/**
 * Ironparse — deterministic COBOL copybook verification engine.
 *
 * This is the *verification layer* of the Ironparse engine: the COBOL parser
 * and the five deterministic quality gates that produce the field-for-field
 * parity guarantee. It runs with zero LLM involvement — same input, same
 * output, every run.
 *
 * In a pilot deployment, a local model running inside the customer's VPC
 * drafts the candidate TypeScript/Zod schema; these same gates then reject any
 * draft whose field count or structural shape does not match the source AST.
 * In this public trace the emitter below produces the reference schema
 * directly from the AST, and the gates verify it. Nothing is mocked: every
 * field count, every diff, and every hash is computed live from the copybook.
 *
 * No dependencies. Runs identically under Node and in a Cloudflare Worker.
 */

// ---------------------------------------------------------------------------
// AST types
// ---------------------------------------------------------------------------

export interface Occurs {
  min: number;
  max: number;
  dependingOn: string | null; // OCCURS DEPENDING ON <field>, else null (fixed)
}

export interface Node {
  level: number;
  name: string;
  pic: string | null; // raw PIC clause, e.g. "S9(9)V99" — null for group items
  comp3: boolean;
  redefines: string | null;
  occurs: Occurs | null;
  children: Node[];
  // populated during analysis:
  path: string; // dotted JSON path of the field in the emitted schema
  storage: string; // human storage description
  tsType: string; // emitted leaf TS/Zod primitive (leaf nodes only)
}

export interface Ast {
  records: Node[]; // top-level 01 records
  leaves: Node[]; // every elementary (PIC-bearing) field, in source order
  redefinesCount: number;
  odoCount: number; // OCCURS DEPENDING ON groups
  fixedOccursCount: number;
}

// ---------------------------------------------------------------------------
// Parser — deterministic, line-oriented. No LLM, no heuristics-on-vibes.
// ---------------------------------------------------------------------------

const LEVEL_RE = /^s*(d{2})s+([A-Z0-9][A-Z0-9-]*)s*(.*?)s*.?s*$/;

function parsePic(rest: string): { pic: string | null; comp3: boolean } {
  const comp3 = /COMP-3/.test(rest) || /COMPUTATIONAL-3/.test(rest);
  const m = rest.match(/PIC(?:TURE)?s+(?:ISs+)?([0-9A-Z()V$S.,+-/]+)/);
  return { pic: m ? m[1] : null, comp3 };
}

function parseOccurs(rest: string): Occurs | null {
  // OCCURS n TIMES                      -> fixed n
  // OCCURS 0 TO m TIMES DEPENDING ON Y  -> variable, depending on Y
  const dep = rest.match(/OCCURSs+(d+)s+TOs+(d+)s+TIMESs+DEPENDINGs+ONs+([A-Z0-9][A-Z0-9-]*)/);
  if (dep) return { min: parseInt(dep[1], 10), max: parseInt(dep[2], 10), dependingOn: dep[3] };
  const fixed = rest.match(/OCCURSs+(d+)s+TIMES/);
  if (fixed) return { min: parseInt(fixed[1], 10), max: parseInt(fixed[1], 10), dependingOn: null };
  return null;
}

function parseRedefines(rest: string): string | null {
  const m = rest.match(/REDEFINESs+([A-Z0-9][A-Z0-9-]*)/);
  return m ? m[1] : null;
}

/** Parse a COBOL copybook into a tree of records keyed by level number. */
export function parseCopybook(text: string): Ast {
  // COBOL copybooks may use fixed-format columns; join continuation logic is
  // out of scope for these ACORD samples (each clause is on one logical line).
  const lines = text
    .split(/
?
/)
    .map((l) => l.replace(/*>.*/, "")) // strip inline comments
    .filter((l) => l.trim().length > 0 && !/^s**/.test(l)); // drop comment lines

  const records: Node[] = [];
  const stack: Node[] = [];

  for (const line of lines) {
    const m = line.match(LEVEL_RE);
    if (!m) continue;
    const level = parseInt(m[1], 10);
    const name = m[2];
    const rest = m[3] || "";

    const { pic, comp3 } = parsePic(rest);
    const node: Node = {
      level,
      name,
      pic,
      comp3,
      redefines: parseRedefines(rest),
      occurs: parseOccurs(rest),
      children: [],
      path: "",
      storage: "",
      tsType: "",
    };

    // Pop the stack until we find a parent with a strictly lower level.
    while (stack.length > 0 && stack[stack.length - 1].level >= level) stack.pop();

    if (stack.length === 0) {
      records.push(node);
    } else {
      stack[stack.length - 1].children.push(node);
    }
    stack.push(node);
  }

  // Collect elementary fields (PIC-bearing leaves) in source order + counts.
  const leaves: Node[] = [];
  let redefinesCount = 0;
  let odoCount = 0;
  let fixedOccursCount = 0;

  const walk = (node: Node, prefix: string) => {
    const seg = toCamel(node.name);
    const isArray = node.occurs !== null;
    const path = prefix ? `${prefix}.${seg}` : seg;
    if (node.redefines) redefinesCount++;
    if (node.occurs?.dependingOn) odoCount++;
    if (node.occurs && !node.occurs.dependingOn) fixedOccursCount++;

    if (node.pic) {
      const { storage, tsType } = mapPic(node.pic, node.comp3);
      node.path = isArray ? `${path}[]` : path;
      node.storage = storage;
      node.tsType = tsType;
      leaves.push(node);
    } else {
      const childPrefix = isArray ? `${path}[]` : path;
      for (const c of node.children) walk(c, childPrefix);
    }
  };
  for (const r of records) walk(r, "");

  return { records, leaves, redefinesCount, odoCount, fixedOccursCount };
}

// ---------------------------------------------------------------------------
// Type mapping — COBOL PIC -> storage description + Zod primitive
// ---------------------------------------------------------------------------

function picLength(pic: string): number {
  // Expand things like X(35) or 9(9)V99 into a character count (display).
  let total = 0;
  const re = /([X9AN])((d+))|([X9AN])/g;
  let mm: RegExpExecArray | null;
  while ((mm = re.exec(pic)) !== null) {
    total += mm[2] ? parseInt(mm[2], 10) : 1;
  }
  return total;
}

export function mapPic(pic: string, comp3: boolean): { storage: string; tsType: string } {
  const signed = /^S/.test(pic);
  const hasDecimal = /V/.test(pic);
  const isNumeric = /9/.test(pic) && !/X/.test(pic) && !/A/.test(pic);
  const len = picLength(pic);

  if (comp3) {
    // Packed decimal — a real number with scale; preserve as numeric.
    return {
      storage: `packed decimal (COMP-3${signed ? ", signed" : ""}${hasDecimal ? ", scaled" : ""})`,
      tsType: hasDecimal ? "z.number()" : "z.number().int()",
    };
  }
  if (isNumeric) {
    // Display numeric: preserve exact digits (leading zeros are significant on
    // a mainframe). Fidelity-first: model as a constrained string.
    return {
      storage: `display numeric${signed ? ", signed" : ""}${hasDecimal ? ", implied decimal" : ""} (${len} digits)`,
      tsType: hasDecimal || signed ? `z.string()` : `z.string().regex(/^\\d{1,${len}}$/)`,
    };
  }
  // Alphanumeric / alphabetic.
  return { storage: `alphanumeric (${len} chars)`, tsType: `z.string().max(${len})` };
}

// ---------------------------------------------------------------------------
// Naming
// ---------------------------------------------------------------------------

export function toCamel(cobolName: string): string {
  const parts = cobolName.toLowerCase().split("-");
  return parts
    .map((p, i) => (i === 0 ? p : p.charAt(0).toUpperCase() + p.slice(1)))
    .join("");
}

// ---------------------------------------------------------------------------
// Emitter — AST -> TypeScript/Zod schema text (deterministic reference output)
// ---------------------------------------------------------------------------

export interface EmitResult {
  code: string;
  primitiveCount: number; // leaf z.<primitive>() emitted — independent of AST leaf count
  unionCount: number; // REDEFINES overlays modelled as z.discriminatedUnion / z.union
  discriminatedUnionCount: number;
  arrayCount: number; // OCCURS modelled as z.array
}

function emitNode(node: Node, indent: string): string {
  // Group item -> z.object({...}); elementary -> its primitive.
  let inner: string;
  if (node.pic) {
    inner = node.tsType;
  } else {
    const pad = indent + "  ";
    const fields = node.children
      .filter((c) => !c.redefines) // redefining siblings handled by union at group level
      .map((c) => `${pad}${toCamel(c.name)}: ${emitNode(c, pad)},`)
      .join("
");
    inner = `z.object({
${fields}
${indent}})`;
  }
  if (node.occurs) {
    const note = node.occurs.dependingOn
      ? ` /* OCCURS 0..${node.occurs.max} DEPENDING ON ${node.occurs.dependingOn} */`
      : ` /* OCCURS ${node.occurs.max} */`;
    inner = `z.array(${inner}).max(${node.occurs.max})${note}`;
  }
  return inner;
}

export function emitZod(ast: Ast): EmitResult {
  const lines: string[] = [];
  lines.push(`import { z } from "zod";`);
  lines.push("");

  // Index records by name to resolve REDEFINES overlays.
  const byName = new Map<string, Node>();
  for (const r of ast.records) byName.set(r.name, r);

  let unionCount = 0;
  let discriminatedUnionCount = 0;

  // A record that is REDEFINES'd by a later record forms a memory-overlay
  // union: the same bytes can be read as either layout.
  const redefinedBy = new Map<string, Node[]>();
  for (const r of ast.records) {
    if (r.redefines) {
      const arr = redefinedBy.get(r.redefines) || [];
      arr.push(r);
      redefinedBy.set(r.redefines, arr);
    }
  }

  const emittedRoots: string[] = [];
  const rootFields: string[] = [];

  for (const r of ast.records) {
    if (r.redefines) continue; // emitted as part of the base record's union
    const overlays = redefinedBy.get(r.name);
    const field = toCamel(r.name);
    if (overlays && overlays.length > 0) {
      // Memory overlay -> union of all layouts that share these bytes.
      const branches = [r, ...overlays];
      const discriminator = sharedDiscriminator(branches);
      const branchCode = branches.map((b) => emitNode(b, "    ")).join(",
    ");
      let unionExpr: string;
      if (discriminator) {
        unionExpr = `z.discriminatedUnion("${discriminator}", [
    ${branchCode}
  ])`;
        discriminatedUnionCount++;
      } else {
        unionExpr = `z.union([
    ${branchCode}
  ])`;
      }
      unionCount++;
      rootFields.push(
        `  // REDEFINES overlay: ${[r.name, ...overlays.map((o) => o.name)].join(" / ")}
  ${field}: ${unionExpr},`,
      );
    } else {
      rootFields.push(`  ${field}: ${emitNode(r, "  ")},`);
    }
  }

  lines.push(`export const RecordSchema = z.object({`);
  lines.push(rootFields.join("
"));
  lines.push(`});`);
  lines.push("");
  lines.push(`export type Record = z.infer<typeof RecordSchema>;`);

  const code = lines.join("
");

  // Independent structural counts parsed back out of the emitted text — this
  // is what Gate 03 / Gate 04 compare against the AST, not a bookkeeping echo.
  const primitiveCount = (code.match(/z.(string|number|boolean)(/g) || []).length;
  const arrayCount = (code.match(/z.array(/g) || []).length;

  return { code, primitiveCount, unionCount, discriminatedUnionCount, arrayCount };
}

/** A discriminator exists if every union branch leads with the same-shaped
 *  1-char indicator field (classic ACORD record-type byte). */
function sharedDiscriminator(branches: Node[]): string | null {
  const firsts = branches.map((b) => b.children[0]).filter(Boolean);
  if (firsts.length !== branches.length) return null;
  const allOneChar = firsts.every((f) => f.pic && /^X(1)$|^X$/.test(f.pic));
  if (!allOneChar) return null;
  // Use each branch's own first field name; discriminatedUnion needs a single
  // shared key, so only valid when the names match. ACORD overlays differ, so
  // we fall back to z.union in that case — modelled honestly, not forced.
  const name = toCamel(firsts[0].name);
  return firsts.every((f) => toCamel(f.name) === name) ? name : null;
}

// ---------------------------------------------------------------------------
// The five deterministic quality gates
// ---------------------------------------------------------------------------

export interface Gate {
  id: string;
  name: string;
  detail: string;
  pass: boolean;
  measure: string; // the actual number that decided it
}

export function runGates(ast: Ast, emit: EmitResult): Gate[] {
  const gates: Gate[] = [];

  // 01 PARSER — Tree extraction produced field-bearing records.
  gates.push({
    id: "01",
    name: "PARSER",
    detail: "Deterministic AST extraction yields a non-empty, field-bearing record set. No LLM involved.",
    pass: ast.records.length > 0 && ast.leaves.length > 0,
    measure: `${ast.records.length} records · ${ast.leaves.length} elementary fields`,
  });

  // 02 SCHEMA_SANITY — Emitted schema is valid, fenced-free Zod.
  const balanced = isBalanced(emit.code);
  const noFence = !/```/.test(emit.code);
  const hasObject = /z.object(/.test(emit.code);
  gates.push({
    id: "02",
    name: "SCHEMA_SANITY",
    detail: "Candidate schema parses as valid Zod — balanced delimiters, no markdown fences, no prose.",
    pass: balanced && noFence && hasObject,
    measure: balanced && noFence && hasObject ? "valid Zod module" : "malformed output",
  });

  // 03 FIELD_PARITY — strict count equality. Math, not judgment.
  gates.push({
    id: "03",
    name: "FIELD_PARITY",
    detail: "len(COBOL elementary fields) === len(emitted schema leaves). Off by one and the build fails.",
    pass: ast.leaves.length === emit.primitiveCount,
    measure: `COBOL ${ast.leaves.length} ⇄ schema ${emit.primitiveCount}`,
  });

  // 04 DARK_CORNER — every overlay/array survives translation.
  const overlayGroups = redefinesUnionGroups(ast);
  const expectedArrays = ast.odoCount + ast.fixedOccursCount;
  const redefinesOk = emit.unionCount >= overlayGroups;
  const odoOk = emit.arrayCount >= expectedArrays;
  gates.push({
    id: "04",
    name: "DARK_CORNER",
    detail:
      "Every REDEFINES overlay compiles to a union (discriminated where a record-type byte exists); every OCCURS to a dynamic array.",
    pass: redefinesOk && odoOk,
    measure: `${emit.unionCount}/${overlayGroups} overlays · ${emit.arrayCount}/${expectedArrays} arrays`,
  });

  // 05 MOCK_STRUCTURE — round-trip a mock document through the shape.
  const mock = buildMock(ast);
  const roundTrip = validateMock(ast, mock);
  gates.push({
    id: "05",
    name: "MOCK_STRUCTURE",
    detail: "A mock document generated from the schema re-validates against it — the schema is internally consistent.",
    pass: roundTrip.ok,
    measure: roundTrip.ok ? `${roundTrip.checked} nodes round-tripped` : roundTrip.reason,
  });

  return gates;
}

// Count of base records that participate in a REDEFINES overlay group.
function redefinesUnionGroups(ast: Ast): number {
  const bases = new Set<string>();
  for (const r of ast.records) if (r.redefines) bases.add(r.redefines);
  return bases.size;
}

// ---------------------------------------------------------------------------
// Mock generation + round-trip validation (Gate 05)
// ---------------------------------------------------------------------------

function mockLeaf(node: Node): unknown {
  if (node.comp3) return node.pic && /V/.test(node.pic) ? 0.0 : 0;
  if (node.pic && /9/.test(node.pic) && !/X|A/.test(node.pic)) return "0";
  return "";
}

function buildMock(ast: Ast): any {
  const byName = new Map<string, Node>();
  for (const r of ast.records) byName.set(r.name, r);
  const redefinedBy = new Map<string, Node[]>();
  for (const r of ast.records) if (r.redefines) {
    const a = redefinedBy.get(r.redefines) || [];
    a.push(r);
    redefinedBy.set(r.redefines, a);
  }

  const buildGroup = (node: Node): any => {
    const obj: any = {};
    for (const c of node.children) {
      if (c.redefines) continue;
      obj[toCamel(c.name)] = buildNode(c);
    }
    return obj;
  };
  const buildNode = (node: Node): any => {
    let val: any;
    if (node.pic) val = mockLeaf(node);
    else val = buildGroup(node);
    if (node.occurs) return [val]; // one representative element
    return val;
  };

  const root: any = {};
  for (const r of ast.records) {
    if (r.redefines) continue;
    root[toCamel(r.name)] = buildNode(r); // base layout chosen for overlays
  }
  return root;
}

function validateMock(ast: Ast, mock: any): { ok: boolean; checked: number; reason: string } {
  let checked = 0;
  const check = (node: Node, val: any): boolean => {
    checked++;
    if (node.occurs) {
      if (!Array.isArray(val)) return false;
      return val.every((v) => checkInner(node, v));
    }
    return checkInner(node, val);
  };
  const checkInner = (node: Node, val: any): boolean => {
    if (node.pic) {
      if (node.comp3) return typeof val === "number";
      return typeof val === "string";
    }
    if (typeof val !== "object" || val === null) return false;
    return node.children.filter((c) => !c.redefines).every((c) => check(c, val[toCamel(c.name)]));
  };
  for (const r of ast.records) {
    if (r.redefines) continue;
    if (!check(r, mock[toCamel(r.name)])) return { ok: false, checked, reason: `node ${r.name} failed` };
  }
  return { ok: true, checked, reason: "" };
}

// ---------------------------------------------------------------------------
// Small utilities
// ---------------------------------------------------------------------------

function isBalanced(code: string): boolean {
  const pairs: Record<string, string> = { ")": "(", "]": "[", "}": "{" };
  const stack: string[] = [];
  for (const ch of code) {
    if (ch === "(" || ch === "[" || ch === "{") stack.push(ch);
    else if (ch === ")" || ch === "]" || ch === "}") {
      if (stack.pop() !== pairs[ch]) return false;
    }
  }
  return stack.length === 0;
}

export async function sha256(text: string): Promise<string> {
  const enc = new TextEncoder().encode(text);
  const digest = await crypto.subtle.digest("SHA-256", enc);
  return [...new Uint8Array(digest)].map((b) => b.toString(16).padStart(2, "0")).join("");
}

// ---------------------------------------------------------------------------
// Receipt — the auditable artifact a pilot delivers per copybook
// ---------------------------------------------------------------------------

export interface FieldMapping {
  cobol: string;
  level: number;
  pic: string;
  storage: string;
  tsPath: string;
  tsType: string;
}

export interface Receipt {
  copybook: string;
  inputSha256: string;
  outputSha256: string;
  recordCount: number;
  fieldCount: number;
  redefinesCount: number;
  odoCount: number;
  fixedOccursCount: number;
  gates: Gate[];
  verdict: "PASS" | "FAIL";
  schema: string;
  mappings: FieldMapping[];
}

export async function buildReceipt(name: string, copybook: string): Promise<Receipt> {
  const ast = parseCopybook(copybook);
  const emit = emitZod(ast);
  const gates = runGates(ast, emit);
  const inputSha256 = await sha256(copybook.replace(/
/g, "
").trimEnd());
  const outputSha256 = await sha256(emit.code);
  const mappings: FieldMapping[] = ast.leaves.map((n) => ({
    cobol: n.name,
    level: n.level,
    pic: (n.comp3 ? `${n.pic} COMP-3` : n.pic) || "",
    storage: n.storage,
    tsPath: n.path,
    tsType: n.tsType,
  }));
  return {
    copybook: name,
    inputSha256,
    outputSha256,
    recordCount: ast.records.length,
    fieldCount: ast.leaves.length,
    redefinesCount: ast.redefinesCount,
    odoCount: ast.odoCount,
    fixedOccursCount: ast.fixedOccursCount,
    gates,
    verdict: gates.every((g) => g.pass) ? "PASS" : "FAIL",
    schema: emit.code,
    mappings,
  };
}