feat(recibos): OCR capture for zona federal (ZOFEMAT Tijuana)

Adds the ZONA FEDERAL TIJUANA parser to the statement intake, measured
against 8 pages of real "Zona Federal Marítimo Terrestre" receipts — the
federal maritime-zone occupancy fee the municipality bills on beachfront
lots. Provider read on 8/8, amount on 8/8 (each verified against the
paper), concession clave on 6/8, period on 8/8, deadline on 2/8.

Four things the corpus forced:

- Tijuana bills predial and zona federal from the same treasury: same
  header, same Paseo del Centenario address, same ATB-541201 RFC. Every
  predial discriminator matches a zona federal page too, so whichever
  rule is asked first wins it. The only words exclusive to this layout
  are "Marítimo Terrestre", so its brand rule is asked ahead of all
  three predial ones — and its structural rule, anchored on the stub's
  "Derechos de ocupación", ahead of theirs.

- FEDERAL_ZONE.accountNumber is an amount, not a reference. It holds
  DATMEX.zfed, whose 77 values include 246.06, 2369.09, 22653.94 and a
  negative -1679, while the concession claves these receipts are keyed
  by appear nowhere in the database. Matching on that column could never
  hit — and because every row already has a value, the `[field]: null`
  guards on learnAccountRefs and on the review blank-service fill would
  never fire either, so every page would return to the queue every
  bimester forever. The clave moves to meterNumber, joining gas and
  Tijuana predial, and the first confirm teaches the match.

- The payable figure is not the printed subtotal. The municipality
  rounds to whole pesos and prints the difference on its own "Ajuste Ley
  Hacienda Mpal" line (-$0.05 against a 591.05 subtotal, $0.21 against
  2,872.79). The "Total a pagar" box carrying the rounded figure sits on
  a grey fill and OCR'd on 1 of 8 pages; the SubTotal row read on 8 of
  8. So the amount is the rounded subtotal, cross-checked against the
  printed box wherever it survives — where it did, it agreed.

- The clave is 2 digits, a letter and 3 digits (12-T -012), not the
  cadastral shape, and the letter is kept as printed: toDigits maps D to
  0, which turns a real 14-D -014 into 140014. It is printed twice,
  which rescued a page whose heading was struck through by the office's
  own highlighter — the failure mode behind both missing claves.

Deriving the deadline from the bimester is deliberately not attempted:
it is the 17th of the month after the bimester closes on a current bill,
but four of these eight are late (a $1,000 Multa) and print a
recalculated date, so a derived date would be wrong on exactly the pages
a human most wants to see.

Re-ran the earlier corpora (25 pages: predial Tijuana/Rosarito/Ensenada,
CFE, CESPT, Telnor) through detection to confirm the new rules steal
nothing — all 25 still read as their original provider, including the
five Tijuana predial pages that share the RFC.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

apps/api/src/statements/parsers/statement-parser.spec.ts

@@ -2,6 +2,7 @@ import type { OcrPage } from "../ocr/ocr.provider";
 import {
   detectProvider,
   normalizeCadastralKey,
+  normalizeZofematKey,
   parseStatement,
 } from "./statement-parser";
 
@@ -192,3 +193,78 @@ describe("parseGas", () => {
     expect(parseStatement(gas("0900003463")).crossChecked).toBeNull();
   });
 });
+
+describe("parseZonaFederal", () => {
+  /**
+   * The Tijuana zona federal receipt, trimmed to the rows the parser reads.
+   * Verbatim from page 7 of the August 2026 batch, including the two ways the
+   * heading OCR'd: the clave line is struck through by the office's own
+   * highlighter, which is what cost two of eight pages their concession clave.
+   */
+  const zf = (clave: string, body = "") =>
+    page(
+      "ESIZ <pYl Av. Independencia y Esq. Paseo del CentenaxiaiiArlhnto de Tijuana, B.C.\n" +
+        "Teléfono: 9737000 R.F.C. ATB-541201-BK2 0070000146 12:54 PM\n" +
+        "Zona Federal Marítimo Terrestre\n" +
+        `${clave} Nombre: DENNIS JOHN SEIN Concesión:\n` +
+        "Periodo Construcción Tasa  Ornato Tasa Impuesto Actualiza. Recargo Multa Importe\n" +
+        "2026-2 / 2026-2 316.40 35.00 0.00 12.11 1,845.66 0.00 27.13 1,000.00 2,872.79\n" +
+        "SubTotal 1,845.66 0.00 27.13 1,000.00 2,872.79\n" +
+        "Concepto: Derechos de ocupación de Zona Federal Marítimo Terrestre\n" +
+        body,
+    );
+
+  it("is not claimed by the predial parser that shares its RFC and header", () => {
+    // Tijuana bills predial and zona federal from the same treasury, so
+    // "Ayuntamiento de Tijuana" and ATB-541201 identify neither on their own.
+    expect(detectProvider("R.F.C. ATB-541201-BK2\nZona Federal Marítimo Terrestre")).toBe(
+      "ZONA FEDERAL TIJUANA",
+    );
+    expect(parseStatement(zf("Clave: 14-D -014")).serviceKind).toBe("FEDERAL_ZONE");
+  });
+
+  it("still recognises the layout when the heading itself did not survive OCR", () => {
+    // Real: page 1 came back as "Zona Ledera) Maritimo Terrestre".
+    expect(
+      detectProvider("Zona Ledera) Maritimo Terrestre\nClave EJ -012% Nombre: STEFAN"),
+    ).toBe("ZONA FEDERAL TIJUANA");
+  });
+
+  it("reads the clave through the loose spacing the receipt prints", () => {
+    expect(parseStatement(zf("Clave: 14-D -014")).accountRef).toBe("14D014");
+    expect(parseStatement(zf("Clave: 14-A-119")).accountRef).toBe("14A119");
+  });
+
+  it("keeps the letter instead of digitising it", () => {
+    // toDigits maps D to 0 and B to 8; a real 14-D -014 must not become 140014.
+    expect(normalizeZofematKey("14-D -014")).toBe("14D014");
+    expect(normalizeZofematKey("12-B -013")).toBe("12B013");
+  });
+
+  it("takes the payable amount from the SubTotal row, rounded to whole pesos", () => {
+    // The municipality rounds and prints the difference as "Ajuste Ley Hacienda
+    // Mpal"; 2,872.79 is charged as $2,873.00.
+    expect(parseStatement(zf("Clave: 14-D -014")).amount).toBe(2873);
+  });
+
+  it("prefers the printed total and cross-checks it against the subtotal", () => {
+    const p = parseStatement(zf("Clave: 14-D -014", "Total a pagar $2,873.00"));
+    expect(p.amount).toBe(2873);
+    expect(p.crossChecked).toBe(true);
+  });
+
+  it("sends a printed total that contradicts the subtotal to review", () => {
+    const p = parseStatement(zf("Clave: 14-D -014", "Total a pagar $2,973.00"));
+    expect(p.crossChecked).toBe(false);
+  });
+
+  it("translates the printed bimester into the ledger's own vocabulary", () => {
+    expect(parseStatement(zf("Clave: 14-D -014")).period).toBe("MAR/APR");
+  });
+
+  it("leaves the clave blank rather than guessing when the marker ate it", () => {
+    const p = parseStatement(zf("Clave EJ -012%"));
+    expect(p.accountRef).toBeNull();
+    expect(p.notes.join(" ")).toContain("clave");
+  });
+});

apps/api/src/statements/parsers/statement-parser.ts

@@ -9,8 +9,8 @@ import type { OcrPage, OcrWord } from "../ocr/ocr.provider";
 export interface ParsedStatement {
   /**
    * "CFE" | "CESPT" | "TELNOR" | "GAS TIJUANA" | "PREDIAL TIJUANA" |
-   * "PREDIAL ROSARITO" | "PREDIAL ENSENADA", or null when no parser claimed
-   * the page.
+   * "PREDIAL ROSARITO" | "PREDIAL ENSENADA" | "ZONA FEDERAL TIJUANA", or null
+   * when no parser claimed the page.
    */
   provider: string | null;
   serviceKind: ServiceKind | null;
@@ -190,6 +190,15 @@ const BRAND: [string, RegExp][] = [
   ["CESPT", /CESPT|COMISI[OÓ]N ESTATAL DE SERVICIOS/i],
   ["TELNOR", /TELNOR|TELEFONOS DEL NOROESTE/i],
   ["GAS TIJUANA", /COMPA[ÑN][IÍ]?A\s*DE\s*GAS\s*DE\s*TIJUANA|bajagas/i],
+  // Ahead of the predial rules on purpose. Tijuana's zona federal receipt is
+  // issued by the same treasury and carries the same header — "Ayuntamiento de
+  // Tijuana", the same address, the same `ATB-541201` RFC — so every predial
+  // discriminator matches it too, and whichever rule is asked first wins the
+  // page. What only the zona federal layout says is "Marítimo Terrestre", which
+  // survived OCR on all eight sample pages even where the heading above it came
+  // back as "Zona Ledera) Maritimo Terrestre" and the printed concession clave
+  // was lost under a highlighter mark.
+  ["ZONA FEDERAL TIJUANA", /ZOFEMAT|Mar[ií]timo\s*Terrestre|ocupaci[oó]n\s*de\s*Zona\s*Federal/i],
   // The municipal RFCs are the single most reliable discriminator on a predial
   // receipt: they are printed in a clean monospaced run on every layout, they
   // never change, and they say which of the three city treasuries issued the
@@ -209,6 +218,11 @@ const BRAND: [string, RegExp][] = [
  * utility ones, so it is the safe first question to ask.
  */
 const LAYOUT: [string, RegExp][] = [
+  // Same reasoning as the brand pass, one rule earlier: the concept line
+  // "Derechos de ocupación de Zona Federal Marítimo Terrestre" is printed on
+  // the stub of every zona federal page and on no other layout, and it read
+  // cleanly on 8 of 8 samples — including the two whose heading did not.
+  ["ZONA FEDERAL TIJUANA", /Derechos\s*de\s*ocupaci[oó]n/i],
   ["PREDIAL TIJUANA", /IMPUESTO\s*PREDIAL[\s\S]*?(?:CERTIFICACION\s*DE\s*CAJA|PASEO\s*DEL\s*CENTENARIO|PAGA\s*TU\s*PREDIAL)/i],
   ["PREDIAL ENSENADA", /(?:IMPUESTO\s*PREDIAL[\s\S]*?TRANSPENINSULAR)|(?:IMPRESION\s*MAQUINA\s*REGISTRADORA)/i],
   ["PREDIAL ROSARITO", /IMPUESTO\s*PREDIAL/i],
@@ -701,6 +715,132 @@ function parsePredialEnsenada(page: OcrPage): ParsedStatement {
   };
 }
 
+// --- ZONA FEDERAL (ZOFEMAT, Tijuana) ----------------------------------------
+
+/**
+ * Normalise the concession clave the zona federal receipt is keyed by.
+ *
+ * It is printed grouped and loosely spaced — `12-T -012`, `14-A-119`,
+ * `14-K -031` — and is a different shape from the cadastral key entirely: two
+ * digits, one letter, three digits. The letter is kept as printed rather than
+ * digitised, for the same reason `normalizeCadastralKey` keeps its third
+ * character: `toDigits` maps `B` to `8` and `D` to `0`, and a real `14-D -014`
+ * run through it becomes `140014`, which is not a clave at all.
+ *
+ * Stored without separators, because nothing on file holds this value yet (see
+ * `parseZonaFederal`) so the canonical form is ours to pick, and a bare run
+ * cannot be broken by the hyphen the scan renders as a dash, a minus or
+ * nothing.
+ */
+export function normalizeZofematKey(raw: string): string | null {
+  const m = raw.match(/^([0-9OIlSBD]{2})\s*-\s*([A-Za-z])\s*-?\s*([0-9OIlSBD]{3})$/);
+  if (!m) return null;
+  const zone = toDigits(m[1]);
+  const lot = toDigits(m[3]);
+  if (zone.length !== 2 || lot.length !== 3) return null;
+  return `${zone}${m[2].toUpperCase()}${lot}`;
+}
+
+/**
+ * The bimester the receipt prints as `2026-2 / 2026-2`, rendered in the
+ * vocabulary the ledger already speaks.
+ *
+ * All 258 legacy FEDERAL ZONE transactions carry a period of `JAN/FEB`,
+ * `MAR/APR`, `MAY/JUN` or `NOV/DEC`, and their payment dates confirm the
+ * ordering — JAN/FEB was paid in March, MAR/APR in May, MAY/JUN in July,
+ * NOV/DEC in January, i.e. always the month after the bimester closes. The
+ * receipts agree: the two `2026-3` samples fall due 17/07/2026 with no
+ * surcharge, which is bimester three, May and June. Writing `2026-3` instead
+ * would leave the OCR-posted rows unsearchable alongside every hand-keyed one.
+ */
+const BIMESTERS = ["JAN/FEB", "MAR/APR", "MAY/JUN", "JUL/AUG", "SEP/OCT", "NOV/DEC"];
+
+/**
+ * Tijuana's "Zona Federal Marítimo Terrestre" — the federal maritime-zone
+ * occupancy fee, billed by the municipality for beachfront lots.
+ *
+ * Nothing on file identifies these. `PropertyService.accountNumber` for
+ * FEDERAL_ZONE holds DATMEX.zfed, which is not a reference at all but an
+ * amount: its 77 values include `246.06`, `2369.09`, `22653.94` and a negative
+ * `-1679`, and the concession claves these receipts are keyed by appear nowhere
+ * in the database. So the clave goes to `meterNumber` (see `scopedRefField`),
+ * every page starts cold, and the first confirm teaches the match — the same
+ * arrangement Tijuana predial needed, for the same reason.
+ *
+ * The amount is taken from the SubTotal row rather than the "Total a pagar"
+ * box, which is printed on a grey fill and OCR'd on only 1 of 8 sample pages
+ * while the SubTotal row read on 8 of 8. The two differ by design: the
+ * municipality rounds to whole pesos and prints the difference on its own
+ * "Ajuste Ley Hacienda Mpal" line — `-$0.05` against a 591.05 subtotal, `$0.21`
+ * against 2,872.79 — so the payable figure is the rounded subtotal, and where
+ * the printed box did read, it agreed.
+ */
+function parseZonaFederal(page: OcrPage): ParsedStatement {
+  const text = page.text;
+  const notes: string[] = [];
+
+  // Printed twice, once on the receipt and once on the stub below it, which is
+  // a free second reading: on one sample the heading was struck through by the
+  // office's own highlighter and only the stub survived.
+  const claves = [
+    ...new Set(
+      allMatches(text, /Clave\s*[:;.]?\s*([0-9OIlSBD]{2}\s*-\s*[A-Za-z]\s*-?\s*[0-9OIlSBD]{3})/i)
+        .map(normalizeZofematKey)
+        .filter((k): k is string => k != null),
+    ),
+  ];
+
+  const accountRef = claves[0] ?? null;
+  let crossChecked: boolean | null = null;
+
+  const subtotalRow = text.split("\n").find((l) => /SubTotal/i.test(l)) ?? "";
+  const figures = allMatches(subtotalRow, /(\d[\d,]*\.\d{2})/);
+  // Impuesto, Actualización, Recargo, Multa, Importe — the payable one is last.
+  const importe = figures.length ? money(figures[figures.length - 1]) : null;
+  const rounded = importe != null ? Math.round(importe) : null;
+  const printed = money(
+    firstMatch(text, [/Total\s*a\s*pagar\s*[:;.]?\s*\$?\s*([\d,]+\.\d{2})/i]),
+  );
+
+  if (printed != null && rounded != null) {
+    crossChecked = Math.abs(printed - rounded) < 0.5;
+    if (!crossChecked) {
+      notes.push(
+        `el total impreso (${printed}) no coincide con el subtotal redondeado (${rounded})`,
+      );
+    }
+  } else if (rounded != null) {
+    notes.push("importe tomado del subtotal, redondeado al peso");
+  } else if (printed == null) {
+    notes.push("no se pudo leer el importe con certeza; capturarlo a mano");
+  }
+
+  // A clave read two different ways means one of the two readings is wrong and
+  // there is no third to break the tie, so the page goes to a human even if the
+  // money cross-checked.
+  if (claves.length > 1) {
+    crossChecked = false;
+    notes.push(`la clave se leyó de ${claves.length} formas distintas (${claves.join(", ")})`);
+  }
+  if (!accountRef) notes.push("no se pudo leer la clave de la concesión");
+
+  const bimester = text.match(/\b(20\d{2})\s*-\s*([1-6])\s*\/\s*20\d{2}\s*-\s*[1-6]/);
+
+  return {
+    provider: "ZONA FEDERAL TIJUANA",
+    serviceKind: "FEDERAL_ZONE",
+    accountRef,
+    cadastralKey: null,
+    amount: printed ?? rounded,
+    dueDate: parseDate(
+      firstMatch(text, [/Vencimiento\s*[:;.]?\s*(\d{2}\/\d{2}\/\d{4})/i]),
+    ),
+    period: bimester ? BIMESTERS[+bimester[2] - 1] : null,
+    crossChecked,
+    notes,
+  };
+}
+
 const PARSERS: Record<string, (page: OcrPage) => ParsedStatement> = {
   CFE: parseCfe,
   CESPT: parseCespt,
@@ -709,6 +849,7 @@ const PARSERS: Record<string, (page: OcrPage) => ParsedStatement> = {
   "PREDIAL TIJUANA": parsePredialTijuana,
   "PREDIAL ROSARITO": parsePredialRosarito,
   "PREDIAL ENSENADA": parsePredialEnsenada,
+  "ZONA FEDERAL TIJUANA": parseZonaFederal,
 };
 
 const EMPTY: ParsedStatement = {

apps/api/src/statements/statement-matcher.service.ts

@@ -40,15 +40,22 @@ export interface MatchResult {
  * When they disagree, a reference gets learned into a column nothing searches,
  * and the same page returns to the review queue every month forever.
  *
- * `meterNumber` is doing double duty for the two kinds whose printed reference
- * DATMEX never held in `accountNumber`:
- *   - GAS, where the number lived in free-text notes, and
+ * `meterNumber` is doing double duty for the three kinds whose printed
+ * reference DATMEX never held in `accountNumber`:
+ *   - GAS, where the number lived in free-text notes,
  *   - PROPERTY_TAX, where `accountNumber` holds DATMEX.predial — a 3-4 digit
  *     office file number that is neither unique nor printed on any statement.
  *     The Tijuana municipal receipt prints an 8-digit account and no clave
  *     catastral at all, so it needs a column of its own; overwriting the legacy
  *     predial numbers to make room would destroy the only link back to the
- *     original records.
+ *     original records, and
+ *   - FEDERAL_ZONE, where `accountNumber` holds DATMEX.zfed, which is not a
+ *     reference of any kind but a peso amount: 3 of its 77 values carry cents
+ *     (`246.06`, `2369.09`, `22653.94`) and one is negative. Searching it for
+ *     the concession clave the receipt prints would never hit, and — worse —
+ *     because every row already has a value, the `[field]: null` guards in
+ *     `learnAccountRefs` and the blank-service fill would never fire either, so
+ *     the same page would return to the review queue every bimester forever.
  */
 export function scopedRefField(
   kind: ServiceKind,
@@ -57,11 +64,11 @@ export function scopedRefField(
     case "ELECTRIC": // CFE "NO. DE SERVICIO" -> DATMEX.rpu
     case "WATER": // CESPT "Cuenta" / "No. DE CUENTA" -> DATMEX.agua
     case "TELEPHONE": // Telnor "Teléfono" (LADA stripped) -> DATMEX.telefono
-    case "FEDERAL_ZONE":
     case "CABLE":
       return "accountNumber";
     case "GAS": // bajagas "Cuenta" -> recovered from notes into meterNumber
     case "PROPERTY_TAX": // Tijuana's 8-digit municipal account
+    case "FEDERAL_ZONE": // ZOFEMAT concession clave, e.g. `12T012`
       return "meterNumber";
     default:
       return null;

apps/web/src/components/StatementIntake.tsx

@@ -34,9 +34,10 @@ const SUPPORTED: ServiceKind[] = [
   "TELEPHONE",
   "GAS",
   "PROPERTY_TAX",
+  "FEDERAL_ZONE",
 ];
 /** Uploadable, but every page will land in review until a parser learns it. */
-const OTHER_KINDS: ServiceKind[] = ["FEDERAL_ZONE", "CABLE"];
+const OTHER_KINDS: ServiceKind[] = ["CABLE"];
 
 const STATUS_LABEL: Record<StatementBatchStatus, string> = {
   UPLOADED: "Recibido",