@abelxiaoxing/cadence 1.8.3 → 2.0.0

package/src/worker-redaction.mjs

@@ -0,0 +1,608 @@
+import {
+  ContractError,
+  canonicalJson,
+  LIMITS,
+  utf8Bytes,
+} from "./worker-public.mjs";
+
+const marker = "[redacted]";
+const sensitive =
+  /^(?:authorization|proxy-authorization|cookie|set-cookie|password|passwd|secret|token|api[-_]?key|access[-_]?token|refresh[-_]?token|credentials|headers|env)$/i;
+/** Own data only. Deliberately never reads a getter, toJSON, exception or prototype method. */
+export function own(value, name) {
+  if (!value || typeof value !== "object") throw new ContractError();
+  const d = Object.getOwnPropertyDescriptor(value, name);
+  if (!d) return undefined;
+  if (!("value" in d)) throw new ContractError();
+  return d.value;
+}
+export function ownArray(value, maximum = LIMITS.dataFrames) {
+  if (!Array.isArray(value)) throw new ContractError();
+  const length = own(value, "length");
+  if (!Number.isSafeInteger(length) || length > maximum)
+    throw new ContractError("oversize");
+  return Array.from({ length }, (_, index) => own(value, String(index)));
+}
+function sourceBytes(value) {
+  let bytes = 0;
+  for (let i = 0; i < value.length; i++) {
+    const code = value.charCodeAt(i);
+    if (code < 0x80) bytes++;
+    else if (code < 0x800) bytes += 2;
+    else if (
+      code >= 0xd800 &&
+      code <= 0xdbff &&
+      value.charCodeAt(i + 1) >= 0xdc00 &&
+      value.charCodeAt(i + 1) <= 0xdfff
+    ) {
+      bytes += 4;
+      i++;
+    } else if (code >= 0xd800 && code <= 0xdfff) throw new ContractError();
+    else bytes += 3;
+  }
+  return bytes;
+}
+/** Charge before cloning. Used at synchronous source ingress as well as public tool values. */
+export function measureData(value, maximum = LIMITS.workerQueueBytes) {
+  let bytes = 0;
+  let nodes = 0;
+  const seen = new Set();
+  function visit(v, depth) {
+    if (++nodes > LIMITS.dataFrames || depth > 64)
+      throw new ContractError("oversize");
+    if (typeof v === "string") bytes += sourceBytes(v) * 6 + 2;
+    else if (v === null || typeof v === "boolean") bytes += 5;
+    else if (typeof v === "number" && Number.isFinite(v)) bytes += 32;
+    else if (v && typeof v === "object") {
+      if (
+        seen.has(v) ||
+        ![Object.prototype, Array.prototype, null].includes(
+          Object.getPrototypeOf(v),
+        )
+      )
+        throw new ContractError();
+      seen.add(v);
+      bytes += 2;
+      const names = Reflect.ownKeys(v);
+      if (names.length > LIMITS.dataFrames) throw new ContractError("oversize");
+      for (const key of names) {
+        if (Array.isArray(v) && key === "length") continue;
+        if (typeof key !== "string") throw new ContractError();
+        bytes += sourceBytes(key) * 6 + 4;
+        visit(own(v, key), depth + 1);
+      }
+      seen.delete(v);
+    } else throw new ContractError();
+    if (bytes > maximum) throw new ContractError("oversize");
+  }
+  visit(value, 0);
+  return bytes;
+}
+
+/** Streaming text sanitizer: literal prefixes, ANSI sequences and heuristic value
+ * states survive chunk boundaries. No truncation-to-success; bounded ambiguous
+ * URL authorities refuse. This is not universal business-secret detection. */
+export class TextSanitizer {
+  constructor(secrets = []) {
+    if (
+      secrets.some((s) => typeof s !== "string") ||
+      secrets.reduce((n, s) => n + utf8Bytes(s), 0) >
+        LIMITS.workerRequestBytes * 2
+    )
+      throw new ContractError("oversize");
+    this.secrets = [];
+    this.addSecrets(secrets);
+    this.pending = "";
+    this.heuristic = "";
+    this.escape = "";
+    this.skipping = false;
+    this.redacted = false;
+    this.ended = false;
+  }
+  /** Owner-only updates. Preserve all stream state and prior literals; bound the
+   * union before retaining any new entries. Never evict a retired secret. */
+  addSecrets(secrets) {
+    if (this.ended || !Array.isArray(secrets)) throw new ContractError();
+    let incoming = 0;
+    for (const s of secrets) {
+      if (typeof s !== "string") throw new ContractError();
+      incoming += utf8Bytes(s);
+      if (incoming > LIMITS.workerRequestBytes * 2)
+        throw new ContractError("oversize");
+    }
+    let bytes = this.secrets.reduce((n, s) => n + utf8Bytes(s), 0);
+    const added = new Set();
+    for (const s of secrets) {
+      if (!s || this.secrets.includes(s) || added.has(s)) continue;
+      bytes += utf8Bytes(s);
+      if (bytes > LIMITS.workerRequestBytes * 2)
+        throw new ContractError("oversize");
+      added.add(s);
+    }
+    this.secrets = this.secrets
+      .concat([...added])
+      .sort((a, b) => b.length - a.length);
+  }
+  get retainedBytes() {
+    return utf8Bytes(this.pending + this.heuristic + this.escape);
+  }
+  write(input, final = false) {
+    if (this.ended || typeof input !== "string") throw new ContractError();
+    let plain = "";
+    for (const ch of input) {
+      if (this.escape) {
+        this.escape += ch;
+        if (this.escape.length > 8192) throw new ContractError("oversize");
+        if (
+          (this.escape.startsWith("\x1b[") &&
+            this.escape.length > 2 &&
+            /[@-~]/.test(ch)) ||
+          (this.escape.startsWith("\x1b]") &&
+            (ch === "\x07" || this.escape.endsWith("\x1b\\"))) ||
+          (this.escape.length === 2 && !"[]".includes(ch))
+        )
+          this.escape = "";
+        this.redacted = true;
+      } else if (ch === "\x1b") this.escape = ch;
+      else if (
+        // biome-ignore lint/suspicious/noControlCharactersInRegex: strip terminal controls before literal matching
+        /[\x00-\x08\x0b-\x1f\x7f-\x9f\u202a-\u202e\u2066-\u2069]/.test(ch)
+      )
+        this.redacted = true;
+      else plain += ch;
+    }
+    this.pending += plain;
+    let literal = "";
+    let at = 0;
+    while (at < this.pending.length) {
+      const rest = this.pending.slice(at);
+      // A longer literal prefix takes precedence over a shorter full match.
+      if (
+        !final &&
+        this.secrets.some((s) => s.length > rest.length && s.startsWith(rest))
+      )
+        break;
+      const match = this.secrets.find((s) => rest.startsWith(s));
+      if (match) {
+        literal += marker;
+        at += match.length;
+        this.redacted = true;
+      } else {
+        const code = this.pending.charCodeAt(at);
+        if (
+          !final &&
+          at + 1 === this.pending.length &&
+          code >= 0xd800 &&
+          code <= 0xdbff
+        )
+          break;
+        const cp = this.pending.codePointAt(at);
+        if (cp >= 0xd800 && cp <= 0xdfff) throw new ContractError();
+        literal += String.fromCodePoint(cp);
+        at += cp > 0xffff ? 2 : 1;
+      }
+    }
+    this.pending = this.pending.slice(at);
+    this.heuristic += literal;
+    let output = "";
+    while (this.heuristic) {
+      if (this.skipping) {
+        const end = this.heuristic.search(/[\s,;"'<>]/);
+        if (end < 0) {
+          this.heuristic = "";
+          break;
+        }
+        this.heuristic = this.heuristic.slice(end);
+        this.skipping = false;
+      }
+      const match =
+        /\b(?:Bearer\s+|(?:api[-_]?key|access[-_]?token|refresh[-_]?token|password|passwd|secret|token|authorization)\s*[:=]\s*)/i.exec(
+          this.heuristic,
+        );
+      const url = /https?:\/\//i.exec(this.heuristic);
+      if (url && (!match || url.index < match.index)) {
+        output += this.heuristic.slice(0, url.index);
+        this.heuristic = this.heuristic.slice(url.index);
+        const start = url[0].length;
+        const end = this.heuristic.slice(start).search(/[/\s?#"'<>]/);
+        if (end < 0 && !final) {
+          if (utf8Bytes(this.heuristic) > 8192)
+            throw new ContractError("oversize");
+          break;
+        }
+        const size = end < 0 ? this.heuristic.length : start + end;
+        const authority = this.heuristic.slice(start, size);
+        const pos = authority.lastIndexOf("@");
+        output +=
+          this.heuristic.slice(0, start) +
+          (pos < 0 ? authority : `${marker}@${authority.slice(pos + 1)}`);
+        this.redacted ||= pos >= 0;
+        this.heuristic = this.heuristic.slice(size);
+      } else if (match) {
+        output += this.heuristic.slice(0, match.index) + match[0] + marker;
+        this.heuristic = this.heuristic.slice(match.index + match[0].length);
+        this.skipping = true;
+        this.redacted = true;
+      } else {
+        // Finite heuristic names plus delimiters. Potentially unbounded whitespace
+        // after a candidate is held/refused, never released as an exposed value.
+        const candidate =
+          /(?:Bearer|api[-_]?key|access[-_]?token|refresh[-_]?token|password|passwd|secret|token|authorization)\s*$/i.exec(
+            this.heuristic,
+          );
+        const keep = final
+          ? 0
+          : Math.max(
+              64,
+              candidate ? this.heuristic.length - candidate.index : 0,
+            );
+        if (keep > 8192) throw new ContractError("oversize");
+        let take = Math.max(0, this.heuristic.length - keep);
+        // The held heuristic suffix is measured in JS code units, but each
+        // emitted field/chunk must remain scalar UTF-8. Retain the whole pair
+        // when the suffix boundary would bisect a supplementary character.
+        const before = this.heuristic.charCodeAt(take - 1);
+        if (before >= 0xd800 && before <= 0xdbff) take--;
+        output += this.heuristic.slice(0, take);
+        this.heuristic = this.heuristic.slice(take);
+        break;
+      }
+    }
+    if (final) {
+      this.ended = true;
+      this.escape = "";
+    }
+    return output;
+  }
+  end() {
+    return this.write("", true);
+  }
+}
+export function sanitizeText(value, secrets = []) {
+  const sanitizer = new TextSanitizer(secrets);
+  return sanitizer.write(value, true);
+}
+/** Bounded own-data structural projection. Sensitive fields are never traversed.
+ * Only tool ordinary data may enter here; SDK/provider envelopes must be selected
+ * by the capture adapter instead of passed wholesale. */
+export function sanitizeData(value, secrets = []) {
+  measureData(value);
+  function visit(v) {
+    if (typeof v === "string") return sanitizeText(v, secrets);
+    if (v === null || typeof v !== "object") return v;
+    if (Array.isArray(v)) return ownArray(v).map(visit);
+    const result = Object.create(null);
+    for (const key of Object.keys(v).sort()) {
+      const safeKey = sanitizeText(key, secrets);
+      if (Object.hasOwn(result, safeKey)) throw new ContractError();
+      result[safeKey] = sensitive.test(key) ? marker : visit(own(v, key));
+    }
+    return result;
+  }
+  return canonicalJson(visit(value), LIMITS.workerQueueBytes);
+}
+
+/** Streaming JSON projection. Escapes are decoded before literal matching,
+ * sensitive values (including subtrees) are replaced without buffering them.
+ * Partial mode is only for genuinely unfinished SDK argument declarations. */
+export class DetailSanitizer {
+  constructor(encoding, secrets = [], partial = false) {
+    this.encoding = encoding;
+    this.secrets = secrets;
+    this.partial = partial;
+    this.limited = false;
+    this.text = new TextSanitizer(secrets);
+    this.stack = [];
+    this.root = "value";
+    this.inString = false;
+    this.escape = "";
+    this.key = "";
+    this.keyMode = false;
+    this.suppressed = false;
+    this.token = "";
+    this.redacted = false;
+  }
+  get retainedBytes() {
+    return (
+      this.text.retainedBytes +
+      utf8Bytes(this.escape + this.key + this.token) +
+      this.stack.reduce(
+        (sum, entry) =>
+          sum +
+          utf8Bytes(entry.lastKey ?? "") +
+          utf8Bytes(entry.lastOutputKey ?? ""),
+        0,
+      )
+    );
+  }
+  write(input, final = false) {
+    if (this.limited) return "";
+    try {
+      return this.project(input, final);
+    } catch (error) {
+      if (
+        !this.partial ||
+        (error instanceof ContractError && error.code === "oversize")
+      )
+        throw error;
+      // Malformed partial argument syntax is an observation, not executable
+      // canonical JSON. Preserve the safe prefix and explicitly omit the
+      // undecodable suffix; never guess a key/value boundary or leak escapes.
+      this.limited = true;
+      this.redacted = true;
+      return "[redacted: unsupported partial arguments]";
+    }
+  }
+  project(input, final = false) {
+    if (this.encoding === "utf8") {
+      const output = this.text.write(input, final);
+      this.redacted = this.text.redacted;
+      return output;
+    }
+    let output = "";
+    const encoded = (s) => JSON.stringify(s).slice(1, -1);
+    const top = () => this.stack.at(-1);
+    const done = () => {
+      if (top()) {
+        top().expect = "comma";
+        top().sensitive = false;
+        top().count++;
+      } else this.root = "done";
+    };
+    const begin = () => {
+      const parent = top();
+      if ((parent?.expect ?? this.root) !== "value") throw new ContractError();
+      const hide = !!parent?.hide || !!parent?.sensitive;
+      if (parent?.sensitive && !parent.hide) {
+        output += JSON.stringify(marker);
+        this.redacted = true;
+      }
+      return hide;
+    };
+    const primitive = () => {
+      if (
+        !/^(?:true|false|null|-?(?:0|[1-9][0-9]*)(?:\.[0-9]+)?(?:[eE][+-]?[0-9]+)?)$/.test(
+          this.token,
+        )
+      ) {
+        if (!this.partial) throw new ContractError();
+        // An invalid unquoted token may contain escaped secrets outside JSON
+        // strings. Do not pretend it is ordinary final data or decode guesses.
+        if (!this.suppressed) output += marker;
+        this.redacted = true;
+        this.limited = true;
+        this.token = "";
+        done();
+        return;
+      }
+      const value = JSON.parse(this.token);
+      if (typeof value === "number" && !Number.isFinite(value))
+        throw new ContractError();
+      if (!this.suppressed) output += this.token;
+      this.token = "";
+      done();
+    };
+    for (const ch of input) {
+      if (this.inString) {
+        let decoded = ch;
+        if (this.escape) {
+          this.escape += ch;
+          if (
+            this.escape === "\\u" ||
+            (this.escape.startsWith("\\u") && this.escape.length < 6)
+          )
+            continue;
+          decoded = JSON.parse(`"${this.escape}"`);
+          this.escape = "";
+        } else if (ch === "\\") {
+          this.escape = ch;
+          continue;
+        } else if (ch === '"') {
+          if (this.keyMode) {
+            const parent = top();
+            if (
+              !this.partial &&
+              parent.lastKey !== undefined &&
+              this.key <= parent.lastKey
+            )
+              throw new ContractError();
+            parent.lastKey = this.key;
+            parent.sensitive = sensitive.test(this.key);
+            parent.expect = "colon";
+            if (!parent.hide) {
+              const safeKey = sanitizeText(this.key, this.secrets);
+              if (
+                !this.partial &&
+                parent.lastOutputKey !== undefined &&
+                safeKey <= parent.lastOutputKey
+              )
+                throw new ContractError();
+              parent.lastOutputKey = safeKey;
+              this.redacted ||= safeKey !== this.key;
+              output += JSON.stringify(safeKey);
+            }
+            this.key = "";
+          } else {
+            if (!this.suppressed) output += `${encoded(this.text.end())}"`;
+            this.redacted ||= this.text.redacted;
+            done();
+          }
+          this.inString = false;
+          continue;
+        }
+        if (this.keyMode) {
+          this.key += decoded;
+          if (utf8Bytes(this.key) > 8192) throw new ContractError("oversize");
+        } else if (!this.suppressed)
+          output += encoded(this.text.write(decoded));
+        continue;
+      }
+      if (this.token && /[\s,\]}]/.test(ch)) primitive();
+      if (this.token) {
+        this.token += ch;
+        if (this.token.length > 128) throw new ContractError("oversize");
+        continue;
+      }
+      if (/\s/.test(ch)) continue;
+      const parent = top();
+      if (ch === '"') {
+        this.keyMode = parent?.kind === "object" && parent.expect === "key";
+        this.suppressed = this.keyMode ? parent.hide : begin();
+        this.inString = true;
+        this.text = new TextSanitizer(this.secrets);
+        if (!this.keyMode && !this.suppressed) output += '"';
+      } else if (ch === "{" || ch === "[") {
+        const hide = begin();
+        if (!hide) output += ch;
+        if (this.stack.length >= 64) throw new ContractError("oversize");
+        this.stack.push({
+          kind: ch === "{" ? "object" : "array",
+          expect: ch === "{" ? "key" : "value",
+          hide,
+          sensitive: false,
+          count: 0,
+          lastKey: undefined,
+          lastOutputKey: undefined,
+        });
+      } else if (ch === "}" || ch === "]") {
+        if (!parent || (ch === "}") !== (parent.kind === "object"))
+          throw new ContractError();
+        if (
+          parent.expect !== "comma" &&
+          !(
+            parent.count === 0 &&
+            parent.expect === (parent.kind === "object" ? "key" : "value")
+          )
+        )
+          throw new ContractError();
+        if (!parent.hide) output += ch;
+        this.stack.pop();
+        done();
+      } else if (ch === ":") {
+        if (parent?.expect !== "colon") throw new ContractError();
+        parent.expect = "value";
+        if (!parent.hide) output += ch;
+      } else if (ch === ",") {
+        if (parent?.expect !== "comma") throw new ContractError();
+        parent.expect = parent.kind === "object" ? "key" : "value";
+        parent.count = Math.max(1, parent.count);
+        if (!parent.hide) output += ch;
+      } else {
+        this.suppressed = begin();
+        this.token = ch;
+      }
+    }
+    if (final) {
+      if (this.token) primitive();
+      if (this.partial) {
+        // This field is UTF-8 partial source, not canonical final arguments.
+        // Retain unfinished strings/keys without manufacturing closing syntax.
+        if (this.inString && !this.suppressed) {
+          output += this.keyMode
+            ? JSON.stringify(sanitizeText(this.key, this.secrets)).slice(0, -1)
+            : encoded(this.text.end());
+        }
+        if (this.escape) {
+          output += marker;
+          this.redacted = true;
+          this.limited = true;
+        }
+      } else if (
+        this.inString ||
+        this.escape ||
+        this.stack.length ||
+        this.root !== "done"
+      )
+        throw new ContractError();
+      this.redacted ||= this.text.redacted;
+    }
+    return output;
+  }
+  end() {
+    return this.write("", true);
+  }
+}
+
+/** Canonical ordinary JSON without materializing a whole large tool body.
+ * The caller holds the TextSanitizer prefix/expansion allowance. Traversal tables,
+ * sanitized keys and source/output slices are charged BEFORE copying, and held
+ * across each yield (including the consumer's original awaited writes). Iterator
+ * close releases only these no-longer-used copies, never an emitter write lease. */
+export function* sanitizedJsonChunks(value, secrets, reserve) {
+  measureData(value, LIMITS.detailBytes * 8);
+  function* visit(v) {
+    if (typeof v === "string") {
+      yield '"';
+      const sanitizer = new TextSanitizer(secrets);
+      for (let at = 0; at < v.length; at += 1024) {
+        // At most three UTF-8 bytes per UTF-16 unit, with conservative literal
+        // replacement/JSON escaping expansion. Reserve before slice/projection.
+        const slice = reserve(Math.min(1024, v.length - at) * 36 + 64);
+        try {
+          const clean = sanitizer.write(v.slice(at, at + 1024));
+          yield JSON.stringify(clean).slice(1, -1);
+        } finally {
+          slice.release();
+        }
+      }
+      // Remaining known-literal/heuristic prefixes and their expansion remain
+      // covered by the caller's prefix lease until this final yield settles.
+      yield JSON.stringify(sanitizer.end()).slice(1, -1);
+      yield '"';
+    } else if (v === null || typeof v !== "object") yield JSON.stringify(v);
+    else if (Array.isArray(v)) {
+      yield "[";
+      const length = own(v, "length");
+      for (let index = 0; index < length; index++) {
+        if (index) yield ",";
+        yield* visit(own(v, String(index)));
+      }
+      yield "]";
+    } else {
+      const table = reserve(64);
+      let previous;
+      try {
+        // No Object.keys/sort copy before reservation. These are validated own
+        // data objects; key references and their UTF-16 storage are conservative
+        // traversal charges, not a claim about all SDK/V8 memory.
+        const keys = [];
+        for (const key in v) {
+          if (!Object.hasOwn(v, key)) continue;
+          const bytes = sourceBytes(key);
+          // Same completed-key bound as the owner/partial lexical projection.
+          if (bytes > 8192) throw new ContractError("oversize");
+          table.grow(bytes * 2 + 16);
+          keys.push(key);
+        }
+        keys.sort();
+        yield "{";
+        let lastKey;
+        for (let index = 0; index < keys.length; index++) {
+          const key = keys[index];
+          // Whole-key sanitization, escaped output and retained comparison key.
+          // Keep the old comparison lease until the replacement is allocated.
+          const current = reserve(sourceBytes(key) * 24 + 64);
+          try {
+            const safe = sanitizeText(key, secrets);
+            if (lastKey !== undefined && safe <= lastKey)
+              throw new ContractError();
+            lastKey = safe;
+            previous?.release();
+            previous = current;
+            if (index) yield ",";
+            yield `${JSON.stringify(safe)}:`;
+            if (sensitive.test(key)) yield JSON.stringify(marker);
+            else yield* visit(own(v, key));
+          } finally {
+            if (previous !== current) current.release();
+          }
+        }
+        yield "}";
+      } finally {
+        previous?.release();
+        table.release();
+      }
+    }
+  }
+  yield* visit(value);
+}