msfPDF

# msfPDF — Complete Technical Reference for LLMs & Coding Agents (`v0.2.0`)

- **Repository:** https://github.com/mfelisberto90/msfPDF
- **Interactive Showcase:** https://mfelisberto90.github.io/msfPDF/
- **JSON Schema:** https://mfelisberto90.github.io/msfPDF/schemas/document-spec.schema.json
- **Licence:** MIT (Author: Murilo Felisberto, Edinburgh, Scotland, UK)

---

## 1. What Problem `msfPDF` Solves

1. **Eliminates 500 MB Headless Browsers & SSRF:** Generating invoices, receipts, or contracts with Puppeteer/Playwright/wkhtmltopdf consumes `300–800 MB` RAM, adds `400–2,500 ms` latency, and exposes servers to HTML/CSS/SVG Server-Side Request Forgery (`169.254.169.254`) and Local File Inclusion (LFI). `msfPDF` parses strictly typed JSON (`DocumentSpec`) with zero network/DOM surface.
2. **Zero `xref` Byte-Offset Drift Across Scripts (`Latin` + `Русский` + `العربية`):**
   - Standard Latin & Western European characters (`£`, `€`, `ç`, `ã`, `m²`, `—`) are mapped to 7-bit ASCII `/WinAnsiEncoding` octal escapes (`\243`, `\200`, `\347`, `\343`, `\262`, `\227`).
   - **Russian & Slavic Cyrillic** (`А-Я`, `а-я`, `Ё`, `ё`, `₽`, `№`) and **Connected Arabic RTL** (`العربية` contextual `Isolated`/`Initial`/`Medial`/`Final` joining + mandatory `لا`/`لأ`/`لإ`/`لآ` ligatures + UAX #9 BiDi visual reordering) are dynamically subsetted on-demand into `2048-UPM` `%PDF-1.4` Type 3 cubic Bézier vector fonts (`/MsfVectorRegular*`, `/MsfVectorBold*`) with `<HEX> Tj` streams and `/ToUnicode` CMaps.
   - Pure Latin documents emit **zero** Type 3 font objects (`0` byte overhead). Every generated PDF remains **100% 7-bit ASCII** (`0x0A` and `0x20..=0x7E`), guaranteeing exact `xref` and `/Length` offsets in Adobe Acrobat, Ghostscript, Poppler (`pdftoppm`/`pdftotext`), Chrome PDFium, and Apple Preview.
3. **1:1 Byte-Identical Polyglot Parity:** Feeding the same `DocumentSpec` JSON into `msfpdf-core` (Rust), `msfpdf` (TypeScript — `0` runtime `npm` dependencies), or `msfpdf render` (CLI) produces the **exact same SHA-256 `%PDF-1.4` byte buffer**.
4. **Exact Integer Minor-Unit Currency Maths:** Binary floating-point numbers are banned in monetary totals. `format_money_minor` / `formatMoneyMinor` format exact signed integers (`pence`, `cents`, `centavos`, `kopecks`, `fils`, `halalas`, `yen`) across `GBP`, `EUR`, `USD`, `BRL`, `JPY`, `RUB`, `AED`, and `SAR`.
5. **Pure Vector QR Codes (`GF(256)`):** Built-in ISO/IEC 18004 Reed-Solomon encoder (`0x11D`) with 8-mask penalty evaluation and horizontal run-length rectangle merging (`re f`) — sharp at any zoom with zero PNG/raster dependencies.

---

## 2. Measured Performance (Single Core, Linux x86_64)

- **`msfpdf-core` (Rust, standard A4):** `0.16 ms` P50 (`~6,300 docs/s/core`), `8.8 KB` PDF, `~16 KB` RAM buffer.
- **`msfpdf-core` (Rust, A4 + Vector QR):** `0.52 ms` P50 (`~1,900 docs/s/core`), `14.0 KB` PDF.
- **`msfpdf` (TypeScript, Node 22 / Edge V8, A4 + Vector QR):** `0.79 ms` P50 (`~1,180 docs/s/core`), `14.0 KB` PDF.
- **Trilingual English + Russian + Connected Arabic + Vector QR:** `~2.2 ms` TypeScript / `< 1.1 ms` Rust, `100%` SHA-256 identical.

---

## 3. TypeScript / JavaScript Edge & Server Quickstart (`0` Dependencies)

Works unmodified in **Node.js 22+**, **Bun**, **Deno**, **Cloudflare Workers**, **Vercel Edge**, and **Browsers**:

```typescript
import { renderDocument, renderDocumentJson, formatMoneyMinor } from './packages/ts/src/index.ts';
import { TenantPdfVault } from './packages/ts/src/vault.ts';

const pdfBytes: Uint8Array = renderDocument({
  page_format: 'a4', // 'a4' | 'letter' | 'thermal80mm'
  theme: 'navy_fintech', // 'navy_fintech' | 'emerald_editorial' | 'slate_minimal' | 'crimson_compliance'
  metadata: {
    issuer_name: 'Meridian Clearing House Ltd • ميريديان للتسوية',
    document_code: '#GLB-2026-9042',
    issued_at_label: '03 Nov 2026',
  },
  blocks: [
    {
      type: 'hero_header',
      kicker: 'MULTILINGUAL TAX INVOICE • СЧЁТ-ФАКТУРА • فاتورة ضريبية',
      title: 'Meridian Clearing House Ltd',
      subtitle: 'Cross-Border Settlement — London • Dubai DIFC • Алматы',
      badge_label: 'CLEARING REF / №',
      badge_code: '#GLB-2026-9042',
      badge_meta: '03/11/2026 • AED / RUB / GBP',
    },
    {
      type: 'bento_grid',
      columns: 2,
      cards: [
        {
          primary_label: 'MENA COUNTERPARTY • الطرف المتعاقد',
          primary_value: 'شركة الخليج للأنظمة السحابية ذ.م.م',
          secondary_label: 'JURISDICTION',
          secondary_value: 'Dubai International Financial Centre • دبي',
        },
        {
          primary_label: 'EURASIA NODE • КОНТРАГЕНТ',
          primary_value: 'ООО «Северный Контур Технолоджи»',
          secondary_label: 'УЗЕЛ РАСЧЁТОВ',
          secondary_value: 'Пресненская наб. 12 • № 04',
        },
      ],
    },
    {
      type: 'financial_summary',
      currency: 'AED',
      rows: [
        { label: 'Net Clearing Subtotal • Промежуточный итог • المجموع الفرعي', amount_minor: 1_850_000 },
        { label: 'GCC Standard VAT (5.0%) • ضريبة القيمة المضافة', amount_minor: 92_500 },
        { label: 'RUB Spot Reference • Справочный курс', text_value: formatMoneyMinor(51_481_250, 'RUB'), emphasize: true },
      ],
    },
    {
      type: 'callout_banner',
      title: 'TOTAL SETTLEMENT DUE • ИТОГО К ОПЛАТЕ • الإجمالي المستحق',
      subtitle: 'SWIFT / UAEFTS Instant Settlement within 15 calendar days',
      right_value: formatMoneyMinor(1_942_500, 'AED'), // "AED 19,425.00"
      tone: 'accent',
    },
    {
      type: 'qr_verification',
      payload: 'https://verify.meridianclearing.example/doc/GLB-2026-9042',
      headline: 'CRYPTOGRAPHIC VERIFICATION • ПРОВЕРКА • التحقق الرقمي',
      details: 'Scan vector QR to verify SHA-256 ledger digest.',
      code_display: 'SHA256: 4E19-C80A-77B2-D301',
    },
  ],
});

// Optional Node/Bun multi-tenant 0600 atomic disk vault with 160-bit CSPRNG token:
const vault = new TenantPdfVault('./.vault-data', 'invoice');
const issued = await vault.issueDocument('tenant_dubai_01', {
  metadata: { issuer_name: 'Meridian Ltd', document_code: 'INV-1', issued_at_label: '2026-11-03' },
  blocks: [{ type: 'divider' }],
});
// Read back strictly scoped to owning tenant (cross-tenant IDs & '../' fail closed):
const verifiedBytes = await vault.readDocument('tenant_dubai_01', issued.token);

// Cryptographic signing & verification (Ed25519 / RFC 8032 with zero external deps):
import { generateKeypair, signPdf, verifyPdf, embedAuditManifest, extractAuditManifest } from 'msfpdf/crypto';

const { privateKeySeedHex, publicKeyHex } = generateKeypair();
const manifest = signPdf(pdfBytes, privateKeySeedHex, 'Caledonia Compliance Authority');
const signedPdf = embedAuditManifest(pdfBytes, manifest);

// Verify signed PDF (tamper-evident):
const extracted = extractAuditManifest(signedPdf);
const isValid = verifyPdf(signedPdf, publicKeyHex, extracted.signatureHex); // true
```

---

## 4. Rust API (`msfpdf-core`)

```rust
use msfpdf_core::{
    format_money_minor, generate_keypair, render_document_json, sign_pdf, verify_pdf,
    AuditManifest, Currency, SecurityLimits, TenantPdfVault,
};

fn example(raw_json: &[u8]) -> Result<Vec<u8>, msfpdf_core::PdfError> {
    assert_eq!(format_money_minor(45_000_050, Currency::RUB), "450 000,50 ₽");
    assert_eq!(format_money_minor(1_942_500, Currency::AED), "AED 19,425.00");
    let pdf = render_document_json(raw_json, SecurityLimits::default())?;

    // Ed25519 signing:
    let (seed, pubkey) = generate_keypair()?;
    let manifest = sign_pdf(&pdf, &seed, "Treasury Validator", None)?;
    assert!(verify_pdf(&pdf, &pubkey, &manifest.signature_hex.as_bytes()[..64].try_into().unwrap_or([0; 64]))?);
    Ok(pdf)
}
```

---

## 5. Standalone CLI (`msfpdf` — `< 1 MB` Release Binary)

```bash
cargo build --release -p msfpdf-cli

# Compile file or stdin -> stdout for Python / PHP / Go / Ruby backends:
./target/release/msfpdf render --in examples/global-multilingual-invoice.json --out multilingual.pdf
cat examples/uk-b2b-invoice.json | ./target/release/msfpdf render --in - --out invoice.pdf

# Issue into atomic 0600 multi-tenant vault:
./target/release/msfpdf vault-issue --vault ./.vault --prefix inv --tenant tenant_uk_01 --in examples/uk-b2b-invoice.json

# Cryptographic Ed25519 key generation, signing, and verification:
./target/release/msfpdf keygen --out keys.json
SEED=$(jq -r .private_key_seed_hex keys.json)
PUB=$(jq -r .public_key_hex keys.json)

# Sign with embedded %MsfAuditManifest trailer:
./target/release/msfpdf sign --in invoice.pdf --key "$SEED" --identity "Audit Partner" --embed --out invoice-signed.pdf

# Verify integrity and tamper detection:
./target/release/msfpdf verify --in invoice-signed.pdf --pubkey "$PUB"

# Run single-core latency benchmark:
./target/release/msfpdf bench --iterations 5000
```

---

## 6. Complete `DocumentSpec` Block Reference

Top-level `DocumentSpec` properties:
- `page_format`: `"a4"` (default, `595.28 x 841.89 pt`), `"letter"` (`612 x 792 pt`), or `"thermal80mm"` (`226.77 x 510 pt`).
- `theme`: `"navy_fintech"` (default), `"emerald_editorial"`, `"slate_minimal"`, or `"crimson_compliance"`.
- `accent_hex_override`: Optional `"#RRGGBB"` override.
- `metadata`: `{ issuer_name, document_code, issued_at_label, footer_notice?, page_counter_template? }`.
- `continuation_header`: Optional `{ title, subtitle }` rendered at the top of automatically paginated continuation pages (`2..N`).
- `blocks`: Ordered array of layout primitives (`max_blocks: 500`):
  1. `hero_header`: `{ kicker, title, subtitle, badge_label?, badge_code?, badge_meta?, logo_jpeg_base64? }` (validates JPEG SOF0/SOF2 dimensions without libjpeg).
  2. `section_heading`: `{ title }` (enforces keep-with-next orphan protection).
  3. `bento_grid`: `{ columns?: 1|2|3, cards: [{ primary_label, primary_value, secondary_label?, secondary_value? }] }`.
  4. `data_table`: `{ columns: [{ header, weight?, align?: 'left'|'center'|'right', bold_values? }], rows: string[][], zebra?: boolean }` (repeats table headers automatically across page breaks).
  5. `financial_summary`: `{ currency: 'GBP'|'EUR'|'USD'|'BRL'|'JPY'|'RUB'|'AED'|'SAR', rows: [{ label, amount_minor?: integer, currency?, text_value?, emphasize? }] }`.
  6. `callout_banner`: `{ title, subtitle, right_value?, tone?: 'accent'|'warning'|'danger'|'neutral' }`.
  7. `clause_list`: `{ paragraphs: string[] }` (wraps logically; RTL-dominant Arabic clauses right-align automatically inside the editorial card).
  8. `qr_verification`: `{ payload, headline, details, code_display? }`.
  9. `signature_grid`: `{ heading?, signers: [{ role_label, party_name }] }`.
  10. `key_value_list`: `{ rows: [{ label, value, emphasize? }], dashed_dividers?: boolean }`.
  11. `audit_manifest`: `{ manifest: { algorithm: 'Ed25519', document_sha256, signer_identity, public_key_hex, signature_hex, timestamp_iso } }` (renders institutional cryptographic audit seal with SHA-256 and Ed25519 badges).
  12. `divider`: `{}`.
  13. `spacer`: `{ height_pt: 2..200 }`.

---

## 7. Security Ceilings (`SecurityLimits`)

- `max_input_bytes` / `maxInputBytes`: `262,144` (`256 KiB`) checked *before* JSON parsing.
- `max_pages` / `maxPages`: `50` pages.
- `max_blocks` / `maxBlocks`: `500` blocks.
- `max_table_rows` / `maxTableRows`: `1,000` rows per table.
- Embedded JPEG logos: `<= 256 KiB` each, max `8` images, Grayscale/RGB SOF0/SOF2 header validation, `<= 4096 x 4096 px`.