspicious', 'Flagged_Fraud'],\n weights: [0.89, 0.08, 0.03],\n },\n },\n {\n id: 'f-ff-ts',\n name: 'settled_at',\n label: 'Settlement Time',\n type: 'date_range',\n description: 'High-precision event timestamp',\n params: { dateStart: '2025-01-01', dateEnd: '2025-08-20' },\n },\n ],\n },\n 'saas-telemetry': {\n id: 'saas-telemetry',\n name: 'SaaS Product Telemetry & Usage Metrics',\n description: 'B2B subscription telemetry with active seat distribution, recurring revenue, and subscription tiers.',\n category: 'Cloud Software & B2B',\n fields: [\n {\n id: 'f-st-id',\n name: 'org_uuid',\n label: 'Organization UUID',\n type: 'name',\n description: 'Tenant identifier',\n params: {},\n },\n {\n id: 'f-st-name',\n name: 'tenant_name',\n label: 'Tenant Name',\n type: 'name',\n description: 'Synthetic enterprise tenant moniker',\n params: {},\n },\n {\n id: 'f-st-email',\n name: 'admin_email',\n label: 'Admin Email',\n type: 'email',\n description: 'Primary owner email',\n params: { domain: 'syn-saas.io' },\n },\n {\n id: 'f-st-seats',\n name: 'active_seats',\n label: 'Active Seats',\n type: 'gaussian_float',\n description: 'Provisioned enterprise seats (μ=48.0, σ=18.0)',\n params: { mean: 48.0, stdDev: 18.0, unit: 'seats' },\n },\n {\n id: 'f-st-mrr',\n name: 'mrr_usd',\n label: 'Monthly Revenue',\n type: 'gaussian_float',\n description: 'MRR ledger (μ=$2850.00, σ=$950.00)',\n params: { mean: 2850.0, stdDev: 950.0, currency: '$' },\n },\n {\n id: 'f-st-plan',\n name: 'plan_tier',\n label: 'Plan Tier',\n type: 'category_weights',\n description: 'Subscription entitlement level',\n params: {\n categories: ['Starter', 'Growth', 'Enterprise'],\n weights: [0.48, 0.36, 0.16],\n },\n },\n {\n id: 'f-st-date',\n name: 'provisioned_at',\n label: 'Provisioned Date',\n type: 'date_range',\n description: 'Cohort subscription activation',\n params: { dateStart: '2024-01-01', dateEnd: '2025-08-20' },\n },\n ],\n },\n}\n\nexport const FIRST_NAMES = [\n 'Elena',\n 'Malik',\n 'Siddharth',\n 'Claire',\n 'Liam',\n 'Amara',\n 'Kenji',\n 'Sofia',\n 'Devon',\n 'Maya',\n 'Tobias',\n 'Fatima',\n 'Mateo',\n 'Astrid',\n 'Zara',\n]\n\nexport const LAST_NAMES = [\n 'Rostova',\n 'Al-Mansoor',\n 'Patel',\n 'Chen',\n \"O'Connor\",\n 'Okafor',\n 'Sato',\n 'Alvarez',\n 'Vance',\n 'Lin',\n 'Richter',\n 'Zahra',\n 'Silva',\n 'Lindqvist',\n 'Novak',\n]\n\nexport const CARD_BRANDS = ['Visa', 'Mastercard', 'Amex']\n\nexport function boxMullerRandom(\n mean: number,\n stdDev: number,\n seedOffset: number,\n seed: number,\n noiseMechanism: boolean,\n epsilon: number,\n): number {\n const u1 = Math.max(1e-6, Math.abs(Math.sin(seed + seedOffset * 17.3)))\n const u2 = Math.max(1e-6, Math.abs(Math.cos(seed + seedOffset * 31.7)))\n const z0 = Math.sqrt(-2.0 * Math.log(u1)) * Math.cos(2.0 * Math.PI * u2)\n\n const dpNoiseScale = noiseMechanism ? stdDev / (Math.max(0.1, epsilon) * 12) : 0\n const dpNoise = Math.sin(seed * 3.7 + seedOffset * 5.1) * dpNoiseScale\n\n return mean + z0 * stdDev + dpNoise\n}\n", "type": "registry:block", "target": "~/src/app/components/blocks/synthetic-data-generator/synthetic-data-templates.ts" }, { "path": "packages/registry-angular/blocks/synthetic-data-generator/index.ts", "content": "export {\n UiSyntheticDataGeneratorComponent,\n type FieldDataType,\n type FidelityMetrics,\n type GeneratedRecord,\n type SchemaField,\n type SyntheticDataGeneratorProps,\n type TemplatePreset,\n} from './synthetic-data-generator.component'\nexport { UiSyntheticDataPreviewComponent, type SyntheticDataPreviewTab } from './synthetic-data-preview.component'\nexport { TEMPLATES, FIRST_NAMES, LAST_NAMES, CARD_BRANDS, boxMullerRandom } from './synthetic-data-templates'\n", "type": "registry:block", "target": "~/src/app/components/blocks/synthetic-data-generator/index.ts" } ], "dependencies": [ "lucide-angular" ], "devDependencies": [], "registryDependencies": [ "https://uipkge.dev/r/angular/badge.json", "https://uipkge.dev/r/angular/button.json", "https://uipkge.dev/r/angular/card.json", "https://uipkge.dev/r/angular/input.json", "https://uipkge.dev/r/angular/progress.json", "https://uipkge.dev/r/angular/select.json", "https://uipkge.dev/r/angular/separator.json", "https://uipkge.dev/r/angular/slider.json", "https://uipkge.dev/r/angular/switch.json", "https://uipkge.dev/r/angular/table.json", "https://uipkge.dev/r/angular/tabs.json" ], "description": "Gretel and Tonic-style LLM and mathematical synthetic dataset generator with differential privacy epsilon controls, custom schema field builders, K-anonymity policy toggles, live tabular CSV preview, and streaming JSON Lines export.", "categories": [ "ai", "data", "app", "security" ] }