@poe-platform/safe-bash 0.1.511 → 0.1.512

package/dist/safe-bash/core.browser.js

@@ -1482,7 +1482,7 @@ var Parser2 = class {
           await this.ledger.checkpoint(this.signal);
           members[code] = !members[code];
         }
-        return this.node(() => ({ kind: "set", members: Object.freeze(members), nullable: false, captured: false }));
+        return this.node(() => ({ kind: "set", members: Object.freeze(members), nonAscii: negate, nullable: false, captured: false }));
       }
       first = false;
       if (this.at("[") && (this.at(".", this.offset + 1) || this.at("=", this.offset + 1))) {
@@ -1543,6 +1543,28 @@ function resolveEreProgram(program, ledger) {
 
 // packages/safe-bash/src/commands/regex-execution/ere/matcher.ts
 init_platform();
+
+// packages/safe-bash/src/commands/regex-execution/utf8.ts
+init_platform();
+function invalidUtf8() {
+  throw new PublicDiagnostic("bounded regex unsupported: literal patterns and subjects require valid non-NUL UTF-8 bytes");
+}
+async function validateUtf8(input3, ledger, signal) {
+  for (let index = 0; index < input3.length; ) {
+    const first = typeof input3 === "string" ? input3.charCodeAt(index) : input3[index];
+    const width = first < 128 ? 1 : first >= 194 && first <= 223 ? 2 : first >= 224 && first <= 239 ? 3 : first >= 240 && first <= 244 ? 4 : 0;
+    ledger.charge("work", Math.max(width, 1), signal);
+    if (first === 0 || width === 0 || width > input3.length - index) invalidUtf8();
+    for (let continuation = 1; continuation < width; continuation++) {
+      const byte = typeof input3 === "string" ? input3.charCodeAt(index + continuation) : input3[index + continuation];
+      if (byte < 128 || byte > 191 || continuation === 1 && (first === 224 && byte < 160 || first === 237 && byte > 159 || first === 240 && byte < 144 || first === 244 && byte > 143)) invalidUtf8();
+    }
+    index += width;
+    await ledger.checkpoint(signal);
+  }
+}
+
+// packages/safe-bash/src/commands/regex-execution/ere/matcher.ts
 function spanOrder(left, right) {
   if (left === null) return right === null ? 0 : -1;
   if (right === null) return 1;
@@ -1621,15 +1643,47 @@ async function matchEre(program, subject, ledger, signal) {
   await admitAscii(subject, ledger, signal);
   return runMatcher(program, subject, ledger, signal, 0, true);
 }
-async function createEreSpanMatcher(program, subject, ledger, signal) {
+async function prepareUtf8EreSubject(bytes3, ledger, signal) {
   ledger.check(signal);
-  resolveEreProgram(program, ledger);
-  ledger.admitInput("subjectBytes", subject.length, signal);
-  await admitAscii(subject, ledger, signal);
-  ledger.charge("allocationUnits", 2, signal);
-  return async (start) => {
-    if (!Number.isSafeInteger(start) || start < 0 || start > subject.length) throw new RangeError("Invalid ERE search cursor");
-    return runMatcher(program, subject, ledger, signal, start, false);
+  ledger.admitInput("subjectBytes", bytes3.length, signal);
+  ledger.charge("allocationUnits", bytes3.length * 11 + 16, signal);
+  ledger.charge("work", bytes3.length, signal);
+  const owned = new Uint8Array(bytes3);
+  await validateUtf8(owned, ledger, signal);
+  const characters = [];
+  const offsets = [];
+  for (let offset = 0; offset < owned.length; ) {
+    const first = owned[offset];
+    const width = first < 128 ? 1 : first < 224 ? 2 : first < 240 ? 3 : 4;
+    ledger.charge("work", width, signal);
+    await ledger.checkpoint(signal);
+    offsets.push(offset);
+    characters.push(String.fromCharCode(first < 128 ? first : 128));
+    offset += width;
+  }
+  offsets.push(owned.length);
+  ledger.charge("work", characters.length, signal);
+  await ledger.checkpoint(signal);
+  const subject = characters.join("");
+  return (program) => {
+    resolveEreProgram(program, ledger);
+    ledger.charge("allocationUnits", 2, signal);
+    return async (start) => {
+      if (!Number.isSafeInteger(start) || start < 0 || start > owned.length) throw new RangeError("Invalid UTF-8 ERE search cursor");
+      let lower = 0, upper = offsets.length - 1;
+      while (lower < upper) {
+        ledger.charge("work", 1, signal);
+        await ledger.checkpoint(signal);
+        const middle = Math.floor((lower + upper) / 2);
+        if (offsets[middle] < start) lower = middle + 1;
+        else upper = middle;
+      }
+      if (offsets[lower] !== start) throw new RangeError("UTF-8 ERE cursor must be a scalar boundary");
+      const span = await runMatcher(program, subject, ledger, signal, lower, false);
+      if (!span) return void 0;
+      ledger.charge("allocationUnits", 2, signal);
+      return Object.freeze({ start: offsets[span.start], end: offsets[span.end] });
+    };
   };
 }
 async function runMatcher(program, subject, ledger, signal, from, materialize) {
@@ -1700,7 +1754,7 @@ async function runMatcher(program, subject, ledger, signal, from, materialize) {
         case "literal":
         case "set": {
           const code = subject.charCodeAt(state.position);
-          if (state.position < subject.length && (node.kind === "dot" || node.kind === "literal" && node.code === code || node.kind === "set" && node.members[code])) {
+          if (state.position < subject.length && (node.kind === "dot" || node.kind === "literal" && node.code === code || node.kind === "set" && (code < 128 ? node.members[code] : node.nonAscii))) {
             push(state.position + 1, current.next, state.captures, state.histories);
           }
           break;
@@ -1980,20 +2034,6 @@ async function admitBre(pattern, ledger, signal) {
     }
   }
 }
-async function validateUtf8(input3, ledger, signal) {
-  for (let index = 0; index < input3.length; ) {
-    const first = typeof input3 === "string" ? input3.charCodeAt(index) : input3[index];
-    const width = first < 128 ? 1 : first >= 194 && first <= 223 ? 2 : first >= 224 && first <= 239 ? 3 : first >= 240 && first <= 244 ? 4 : 0;
-    ledger.charge("work", Math.max(width, 1), signal);
-    if (first === 0 || width === 0 || width > input3.length - index) fail("unsupported", "literal patterns and subjects require valid non-NUL UTF-8 bytes");
-    for (let continuation = 1; continuation < width; continuation++) {
-      const byte = typeof input3 === "string" ? input3.charCodeAt(index + continuation) : input3[index + continuation];
-      if (byte < 128 || byte > 191 || continuation === 1 && (first === 224 && byte < 160 || first === 237 && byte > 159 || first === 240 && byte < 144 || first === 244 && byte > 143)) fail("unsupported", "literal patterns and subjects require valid non-NUL UTF-8 bytes");
-    }
-    index += width;
-    await ledger.checkpoint(signal);
-  }
-}
 async function literalBytes(pattern, selected, ledger, signal) {
   const { kind } = selected;
   let length = 0;
@@ -2154,27 +2194,18 @@ async function execute(input3, signal) {
   const results = [];
   const usage2 = { count: 0 };
   for (const row of rows2) {
-    ledger.charge("allocationUnits", row.bytes.length * 2 + 2, signal);
-    const characters = [];
-    for (const byte of row.bytes) {
-      ledger.charge("work", 1, signal);
-      if (byte === 0 || byte > 127) fail("unsupported", "subjects require non-NUL ASCII; Unicode and invalid UTF-8 are unsupported");
-      characters.push(String.fromCharCode(byte));
-      await ledger.checkpoint(signal);
-    }
-    const subject = characters.join("");
+    const subject = await prepareUtf8EreSubject(row.bytes, ledger, signal);
     if (row.all) {
       ledger.charge("allocationUnits", programs2.length + 1, signal);
       const finders = [];
-      for (const program of programs2) finders.push(await createEreSpanMatcher(program, subject, ledger, signal));
+      for (const program of programs2) finders.push(subject(program));
       results.push(await enumerate(input3, row, finders, usage2, signal));
       continue;
     }
     let span;
     for (const program of programs2) {
-      const match = await matchEre(program, subject, ledger, signal);
-      if (!match.matched) continue;
-      const candidate = match.captures[0];
+      const candidate = await subject(program)(0);
+      if (!candidate) continue;
       if (selected.kind === "grep") {
         span = candidate;
         break;