@portll/cobolwork 0.5.0 → 0.7.0

package/lib/hlasm/locate.mjs

@@ -0,0 +1,333 @@
+// SPDX-License-Identifier: AGPL-3.0-or-later
+// Where the assembler would put each open-code statement and symbol: location counters per control
+// and dummy section, symbol values, length attributes and types, literal pools, and the external
+// symbols a module defines and needs. Offsets are relative to their section. Nothing after a
+// statement whose length cannot be known (a macro call, a COPY member, a conditional branch, a
+// statement the reader refused) is placed in its section: its location is null and says after which
+// line, since a guessed location would be graded as the assembler's.
+// https://www.ibm.com/docs/en/hla-and-tf/1.6.0?topic=reference-location-counter
+import { readHlasmStatements, parseHlasmStatement } from './read.mjs';
+import { NAME } from './operands.mjs';
+
+const ALIGN = { H: 2, Y: 2, S: 2, F: 4, E: 4, A: 4, V: 4, Q: 4, J: 4, R: 4, D: 8, L: 8 };
+const ALIGN_EXT = { FD: 8, AD: 8, VD: 8, QD: 8, JD: 8, RD: 8, LQ: 16, SY: 1, QY: 1 };
+const FIXED = { F: 4, H: 2, E: 4, D: 8, L: 16, A: 4, Y: 2, S: 2, V: 4, Q: 4, J: 4, R: 4 };
+const FIXED_EXT = { FD: 8, AD: 8, VD: 8, QD: 8, JD: 8, RD: 8, SY: 3, QY: 3 };
+const NO_STORAGE = new Set(['USING', 'DROP', 'PRINT', 'TITLE', 'EJECT', 'SPACE', 'CEJECT', 'AMODE', 'RMODE', 'ENTRY',
+  'EXTRN', 'WXTRN', 'PUSH', 'POP', 'MNOTE', 'ICTL', 'ISEQ', 'PUNCH', 'REPRO', 'OPSYN', 'ACONTROL', 'ADATA', 'EXITCTL',
+  'XATTR', 'ALIAS', 'CATTR', 'MACRO DEFINITION', 'PROTOTYPE', 'MODEL']);
+const SECTION = { CSECT: 'CST', RSECT: 'CST', START: 'CST', COM: 'COM', DSECT: 'DST' };
+// Conditional assembly that only declares or sets variables does not change which statements are read.
+const QUIET_CONDITIONAL = new Set(['ANOP', 'SETA', 'SETB', 'SETC', 'LCLA', 'LCLB', 'LCLC', 'GBLA', 'GBLB', 'GBLC', 'ACTR', 'MHELP']);
+
+const alignUp = (n, a) => Math.ceil(n / a) * a;
+const typeKey = (o) => `${o.type}${o.ext || ''}`;
+
+const EBCDIC = (() => {
+  const t = {};
+  const rows = [[0x40, ' '], [0x4b, '.'], [0x4c, '<'], [0x4d, '('], [0x4e, '+'], [0x50, '&'], [0x5b, '$'], [0x5c, '*'], [0x5d, ')'], [0x5e, ';'],
+    [0x60, '-'], [0x61, '/'], [0x6b, ','], [0x6c, '%'], [0x6d, '_'], [0x6e, '>'], [0x6f, '?'], [0x7a, ':'], [0x7b, '#'], [0x7c, '@'], [0x7d, "'"], [0x7e, '='], [0x7f, '"']];
+  for (const [code, ch] of rows) t[ch] = code;
+  'ABCDEFGHI'.split('').forEach((c, i) => { t[c] = 0xc1 + i; t[c.toLowerCase()] = 0x81 + i; });
+  'JKLMNOPQR'.split('').forEach((c, i) => { t[c] = 0xd1 + i; t[c.toLowerCase()] = 0x91 + i; });
+  'STUVWXYZ'.split('').forEach((c, i) => { t[c] = 0xe2 + i; t[c.toLowerCase()] = 0xa2 + i; });
+  '0123456789'.split('').forEach((c, i) => { t[c] = 0xf0 + i; });
+  return t;
+})();
+
+function selfDefining(sdt) {
+  const text = sdt.text;
+  if (sdt.type === 'X') return /^[0-9A-F]+$/i.test(text) ? parseInt(text, 16) : null;
+  if (sdt.type === 'B') return /^[01]+$/.test(text) ? parseInt(text, 2) : null;
+  if (sdt.type === 'C' || sdt.type === 'CE') {
+    let v = 0;
+    for (const ch of text) { if (EBCDIC[ch] === undefined) return null; v = v * 256 + EBCDIC[ch]; }
+    return v;
+  }
+  return null;
+}
+
+// The values, one per constant, an operand's nominal field holds, and the implicit length of each.
+function nominalLengths(o) {
+  const key = typeKey(o);
+  const n = o.nominal;
+  if (n && n.expressions) return n.expressions.map(() => FIXED_EXT[key] ?? FIXED[o.type]);
+  if (FIXED_EXT[key] || FIXED[o.type]) {
+    const count = n && n.quoted !== undefined ? n.quoted.split(',').length : 1;
+    return Array(count).fill(FIXED_EXT[key] ?? FIXED[o.type]);
+  }
+  if (!n) return [1];
+  const q = n.quoted;
+  if (o.type === 'C') return [o.ext === 'U' ? q.length * 2 : q.length];
+  if (o.type === 'X') return q.split(',').map((v) => Math.ceil(v.replace(/\s/g, '').length / 2));
+  if (o.type === 'B') return q.split(',').map((v) => Math.ceil(v.replace(/\s/g, '').length / 8));
+  if (o.type === 'P') return q.split(',').map((v) => Math.ceil((v.replace(/[^0-9]/g, '').length + 1) / 2));
+  if (o.type === 'Z') return q.split(',').map((v) => v.replace(/[^0-9]/g, '').length);
+  return null;
+}
+
+// COPY members read in where they are copied, each statement naming the member it came from, up to
+// five members deep. copy(name) gives a member's text, or null when the tree does not hold it.
+function expand(statements, copy, depth = 0, file = null) {
+  const out = [];
+  for (const st of statements) {
+    const name = st.kind === 'statement' && st.operation === 'COPY' && !st.inMacro ? (st.field || '').trim().toUpperCase() : null;
+    const member = name && copy && depth < 5 && NAME.test(name) ? copy(name) : null;
+    out.push({ ...st, file, expanded: member != null });
+    if (member != null) out.push(...expand(readHlasmStatements(member).statements, copy, depth + 1, name));
+  }
+  return out;
+}
+
+export function locate(text, { copy = null } = {}) {
+  const statements = expand(readHlasmStatements(text).statements, copy);
+  const sections = new Map();
+  const order = [];
+  const symbols = new Map();
+  const placed = [];
+  const literals = { pending: [], seen: new Map() };
+  const esd = { entries: [], externals: [] };
+  let current = null;
+  let lost = null;
+  const unknown = [];
+
+  const section = (name, type) => {
+    if (!sections.has(name)) {
+      sections.set(name, { name, type, lc: 0, high: 0, after: null, line: null });
+      order.push(name);
+    }
+    return sections.get(name);
+  };
+  const here = () => current || (current = section('', 'CST'));
+  const advance = (s, to) => { s.lc = to; s.high = Math.max(s.high, to); };
+  const markUnknown = (s, line, why) => { if (s && s.after === null) { s.after = line; unknown.push({ line, why }); } };
+
+  function value(e, s) {
+    if (!e) return null;
+    switch (e.t) {
+      case 'num': return { abs: e.v };
+      case 'sdt': { const v = selfDefining(e); return v === null ? null : { abs: v }; }
+      case 'loc': return s && s.after === null ? { sec: s.name, off: s.lc } : null;
+      case 'paren': return value(e.e, s);
+      case 'unary': { const v = value(e.e, s); return v && v.abs !== undefined ? { abs: e.op === '-' ? -v.abs : v.abs } : null; }
+      case 'sym': {
+        const y = symbols.get(e.name);
+        if (!y || y.value === null) return null;
+        return y.value;
+      }
+      case 'attr': {
+        if (e.attr !== 'L') return null;
+        if (e.of.t === 'loc') return { abs: 1 };
+        const y = e.of.t === 'sym' ? symbols.get(e.of.name) : null;
+        return y && y.len !== null ? { abs: y.len } : null;
+      }
+      case 'bin': {
+        const l = value(e.l, s);
+        const r = value(e.r, s);
+        if (!l || !r) return null;
+        if (e.op === '+' || e.op === '-') {
+          if (l.abs !== undefined && r.abs !== undefined) return { abs: e.op === '+' ? l.abs + r.abs : l.abs - r.abs };
+          if (l.sec !== undefined && r.abs !== undefined) return { sec: l.sec, off: e.op === '+' ? l.off + r.abs : l.off - r.abs };
+          if (e.op === '+' && l.abs !== undefined && r.sec !== undefined) return { sec: r.sec, off: r.off + l.abs };
+          if (e.op === '-' && l.sec !== undefined && r.sec !== undefined && l.sec === r.sec) return { abs: l.off - r.off };
+          return null;
+        }
+        if (l.abs === undefined || r.abs === undefined) return null;
+        if (e.op === '*') return { abs: l.abs * r.abs };
+        return r.abs === 0 ? { abs: 0 } : { abs: Math.trunc(l.abs / r.abs) };
+      }
+      default: return null;
+    }
+  }
+  const absolute = (e, s) => { const v = value(e, s); return v && v.abs !== undefined ? v.abs : null; };
+
+  // Length, alignment and size of one DC or DS operand, or null when one of them cannot be known.
+  function constant(o, s) {
+    const dup = o.dup ? absolute(o.dup, s) : 1;
+    if (dup === null || o.bitLength) return null;
+    let lengths;
+    let align = 1;
+    if (o.length) {
+      const n = absolute(o.length, s);
+      if (n === null) return null;
+      const count = o.nominal ? (o.nominal.expressions ? o.nominal.expressions.length : (['C', 'G'].includes(o.type) ? 1 : o.nominal.quoted.split(',').length)) : 1;
+      lengths = Array(count).fill(n);
+    } else {
+      lengths = nominalLengths(o);
+      if (!lengths) return null;
+      align = ALIGN_EXT[typeKey(o)] ?? ALIGN[o.type] ?? 1;
+    }
+    return { align, length: lengths[0], size: dup * lengths.reduce((a, b) => a + b, 0) };
+  }
+
+  function define(st, s, off, len, type, how) {
+    if (!st.name || st.name.startsWith('&') || st.name.startsWith('.')) return;
+    const name = st.name.toUpperCase();
+    const known = s.after === null && off !== null;
+    symbols.set(name, { name, line: st.line, file: st.file, section: s.name, value: known ? { sec: s.name, off } : null, len, type, how, after: known ? null : s.after });
+  }
+
+  // An EQU with no length operand takes the length of its value's leftmost term: 1 for *, a
+  // self-defining term, L' or a symbol naming an assembler instruction other than DC, DS or CCW.
+  // https://www.ibm.com/docs/en/hla-and-tf/1.6.0?topic=instructions-equ-instruction
+  function leftmostLength(e) {
+    while (e && (e.t === 'bin' || e.t === 'paren' || e.t === 'unary')) e = e.t === 'bin' ? e.l : e.e;
+    if (!e) return null;
+    if (e.t !== 'sym') return 1;
+    const y = symbols.get(e.name);
+    return y ? y.len : null;
+  }
+
+  // Each literal a statement names, and whether an instruction names it by relative address (an RI
+  // operand), which aligns an odd-length one to a halfword in the pool.
+  function literalsIn(node, relative = false, out = []) {
+    if (!node || typeof node !== 'object') return out;
+    if (node.t === 'lit') out.push({ dc: node.dc, relative });
+    if (node.kind === 'INSTRUCTION') { for (const op of node.operands) literalsIn(op, /^RI/.test(op.operand), out); return out; }
+    for (const v of Object.values(node)) if (v && typeof v === 'object') literalsIn(v, relative, out);
+    return out;
+  }
+
+  function pool(s, line) {
+    if (!literals.pending.length) return;
+    if (s.after !== null) { literals.pending = []; return; }
+    // Five segments in the order literals were written: lengths that are multiples of 16, of 8, of 4,
+    // even, then odd. https://www.ibm.com/docs/en/hla-and-tf/1.6.0?topic=instructions-ltorg-instruction
+    const segment = (x) => { const i = [16, 8, 4, 2].findIndex((n) => x.size % n === 0); return i < 0 ? 4 : i; };
+    const sized = literals.pending.map((lit) => ({ ...lit, ...constant(lit.dc, s) }));
+    if (sized.some((x) => x.size === undefined || x.size === null)) { literals.pending = []; markUnknown(s, line, 'literal of unknown length'); return; }
+    let at = alignUp(s.lc, 8);
+    for (const g of [0, 1, 2, 3, 4]) {
+      for (const x of sized.filter((y) => segment(y) === g)) { if (g === 4 && x.relative) at = alignUp(at, 2); at += x.size; }
+    }
+    advance(s, at);
+    literals.pending = [];
+    literals.seen = new Map();
+  }
+
+  let first = null;
+  for (const st of statements) {
+    if (st.inMacro) continue;
+    let r = parseHlasmStatement(st);
+    // An open-code statement with variable symbols is placed as what it reads as, when its lengths
+    // do not depend on a variable's value; its label is a variable and defines nothing here.
+    if (r.kind === 'SUBSTITUTED' && r.status === 'parsed' && r.node.exact) r = { kind: r.node.as, status: 'parsed', node: r.node.node };
+    const kind = r.kind;
+    if (kind === 'UNREADABLE' || kind === 'UNKNOWN') { markUnknown(here(), st.line, `${kind}`); continue; }
+    if (kind === 'CONDITIONAL') {
+      if (!QUIET_CONDITIONAL.has(st.operation)) { lost = lost ?? st.line; for (const s of sections.values()) markUnknown(s, st.line, st.operation); markUnknown(here(), st.line, st.operation); }
+      continue;
+    }
+    if (SECTION[kind]) {
+      const name = (st.name || '').toUpperCase();
+      current = section(name, SECTION[kind]);
+      if (kind !== 'DSECT' && first === null) first = current;
+      if (current.line === null) current.line = st.line;
+      if (r.status === 'parsed' && kind === 'START' && r.node.origin) {
+        const o = absolute(r.node.origin, current);
+        if (o === null) markUnknown(current, st.line, 'START origin'); else advance(current, o);
+      }
+      if (lost !== null) markUnknown(current, lost, 'conditional assembly');
+      placed.push({ line: st.line, file: st.file, section: current.name, loc: current.after === null ? current.lc : null });
+      if (!symbols.has(name) && name) symbols.set(name, { name, line: st.line, section: name, value: { sec: name, off: 0 }, len: 1, type: SECTION[kind], how: 'SECTION', after: null });
+      continue;
+    }
+    const s = here();
+    const at = () => (s.after === null ? s.lc : null);
+    // Statements that take no storage leave the location counter where it is, read or not; a COPY
+    // whose member was read in is one, its statements following it.
+    if (NO_STORAGE.has(kind) || (kind === 'COPY' && st.expanded) || kind === 'ENTRY' || kind === 'EXTRN' || kind === 'WXTRN' || kind === 'END') {
+      placed.push({ line: st.line, file: st.file, section: s.name, loc: at() });
+      if (r.status === 'parsed' && kind === 'ENTRY') for (const n of r.node.names || []) esd.entries.push({ name: n, line: st.line });
+      if (r.status === 'parsed' && (kind === 'EXTRN' || kind === 'WXTRN')) for (const n of r.node.names || []) esd.externals.push({ name: n, how: kind, line: st.line });
+      if (kind === 'END') break;
+      continue;
+    }
+    if (kind === 'LTORG') { placed.push({ line: st.line, file: st.file, section: s.name, loc: at() }); define(st, s, s.lc, 1, 'REL', 'LTORG'); pool(s, st.line); continue; }
+    // CALL's macro names its entry in a V-type constant, so the entry is an external the module needs.
+    if (kind === 'CALL' && r.status === 'parsed' && r.node.entry) esd.externals.push({ name: r.node.entry.toUpperCase(), how: 'CALL', line: st.line });
+    if (r.status !== 'parsed') { placed.push({ line: st.line, file: st.file, section: s.name, loc: null }); markUnknown(s, st.line, `${kind} ${r.status}`); continue; }
+    const node = r.node;
+    for (const { dc, relative } of literalsIn(node)) {
+      const key = JSON.stringify(dc);
+      if (!literals.seen.has(key)) { literals.seen.set(key, { dc, relative }); literals.pending.push(literals.seen.get(key)); }
+      else if (relative) literals.seen.get(key).relative = true;
+      if (dc.type === 'V' && dc.nominal && dc.nominal.expressions) for (const e of dc.nominal.expressions) if (e.t === 'sym') esd.externals.push({ name: e.name, how: 'V', line: st.line });
+    }
+    if (node.kind === 'INSTRUCTION') {
+      const at = alignUp(s.lc, 2);
+      placed.push({ line: st.line, file: st.file, section: s.name, loc: s.after === null ? at : null });
+      define(st, s, at, node.length, 'REL', 'INSTRUCTION');
+      advance(s, at + node.length);
+      continue;
+    }
+    if (kind === 'DC' || kind === 'DS') {
+      let at = s.lc;
+      let firstAt = null;
+      let firstLen = null;
+      let known = s.after === null;
+      for (const o of node.operands) {
+        if (kind === 'DC' && o.type === 'V' && o.nominal && o.nominal.expressions) for (const e of o.nominal.expressions) if (e.t === 'sym') esd.externals.push({ name: e.name, how: 'V', line: st.line });
+        const c = known ? constant(o, s) : null;
+        if (!c) { known = false; continue; }
+        at = alignUp(at, c.align);
+        if (firstAt === null) { firstAt = at; firstLen = c.length; }
+        at += c.size;
+      }
+      if (!known) { placed.push({ line: st.line, file: st.file, section: s.name, loc: null }); markUnknown(s, st.line, `${kind} of unknown length`); define(st, s, null, null, 'REL', kind); continue; }
+      placed.push({ line: st.line, file: st.file, section: s.name, loc: firstAt });
+      define(st, s, firstAt, firstLen, 'REL', kind);
+      advance(s, at);
+      continue;
+    }
+    if (kind === 'EQU') {
+      const v = value(node.value, s);
+      const len = node.length ? absolute(node.length, s) : leftmostLength(node.value);
+      const name = (st.name || '').toUpperCase();
+      if (name) symbols.set(name, { name, line: st.line, section: v && v.sec !== undefined ? v.sec : null, value: v, len, type: v && v.abs !== undefined ? 'ABS' : 'REL', how: 'EQU', after: v ? null : s.after });
+      placed.push({ line: st.line, file: st.file, section: s.name, loc: s.after === null ? s.lc : null });
+      continue;
+    }
+    if (kind === 'ORG') {
+      placed.push({ line: st.line, file: st.file, section: s.name, loc: s.after === null ? s.lc : null });
+      if (s.after !== null) continue;
+      if (!node.value) { advance(s, s.high); s.lc = s.high; continue; }
+      const v = value(node.value, s);
+      if (!v || v.sec !== s.name) { markUnknown(s, st.line, 'ORG to an unknown location'); continue; }
+      let to = v.off;
+      if (node.boundary) to = alignUp(to, node.boundary) + (node.offset ? absolute(node.offset, s) ?? 0 : 0);
+      s.lc = to;
+      s.high = Math.max(s.high, to);
+      continue;
+    }
+    if (kind === 'CNOP') {
+      const b = absolute(node.byte, s);
+      const w = absolute(node.boundary, s);
+      // CNOP pads with two-byte NOPs, so the statement starts at the next halfword.
+      placed.push({ line: st.line, file: st.file, section: s.name, loc: s.after === null ? alignUp(s.lc, 2) : null });
+      if (b === null || !w) { markUnknown(s, st.line, 'CNOP'); continue; }
+      let at = alignUp(s.lc, 2);
+      while (at % w !== b) at += 2;
+      advance(s, at);
+      continue;
+    }
+    if (kind === 'CCW' || kind === 'CCW0' || kind === 'CCW1') {
+      const at = alignUp(s.lc, 8);
+      placed.push({ line: st.line, file: st.file, section: s.name, loc: s.after === null ? at : null });
+      define(st, s, at, 8, 'REL', 'CCW');
+      advance(s, at + 8);
+      continue;
+    }
+    placed.push({ line: st.line, file: st.file, section: s.name, loc: null });
+    markUnknown(s, st.line, kind);
+  }
+  if (literals.pending.length) pool(first || here(), null);
+
+  return {
+    sections: order.map((n) => { const s = sections.get(n); return { name: n, type: s.type, line: s.line, length: s.after === null ? s.high : null, after: s.after }; }),
+    symbols: [...symbols.values()].map((y) => ({ name: y.name, line: y.line, file: y.file ?? null, section: y.section, loc: y.value && y.value.sec !== undefined ? y.value.off : y.value ? y.value.abs : null, len: y.len, type: y.type, how: y.how, after: y.after })),
+    statements: placed,
+    esd,
+    unknown,
+  };
+}

package/lib/hlasm/macro/authorization.mjs

@@ -0,0 +1,146 @@
+// SPDX-License-Identifier: AGPL-3.0-or-later
+// Parsers for the z/OS authorization macros MODESET, TESTAUTH, and RACROUTE.
+import { HlasmSyntax, NAME, remoteForm } from '../operands.mjs';
+import { MVS38_EXTKEYS, mvs38Form } from '../mvs38.mjs';
+
+const MODESET_EXTKEYS = new Set(['ZERO', 'TCB', 'RBT1', 'RBT234', 'KEY2', 'KEY3', 'KEY4', 'KEY7']);
+const MODESET_KEYS = new Set(['ZERO', 'NZERO']);
+const MODESET_MODES = new Set(['PROB', 'SUP']);
+const MODESET_KNOWN = new Set(['EXTKEY', 'KEYADDR', 'KEYREG', 'SAVEKEY', 'WORKREG', 'RELATED', 'KEY', 'MODE', 'MF']);
+// The list and execute forms of the MODESET that generates an SVC (z/OS 3.1 MVS Programming:
+// Authorized Assembler Services Reference, Volume 3): the list form holds KEY and MODE, the execute
+// form names the list.
+const MODESET_LIST_KNOWN = new Set(['KEY', 'MODE', 'RELATED', 'MF']);
+const MODESET_EXECUTE_KNOWN = new Set(['RELATED', 'MF']);
+
+const TESTAUTH_KNOWN = new Set(['FCTN', 'STATE', 'KEY', 'RBLEVEL', 'BRANCH']);
+const TESTAUTH_STATES = new Set(['YES', 'NO']);
+
+const RACROUTE_REQUESTS = new Set(['AUTH', 'FASTAUTH', 'DEFINE', 'DIRAUTH', 'EXTRACT', 'LIST', 'SIGNON', 'STAT', 'TOKENBLD', 'TOKENMAP', 'TOKENXTR', 'VERIFY', 'VERIFYX', 'AUDIT']);
+
+function parseModeset(st, ops, opts) {
+  const { keywords } = ops;
+  const node = { kind: 'MODESET', key: null, mode: null, extkey: null, keywords: {} };
+
+  for (const [k, v] of keywords) {
+    if (!MODESET_KNOWN.has(k)) {
+      throw new HlasmSyntax(`MODESET: unknown keyword ${k}`);
+    }
+    node.keywords[k] = v;
+  }
+
+  if (keywords.has('MF')) {
+    const { form, list } = remoteForm(keywords.get('MF'), 'MF');
+    const known = form === 'L' ? MODESET_LIST_KNOWN : MODESET_EXECUTE_KNOWN;
+    for (const k of keywords.keys()) {
+      if (!known.has(k)) throw new HlasmSyntax(`MODESET: the ${form === 'L' ? 'list' : 'execute'} form does not take ${k}`);
+    }
+    if (form === 'L' && !keywords.has('KEY') && !keywords.has('MODE')) throw new HlasmSyntax('MODESET: the list form needs KEY or MODE');
+    node.mf = form;
+    if (list) node.list = list;
+  }
+
+  if (keywords.has('EXTKEY')) {
+    const v = keywords.get('EXTKEY').trim().toUpperCase();
+    if (MVS38_EXTKEYS.has(v)) {
+      mvs38Form(opts, node, 'MODESET', `EXTKEY=${v}`);
+      node.pswKey = MVS38_EXTKEYS.get(v);
+    } else if (!MODESET_EXTKEYS.has(v)) {
+      throw new HlasmSyntax(`MODESET: invalid EXTKEY value ${v}`);
+    }
+    node.extkey = v;
+  }
+
+  if (keywords.has('KEY')) {
+    const v = keywords.get('KEY').trim().toUpperCase();
+    if (!MODESET_KEYS.has(v)) {
+      throw new HlasmSyntax(`MODESET: invalid KEY value ${v}`);
+    }
+    node.key = v;
+  }
+
+  if (keywords.has('MODE')) {
+    const v = keywords.get('MODE').trim().toUpperCase();
+    if (!MODESET_MODES.has(v)) {
+      throw new HlasmSyntax(`MODESET: invalid MODE value ${v}`);
+    }
+    node.mode = v;
+  }
+
+  return node;
+}
+
+function parseTestauth(st, ops) {
+  const { keywords } = ops;
+  const node = { kind: 'TESTAUTH', keywords: {} };
+
+  for (const [k, v] of keywords) {
+    if (!TESTAUTH_KNOWN.has(k)) {
+      throw new HlasmSyntax(`TESTAUTH: unknown keyword ${k}`);
+    }
+    node.keywords[k] = v;
+  }
+
+  if (keywords.has('STATE')) {
+    const v = keywords.get('STATE').trim().toUpperCase();
+    if (!TESTAUTH_STATES.has(v)) {
+      throw new HlasmSyntax(`TESTAUTH: invalid STATE value ${v}`);
+    }
+  }
+
+  if (keywords.has('KEY')) {
+    const v = keywords.get('KEY').trim().toUpperCase();
+    if (!TESTAUTH_STATES.has(v)) {
+      throw new HlasmSyntax(`TESTAUTH: invalid KEY value ${v}`);
+    }
+  }
+
+  if (keywords.has('RBLEVEL')) {
+    const v = keywords.get('RBLEVEL').trim();
+    if (v !== '1' && v !== '2') {
+      throw new HlasmSyntax(`TESTAUTH: invalid RBLEVEL value ${v}`);
+    }
+  }
+
+  if (keywords.has('BRANCH')) {
+    const v = keywords.get('BRANCH').trim().toUpperCase();
+    if (!TESTAUTH_STATES.has(v)) {
+      throw new HlasmSyntax(`TESTAUTH: invalid BRANCH value ${v}`);
+    }
+  }
+
+  return node;
+}
+
+function parseRacroute(st, ops) {
+  const { keywords } = ops;
+  const node = { kind: 'RACROUTE', request: null, class: null, entity: null, attr: null, keywords: {} };
+
+  const req = keywords.get('REQUEST');
+  if (!req) {
+    throw new HlasmSyntax('RACROUTE: REQUEST is required');
+  }
+  const reqVal = req.trim().toUpperCase();
+  if (!RACROUTE_REQUESTS.has(reqVal)) {
+    throw new HlasmSyntax(`RACROUTE: invalid REQUEST value ${reqVal}`);
+  }
+  node.request = reqVal;
+
+  for (const [k, v] of keywords) {
+    if (!NAME.test(k)) {
+      throw new HlasmSyntax(`RACROUTE: invalid keyword name ${k}`);
+    }
+    node.keywords[k] = v;
+    if (k === 'CLASS') node.class = v;
+    if (k === 'ENTITY') node.entity = v;
+    if (k === 'ATTR') node.attr = v;
+  }
+
+  return node;
+}
+
+export const parsers = {
+  MODESET: parseModeset,
+  TESTAUTH: parseTestauth,
+  RACROUTE: parseRacroute
+};

package/lib/hlasm/macro/datasets.mjs

@@ -0,0 +1,107 @@
+// SPDX-License-Identifier: AGPL-3.0-or-later
+// Parser for z/OS data-set macros: OPEN, CLOSE, and DCB.
+import { HlasmSyntax, NAME, remoteForm, splitOperands } from '../operands.mjs';
+import { parseAddress } from '../expr.mjs';
+import { MVS38_KEYWORDS, mvs38Form } from '../mvs38.mjs';
+
+const OPEN_OPTS = new Set(['INPUT', 'OUTPUT', 'UPDAT', 'EXTEND', 'OUTIN', 'OUTINX', 'INOUT', 'RDBACK', 'DISP', 'LEAVE', 'REREAD']);
+const CLOSE_OPTS = new Set(['REREAD', 'LEAVE', 'REWIND', 'FREE', 'DISP']);
+const DCB_KEYWORDS = new Set(['BFALN', 'BFTEK', 'BLKSIZE', 'BUFCB', 'BUFL', 'BUFNO', 'BUFOFF', 'DCBE', 'DDNAME', 'DEVD', 'DSORG', 'EODAD', 'EROPT', 'EXLST', 'LRECL', 'MACRF', 'NCP', 'OPTCD', 'RECFM', 'SYNAD', 'KEYLEN', 'LIMCT', 'RKP', 'DEN', 'TRTCH', 'PRTSP', 'STACK', 'FUNC', 'MODE', 'CODE',
+  // EXCP (z/OS 3.1 DFSMSdfp Advanced Services, Data Control Block (DCB) Fields).
+  'REPOS', 'EOEA', 'PCIA', 'SIOA', 'CENDA', 'XENDA', 'IMSK', 'PGFIX', 'IOBAD']);
+
+// (dcb,(options),dcb,(options),...): each DCB address followed by its options, written in parentheses,
+// as one bare word or omitted. A single DCB may be written without the outer parentheses. The list
+// form (MF=L) and the execute form (MF=(E,addr)) may leave a DCB address empty, to be filled in by
+// the other (z/OS 3.1 DFSMS Macro Instructions for Data Sets, OPEN and CLOSE list and execute forms).
+function parseDcbList(raw, validOpts, emptyAllowed = false) {
+  const text = raw.trim();
+  const items = text.startsWith('(') && text.endsWith(')') ? splitOperands(text.slice(1, -1)) : [text];
+  const dcbs = [];
+  for (let i = 0; i < items.length; i++) {
+    const dcbRaw = items[i].trim();
+    if (!dcbRaw && !emptyAllowed) throw new HlasmSyntax('empty dcb address in list');
+    let dcb = null;
+    if (dcbRaw) try { dcb = parseAddress(dcbRaw); } catch { throw new HlasmSyntax(`invalid dcb address: ${dcbRaw}`); }
+    const next = (items[i + 1] ?? '').trim();
+    const word = next.toUpperCase();
+    let options = [];
+    if (next.startsWith('(') && next.endsWith(')')) options = splitOperands(next.slice(1, -1)).map((o) => o.trim().toUpperCase()).filter(Boolean);
+    else if (validOpts.has(word)) options = [word];
+    else if (next !== '' || i + 1 >= items.length) { dcbs.push({ dcb, options }); continue; }
+    for (const o of options) if (!validOpts.has(o)) throw new HlasmSyntax(`unknown option: ${o}`);
+    dcbs.push({ dcb, options });
+    i++;
+  }
+  return dcbs;
+}
+
+// The DCB list of OPEN or CLOSE and the form MF= gives it; the execute form may omit the list.
+function dcbOperand(macro, ops, validOpts) {
+  const mf = ops.keywords.has('MF') ? remoteForm(ops.keywords.get('MF'), 'MF') : null;
+  if (ops.positional.length === 0 && mf?.form === 'E') return { dcbs: [], mf };
+  if (ops.positional.length !== 1) throw new HlasmSyntax(`${macro} requires exactly one positional operand`);
+  return { dcbs: parseDcbList(ops.positional[0], validOpts, Boolean(mf)), mf };
+}
+
+export const parsers = {
+  OPEN(st, ops) {
+    const { dcbs, mf } = dcbOperand('OPEN', ops, OPEN_OPTS);
+    const keywords = {};
+    for (const [k, v] of ops.keywords) {
+      if (k === 'TYPE') {
+        if (v.toUpperCase() !== 'J') throw new HlasmSyntax('OPEN TYPE must be J');
+        keywords.TYPE = v;
+      } else if (k === 'MODE') {
+        if (v !== '24' && v !== '31') throw new HlasmSyntax('OPEN MODE must be 24 or 31');
+        keywords.MODE = v;
+      } else if (k === 'MF') {
+        keywords.MF = v;
+      } else {
+        throw new HlasmSyntax(`unknown keyword for OPEN: ${k}`);
+      }
+    }
+    return { kind: 'OPEN', dcbs, ...(mf ? { mf: mf.form === 'L' ? 'L' : { form: 'E', list: mf.list } } : {}), keywords };
+  },
+  CLOSE(st, ops) {
+    const { dcbs, mf } = dcbOperand('CLOSE', ops, CLOSE_OPTS);
+    const keywords = {};
+    for (const [k, v] of ops.keywords) {
+      if (k === 'TYPE') {
+        if (v.toUpperCase() !== 'T') throw new HlasmSyntax('CLOSE TYPE must be T');
+        keywords.TYPE = v;
+      } else if (k === 'MODE') {
+        if (v !== '24' && v !== '31') throw new HlasmSyntax('CLOSE MODE must be 24 or 31');
+        keywords.MODE = v;
+      } else if (k === 'MF') {
+        keywords.MF = v;
+      } else {
+        throw new HlasmSyntax(`unknown keyword for CLOSE: ${k}`);
+      }
+    }
+    return { kind: 'CLOSE', dcbs, ...(mf ? { mf: mf.form === 'L' ? 'L' : { form: 'E', list: mf.list } } : {}), keywords };
+  },
+  DCB(st, ops, opts) {
+    if (ops.positional.some((p) => p.trim() !== '')) throw new HlasmSyntax('DCB takes no positional operands');
+    const keywords = {};
+    let ddname = null;
+    let dsorg = null;
+    let macrf = null;
+    let recfm = null;
+    let lrecl = null;
+    let blksize = null;
+    const legacy = {};
+    for (const [k, v] of ops.keywords) {
+      if (MVS38_KEYWORDS.DCB.has(k)) mvs38Form(opts, legacy, 'DCB', `${k}=${String(v).trim()}`);
+      else if (!DCB_KEYWORDS.has(k)) throw new HlasmSyntax(`unknown keyword for DCB: ${k}`);
+      keywords[k] = v;
+      if (k === 'DDNAME') ddname = v.toUpperCase();
+      else if (k === 'DSORG') dsorg = v;
+      else if (k === 'MACRF') macrf = v;
+      else if (k === 'RECFM') recfm = v;
+      else if (k === 'LRECL') lrecl = v;
+      else if (k === 'BLKSIZE') blksize = v;
+    }
+    return { kind: 'DCB', ...legacy, name: st.name, ddname, dsorg, macrf, recfm, lrecl, blksize, keywords };
+  }
+};