redac---n 0.0.4 → 0.0.5

package/dist/transcript.js

@@ -1,10 +1,10 @@
 /**
  * Redacting a whole transcript or request body: any JSON value, typically a list of chat messages in
- * the Anthropic or OpenAI format. Two passes:
+ * the Anthropic, OpenAI or Gemini format. Two passes:
  *
  * 1. Every string is scanned, with the command or path of the tool call that produced it as a hint
- *    (a `tool_result` is matched to its `tool_use` by id), and every confident finding is learned as a
- *    known value.
+ *    (a `tool_result` is matched to its `tool_use` by id, a Gemini `functionResponse` to its `functionCall` too), and
+ *    every confident finding is learned as a known value.
  * 2. Every string is redacted, so a secret found once (`API_KEY=…` in a `cat .env`) is also hidden where
  *    it appears with no context at all (the model echoing it, a later `curl` using it).
  *
@@ -24,18 +24,24 @@ import { isLearnable, isNeverSecretShape, loginPassword, merge, secretValue } fr
 import { occurrences } from "./repeats.js";
 import { isCredentialName } from "./names.js";
 import { isNamedSecret } from "./named.js";
+import { KnownValues } from "./known.js";
 /** Keys whose values are IDs the provider minted (`toolu_01Ab…`, `chatcmpl-B9…`): one token, no spaces. */
-const ID_KEYS = new Set(["id", "tool_use_id", "tool_call_id", "call_id"]);
+const ID_KEYS = new Set(["id", "tool_use_id", "tool_call_id", "call_id", "file_id"]);
 /** Keys whose values are opaque blobs the provider signed or encrypted, which no one reads. */
 const OPAQUE_KEYS = new Set(["signature", "thoughtSignature", "thought_signature", "encrypted_content"]);
-/** Keys whose values are a short enum or a model name (`assistant`, `end_turn`, `image/png`, `claude-sonnet-4-5-20250929`). */
+/**
+ * Keys whose values are a short enum, a model's or a tool's name (`assistant`, `end_turn`, `image/png`,
+ * `claude-sonnet-4-5-20250929`, `run_shell_command`).
+ */
 const ENUM_KEYS = new Set([
     "type",
     "role",
     "model",
+    "name",
     "stop_reason",
     "finish_reason",
     "media_type",
+    "mimeType",
     "object",
     "status",
     "mime_type",
@@ -111,8 +117,17 @@ function isEnvelope(o) {
         (typeof o.type === "string" && BLOCK_TYPES.has(o.type)) ||
         typeof o.tool_call_id === "string" ||
         typeof o.tool_use_id === "string" ||
+        // (OpenAI Responses' items of a tool call and its output, whatever the tool: `shell_call`, `custom_tool_call_output`.)
+        typeof o.call_id === "string" ||
         (typeof o.object === "string" && typeof o.model === "string"));
 }
+/**
+ * Keys of an envelope's object that hold its blocks: a tool result's `content` (MCP's `resource` and `resource_link`
+ * blocks, a server tool's `bash_code_execution_result`), a document's `source`, an MCP resource block's `resource`, a
+ * Responses output's `output`. An object there is a block too when it has a `type` of an enum's shape, whatever the type
+ * (an allow-list of them falls behind every new tool), and so is an MCP `resource`, which has none.
+ */
+const BLOCK_KEYS = new Set(["content", "source", "resource", "output"]);
 /** Keys of tool inputs that hold what a tool writes to the file at its path (`Write`, `create_file`, `str_replace`). */
 const CONTENT_KEYS = new Set(TOOL_CONTENT_KEYS);
 /** Files a patch adds or changes: `*** Add File: x` (apply_patch), `+++ b/x` (unified diff). */
@@ -171,7 +186,8 @@ function toolHints(value) {
         const o = v;
         // Anthropic: {type: "tool_use", id, input: {...}}. OpenAI: {id, type: "function", function: {arguments: "<json>"}}.
         // OpenAI Responses: {type: "function_call", call_id, arguments}; {type: "local_shell_call", call_id, action: {command: [...]}}.
-        let input = o.input ?? o.action ?? o.arguments;
+        // Gemini: {functionCall: {id, name, args: {...}}} (the id is the client's: Gemini CLI and pydantic-ai send one).
+        let input = o.input ?? o.action ?? o.arguments ?? o.args;
         const fn = o.function && typeof o.function === "object" ? o.function.arguments : undefined;
         if (fn !== undefined && fn !== null)
             input = fn;
@@ -235,11 +251,13 @@ function encodeBase64Text(text, like) {
  */
 function base64Text(o, key, x) {
     const uri = BASE64_URI.exec(x);
-    const textual = uri
-        ? TEXT_MEDIA_TYPE.test(uri[1])
-        : key === "data" && typeof o.media_type === "string" && TEXT_MEDIA_TYPE.test(o.media_type.split(";")[0].trim());
+    const type = uri
+        ? uri[1]
+        : key === "data" && typeof o.media_type === "string"
+            ? o.media_type.split(";")[0].trim()
+            : "";
     const payload = uri ? uri[3] : x;
-    const text = textual ? decodeBase64Text(payload) : undefined;
+    const text = TEXT_MEDIA_TYPE.test(type) ? decodeBase64Text(payload) : undefined;
     return text === undefined
         ? undefined
         : { text, encode: (redacted) => (uri ? `data:${uri[1]}${uri[2]}` : "") + encodeBase64Text(redacted, payload) };
@@ -303,6 +321,7 @@ function collect(value, hints, keepThinking) {
                 hint,
                 key,
                 parent,
+                lone: scope.lone,
             });
             return;
         }
@@ -311,11 +330,17 @@ function collect(value, hints, keepThinking) {
         const o = v;
         if (keepThinking && (o.type === "thinking" || o.type === "redacted_thinking"))
             return "skip";
-        // A tool result inherits the hint of its call; OpenAI tool messages carry `tool_call_id`.
+        // A tool result inherits the hint of its call; OpenAI tool messages carry `tool_call_id`, a Responses output
+        // (`…_call_output`) its call's `call_id`.
         const id = (o.tool_use_id ??
             o.tool_call_id ??
-            (o.type === "function_call_output" || o.type === "local_shell_call_output" ? o.call_id : undefined));
-        const hint = typeof id === "string" && hints.has(id) ? hints.get(id) : scope.hint;
+            (typeof o.type === "string" && o.type.endsWith("_call_output") ? o.call_id : undefined));
+        let hint = typeof id === "string" && hints.has(id) ? hints.get(id) : scope.hint;
+        // Gemini's `functionResponse`: its call's, by id. (One with no id gets none: the clients sending these requests,
+        // Gemini CLI and pydantic-ai, give each call an id.)
+        const gemini = key === "functionCall" || key === "functionResponse";
+        if (key === "functionResponse" && typeof o.id === "string" && hints.has(o.id))
+            hint = hints.get(o.id);
         // Base64 media: `{type: "base64", media_type, data}`, `image_url: {url: "data:…"}`.
         const media = o.type === "base64" ||
             o.type === "image" ||
@@ -324,7 +349,18 @@ function collect(value, hints, keepThinking) {
             o.type === "document";
         // A tool input that writes a file: its content is judged as that file (`.env` content as a secret file).
         const writes = TOOL_PATH_KEYS.map((k) => o[k]).find((x) => typeof x === "string");
-        const envelope = !scope.free && isEnvelope(o);
+        const envelope = !scope.free &&
+            (isEnvelope(o) ||
+                gemini ||
+                (!!scope.block && (typeof o.type === "string" ? ENUM_VALUE.test(o.type) : key === "resource")));
+        // (An object's `input`, `arguments` or `action` is a tool's input when the object is a tool call, which has a name or
+        // a type: a Responses request's `input` is its list of messages.)
+        const call = typeof o.name === "string" || typeof o.type === "string";
+        // (How many members are text that isn't structure: one alone is the envelope's text, a message's or a block's.)
+        const texts = envelope
+            ? Object.entries(o).filter(([k, x]) => (typeof x === "string" || typeof x === "number") && !isStructure(k, x))
+                .length
+            : Infinity;
         // (An attached file's name: Anthropic's `document` block, OpenAI's `input_file` part and Chat Completions' `file`.)
         const name = !hint.path && (o.type === "document" || o.type === "input_file" || "file_data" in o)
             ? documentName(o)
@@ -362,7 +398,13 @@ function collect(value, hints, keepThinking) {
             if ((media && k === "data" && typeof x === "string" && BASE64_VALUE.test(x.replace(/\s/g, ""))) ||
                 (typeof x === "string" && MEDIA_URI.test(x)))
                 return undefined;
-            return { hint: childHint, free: scope.free || TOOL_INPUT_KEYS.has(k) };
+            return {
+                hint: childHint,
+                free: scope.free || (call && TOOL_INPUT_KEYS.has(k)),
+                // (Down through what a block holds: a shell output's `{stdout, outcome: {type: "exit"}}`.)
+                block: envelope ? BLOCK_KEYS.has(k) : scope.block,
+                lone: texts <= 1 && !isCredentialName(k),
+            };
         };
     });
     return leaves;
@@ -489,26 +531,31 @@ export function redactValue(redactor, value, options = {}) {
     const inParents = new Map();
     const found = (leaf) => {
         const alone = redactor.detect(leaf.text, { hint: leaf.hint, guessing: options.guessing });
-        const ctx = leaf.hint.command === undefined && leaf.hint.path === undefined ? leafContext(leaf, serialized) : undefined;
+        const ctx = !leaf.lone && leaf.hint.command === undefined && leaf.hint.path === undefined
+            ? leafContext(leaf, serialized)
+            : undefined;
         if (!ctx)
             return alone;
         let inParent = inParents.get(ctx.text);
         if (!inParent)
             inParents.set(ctx.text, (inParent = redactor.detect(ctx.text, { guessing: options.guessing })));
-        const end = ctx.at + leaf.text.length;
+        const end = ctx.at + ctx.length;
         const mine = inParent
             .filter((f) => f.start >= ctx.at && f.end <= end)
-            .map((f) => ({ ...f, start: f.start - ctx.at, end: f.end - ctx.at }));
+            .map((f) => ({ ...f, start: ctx.toLeaf(f.start - ctx.at, false), end: ctx.toLeaf(f.end - ctx.at, true) }));
         return mine.length ? merge([...alone, ...mine]) : alone;
     };
     const judged = (leaf) => {
-        const ctx = leafContext(leaf, serialized);
+        // (Placed once a finding is to be judged: most leaves have none.)
+        let ctx = null;
         // (The key names only a finding that is the whole leaf; inside a longer text, the name is the text's own.)
         return (f) => {
+            if (ctx === null)
+                ctx = leafContext(leaf, serialized);
             const name = f.name ?? (f.start === 0 && f.end === leaf.text.length ? leaf.key : undefined);
             const g = name ? { ...f, name } : f;
             return ctx
-                ? redactor.judge(ctx.text, { ...g, start: f.start + ctx.at, end: f.end + ctx.at })
+                ? redactor.judge(ctx.text, { ...g, start: ctx.at + ctx.toContext(f.start), end: ctx.at + ctx.toContext(f.end) })
                 : redactor.judge(leaf.text, g);
         };
     };
@@ -516,11 +563,14 @@ export function redactValue(redactor, value, options = {}) {
     // this value's other leaves only.
     // (Learned all at once after the pass: adding values rebuilds the short-value matcher, once per batch.)
     const learned = [];
+    const first = [];
     for (const leaf of leaves) {
         const keep = judged(leaf);
         if (isNamedKeySecret(leaf, keep) && !isNeverSecretShape(leaf.text))
             learned.push([leaf.text, leaf.key]);
-        for (const f of found(leaf).filter(keep)) {
+        const here = found(leaf);
+        first.push(here);
+        for (const f of here.filter(keep)) {
             const secret = secretValue(leaf.text, f);
             if (!isLearnable(f, secret))
                 continue;
@@ -535,12 +585,20 @@ export function redactValue(redactor, value, options = {}) {
         for (const [value, label] of learned)
             redactor.addKnown(value, label);
     const seen = remember ? undefined : new Map(learned.map(([value, kind]) => [value, { kind, value }]));
+    // (A leaf is detected again only where a value learned here turns up in it, in any form: elsewhere, what pass 1 found
+    // is what detection finds now.)
+    let fresh;
+    if (remember && learned.length) {
+        fresh = new KnownValues();
+        for (const [value, label] of learned)
+            fresh.add(value, label);
+    }
     // Pass 2: redact, into one deep copy of the value.
     let out = leaves.length ? cloneJson(value) : value;
     const findings = [];
     // (Object keys hidden: the key's text, and what it is replaced by.)
     const renames = new Map();
-    for (const leaf of leaves) {
+    for (const [i, leaf] of leaves.entries()) {
         const keep = judged(leaf);
         if (isNamedKeySecret(leaf, keep)) {
             findings.push({
@@ -555,7 +613,7 @@ export function redactValue(redactor, value, options = {}) {
             out = setAt(out, leaf.path, redactor.placeholderFor({ start: 0, end: leaf.text.length, kind: leaf.key, rule: "transcript:named-key", confidence: 0.95 }, leaf.text));
             continue;
         }
-        const here = found(leaf).filter(keep);
+        const here = (fresh?.find(leaf.text).length ? found(leaf) : first[i]).filter(keep);
         const r = redactor.apply(leaf.text, seen?.size ? merge([...here, ...occurrences(leaf.text, seen, here)]) : here);
         if (!r.findings.length)
             continue;
@@ -582,14 +640,42 @@ export function redactValue(redactor, value, options = {}) {
         renameKeys(out, renames);
     return { value: out, findings };
 }
+/** A leaf's text placed at `at` in its context, as JSON writes it (`written`, without its quotes). */
+function placed(text, context, at, written) {
+    if (written.length === text.length)
+        return { text: context, at, length: text.length, toContext: (i) => i, toLeaf: (i) => i };
+    // (Where each character is written: a code point as JSON writes it, a valid surrogate pair as it is, a lone one escaped.)
+    const starts = [];
+    let w = 0;
+    for (const ch of text) {
+        starts.push(w);
+        if (ch.length === 2)
+            starts.push(w + 1);
+        w += JSON.stringify(ch).length - 2;
+    }
+    starts.push(w);
+    return {
+        text: context,
+        at,
+        length: written.length,
+        toContext: (i) => starts[i],
+        // (The last leaf position written at or before `i`; with `after`, the first at or after it.)
+        toLeaf: (i, after) => {
+            const j = starts.findLastIndex((s) => s <= i);
+            return after && starts[j] < i ? j + 1 : j;
+        },
+    };
+}
 /**
  * The leaf's object serialized, or its list of strings (an argv: `["docker", "login", "-p", "…"]`), and where the leaf's
- * text starts in it; undefined when it can't be placed exactly.
+ * text stands in it; undefined when it can't be placed exactly.
  */
 function leafContext(leaf, serialized) {
     const parent = leaf.parent;
-    if (!parent || JSON.stringify(leaf.text) !== `"${leaf.text}"`)
+    if (!parent)
         return undefined;
+    const literal = JSON.stringify(leaf.text);
+    const written = literal.slice(1, -1);
     if (Array.isArray(parent)) {
         const index = leaf.path.at(-1);
         if (typeof index !== "number" || parent.length > 200 || !parent.every((x) => typeof x === "string"))
@@ -598,7 +684,9 @@ function leafContext(leaf, serialized) {
             serialized.set(parent, JSON.stringify(parent));
         const text = serialized.get(parent);
         const at = 1 + parent.slice(0, index).reduce((n, x) => n + JSON.stringify(x).length + 1, 0) + 1;
-        return text.length <= 20_000 && text.slice(at, at + leaf.text.length) === leaf.text ? { text, at } : undefined;
+        return text.length <= 20_000 && text.slice(at, at + written.length) === written
+            ? placed(leaf.text, text, at, written)
+            : undefined;
     }
     if (leaf.key === undefined)
         return undefined;
@@ -623,9 +711,9 @@ function leafContext(leaf, serialized) {
     const text = serialized.get(parent);
     if (text === undefined)
         return undefined;
-    const needle = `${JSON.stringify(leaf.key)}:"${leaf.text}"`;
+    const needle = `${JSON.stringify(leaf.key)}:${literal}`;
     const i = text.indexOf(needle);
-    return i < 0 ? undefined : { text, at: i + needle.length - 1 - leaf.text.length };
+    return i < 0 ? undefined : placed(leaf.text, text, i + needle.length - 1 - written.length, written);
 }
 /**
  * The `arguments` or `output` JSON string `original` with the scalars that redaction changed (`redacted` is the value it parsed

package/dist/unnamed.js

@@ -16,7 +16,7 @@ import { base64Binary, base64Bytes } from "./base64.js";
 import { naturalness } from "./charlm.js";
 import { contextOf, Doc } from "./context.js";
 import { isCredentialName, isVendorClient, mentionsCredential } from "./names.js";
-import { isPlaceholder } from "./public.js";
+import { isPlaceholder, isTruncated } from "./public.js";
 import { alphabetOf, entropyZ, shape } from "./random.js";
 import { trailingRun } from "./lines.js";
 import { ENCRYPTION_WRAPPER, ANALYTICS_COOKIE, isIdentifierShaped, isInert, isPodName, PASSWORD_HASH_IN_TEXT, isWordsValue, readsAsWords, } from "./values.js";
@@ -36,7 +36,7 @@ const API_OBJECT_ID = /^(?:(?:vpc|subnet|sg|i|ami|vol|snap|eni|igw|rtb|acl|nat|e
  * Prefixes of payment, messaging and SaaS object IDs (Stripe `cus_`, `pi_`, `ch_`, Twilio, Svix…). Some
  * are also prefixes of keys (`re_` is Resend's), so these are public only outside a key's place (keyPlace).
  */
-const OBJECT_ID = /^(?:env|usr|svc|srv|db|setl|order|job|raysubmit|ins|client|sia|sms|idn|pri|pro|sub|txn|ctm|add|biz|dsc|dpl|prj|team|conn|directory|user|ap|in|ta|fu|vol|sb|sess(?=_)|we|promo|qt|cs|dp|cus|pi|ch|il|ii|price|prod|evt|acct|tr|po|pm|seti|src|card|ba|btok|tok|re|plink|si|pmc|ipi|cbtxn|ich|ic|fee|link|req|test_clock|whe|wh|ca|cn|du|py|mandate|pyr|trr|setatt|sched|subi|ri|gsk_live_?)[-_]/;
+const OBJECT_ID = /^(?:env|usr|svc|srv|db|setl|order|org|job|raysubmit|ins|client|sia|sms|idn|pri|pro|sub|txn|ctm|add|biz|dsc|dpl|prj|team|conn|directory|user|ap|in|ta|fu|vol|sb|sess(?=_)|we|promo|qt|cs|dp|cus|pi|ch|il|ii|price|prod|evt|acct|tr|po|pm|seti|src|card|ba|btok|tok|re|plink|si|pmc|ipi|cbtxn|ich|ic|fee|link|req|test_clock|whe|wh|ca|cn|du|py|mandate|pyr|trr|setatt|sched|subi|ri|gsk_live_?)[-_]/;
 /** Right after an encryption wrapper (values.ts ENCRYPTION_WRAPPER), or SOPS's `,iv:` and `,tag:` fields, or Splunk's `$7$`. */
 const AFTER_ENCRYPTION_WRAPPER = new RegExp(String.raw `(?:${ENCRYPTION_WRAPPER}|,iv:|,tag:|(?:^|[\s=:"'])\$[17]\$)$`);
 /** Shapes that are IDs or encodings, not keys. */
@@ -59,6 +59,10 @@ const CANDIDATE = new RegExp(`^[A-Za-z0-9+/=_-]{${MIN_TOKEN},}$`);
  * `%40scope%2Fname`). A base64 value's `/`, `+` and `=` are escaped in one.
  */
 const PERCENT_RUN = /(?:[A-Za-z0-9_.~-]|%[0-9A-Fa-f]{2}){16,}/g;
+/** A run of token characters with JSON's escaped slashes in it (PHP's and Laravel's `Zx9…\/Hd1…`). */
+const ESCAPED_SLASH_RUN = /(?:[A-Za-z0-9+=_-]|\\\/){16,}/g;
+/** A letter beyond ASCII in an alphabet with cases (Latin's, Greek's, Cyrillic's): `ð`, `é`, `ж`. */
+const NON_ASCII_CASED = /(?![\x00-\x7f])[\p{Lu}\p{Ll}]/u;
 /**
  * Runs of token characters long enough to be a key. A run that is a path (`/api/v1/<id>/users`) or
  * dotted (`a.b.c`) is split into its segments, each judged alone. A percent-encoded run that decodes to one token is
@@ -83,6 +87,20 @@ export function candidates(text) {
             encoded.push([m.index, m.index + m[0].length]);
         }
     }
+    // A base64 value whose slashes JSON escaped, as PHP's json_encode does every one, is judged whole too, as its
+    // decoded value; a path's (`api.example.io\/v1\/me`) is split as a path is.
+    if (text.includes("\\/")) {
+        for (const m of text.matchAll(ESCAPED_SLASH_RUN)) {
+            if (!m[0].includes("\\/") || backslashesBefore(text, m.index) % 2)
+                continue;
+            const token = m[0].replace(/\\\//g, "/");
+            if (token.length > 256 || !CANDIDATE.test(token) || !isBase64Run(token, text, m.index))
+                continue;
+            out.push({ start: m.index, end: m.index + m[0].length, token });
+            encoded.push([m.index, m.index + m[0].length]);
+        }
+        encoded.sort((x, y) => x[0] - y[0]);
+    }
     // (Both in the text's order: the encoded run a token run may be in is the first not ended before it.)
     let k = 0;
     for (const [runStart, runEnd] of longRuns(text, CANDIDATE_SET, MIN_TOKEN)) {
@@ -92,6 +110,11 @@ export function candidates(text) {
             continue;
         if (runEnd - runStart > 256)
             continue; // blobs: images, certificates, bundles; longer tokens are caught by format or header
+        // The ASCII part of a word in a cased alphabet, cut at its first letter beyond ASCII: Icelandic's
+        // `Endurheimtulykilorð`, `endurheimtulykilorðið`. (Not after a letter of a script without cases: Chinese writes a
+        // token against its words, `令牌是Xk9…`.)
+        if (NON_ASCII_CASED.test(text[runStart - 1] ?? "") || NON_ASCII_CASED.test(text[runEnd] ?? ""))
+            continue;
         const run = text.slice(runStart, runEnd);
         // `name=value` inside a run (query strings, form bodies): split at an `=` that isn't padding.
         for (const [eqOff, piece] of splitKeep(run, /=(?=[A-Za-z0-9+/_-])/g)) {
@@ -310,21 +333,30 @@ const isUserinfo = (f) => /^[^\s"'<>@?#/]*@[\w-]+(?:\.[\w-]+)+/.test(f.ctx.right
 /**
  * A URL path segment that names an object: under a collection (`/v1/predictions/<id>`), or anywhere when its shape
  * says ID (a file name, `<key>~noop.jpeg`, `<id>.png`; short; one case). A mixed-case random run of 20 or more is a
- * token whatever path it's in, unless the host serves media (`inews.gtimg.com`) or a content word is its parent
- * (`/article/<id>`). Never a hook's URL, the user part before an `@`, or a secret file's value.
+ * token whatever path it's in (24 or more under a collection: `/integrations/<token>`, `/endpoints/<token>`), unless
+ * the host serves media (`inews.gtimg.com`) or a content word is its parent (`/article/<id>`). Never a hook's URL, the
+ * user part before an `@`, or a secret file's value.
  */
 function urlPathId(f) {
     if (f.ctx.syntax !== "path" || f.secretFile || isHookUrl(f) || isUserinfo(f))
         return undefined;
-    if (/^[a-z][a-z-]{2,30}s$/i.test(f.ctx.pathParent) && !isHookSegment(f.ctx.pathParent))
-        return "collection-id";
-    if (f.ctx.nameClass === "credential")
+    const collection = /^[a-z][a-z-]{2,30}s$/i.test(f.ctx.pathParent) && !isHookSegment(f.ctx.pathParent);
+    if (f.ctx.nameClass === "credential" && !collection)
         return undefined;
     const mediaHost = /:\/\/(?:[\w-]+\.)*[\w-]*(?:img|image|images|cdn|static|media|pic|pics|photo|photos|thumb|thumbs|resiz|assets?)[\w-]*\.[\w.-]+\/[^\s"']*$/i.test(f.ctx.left);
     const contentParent = /^(?:article|articles|post|posts|p|story|stories|news|item|items|video|videos|watch|status|statuses|question|questions|answer|issue|issues|pull|commit|blob|tree|wiki|doc|docs|d|album|track|playlist|episode|product|products)$/i.test(f.ctx.pathParent);
     const t = f.token;
     const tokenShaped = t.length >= 20 && /[a-z]/.test(t) && /[A-Z]/.test(t) && /\d/.test(t) && !mediaHost && !contentParent;
-    return /^[~.][A-Za-z0-9]{2,6}\b|^~|^=[swhc]\d+/.test(f.ctx.right) || !tokenShaped ? "url-path-id" : undefined;
+    // (Under a collection, longer than the IDs services generate in that alphabet: Firestore's 20, nanoid's 21, Spotify's
+    // 22, ElevenLabs' voice 20.)
+    if (collection)
+        return tokenShaped && t.length >= 24 ? undefined : "collection-id";
+    // (A file name: an image's or a page asset's, named by its ID on any host, `<key>~noop.jpeg`, `<id>.png`, an image
+    // server's sizing, `<id>=s1200`. Not a feed, a calendar or a download, which a private link names by its token:
+    // `/subscribe/<token>.ics`, `/feeds/<token>.rss`.)
+    if (/^~|^=[swhc]\d+|^\.(?:jpe?g|png|gif|webp|avif|svg|ico|bmp|tiff?|heic|mp[34]|webm|mov|m4[av]|ogg|wav|m3u8|ts|js|mjs|css|map|woff2?|ttf|html?)\b/i.test(f.ctx.right))
+        return "url-path-id";
+    return tokenShaped ? undefined : "url-path-id";
 }
 /**
  * Hex, hashes and IDs until a key context says otherwise: under a credential's name, or (32 characters or more) a
@@ -389,6 +421,15 @@ const VETOES = [
     // The token's own shape.
     ["digits", (_, t) => /^\d+$/.test(t)],
     ["placeholder", (_, t) => isPlaceholder(t)],
+    // A key shown cut short, its front before an ellipsis (`sk-ant-api03-Zx9K...`, `…QAAA` after it): the token stops at
+    // the dots, the display doesn't.
+    [
+        "placeholder",
+        (f, t) => {
+            const cut = /^(?:\.\.\.|…)[\w+/=-]{0,40}/.exec(f.ctx.right)?.[0];
+            return cut !== undefined && isTruncated(`${f.lead ?? ""}${t}${cut}`);
+        },
+    ],
     // Hex of one short unit written over and over (`Key = ef012345ef012345ef012345ef012345`): a filler someone typed, not
     // key bytes, which never repeat in step. (Other alphabets' fillers, `foobarfoobar…`, are in the model's training
     // negatives, which a veto would change.)
@@ -416,6 +457,44 @@ const VETOES = [
         (f) => /[?&](?:pf_rd_\w+|pd_rd_\w+|utm_\w+|fbclid|gclid|msclkid|dclid|mc_eid|igshid|_hsenc|crid|qid|sprefix|ved|usg|ei|sei|iflsig|refRID|sca_esv|sca_upv|sxsrf|oq|aqs|gs_lcp|gs_lp|uact|rlz|ictx|vin)=$/i.test(f.ctx.left),
     ],
     ["build-id", (f) => /\bBuildID=\S*$/.test(f.ctx.left)],
+    // The name a declaration gives, an identifier in its letters' case: C++'s `class XiaomiLYWSD03MMC final`, `struct
+    // QMI8658Config {`, `enum`, `interface`.
+    [
+        "declared-name",
+        (f, t) => /(?:^|[\s;{])(?:class|struct|enum|union|interface|trait|namespace|record)[ \t]+$/.test(f.ctx.left) &&
+            /^[A-Z][A-Za-z0-9_]*$/.test(t) &&
+            /^[ \t]*(?:final\b|[:{<(]|extends\b|implements\b|\r?$)/m.test(f.ctx.right),
+    ],
+    // A cell of a row of a command's table, six cells or more and three of the others counts, sizes or dates, with no
+    // credential's word: an ID (Elasticsearch's `_cat/indices`, `green open chunks-2026.10 <uuid> 3 1 1820331 0 14.2gb`).
+    [
+        "table-row",
+        (f, t, cred) => {
+            if (cred || f.ctx.lineCredential)
+                return false;
+            const cells = `${f.ctx.left}${f.lead ?? ""}${t}${f.ctx.right}`.trim().split(/[ \t]+/);
+            if (cells.length < 6 || !/(?:^|[ \t])$/.test(f.ctx.left) || !/^(?:[ \t]|$)/.test(f.ctx.right))
+                return false;
+            const counts = cells.filter((c) => /^\d+(?:\.\d+)?(?:[kmgtp]?i?b|[kmg]|%|ms|s|m|h|d)?$|^\d{4}-\d\d-\d\d(?:T[\d:.]+Z?)?$/i.test(c));
+            return counts.length >= 3 && !/key|token|secret|passw|auth|credential/i.test(f.ctx.left + f.ctx.right);
+        },
+    ],
+    // The object a lookup is given, its first argument: `Subscription::retrieve("<id>")`, `stripe.customers.retrieve(…)`,
+    // `Model.findById(…)`. (Not a secret's lookup, `get_secret`, `Secret.fetch`: there the argument names the secret.)
+    [
+        "lookup-id",
+        (f, _, cred) => !cred &&
+            f.ctx.syntax === "call" &&
+            /(?:^|[.:])(?:retrieve|find|findById|findOne|fetch|lookup|describe|delete|destroy|cancel|getById|get_by_id|find_by_id)$/i.test(f.ctx.name) &&
+            /(?:::|\.|->)\w+\([ \t]*["'`]$/.test(f.ctx.left) &&
+            !/secret|token|key|passw|credential|vault/i.test(/[\w$\\:.>-]*\([ \t]*["'`]$/.exec(f.ctx.left)?.[0] ?? ""),
+    ],
+    // A static file's cache buster, a digest of its content: `site.css?v=<base64url SHA-256>` (ASP.NET's
+    // `asp-append-version`), and as IIS logs it, the query after the path's field, `/css/site.css v=…`.
+    [
+        "cache-buster",
+        (f) => /\.(?:css|js|mjs|png|svg|jpe?g|gif|webp|avif|woff2?|ttf|otf|eot|ico|map)(?:\?|[ \t]+)(?:[\w.-]+=[^\s&=]*&)*(?:v|ver|version|hash|h|cb|rev|_)=$/i.test(f.ctx.left),
+    ],
     ["public-id", (f) => inPublicNamespace(f)],
     // A hashed host in `~/.ssh/known_hosts` (`|1|<salt>|<HMAC-SHA1 of the host name>`, both base64 of 20 bytes): public, as the host names are.
     // (With the `+` a salt can start with, which the token drops: `|1|+bsj…=|`.)
@@ -528,12 +607,15 @@ const VETOES = [
             /^(?:data|embeddings?|vectors?|payload|blob|frame(?:_?\d+)?|thumbnail|bytes|mac|cnonce|opaque|protected|aad)$/i.test(f.ctx.name),
     ],
     // Where it sits: a file path, a PEM block of public material, a line of a wrapped blob.
-    ["fs-path", (f, _, cred) => f.ctx.fsPath && !cred],
+    // (Not a request target in a log under a hook's segment, `"POST /hooks/<token> HTTP/1.1"`: a URL's path.)
+    ["fs-path", (f, _, cred) => f.ctx.fsPath && !cred && !isHookSegment(f.ctx.pathParent)],
     ["armored-public", (f) => f.ctx.armored],
     // A PEM block's label on the line above it, a name of parts (`ALICE_cf_c2pnb163v1`, `PRIME192V1_RFC5114-Peer`): the
     // block holds the key.
     ["block-label", (f, t) => f.ctx.labelsBlock && /^[A-Za-z][A-Za-z0-9]*(?:[_-][A-Za-z0-9]+)+$/.test(t)],
-    ["blob-line", (f, _, cred) => f.ctx.blobLine && !cred],
+    // (Not lines a handover heads, keys listed one per line: "Here are your new API keys:" over three `openssl rand
+    // -base64 32`s, of one width as a blob's lines are.)
+    ["blob-line", (f, _, cred) => f.ctx.blobLine && !cred && !f.ctx.handover],
     ["ssh-public-key", (_, t) => /^AAAA[A-Z]\d?Nza|^AAAAE2Vj/.test(t)],
     // (After a version's number, `1.2.`, not after the last digit of a token cut at a dot: `<id>7.<secret>`.)
     ["version", (f, t) => /(?:^|[^\w.])v?\d{1,4}(?:\.\d{1,6})*\.$/.test(f.ctx.left) || /\+g[0-9a-f]{7,40}$/.test(t)],
@@ -559,7 +641,12 @@ const VETOES = [
             !cred &&
             !keyPlace(f),
     ],
-    ["filename", (f, _, cred) => f.ctx.fileish && !cred],
+    // (A web link's path segment is read by urlPathId, which knows a private feed's or download's link from a file's
+    // name; a `file://` URL's are a file's path.)
+    [
+        "filename",
+        (f, _, cred) => f.ctx.fileish && !cred && (f.ctx.syntax !== "path" || /file:\/\/[^\s"']*$/i.test(f.ctx.left)),
+    ],
     // (Elsewhere a UUID is hex in groups: hexVeto.)
     ["uuid", (f, t, cred) => cred && UUID.test(t) && /session[_-]?id$|sessionId$/i.test(f.ctx.name)],
     // (A ULID's and a cuid's own timestamp says ID wherever it stands; a key's random characters rarely make one.)
@@ -605,10 +692,13 @@ const VETOES = [
     ],
     // The local part of an email address or Message-ID (`<CAOS…@mail.gmail.com>`, its first piece when an `=` in it
     // split it, `CAPig+…=JtAj…@mail.gmail.com`), a push token's body. (Not a URL's userinfo, `https://<token>@host`: a
-    // password or a token there.)
+    // password or a token there; nor the password after a user and a colon with the scheme left off, as proxy variables
+    // are written, `HTTPS_PROXY=svc:<password>@proxy:8080`: an address's local part holds no colon.)
     [
         "local-part",
-        (f) => (/^(?:=[\w+=.-]*)?@/.test(f.ctx.right) && !/:\/\/[^\s/@]*$/.test(f.ctx.left)) ||
+        (f) => (/^(?:=[\w+=.-]*)?@/.test(f.ctx.right) &&
+            !/:\/\/[^\s/@]*$/.test(f.ctx.left) &&
+            !/(?:^|[\s="'])[\w.%+-]{1,64}:$/.test(f.ctx.left)) ||
             /ExponentPushToken\[$|ExpoPushToken\[$/.test(f.ctx.left),
     ],
     // An Android app's signature hash in MSAL's redirect URI (`msauth://com.contoso.app/<base64 SHA-1>`, encoded or in
@@ -632,12 +722,16 @@ const VETOES = [
     // A fragment of a percent-encoded value (`…%2fAx%2bTHQ…`): the whole value, not a key, if anything.
     ["percent-fragment", (f) => /%[0-9A-Fa-f]{2}$/.test(f.ctx.left) || /^%[0-9A-Fa-f]{2}/.test(f.ctx.right)],
     // Cloud resource IDs: `/subscriptions/<uuid>/resourceGroups/<name>/providers/…`, ARNs.
-    // (The path ends at a query: a parameter after `?` or `&` is its own, and a credential's name keeps it.)
+    // (The path ends at a query: a parameter after `?` or `&` is its own, and a credential's name keeps it. And a hook's
+    // segment after the resource starts a secret's path: `/projects/42/webhooks/<token>`.)
     [
         "resource-id",
-        (f, _, cred) => !cred &&
-            (/\/(?:subscriptions|resourceGroups|resourcegroups|providers|projects|locations|instances|clusters|namespaces)\/[^\s"'?&#]*$/.test(f.ctx.left) ||
-                /\barn:aws[\w-]*:[^\s"'?&#]*$/.test(f.ctx.left)),
+        (f, _, cred) => {
+            if (cred)
+                return false;
+            const path = /\/(?:subscriptions|resourceGroups|resourcegroups|providers|projects|locations|instances|clusters|namespaces)\/([^\s"'?&#]*)$/.exec(f.ctx.left)?.[1];
+            return ((path !== undefined && !path.split("/").some(isHookSegment)) || /\barn:aws[\w-]*:[^\s"'?&#]*$/.test(f.ctx.left));
+        },
     ],
     // IDs that frameworks generate: cuid2 (24 lowercase characters starting with a letter, random all through: only
     // where IDs are written says it's one).
@@ -649,10 +743,15 @@ const VETOES = [
         "public-browser-key",
         (f) => f.ctx.syntax === "call" && (PUBLIC_KEY_CALLEE.test(f.ctx.name) || FULLSTORY_GLOBAL.test(f.ctx.name)),
     ],
-    // A Solana address near Solana's words. (Other blockchain addresses and content IDs are `inert`'s: values.ts.)
+    // A Solana address near Solana's words. (Other blockchain addresses and content IDs are `inert`'s: values.ts.) Not
+    // under a credential's name, nor where an RPC provider's URL takes its key: after the API's version
+    // (`solana-devnet.g.alchemy.com/v2/<key>`). (A key's parameter, `?api-key=`, is a credential's name.)
     [
         "public-address",
-        (f, t) => /^[1-9A-HJ-NP-Za-km-z]{32,44}$/.test(t) && /solana|devnet|mainnet-beta|lamports|phantom|pubkey/i.test(f.ctx.near),
+        (f, t, cred) => /^[1-9A-HJ-NP-Za-km-z]{32,44}$/.test(t) &&
+            /solana|devnet|mainnet-beta|lamports|phantom|pubkey/i.test(f.ctx.near) &&
+            !cred &&
+            !(f.ctx.inUrl && /\/v\d{1,2}\/$/.test(f.ctx.left)),
     ],
     ["timestamp", (_, t) => /^\d{8}T\d{6}Z$/.test(t)],
     // A Google Drive file's ID near Drive's words.
@@ -666,7 +765,16 @@ const VETOES = [
     ["name-hash", (f) => f.ctx.nameClass === "hash"],
     ["name-public", (f) => f.ctx.nameClass === "public"],
     ["name-identifier", (f) => f.ctx.nameClass === "identifier"],
-    ["base64-text", (f, t, cred) => f.decodes === "text" && f.length >= 24 && !cred && decodedIsNatural(t)],
+    // (Shorter, only words or a JWT's header: a tokenizer's vocabulary, `IG5lY2Vzc2FyaWx5` for ` necessarily`;
+    // `eyJhbGciOiJFUzI1NiJ9`, `{"alg":"ES256"}`. A pair, `user:pass`, is a credential's, and other JSON a cookie's,
+    // Laravel's `{"iv":…}`.)
+    [
+        "base64-text",
+        (f, t, cred) => f.decodes === "text" &&
+            !cred &&
+            decodedIsNatural(t) &&
+            (f.length >= 24 || /^(?:[A-Za-z]+(?: [A-Za-z]+)*|\{"(?:alg|typ)":"[\w-]+"\})$/.test(base64Binary(t).trim())),
+    ],
     // Words, and words around a digit or two: package and module names (`qtquick3dphysics`, `oauth2client`).
     [
         "words",

package/dist/urls.d.ts

@@ -15,9 +15,11 @@ export declare const URL_SCHEME: string;
  * Where a URL that starts at `at` ends: at a blank, a quote, an angle bracket, a backtick, a backslash (a JSON
  * string's escaped quote), or a closing bracket it didn't open (a Markdown link's `)`); sentence punctuation at its
  * end isn't part of it (`…/reset/<token>.`). A password's brackets are (`app:p)dgbs@db`, `admin:<Harbor2291x>@host`).
- * However long the password: each look-ahead for its `@` is remembered to where it stopped (`reachesAt`).
+ * However long the password: each look-ahead for its `@` is remembered to where it stopped (`reachesAt`). With `enough`,
+ * it stops once it has read past that point (the answer is then only known to be past it): urlAround asks whether a
+ * URL reaches a position, and a line with no blank in it (a comma-separated list of hosts) reads, as one URL, to its end.
  */
-export declare function urlEnd(text: string, at: number): number;
+export declare function urlEnd(text: string, at: number, enough?: number): number;
 /** Whether `value` starts with a URL with a scheme. */
 export declare const isUrl: (value: string) => boolean;
 /** The URL with a scheme that `value` starts with, read to where urlEnd ends it; undefined if it starts with none. */
@@ -49,7 +51,10 @@ export interface Param {
     name: string;
     value: Span;
 }
-/** A parameter's start and name: after `?`, `&` (or HTML's `&amp;`) or `#`, a name with PHP's `[]` allowed. */
+/**
+ * A parameter's start and name: after `?`, `&` (or HTML's `&amp;`, or the `\u0026` and `\x26` that JSON encoders and JS
+ * write for it: Go's encoding/json escapes every `&`) or `#`, a name with PHP's `[]` allowed.
+ */
 export declare const PARAM_START: string;
 export declare const PARAM_NAME: string;
 /**