parseman 0.47.0 → 0.47.1

package/dist/cli/index.js

@@ -52,6 +52,13 @@ function empty() {
 function matchesEmpty(p, seen = /* @__PURE__ */ new Set(), resolve2) {
   if (seen.has(p)) return true;
   seen.add(p);
+  try {
+    return matchesEmptyBody(p, seen, resolve2);
+  } finally {
+    seen.delete(p);
+  }
+}
+function matchesEmptyBody(p, seen, resolve2) {
   const me = (c) => matchesEmpty(c, seen, resolve2);
   const d = p._def;
   switch (d.tag) {
@@ -153,6 +160,13 @@ function sequenceFirstSet(parsers) {
 function firstSetOf(p, seen = /* @__PURE__ */ new Set(), resolve2) {
   if (seen.has(p)) return any();
   seen.add(p);
+  try {
+    return firstSetBody(p, seen, resolve2);
+  } finally {
+    seen.delete(p);
+  }
+}
+function firstSetBody(p, seen, resolve2) {
   const fs = (c) => firstSetOf(c, seen, resolve2);
   const empties = (c) => matchesEmpty(c, /* @__PURE__ */ new Set(), resolve2);
   const d = p._def;
@@ -2501,6 +2515,23 @@ function rollbackTrivia(ctx, mark) {
   if (ctx._triviaLog && ctx._triviaLog.length !== mark.log) ctx._triviaLog.length = mark.log;
   if (ctx._rootTriviaLog && ctx._rootTriviaLog.length !== mark.rootLog) ctx._rootTriviaLog.length = mark.rootLog;
 }
+function removeNumberRange(values, start, end) {
+  if (values === void 0 || end <= start) return;
+  if (values.length > end) values.copyWithin(start, end);
+  values.length -= end - start;
+}
+function rollbackScannedTriviaAt(ctx, tlogStart, tlogEnd, logStart, logEnd, rootLogStart, rootLogEnd) {
+  const b = ctx._cstBuf;
+  if (b !== void 0) {
+    const tlog = b.tl;
+    removeNumberRange(tlog, tlogStart, tlogEnd);
+    if (tlog !== void 0 && tlog.length === 0) b.tl = void 0;
+  } else {
+    removeNumberRange(ctx._cstTriviaLog, tlogStart, tlogEnd);
+  }
+  removeNumberRange(ctx._triviaLog, logStart, logEnd);
+  removeNumberRange(ctx._rootTriviaLog, rootLogStart, rootLogEnd);
+}
 function parseTriviaNoCapture(triviaP, input, cur, ctx) {
   const probeCtx = createDetachedParseContext(ctx.trackLines, ctx.state);
   probeCtx._lineIndex = ctx._lineIndex;
@@ -3204,7 +3235,9 @@ function sequence(...parsers) {
           continue;
         }
         if (ctx.trivia && i > 0) {
-          const mark = saveTriviaMark(ctx);
+          const mTlog = cstTlLen(ctx);
+          const mLog = ctx._triviaLog?.length ?? 0;
+          const mRootLog = ctx._rootTriviaLog?.length ?? 0;
           let scanEnd;
           if (needsDeferredTriviaCommit(ctx)) {
             const scan = scanTrivia(input, cur, ctx);
@@ -3213,10 +3246,13 @@ function sequence(...parsers) {
           } else {
             scanEnd = advanceTrivia(input, cur, ctx);
           }
+          const scanTlog = cstTlLen(ctx);
+          const scanLog = ctx._triviaLog?.length ?? 0;
+          const scanRootLog = ctx._rootTriviaLog?.length ?? 0;
           const result2 = parsers[i].parse(input, scanEnd, ctx);
           if (!result2.ok) return result2;
           if (result2.span.end > scanEnd) cur = result2.span.end;
-          else rollbackTrivia(ctx, mark);
+          else rollbackScannedTriviaAt(ctx, mTlog, scanTlog, mLog, scanLog, mRootLog, scanRootLog);
           if (values !== void 0) values.push(result2.value);
           continue;
         }
@@ -3247,7 +3283,9 @@ function sequence(...parsers) {
       }
       if (ctx.trivia && i > 0) {
         let scanEnd;
-        const mark = saveTriviaMark(ctx);
+        const mTlog = cstTlLen(ctx);
+        const mLog = ctx._triviaLog?.length ?? 0;
+        const mRootLog = ctx._rootTriviaLog?.length ?? 0;
         if (needsDeferredTriviaCommit(ctx)) {
           const scan = scanTrivia(input, cur, ctx);
           scan.commit();
@@ -3255,10 +3293,13 @@ function sequence(...parsers) {
         } else {
           scanEnd = advanceTrivia(input, cur, ctx);
         }
+        const scanTlog = cstTlLen(ctx);
+        const scanLog = ctx._triviaLog?.length ?? 0;
+        const scanRootLog = ctx._rootTriviaLog?.length ?? 0;
         const result2 = parsers[i].parse(input, scanEnd, ctx);
         if (!result2.ok) return result2;
         if (result2.span.end > scanEnd) cur = result2.span.end;
-        else rollbackTrivia(ctx, mark);
+        else rollbackScannedTriviaAt(ctx, mTlog, scanTlog, mLog, scanLog, mRootLog, scanRootLog);
         if (values !== void 0) values.push(result2.value);
         continue;
       }
@@ -3286,7 +3327,9 @@ function sequence(...parsers) {
       for (let i = 0; i < parsers.length; i++) {
         if (ctx.trivia && i > 0) {
           let scanEnd;
-          let mark = saveTriviaMark(ctx);
+          const mTlog = cstTlLen(ctx);
+          const mLog = ctx._triviaLog?.length ?? 0;
+          const mRootLog = ctx._rootTriviaLog?.length ?? 0;
           if (needsDeferredTriviaCommit(ctx)) {
             const scan = scanTrivia(input, cur, ctx);
             scan.commit();
@@ -3294,12 +3337,15 @@ function sequence(...parsers) {
           } else {
             scanEnd = advanceTrivia(input, cur, ctx);
           }
+          const scanTlog = cstTlLen(ctx);
+          const scanLog = ctx._triviaLog?.length ?? 0;
+          const scanRootLog = ctx._rootTriviaLog?.length ?? 0;
           const result2 = parsers[i].parse(input, scanEnd, ctx);
           if (!result2.ok) return result2;
           if (result2.span.end > scanEnd) {
             cur = result2.span.end;
           } else {
-            rollbackTrivia(ctx, mark);
+            rollbackScannedTriviaAt(ctx, mTlog, scanTlog, mLog, scanLog, mRootLog, scanRootLog);
           }
           if (values !== void 0) values.push(result2.value);
           continue;
@@ -5293,6 +5339,7 @@ var OP_LIVE = 37;
 var OP_ATTEMPT = 38;
 var OP_LABEL = 39;
 var OP_COV = 40;
+var OP_SCOPE_PLAIN = 41;
 var OP_DISPATCH = 25;
 var OP_ROUTED = 26;
 var OP_NAMES = {
@@ -5335,7 +5382,8 @@ var OP_NAMES = {
   [OP_LIVE]: "LIVE",
   [OP_ATTEMPT]: "ATTEMPT",
   [OP_LABEL]: "LABEL",
-  [OP_COV]: "COV"
+  [OP_COV]: "COV",
+  [OP_SCOPE_PLAIN]: "SCOPE_PLAIN"
 };
 
 // src/table/program.ts
@@ -5393,6 +5441,15 @@ function expandCompact(p) {
     ...p.a === void 0 ? {} : { asm: p.a }
   });
 }
+function validateDispatchSpec(spec, arity, fallbackRouted) {
+  const invalid = (arms) => {
+    for (const arm of arms) if (!Number.isInteger(arm) || arm < 0 || arm >= arity) return true;
+    return false;
+  };
+  if (spec === void 0 || !Number.isInteger(arity) || arity < 0 || fallbackRouted !== 0 && fallbackRouted !== 1 || spec.routed.length !== arity || spec.routed.some((value) => value !== 0 && value !== 1) || invalid(spec.byKey.values()) || invalid(spec.byFold.values()) || spec.match.some((matcher) => invalid([matcher[3]]))) {
+    throw new TypeError("table: malformed dispatch");
+  }
+}
 var TABLE_RUNTIME = /* @__PURE__ */ Symbol("parseman.tableRuntime");
 function ownTableProgram(prog, resolved) {
   return { ...prog, [TABLE_RUNTIME]: { resolved } };
@@ -5951,7 +6008,7 @@ var Encoder = class {
     this.carryTriviaMeta(amb);
     this.carryTriviaMeta(p);
     if (p._def.tag === "grammar") this.carryTriviaMeta(p._def.triviaParser);
-    this.rules[name] = amb === void 0 ? body : this.emit(OP_SCOPE, this.triviaSlot(amb), body);
+    this.rules[name] = amb === void 0 ? body : this.emit(OP_SCOPE_PLAIN, this.triviaSlot(amb), body);
   }
   /**
    * The row a cross-rule `g.X` reference jumps to — the target's body, RE-WRAPPED
@@ -5994,7 +6051,7 @@ var Encoder = class {
     const amb = p._meta.grammarTrivia ?? target._meta.grammarTrivia;
     if (amb === void 0 || this.activeTrivia !== void 0) return ip;
     if (!hasOwnTriviaBoundary(target)) return ip;
-    return this.emit(OP_SCOPE, this.triviaSlot(amb), ip);
+    return this.emit(OP_SCOPE_PLAIN, this.triviaSlot(amb), ip);
   }
   node(p) {
     const lazyRef = p._def.tag === "lazy";
@@ -6399,6 +6456,7 @@ var Encoder = class {
           return [ip + 1];
         case OP_SCOPE:
         case OP_SCOPE_CAP:
+        case OP_SCOPE_PLAIN:
         case OP_WITHCTX:
         case OP_XFORM:
         case OP_LEAF:
@@ -7589,6 +7647,7 @@ function childSlots(code, ip, out) {
     case OP_GATE:
     case OP_SCOPE:
     case OP_SCOPE_CAP:
+    case OP_SCOPE_PLAIN:
     case OP_WITHCTX:
     case OP_XFORM:
     case OP_NODE:
@@ -7677,7 +7736,7 @@ function reachableSites(code, roots) {
 }
 function transfer(code, ip, at, hostCst) {
   const op = code[ip];
-  if (op === OP_SCOPE || op === OP_SCOPE_CAP) {
+  if (op === OP_SCOPE || op === OP_SCOPE_CAP || op === OP_SCOPE_PLAIN) {
     const ki = code[ip + 1];
     const tri = ki < 0 ? TRI_NONE : ki;
     const cap = op === OP_SCOPE_CAP ? CAP_ON : at.cap;
@@ -7758,6 +7817,7 @@ var EMITTED_PARAMS = [
   "pushCstLeaf",
   "pushCstChild",
   "rollbackTriviaAt",
+  "rollbackScannedTriviaAt",
   "failAt",
   "classHas",
   "consumeTrivia",
@@ -7901,19 +7961,24 @@ if(ctx._errors!==undefined&&ctx._errors.length!==${t2}er)ctx._errors.length=${t2
   if (!s.fd && !s.er) return `if(${t2}n)rollbackTriviaAt(ctx,${t2}raw,${t2}tl,${t2}lv,0,undefined,${t2}lg,${t2}rt)`;
   return `if(${t2}n){rollbackTriviaAt(ctx,${t2}raw,${t2}tl,${t2}lv,${s.fd ? `${t2}fd` : "0"},${s.er ? `${t2}er` : "undefined"},${t2}lg,${t2}rt)}`;
 }
-function emitTerm(callee, dst, t2, l, skip, s) {
+function emitTerm(callee, dst, t2, l, skip) {
   const fast = `const ${t2}v=${callee}(input,cur,ctx)
 if(${t2}v===FAIL)return FAIL
 ${dst}=${t2}v
 cur=EC.e`;
   if (l.tri === TRI_NONE) return fast;
-  const scanned = `${emitMark(t2, l.buf, s)}
+  const scanned = `const ${t2}tl=ctx._cstBuf!==undefined?(ctx._cstBuf.tl!==undefined?ctx._cstBuf.tl.length:0):(ctx._cstTriviaLog!==undefined?ctx._cstTriviaLog.length:0)
+const ${t2}lg=ctx._triviaLog!==undefined?ctx._triviaLog.length:0
+const ${t2}rt=ctx._rootTriviaLog!==undefined?ctx._rootTriviaLog.length:0
 const ${t2}s=${skip}(input,cur,ctx)
+const ${t2}stl=ctx._cstBuf!==undefined?(ctx._cstBuf.tl!==undefined?ctx._cstBuf.tl.length:0):(ctx._cstTriviaLog!==undefined?ctx._cstTriviaLog.length:0)
+const ${t2}slg=ctx._triviaLog!==undefined?ctx._triviaLog.length:0
+const ${t2}srt=ctx._rootTriviaLog!==undefined?ctx._rootTriviaLog.length:0
 const ${t2}v=${callee}(input,${t2}s,ctx)
 if(${t2}v===FAIL)return FAIL
 ${dst}=${t2}v
 if(EC.e>${t2}s)cur=EC.e
-else{${emitRollback(t2, l.buf, s)}}`;
+else{rollbackScannedTriviaAt(ctx,${t2}tl,${t2}stl,${t2}lg,${t2}slg,${t2}rt,${t2}srt)}`;
   if (l.tri !== TRI_UNKNOWN) return scanned;
   return `if(ctx.trivia===undefined){
 ${fast}
@@ -8076,7 +8141,7 @@ ${cfg.probe ? `failAt(ctx,${xf},pos)
     const op = code[ip];
     if (op === OP_GATE && (cfg.tolerant || cfg.probe)) return code[ip + 2];
     if (op === OP_RULE) return code[ip + 1];
-    if (op === OP_SCOPE && code[ip + 3] === 0 && code[ip + 1] >= 0 && labels.at(ip).tri === code[ip + 1]) {
+    if ((op === OP_SCOPE_PLAIN || op === OP_SCOPE && code[ip + 3] === 0) && code[ip + 1] >= 0 && labels.at(ip).tri === code[ip + 1]) {
       return code[ip + 2];
     }
     return void 0;
@@ -8244,10 +8309,11 @@ return r.value
 }`;
       }
       case OP_SCOPE:
-      case OP_SCOPE_CAP: {
+      case OP_SCOPE_CAP:
+      case OP_SCOPE_PLAIN: {
         const ki = code[ip + 1];
         const cap = op === OP_SCOPE_CAP;
-        const flags = code[ip + 3];
+        const flags = op === OP_SCOPE_PLAIN ? 0 : code[ip + 3];
         const child = link(code[ip + 2]);
         const scanFor = swapLegal && ki >= 0 && !triviaLabelled[ki] ? hoist("ts", `TRIVIASCAN[${ki}]`) : "null";
         const tri = ki < 0 ? "undefined" : hoist("tv", `TRIVIA[${ki}]`);
@@ -8401,7 +8467,7 @@ return v
           const vn = `v${i}`;
           parts.push(`let ${vn}`);
           if (REC) parts.push(pub(i));
-          parts.push(emitTerm(kids[i], vn, tmp(), L, skipFor(L), sinks));
+          parts.push(emitTerm(kids[i], vn, tmp(), L, skipFor(L)));
           names.push(vn);
         }
         parts.push("EC.e=cur");
@@ -8502,11 +8568,11 @@ return FAIL
         const other = otherIp >= 0 ? link(otherIp) : void 0;
         const otherRouted = code[ip + 4] === 1;
         const n = code[ip + 5];
+        validateDispatchSpec(spec, n, code[ip + 4]);
         const armBase = ip + 6;
         const arms = [];
         for (let i = 0; i < n; i++) arms.push(link(code[armBase + i]));
         const bk = hoist("bk", `DSP[${di}].byKey`);
-        const rt = hoist("rt", `DSP[${di}].routed`);
         const dx = hoist("dx", `DSP[${di}].expected`);
         const claims = (m) => {
           switch (m[0]) {
@@ -8530,6 +8596,15 @@ ${spec.match.map((m, i) => `${i === 0 ? "" : "else "}if(${claims(m)})arm=${m[3]}
 ` : "";
         const m1 = tmp();
         const m2 = tmp();
+        const routedCall = (target) => `{const savedRouted=ctx._routed
+${emitRollback(m1, L.buf, sinks)}
+${emitMark(m2, L.buf, sinks, false)}
+ctx._routed={value:key,span:{start:pos,end:selEnd}}
+try{v=${target}(input,pos,ctx)}finally{ctx._routed=savedRouted}
+break}`;
+        const plainCall = (target) => `v=${target}(input,selEnd,ctx);break`;
+        const armCases = arms.map((target, i) => `case ${i}:${spec.routed[i] === 1 ? routedCall(target) : plainCall(target)}`).join("\n");
+        const fallbackCase = other === void 0 ? "" : `default:${otherRouted ? routedCall(other) : plainCall(other)}`;
         return `${head}
 ${emitMark(m1, L.buf, sinks)}
 const sv=${selector}(input,pos,ctx)
@@ -8537,24 +8612,16 @@ if(sv===FAIL)return FAIL
 const selEnd=EC.e
 const key=sv
 let arm=${bk}.get(key)
-${fold}${chain}let ur
+${fold}${chain}
 if(arm===undefined){
-${other === void 0 ? `ctx._fe=selEnd;ctx._fx=${dx};return FAIL` : `ur=${String(otherRouted)}`}
-}else ur=${rt}[arm]===1
-const savedRouted=ctx._routed
-${emitMark(m2, L.buf, sinks)}
-if(ur){
-${emitRollback(m1, L.buf, sinks)}
-${emitMark(m2, L.buf, sinks, false)}
-ctx._routed={value:key,span:{start:pos,end:selEnd}}
+${other === void 0 ? `ctx._fe=selEnd;ctx._fx=${dx};return FAIL` : ""}
 }
-const at=ur?pos:selEnd
+${emitMark(m2, L.buf, sinks)}
 let v
 switch(arm){
-${arms.map((a, i) => `case ${i}:v=${a}(input,at,ctx);break`).join("\n")}
-${other === void 0 ? "" : `default:v=${other}(input,at,ctx)`}
+${armCases}
+${fallbackCase}
 }
-if(ur)ctx._routed=savedRouted
 if(v===FAIL){
 ${emitRollback(m2, L.buf, sinks)}
 ctx._fc=true
@@ -8879,7 +8946,7 @@ function committed(c) {
 }
 function stampRuleMap(prog, d, artifactMetadata = {}, resolved = resolveTable(prog)) {
   const triviaOfRule = (ip) => {
-    if (prog.code[ip] !== OP_SCOPE) return void 0;
+    if (prog.code[ip] !== OP_SCOPE_PLAIN && prog.code[ip] !== OP_SCOPE) return void 0;
     const slot = prog.code[ip + 1];
     return slot < 0 ? void 0 : resolved.trivia[slot];
   };
@@ -9093,20 +9160,19 @@ function assemble(t2, prog, cfg) {
       TERMV = v2;
       return EC.e > scanEnd2 ? EC.e : cur;
     }
-    const need = markCst(ctx);
-    const mRaw = MRAW;
-    const mTl = MTL;
-    const mLv = MLV;
-    const mFl = need ? ctx._fields?.length ?? 0 : 0;
-    const mEr = need ? ctx._errors?.length ?? 0 : 0;
+    const need = rollbackNeeded(ctx);
+    const mTl = need ? cstTlLen(ctx) : 0;
     const mLog = need ? ctx._triviaLog?.length ?? 0 : 0;
     const mRoot = need ? ctx._rootTriviaLog?.length ?? 0 : 0;
     const scanEnd = skipTrivia(input, cur, ctx);
+    const scanTl = need ? cstTlLen(ctx) : 0;
+    const scanLog = need ? ctx._triviaLog?.length ?? 0 : 0;
+    const scanRoot = need ? ctx._rootTriviaLog?.length ?? 0 : 0;
     const v = child(input, scanEnd, ctx);
     if (v === FAIL) return -1;
     TERMV = v;
     if (EC.e > scanEnd) return EC.e;
-    if (need) rollbackTriviaAt(ctx, mRaw, mTl, mLv, mFl, mEr, mLog, mRoot);
+    if (need) rollbackScannedTriviaAt(ctx, mTl, scanTl, mLog, scanLog, mRoot, scanRoot);
     return cur;
   }
   function committed2(c) {
@@ -9520,7 +9586,8 @@ function assemble(t2, prog, cfg) {
         };
       }
       case OP_SCOPE:
-      case OP_SCOPE_CAP: {
+      case OP_SCOPE_CAP:
+      case OP_SCOPE_PLAIN: {
         const ki = code[ip + 1];
         const scopeTrivia = ki < 0 ? void 0 : trivia2[ki];
         const scopeLabels = scopeTrivia?._meta.triviaKindLabels;
@@ -9554,7 +9621,7 @@ function assemble(t2, prog, cfg) {
           SCAN = savedScan;
           return v;
         };
-        return scopeRootPolicy(scopePiece, code[ip + 3]);
+        return scopeRootPolicy(scopePiece, code[ip] === OP_SCOPE_PLAIN ? 0 : code[ip + 3]);
       }
       case OP_ROUTED: {
         const fb = code[ip + 1];
@@ -10470,15 +10537,21 @@ function assemble(t2, prog, cfg) {
             target = arms[arm];
             usesRouted = routed2[arm] === 1;
           }
-          const savedRouted = ctx._routed;
           let mark = saveTriviaMark(ctx);
+          let v;
           if (usesRouted) {
+            const savedRouted = ctx._routed;
             rollbackTrivia(ctx, selectorMark);
             mark = saveTriviaMark(ctx);
             ctx._routed = { value: key, span: { start: pos, end: selEnd } };
+            try {
+              v = target(input, pos, ctx);
+            } finally {
+              ctx._routed = savedRouted;
+            }
+          } else {
+            v = target(input, selEnd, ctx);
           }
-          const v = target(input, usesRouted ? pos : selEnd, ctx);
-          if (usesRouted) ctx._routed = savedRouted;
           if (v === FAIL) {
             rollbackTrivia(ctx, mark);
             ctx._fc = true;
@@ -10624,6 +10697,11 @@ function assemble(t2, prog, cfg) {
       if (s.sentinel !== void 0) extraIps.push(s.sentinel[0]);
     }
     for (const set of prog.scanSkip ?? []) for (const r of set) extraIps.push(r[0]);
+    const validationRoots = [...Object.values(prog.rules), ...extraIps];
+    for (const ip of reachableSites(code, validationRoots)) {
+      if (code[ip] !== OP_DISPATCH) continue;
+      validateDispatchSpec(dsp[code[ip + 2]], code[ip + 5], code[ip + 4]);
+    }
     const pre = cfg.hostCaptureTrivia === void 0 ? prog.asm?.find((a) => a.key === cfgKey(cfg)) : void 0;
     if (pre !== void 0) {
       const pools = rebuildPools(t2.cc, t2.fx, pre.plan);
@@ -10650,6 +10728,7 @@ function assemble(t2, prog, cfg) {
         pushCstLeaf,
         pushCstChild,
         rollbackTriviaAt,
+        rollbackScannedTriviaAt,
         failAt,
         classHas,
         consumeTrivia,
@@ -10720,6 +10799,7 @@ function assemble(t2, prog, cfg) {
         pushCstLeaf,
         pushCstChild,
         rollbackTriviaAt,
+        rollbackScannedTriviaAt,
         failAt,
         classHas,
         consumeTrivia,
@@ -10952,6 +11032,14 @@ function tableRules(source, artifactMetadata = {}) {
 function jsString(s) {
   return JSON.stringify(s);
 }
+function emitClassPool(classes) {
+  const array = `[${classes.map(jsString).join(",")}]`;
+  if (classes.length < 2) return array;
+  const delimiter = "|/~?^`".split("").find((ch2) => classes.every((spec) => !spec.includes(ch2)));
+  if (delimiter === void 0) return array;
+  const split = `${jsString(classes.join(delimiter))}.split(${jsString(delimiter)})`;
+  return split.length < array.length ? split : array;
+}
 function emitConst(v) {
   if (v instanceof RegExp) return `/${v.source}/${v.flags}`;
   if (typeof v === "string") return jsString(v);
@@ -11030,7 +11118,7 @@ function programFields(prog, fns, opts = {}) {
     ...emitAssemblies(prog, opts.assemblies ?? []),
     `c:[${prog.code.join(",")}],`,
     `k:[${prog.k.map(emitConst).join(",")}],`,
-    `x:[${prog.cc.map(jsString).join(",")}],`,
+    `x:${emitClassPool(prog.cc)},`,
     `e:[${prog.fx.map((f) => `[${f.map(jsString).join(",")}]`).join(",")}],`,
     `d:[${prog.disp.map((a) => `[${a.join(",")}]`).join(",")}],`,
     `r:${JSON.stringify(prog.rules)},`,