polite-crawler 0.1.2 → 0.1.4

package/build/cli.mjs

@@ -1,5 +1,6 @@
 #!/usr/bin/env node
-import { i as VERSION, t as durationMs } from "./duration-iIuMMofi.mjs";
+import { i as durationMs, r as VERSION } from "./version-BPem-XZd.mjs";
+import { t as isLoopbackHost } from "./loopback-C_QYowSf.mjs";
 import { join, resolve } from "node:path";
 import { readFileSync } from "node:fs";
 import { parseArgs } from "node:util";
@@ -80,30 +81,85 @@ const OPTIONS = {
 	"archive-despite-robots": { type: "boolean" },
 	"sign-key": { type: "string" },
 	"signature-agent": { type: "string" },
+	tick: { type: "string" },
+	jitter: { type: "string" },
+	"tick-budget": { type: "string" },
+	"tick-pages": { type: "string" },
+	grace: { type: "string" },
+	events: { type: "string" },
+	"on-change": { type: "string" },
+	"mcp-http": { type: "boolean" },
+	"mcp-port": { type: "string" },
+	"mcp-host": { type: "string" },
+	"mcp-token-file": { type: "string" },
+	"allow-remote": { type: "boolean" },
 	field: {
 		type: "string",
 		multiple: true
 	},
 	capture: { type: "string" },
+	embedder: { type: "string" },
+	"embed-model": { type: "string" },
+	"ollama-url": { type: "string" },
+	"allow-remote-ollama": { type: "boolean" },
+	"max-docs": { type: "string" },
+	name: { type: "string" },
+	"require-name": { type: "boolean" },
+	address: { type: "string" },
+	lat: { type: "string" },
+	lon: { type: "string" },
+	osm: { type: "string" },
+	pages: { type: "string" },
+	compare: { type: "string" },
+	"on-behalf-of": {
+		type: "string",
+		multiple: true
+	},
 	...EXPORT_OPTIONS
 };
+/** These options belong to `daemon`, and not to `daemon status`. */
+const DAEMON_RUN_OPTIONS = [
+	"tick",
+	"jitter",
+	"tick-budget",
+	"tick-pages",
+	"grace",
+	"events",
+	"on-change",
+	"mcp-http",
+	"mcp-port",
+	"mcp-host",
+	"mcp-token-file",
+	"allow-remote"
+];
+/** These options belong to `venue` only. */
+const VENUE_OPTIONS = [
+	"name",
+	"require-name",
+	"address",
+	"lat",
+	"lon",
+	"osm",
+	"pages",
+	"compare"
+];
 /** These options belong to one command, or to the commands that the text names. */
 const OWNER = {
-	render: "fetch crawl watch",
-	"max-age": "fetch crawl discover",
+	render: "fetch crawl watch venue product",
+	"max-age": "fetch crawl discover venue product",
 	limit: "find",
-	kind: "find",
+	kind: "find embed",
 	rank: "find",
 	from: "read diff",
 	length: "read",
-	"user-provided": "read extract",
-	"allow-user-provided": "mcp",
+	"user-provided": "read extract venue product",
+	"allow-user-provided": "mcp daemon",
 	...Object.fromEntries(Object.keys(EXPORT_OPTIONS).map((option) => [option, "export"])),
 	"max-urls": "discover",
 	feed: "discover",
 	"max-pages": "crawl watch",
 	"max-depth": "crawl",
-	budget: "crawl watch",
+	budget: "crawl watch embed venue product",
 	include: "crawl",
 	exclude: "crawl",
 	sitemaps: "crawl",
@@ -114,12 +170,20 @@ const OWNER = {
 	confirm: "watch",
 	scope: "watch",
 	label: "watch",
-	"archive-fallback": "fetch",
-	"archive-despite-robots": "fetch",
-	"sign-key": "fetch discover crawl watch mcp key-directory",
-	"signature-agent": "fetch discover crawl watch mcp key-directory",
+	"archive-fallback": "fetch venue product",
+	"archive-despite-robots": "fetch venue product",
+	"sign-key": "fetch discover crawl watch mcp key-directory daemon venue product",
+	"signature-agent": "fetch discover crawl watch mcp key-directory daemon venue product",
+	...Object.fromEntries(DAEMON_RUN_OPTIONS.map((option) => [option, "daemon"])),
 	field: "extract",
-	capture: "extract"
+	capture: "extract",
+	embedder: "find embed mcp",
+	"embed-model": "find embed mcp",
+	"ollama-url": "find embed mcp",
+	"allow-remote-ollama": "find embed mcp",
+	"max-docs": "embed",
+	...Object.fromEntries(VENUE_OPTIONS.map((option) => [option, "venue"])),
+	"on-behalf-of": "venue product"
 };
 /** These are the commands, with the smallest and the largest count of arguments. */
 const COMMANDS = {
@@ -178,6 +242,11 @@ const COMMANDS = {
 		max: Infinity,
 		wants: "watch wants add, remove, list or run"
 	},
+	daemon: {
+		min: 0,
+		max: 1,
+		wants: "daemon takes no argument, or status"
+	},
 	"key-directory": {
 		min: 0,
 		max: 0,
@@ -187,6 +256,21 @@ const COMMANDS = {
 		min: 1,
 		max: 1,
 		wants: "extract wants one document id and --field name=value"
+	},
+	embed: {
+		min: 0,
+		max: 0,
+		wants: "embed takes no argument: give the model with --embedder"
+	},
+	venue: {
+		min: 1,
+		max: 1,
+		wants: "venue wants the URL of the site of one venue"
+	},
+	product: {
+		min: 1,
+		max: 1,
+		wants: "product wants the URL of one product page"
 	}
 };
 const parse = (args) => {
@@ -424,6 +508,102 @@ const describeEvent = (event) => {
 	].join("\n");
 };
 const describeWatchSummary = (summary) => `${String(summary.checked)} checked: ${String(summary.new)} new, ${String(summary.changed)} changed, ${String(summary.unchanged)} unchanged, ${String(summary.gone)} gone, ${String(summary.refused)} refused, ${String(summary.error)} errors. ${String(summary.suspected)} suspected, ${String(summary.confirmed)} confirmed, ${String(summary.due)} still due. Stopped (${summary.stopped}): ${summary.reason}.`;
+/** This function reads a file name of the command line: a leading `~` is the home folder. */
+const fileOf = (text) => text === "~" ? homedir() : /^~[\\/]/.test(text) ? resolve(homedir(), text.slice(2)) : resolve(text);
+/** This function reads a duration of `daemon`, with a floor in milliseconds. */
+const daemonDuration = (name, text, floor) => {
+	if (text === void 0) return void 0;
+	let ms;
+	try {
+		ms = durationMs(text);
+	} catch (error) {
+		throw new UsageError(`--${name}: ${error.message}`, { cause: error });
+	}
+	if (ms < floor) throw new UsageError(floor === 0 ? `--${name} wants 0 or more` : `--${name} wants ${String(floor / 1e3)}s or more`);
+	return ms;
+};
+/** These options of `daemon` belong to its MCP server, so they need `--mcp-http`. */
+const MCP_OPTIONS = [
+	"mcp-port",
+	"mcp-host",
+	"mcp-token-file",
+	"allow-remote"
+];
+/** This function checks the options of `daemon` and of `daemon status`. */
+const daemonLineOf = (values, rest) => {
+	const [action] = rest;
+	if (action !== void 0 && action !== "status") throw new UsageError(`daemon takes no argument, or status, not "${action}"`);
+	if (action === "status") {
+		const run = [
+			...DAEMON_RUN_OPTIONS,
+			"allow-user-provided",
+			"sign-key",
+			"signature-agent"
+		];
+		for (const option of run) if (values[option] !== void 0) throw new UsageError(`--${option} is an option of daemon, not of daemon status`);
+	}
+	if (values["mcp-http"] !== true) {
+		for (const option of MCP_OPTIONS) if (values[option] !== void 0) throw new UsageError(`--${option} goes with --mcp-http`);
+	}
+	const host = (values["mcp-host"] ?? "127.0.0.1").trim();
+	if (host === "") throw new UsageError("--mcp-host wants a host");
+	const allowRemote = values["allow-remote"] === true;
+	if (values["mcp-http"] === true && !isLoopbackHost(host) && !allowRemote) throw new UsageError(`--mcp-host ${host} is not a loopback address: add --allow-remote to serve other computers, after you read the warning in the README`);
+	const port = whole("mcp-port", values["mcp-port"], 9307, 0);
+	if (port > 65535) throw new UsageError("--mcp-port wants a port from 0 to 65535");
+	const onChange = values["on-change"];
+	if (onChange !== void 0 && onChange.trim() === "") throw new UsageError("--on-change wants a command");
+	const events = values.events;
+	if (events !== void 0 && events.trim() === "") throw new UsageError("--events wants a file");
+	const tokenFile = values["mcp-token-file"];
+	if (tokenFile !== void 0 && tokenFile.trim() === "") throw new UsageError("--mcp-token-file wants a file");
+	return {
+		action: action === "status" ? "status" : "run",
+		tickMs: daemonDuration("tick", values.tick, 1e3),
+		jitterMs: daemonDuration("jitter", values.jitter, 0),
+		budgetMs: daemonDuration("tick-budget", values["tick-budget"], 1e3),
+		maxPages: values["tick-pages"] === void 0 ? void 0 : whole("tick-pages", values["tick-pages"], 1, 1),
+		graceMs: daemonDuration("grace", values.grace, 0),
+		events: events === void 0 ? void 0 : fileOf(events),
+		onChange,
+		mcp: values["mcp-http"] === true ? {
+			port,
+			host,
+			allowRemote,
+			tokenFile: tokenFile === void 0 ? void 0 : fileOf(tokenFile)
+		} : void 0
+	};
+};
+/** This function gives the first line of `daemon status`: the daemon runs, stopped, or stopped with no clean stop. */
+const daemonHead = (state) => state.alive ? `the daemon runs (process ${String(state.pid)}) since ${state.started_at}, on the store ${state.store}` : state.running ? `the daemon of process ${String(state.pid)} does not run: it stopped with no clean stop. Its state is of ${state.updated_at}` : `the daemon stopped at ${state.stopped_at ?? state.updated_at}`;
+/** This function gives the line of the last tick of `daemon status`. */
+const lastTick = (state) => {
+	const tick = state.last_tick;
+	if (tick === null) return "last tick: none yet";
+	if (tick.summary === null) return `last tick: ${tick.ended_at}, failed: ${tick.error ?? "error"}`;
+	const summary = tick.summary;
+	return `last tick: ${tick.ended_at}, ${String(summary.checked)} checked, ${String(summary.confirmed)} changes, ${String(summary.refused)} refused, ${String(summary.error)} errors (${summary.stopped})`;
+};
+/** This function gives the text of `daemon status`. */
+const describeDaemonStatus = (state, dir) => {
+	if (state === void 0) return `no daemon has run on the store ${dir}`;
+	const lines = [daemonHead(state), lastTick(state)];
+	if (state.alive && state.next_tick_at !== null) lines.push(`next tick: ${state.next_tick_at}`);
+	lines.push(`${String(state.ticks)} ticks, ${String(state.checked)} checks, ${String(state.changes)} changes`);
+	if (state.next_due.length === 0) lines.push("next due: no watch");
+	else {
+		lines.push("next due:");
+		for (const due of state.next_due) lines.push(`  ${due.next_at}  ${due.url}  (every ${everyText(due.every_ms)}, last ${due.last_kind ?? "none"})`);
+	}
+	if (state.mcp !== null) {
+		const where = state.mcp.token_file === null ? "" : ` (the token is in ${state.mcp.token_file})`;
+		lines.push(`MCP: ${state.mcp.url}${where}`);
+		lines.push(state.tasks.length === 0 ? "tasks: none running" : `tasks: ${String(state.tasks.length)} running`);
+		for (const task of state.tasks) lines.push(`  ${task.id}  ${task.tool}  ${task.status}${task.message === null ? "" : `: ${task.message}`}`);
+	}
+	lines.push(state.last_error === null ? "last error: none" : `last error: ${state.last_error.at}: ${state.last_error.message}`);
+	return lines.join("\n");
+};
 /** These are the kinds of a field (`FIELD_KINDS` of `extract.ts`). The command line knows them before it loads the crawler. */
 const KINDS = [
 	"text",
@@ -463,6 +643,33 @@ const fieldsOf = (given) => {
 	}
 	return fields;
 };
+/**
+* This function reads the embedder of the command line, or of the environment for the commands
+* that use one. `--embed-model` and `--ollama-url` need `--embedder`.
+*/
+const embedderOf = (values, env, command) => {
+	const uses = OWNER["embedder"].split(" ").includes(command);
+	const fromEnv = (name) => uses ? env[name] || void 0 : void 0;
+	const name = values.embedder ?? fromEnv("POLITE_CRAWLER_EMBEDDER");
+	const model = values["embed-model"] ?? fromEnv("POLITE_CRAWLER_EMBED_MODEL");
+	const url = values["ollama-url"] ?? fromEnv("POLITE_CRAWLER_OLLAMA_URL");
+	const allowRemote = values["allow-remote-ollama"] === true || uses && envFlag(env, "POLITE_CRAWLER_ALLOW_REMOTE_OLLAMA");
+	if (name === void 0) {
+		if (values["embed-model"] !== void 0 || values["ollama-url"] !== void 0) throw new UsageError("--embed-model and --ollama-url go with --embedder transformers or ollama");
+		return;
+	}
+	if (name !== "transformers" && name !== "ollama") throw new UsageError(`--embedder is transformers or ollama, not "${name}"`);
+	if (model !== void 0 && model.trim() === "") throw new UsageError("--embed-model wants a model");
+	if (values["ollama-url"] !== void 0 && name !== "ollama") throw new UsageError("--ollama-url goes with --embedder ollama");
+	return {
+		kind: name,
+		model,
+		url: name === "ollama" ? url : void 0,
+		allowRemote
+	};
+};
+/** This function gives the text of `embed`: the counts and the reason of the stop. */
+const describeEmbed = (result) => [`${String(result.embedded)} texts got vectors (${String(result.chunks)} chunks); ${String(result.reused)} documents had vectors; ${String(result.held_back)} held back, so not sent to the model`, `${String(result.docs)} documents examined in ${String(result.ms)} ms with ${result.embedder}. Stopped (${result.stopped})${result.stopped === "done" ? "" : ": start embed again for the rest"}.`].join("\n");
 /** This function gives the pages of `fetch`: a rejected fetch is an error of its URL only. A page that needs a browser names the reason when rendering is off. */
 const fetchedPages = (urls, settled, renderOff) => settled.map((result, at) => {
 	const url = urls[at];
@@ -477,7 +684,7 @@ const fetchedPages = (urls, settled, renderOff) => settled.map((result, at) => {
 	} : page;
 });
 /** This function gives the text of the results of `find`. An archived copy says so. */
-const describeFound = (found) => found.length === 0 ? "nothing stored matches" : found.map((hit) => `${hit.doc}  ${hit.title}${hit.archived === void 0 ? "" : `  [archived copy, captured ${hit.archived.timestamp}]`}\n  ${hit.snippet.replace(/\s+/g, " ")}`).join("\n");
+const describeFound = (found) => found.length === 0 ? "nothing stored matches" : found.map((hit) => `${hit.doc}  ${hit.title}${hit.archived === void 0 ? "" : `  [archived copy, captured ${hit.archived.timestamp}]`}${hit.passage === void 0 ? "" : `  [passage ${String(hit.passage.start)}–${String(hit.passage.end)}]`}\n  ${hit.snippet.replace(/\s+/g, " ")}`).join("\n");
 /** This function gives the text of `read`. With `--at`, the head names the capture and its fetch time. */
 const describeRead = (read, at) => {
 	const version = at === void 0 ? "" : ` ${read.capture} (${read.fetched_at})`;
@@ -520,6 +727,311 @@ const extractCode = (result) => {
 	const done = result.summary.verified + result.summary.normalized;
 	return done === result.summary.fields && done > 0 ? 0 : 1;
 };
+/** This function reads a latitude or a longitude of the command line. */
+const coordinateOf = (name, text, limit) => {
+	const value = /^\s*-?\d+(?:\.\d+)?\s*$/.test(text) ? Number(text) : NaN;
+	if (!Number.isFinite(value) || Math.abs(value) > limit) throw new UsageError(`--${name} wants a number from -${String(limit)} to ${String(limit)}, not "${text}"`);
+	return value;
+};
+/** This function checks the options of `venue` and `product`, and gives the options of the recipe. */
+const recipeLineOf = (recipe, values, rest, page) => {
+	const url = rest[0] ?? "";
+	if (!/^https?:\/\//i.test(url) || !URL.canParse(url)) throw new UsageError(`${recipe} wants an http or https URL, not "${url}"`);
+	for (const agent of values["on-behalf-of"] ?? []) if (!/^[A-Za-z0-9_-]+$/.test(agent)) throw new UsageError(`--on-behalf-of wants the robots.txt token of an agent, such as Claude-User, not "${agent}"`);
+	const name = values.name;
+	if (name !== void 0 && name.trim() === "") throw new UsageError("--name wants the name of the venue");
+	if (values["require-name"] === true && name === void 0) throw new UsageError("--require-name goes with --name");
+	const address = values.address;
+	if (address !== void 0 && address.trim() === "") throw new UsageError("--address wants a street address");
+	if (values.lat === void 0 !== (values.lon === void 0)) throw new UsageError("--lat and --lon go together");
+	const lat = values.lat === void 0 ? void 0 : coordinateOf("lat", values.lat, 90);
+	const lon = values.lon === void 0 ? void 0 : coordinateOf("lon", values.lon, 180);
+	const osm = values.osm;
+	if (osm !== void 0 && osm.trim() === "") throw new UsageError("--osm wants an opening_hours value, such as \"Mo-Fr 08:00-18:00\"");
+	const pages = values.pages === void 0 ? void 0 : whole("pages", values.pages, 0, 0);
+	if (pages !== void 0 && pages > 8) throw new UsageError(`--pages wants a whole number from 0 to ${String(8)}`);
+	const compare = values.compare;
+	if (compare !== void 0 && compare.trim() === "") throw new UsageError("--compare wants a file");
+	return {
+		recipe,
+		url,
+		render: page.render,
+		maxAge: page.maxAge,
+		budget: page.budget,
+		onBehalfOf: values["on-behalf-of"],
+		userProvided: values["user-provided"] === true,
+		fallback: values["archive-fallback"] === true ? "archive" : void 0,
+		archiveDespiteRobots: values["archive-despite-robots"] === true,
+		name,
+		requireName: values["require-name"] === true,
+		location: address === void 0 && lat === void 0 ? void 0 : {
+			...address === void 0 ? {} : { address },
+			...lat === void 0 ? {} : {
+				lat,
+				lon
+			}
+		},
+		osm,
+		pages,
+		compare
+	};
+};
+/** The days of a week in their order, as `DAYS` of `hours.ts`. */
+const WEEK_DAYS = [
+	"mon",
+	"tue",
+	"wed",
+	"thu",
+	"fri",
+	"sat",
+	"sun"
+];
+const DAY_LABELS = {
+	mon: "Mon",
+	tue: "Tue",
+	wed: "Wed",
+	thu: "Thu",
+	fri: "Fri",
+	sat: "Sat",
+	sun: "Sun"
+};
+const isRecord = (value) => typeof value === "object" && value !== null && !Array.isArray(value);
+/** This function tells if a value is a time of a kept week: `HH:MM` or `H:MM`, to `24:00`. */
+const isClock = (value) => {
+	if (typeof value !== "string") return false;
+	const match = /^(\d{1,2}):(\d{2})$/.exec(value);
+	return match !== null && Number(match[1]) <= 24 && Number(match[2]) < 60;
+};
+const intervalOf = (value) => Array.isArray(value) && value.length === 2 && isClock(value[0]) && isClock(value[1]) ? [value[0], value[1]] : void 0;
+/**
+* This function reads the kept week of `--compare`, a JSON file. The file is a week, or an object
+* with the field `week` (a reading of `venue --json`) or `hours_week`. Each day gives a list of
+* intervals (`[]` for closed), one interval, or null for closed. An error names the file.
+*/
+const keptWeekOf = (path) => {
+	const file = fileOf(path);
+	let text;
+	try {
+		text = readFileSync(file, "utf8");
+	} catch (error) {
+		const code = error.code;
+		throw new UsageError(`--compare: the file ${file} does not open (${typeof code === "string" ? code : "error"})`, { cause: error });
+	}
+	let data;
+	try {
+		data = JSON.parse(text.charCodeAt(0) === 65279 ? text.slice(1) : text);
+	} catch (error) {
+		throw new UsageError(`--compare: the file ${file} is not JSON`, { cause: error });
+	}
+	if (isRecord(data) && (isRecord(data["week"]) || isRecord(data["hours_week"]))) data = isRecord(data["week"]) ? data["week"] : data["hours_week"];
+	const wants = `--compare: the file ${file} is not a week. Write a list of intervals, one interval or null for each day, such as {"mon": [["09:00", "17:00"]], "sun": []}`;
+	if (!isRecord(data)) throw new UsageError(wants);
+	const week = {};
+	for (const [key, value] of Object.entries(data)) {
+		const day = WEEK_DAYS.find((one) => one === key);
+		if (day === void 0) throw new UsageError(`--compare: the file ${file} names "${key}", not a day: use mon, tue, wed, thu, fri, sat or sun`);
+		const one = intervalOf(value);
+		const intervals = value === null ? [] : one !== void 0 ? [one] : Array.isArray(value) ? value.map(intervalOf) : [void 0];
+		if (intervals.some((interval) => interval === void 0)) throw new UsageError(`${wants} (the value of ${key} is not)`);
+		week[day] = intervals;
+	}
+	return {
+		file,
+		week
+	};
+};
+/** This function writes the intervals of one day, as `describeWeek` of `hours.ts` does. */
+const dayText = (intervals) => intervals === void 0 ? "not stated" : intervals.length === 0 ? "closed" : intervals.map(([open, close]) => open === "00:00" && close === "00:00" ? "open 24 hours" : `${open}–${close}`).join(", ");
+/** This function gives the days of a comparison in their order. A day that no week states is not there. */
+const compareDays = (kept, read, comparison) => {
+	const days = [];
+	for (const day of WEEK_DAYS) {
+		const one = {
+			day,
+			kept: kept[day],
+			read: read[day]
+		};
+		if (one.kept === void 0 && one.read === void 0) continue;
+		const status = comparison.status === "new" || comparison.added.includes(day) ? "new" : comparison.days.includes(day) ? "changed" : comparison.skipped.includes(day) ? "skipped" : one.read === void 0 ? "not read" : "same";
+		if (comparison.status === "new" && one.read === void 0) continue;
+		days.push({
+			...one,
+			status
+		});
+	}
+	return days;
+};
+const QUOTE_CHARS = 200;
+const shortText = (text) => text.length > QUOTE_CHARS ? `${text.slice(0, QUOTE_CHARS)}…` : text;
+/** This function gives the lines of one reading: its rank, its source, its week and its quote. */
+const readingLines = (one) => {
+	const marks = [
+		one.schedule === "other" ? "not the opening hours" : null,
+		one.inferred ? "am or pm guessed" : null,
+		one.partial ? "a rule left out" : null,
+		one.season === null ? null : `season ${one.season}`
+	].filter((mark) => mark !== null);
+	const where = [
+		one.url,
+		one.path,
+		one.capture === null ? null : `capture ${one.capture}`,
+		one.archived === void 0 ? null : `archived copy (${one.archived.source}, captured ${one.archived.timestamp})`
+	].filter((part) => part !== null);
+	return [
+		`  rank ${String(one.rank)}  ${one.from}${one.label === null ? "" : ` "${one.label}"`}  ${one.hours === "" ? "no day" : one.hours}${marks.length > 0 ? ` (${marks.join(", ")})` : ""}`,
+		...where.length > 0 ? [`    ${where.join(" · ")}`] : [],
+		one.from === "text" ? `    quote: ${JSON.stringify(shortText(one.raw))}` : `    value: ${shortText(one.raw.replace(/\s+/g, " "))}`
+	];
+};
+const pageLine = (page) => {
+	const facts = [
+		`${page.outcome} ${String(page.status)}`,
+		page.from_cache ? "from the store" : null,
+		page.rendered ? "rendered" : null,
+		page.held_back ? "held back" : null,
+		page.archived === void 0 ? null : `archived copy (${page.archived.source})`
+	].filter((fact) => fact !== null);
+	return `  ${page.outcome === "ok" || page.outcome === "not-modified" || page.archived !== void 0 ? "✓" : "✗"} ${page.url}  ${facts.join(" · ")}`;
+};
+const WALL_OUTCOMES = /* @__PURE__ */ new Set([
+	"blocked",
+	"challenge",
+	"payment-required"
+]);
+/**
+* This function tells if a site refused the crawler, with the rule of `wall` of `readProduct`. The
+* causes are a 401, a 402 or a 403, a challenge, a bot check, robots.txt or a stop of the host.
+*/
+const refusedPage = (page) => WALL_OUTCOMES.has(page.outcome) || page.outcome === "skipped" && /^(?:robots\.txt disallows|host stopped)/.test(page.note ?? "") || page.doc === null && /\bbot check\b/.test(page.note ?? "");
+/**
+* This function gives the exit code of `venue`. It is 0 when the site gave a reading of the opening
+* hours. It is `WALL` when the home page refused the crawler, and `NOTHING_READ` for a different
+* result. An OpenStreetMap value alone does not count.
+*/
+const venueCode = (result) => {
+	if (result.readings.some((one) => one.from !== "osm" && one.schedule === "opening")) return 0;
+	const home = result.pages[0];
+	return home !== void 0 && refusedPage(home) ? 4 : 3;
+};
+/** This function gives the exit code of `product`: 0 for Product data, `WALL` for a refusal, else `NOTHING_READ`. */
+const productCode = (result) => result.found ? 0 : result.wall ? 4 : 3;
+/** This line tells how to read a held-back page. */
+const HELD_BACK_LINE = "Add --user-provided if you gave this URL yourself. The crawler logs that read.";
+/** This function gives the line of a page that needs a browser, when rendering is off. */
+const renderLine = (notes, renderOff) => renderOff !== void 0 && notes.some((note) => note !== null && /no browser/i.test(note)) ? [`rendering is off: ${renderOff}`] : [];
+/**
+* This function gives the lines of `venue`: the week of the site, and each reading with its source.
+* Then come the days where the readings disagree, the menus, the phone, the pages and the notes.
+* With `--compare`, the lines give each day of the comparison.
+*/
+const describeVenue = (result, compare, renderOff) => {
+	const code = venueCode(result);
+	const lines = [`${code === 0 ? "✓" : "✗"} ${result.home ?? result.url}`];
+	const site = result.from_site;
+	const home = result.pages[0];
+	if (site !== null) {
+		const source = [
+			`${site.from}, rank ${String(site.rank)}`,
+			site.url,
+			site.capture === null ? null : `capture ${site.capture}`
+		].filter((part) => part !== null);
+		lines.push(`  the site: ${site.hours}`, `  from ${source.join(", ")}`);
+	} else if (result.best !== null) lines.push(`  no reading of the site with no guess. The best reading (${result.best.from}, rank ${String(result.best.rank)}): ${result.best.hours}`);
+	else if (code === 4 && home !== void 0) lines.push(`  wall: the site refused the crawler (${home.outcome} ${String(home.status)}). The crawler does not try a different identity.`);
+	else lines.push("  no opening hours found");
+	if (result.readings.length > 0) {
+		lines.push("readings, the best first:");
+		for (const one of result.readings) lines.push(...readingLines(one));
+	}
+	for (const day of result.disagree) {
+		const said = result.readings.filter((one) => one.rank <= 2 && one.week[day] !== void 0).map((one) => `${dayText(one.week[day])} (rank ${String(one.rank)} ${one.from})`);
+		lines.push(`the readings disagree on ${DAY_LABELS[day]}: ${said.join(" · ")}`);
+	}
+	if (result.menus.length > 0) lines.push("menus:", ...result.menus.map((menu) => `  ${menu.url}${menu.text === "" ? "" : `  ${menu.text}`}`));
+	if (result.phone !== null) lines.push(`phone: ${result.phone}`);
+	if (result.moved !== null) lines.push(`moved: the site now goes to ${result.moved}`);
+	if (result.names_venue !== null) lines.push(result.names_venue ? "names the venue: yes" : "names the venue: no. The domain can belong to a different business now.");
+	if (result.held_back) lines.push(`held back: a page asked not to be AI input, so the recipe read nothing from it. ${HELD_BACK_LINE}`);
+	if (result.pages.length > 0) lines.push(`pages (${String(result.pages.length)}):`, ...result.pages.map(pageLine));
+	if (compare !== void 0) {
+		const comparison = compare.comparison;
+		if (comparison === null || site === null) lines.push(`compare with ${compare.file}: no reading of the site with no guess, so nothing to compare`);
+		else {
+			const changed = comparison.days.length > 0 ? ` (${comparison.days.map((day) => DAY_LABELS[day]).join(", ")})` : "";
+			lines.push(`compare with ${compare.file}: ${comparison.status}${changed}`);
+			for (const one of compareDays(compare.kept, site.week, comparison)) {
+				const text = one.status === "changed" || one.status === "skipped" ? `${dayText(one.kept)} → ${dayText(one.read)}` : one.status === "not read" ? `kept ${dayText(one.kept)}` : dayText(one.read);
+				lines.push(`  ${DAY_LABELS[one.day]}  ${one.status.padEnd(8)}  ${text}`);
+			}
+		}
+	}
+	if (result.note !== null) lines.push(`note: ${result.note}`);
+	lines.push(...renderLine([result.note], renderOff));
+	return lines;
+};
+/** This function gives the number of digits after the decimal point of a currency: 2 for EUR, 0 for JPY. */
+const fractionOf = (currency) => {
+	try {
+		return new Intl.NumberFormat("en", {
+			style: "currency",
+			currency
+		}).resolvedOptions().maximumFractionDigits ?? 2;
+	} catch {
+		return 2;
+	}
+};
+/** This function writes an amount of money with its ISO 4217 code: `1,299.00 EUR`, `980 JPY`. */
+const moneyText = (money) => {
+	const digits = fractionOf(money.currency);
+	return `${new Intl.NumberFormat("en", {
+		minimumFractionDigits: digits,
+		maximumFractionDigits: Math.max(digits, 4)
+	}).format(money.amount)} ${money.currency}`;
+};
+/**
+* This function gives the lines of `product`. They are the name, the brand and the codes, the price
+* as money, the availability, the variant, the seller, the page and the notes. A wall and a
+* held-back page say so.
+*/
+const describeProduct = (result, renderOff) => {
+	const code = productCode(result);
+	const lines = [`${code === 0 ? "✓" : "✗"} ${result.name ?? result.url}`];
+	if (result.name !== null) lines.push(`  ${result.url}`);
+	if (code === 4) lines.push(`  wall: the shop refused the crawler (${result.outcome} ${String(result.status)}). The crawler does not try a different identity.`);
+	if (result.held_back) lines.push(`  held back: the page asked not to be AI input. ${HELD_BACK_LINE}`);
+	const codes = [
+		result.brand === null ? null : `brand ${result.brand}`,
+		result.sku === null ? null : `sku ${result.sku}`,
+		result.gtin === null ? null : `gtin ${result.gtin}`
+	].filter((part) => part !== null);
+	if (codes.length > 0) lines.push(`  ${codes.join(" · ")}`);
+	if (result.price !== null) lines.push(result.price_to === null ? `  price ${moneyText(result.price)}` : `  price ${moneyText(result.price)} to ${moneyText(result.price_to)}, the range of the offers`);
+	if (result.availability !== null) lines.push(`  availability ${result.availability}${result.availabilities.length > 1 ? ` (the offers: ${result.availabilities.join(", ")})` : ""}`);
+	const variant = result.variant;
+	if (variant !== null) {
+		const parts = [
+			variant.name,
+			variant.sku === null ? null : `sku ${variant.sku}`,
+			variant.price === null ? null : moneyText(variant.price),
+			variant.availability
+		].filter((part) => part !== null);
+		lines.push(`  variant ${parts.length > 0 ? parts.join(" · ") : "with no data"}`);
+	}
+	if (result.seller !== null) lines.push(`  seller ${result.seller}`);
+	if (result.image !== null) lines.push(`  image ${result.image}`);
+	if (result.archived !== void 0) lines.push(`  archived copy (${result.archived.source}, captured ${result.archived.timestamp}): ${result.archived.url}`);
+	const facts = [
+		`${result.outcome} ${String(result.status)}`,
+		result.doc === null ? null : `doc ${result.doc}`,
+		result.capture === null ? null : `capture ${result.capture}`,
+		result.rendered ? "rendered" : null,
+		result.from_cache ? "from the store" : null
+	].filter((fact) => fact !== null);
+	lines.push(`  ${facts.join(" · ")}`);
+	if (result.note !== null) lines.push(`  note: ${result.note}`);
+	lines.push(...renderLine([result.note], renderOff).map((line) => `  ${line}`));
+	return lines;
+};
 /** This function checks the options of `export`, and gives its format and its options. */
 const exportPlan = (values) => {
 	const out = values.out;
@@ -577,7 +1089,11 @@ const readLine = (args, env) => {
 	const render = values.render ?? "auto";
 	if (render !== "auto" && render !== "never" && render !== "always") throw new UsageError(`--render is auto, never or always, not "${render}"`);
 	const rank = values.rank ?? "exact";
-	if (rank !== "exact" && rank !== "fast") throw new UsageError(`--rank is exact or fast, not "${rank}"`);
+	if (rank !== "exact" && rank !== "fast" && rank !== "hybrid") throw new UsageError(`--rank is exact, fast or hybrid, not "${rank}"`);
+	const embedder = embedderOf(values, env, command);
+	if (rank === "hybrid" && embedder === void 0) throw new UsageError("--rank hybrid wants --embedder transformers or ollama");
+	if (command === "embed" && embedder === void 0) throw new UsageError("embed wants --embedder transformers or ollama");
+	const maxDocs = values["max-docs"] === void 0 ? void 0 : whole("max-docs", values["max-docs"], 1, 1);
 	const limit = whole("limit", values.limit, 10, 1);
 	const from = command === "read" ? whole("from", values.from, 0, 0) : 0;
 	const length = whole("length", values.length, 4e3, 1);
@@ -587,11 +1103,12 @@ const readLine = (args, env) => {
 	const diffTo = versionOf("to", values.to);
 	const watchLine = command === "watch" ? watchLineOf(values, rest) : void 0;
 	const extractFields = command === "extract" ? fieldsOf(values.field) : void 0;
+	const daemonLine = command === "daemon" ? daemonLineOf(values, rest) : void 0;
 	const dir = storeDir(values.dir ?? (env["POLITE_CRAWLER_DIR"] || void 0));
 	const allowPrivate = values["allow-private"] === true || envFlag(env, "POLITE_CRAWLER_ALLOW_PRIVATE");
-	const allowUserProvided = values["allow-user-provided"] === true || command === "mcp" && envFlag(env, "POLITE_CRAWLER_ALLOW_USER_PROVIDED");
+	const allowUserProvided = values["allow-user-provided"] === true || (command === "mcp" || daemonLine?.action === "run") && envFlag(env, "POLITE_CRAWLER_ALLOW_USER_PROVIDED");
 	if (values["archive-despite-robots"] === true && values["archive-fallback"] !== true) throw new UsageError("--archive-despite-robots works only with --archive-fallback");
-	const signs = OWNER["sign-key"].split(" ").includes(command) && (watchLine === void 0 || watchLine.action === "run");
+	const signs = OWNER["sign-key"].split(" ").includes(command) && (watchLine === void 0 || watchLine.action === "run") && (daemonLine === void 0 || daemonLine.action === "run");
 	const signKey = values["sign-key"] ?? (signs ? env["POLITE_CRAWLER_SIGN_KEY"] || void 0 : void 0);
 	const signatureAgent = values["signature-agent"] ?? (signs ? env["POLITE_CRAWLER_SIGNATURE_AGENT"] || void 0 : void 0);
 	if (command === "key-directory") {
@@ -603,6 +1120,11 @@ const readLine = (args, env) => {
 	} catch (error) {
 		throw new UsageError(`--max-age: ${error.message}`, { cause: error });
 	}
+	const recipeLine = command === "venue" || command === "product" ? recipeLineOf(command, values, rest, {
+		render,
+		maxAge,
+		budget: values.budget === void 0 ? void 0 : crawlLimits.budgetMs
+	}) : void 0;
 	return {
 		kind: "command",
 		command,
@@ -619,6 +1141,7 @@ const readLine = (args, env) => {
 		diffFrom,
 		diffTo,
 		watchLine,
+		daemonLine,
 		extractFields,
 		exportPlan: command === "export" ? exportPlan(values) : void 0,
 		dir,
@@ -626,7 +1149,10 @@ const readLine = (args, env) => {
 		allowPrivate,
 		allowUserProvided,
 		signKey,
-		signatureAgent
+		signatureAgent,
+		embedder,
+		maxDocs,
+		recipeLine
 	};
 };
 //#endregion
@@ -656,9 +1182,15 @@ Usage:
   polite-crawler watch remove <url...> stop the watch of these pages
   polite-crawler watch list            print the watch list
   polite-crawler watch run             check the pages that are due, within a budget
+  polite-crawler daemon                check the watch list on a schedule, until SIGINT or SIGTERM;
+                                       with --mcp-http, serve the MCP tools over local HTTP too
+  polite-crawler daemon status         print the state of the daemon of the store
   polite-crawler key-directory         print the key directory (JWKS) of --sign-key, to host
   polite-crawler extract <doc> --field name=value[:kind]...
                                        check proposed field values against a stored page
+  polite-crawler embed --embedder <name>  give vectors to the stored pages, for find --rank hybrid
+  polite-crawler venue <url>           read the opening hours of a venue from its own site
+  polite-crawler product <url>         read the schema.org Product data of a product page
 
 Options:
   --dir <path>       the store (default $POLITE_CRAWLER_DIR, or ~/.polite-crawler)
@@ -670,6 +1202,13 @@ Options:
                      ($POLITE_CRAWLER_SIGN_KEY)
   --signature-agent <url>  the https origin of your key directory, or the URL of a
                      JSON Web Key Set on a path, for --sign-key ($POLITE_CRAWLER_SIGNATURE_AGENT)
+  --embedder <name>  the local model of hybrid search, for find, embed and mcp: transformers
+                     (transformers.js in this process) or ollama ($POLITE_CRAWLER_EMBEDDER)
+  --embed-model <m>  the model: Xenova/bge-small-en-v1.5 (transformers) or nomic-embed-text
+                     (ollama) by default ($POLITE_CRAWLER_EMBED_MODEL)
+  --ollama-url <url> the Ollama server, http://127.0.0.1:11434 by default ($POLITE_CRAWLER_OLLAMA_URL)
+  --allow-remote-ollama  permit an Ollama server on a different computer: the page text goes
+                     to it ($POLITE_CRAWLER_ALLOW_REMOTE_OLLAMA=1)
   --json             print JSON
   -v, --version      print the version
   -h, --help         print this help
@@ -683,7 +1222,8 @@ Options:
   find     --limit <n>       how many results (default 10)
            --kind <kind>     only this kind of document, such as page or pdf
            --rank <mode>     exact (default): BM25 over each match; fast: leave common words
-                             out of the one-word part, faster for a long question
+                             out of the one-word part, faster for a long question; hybrid: BM25
+                             and the meaning of the text, fused (needs --embedder)
   read     --from <n>        where the range starts (default 0)
            --length <n>      how many characters (default 4000)
            --user-provided   you gave this URL yourself: read a page that refuses AI input (logged)
@@ -720,26 +1260,66 @@ Options:
              --render <mode> as for fetch
   watch run  --budget <time>  the longest time (default 10m); no request starts after it
              --max-pages <n>  how many pages to check (default 100)
+  daemon   --tick <time>     the longest wait between two ticks: 1s or more (default 1m)
+           --jitter <time>   a random wait from 0 to this after each tick (default 10s)
+           --tick-budget <time>  the longest time of the checks of one tick (default 10m)
+           --tick-pages <n>  how many pages one tick checks at most (default 100)
+           --grace <time>    at a stop, the checks in progress get this time to end (default 10s)
+           --events <file>   write each change as one JSON line to this file (default: the
+                             standard output)
+           --on-change <command>  start this command of yours for each change: it gets the
+                             event as one JSON line on its standard input
+           --mcp-http        serve the MCP tools over HTTP; each request needs the bearer token
+           --mcp-port <n>    the port (default 9307)
+           --mcp-host <host> the host to bind (default 127.0.0.1)
+           --mcp-token-file <file>  read the token from this file (default: a new token at each
+                             start, in <store>/daemon/mcp-token)
+           --allow-remote    permit a --mcp-host that is not loopback: read the README first
+           --allow-user-provided  as for mcp
+  daemon status              exit code 0 when a daemon runs on the store, 1 when none runs
   extract  --field <name=value[:kind]>  a value to check, never made up by the crawler (repeat it);
                              kind: text (default), number, money, date, url, email, phone,
                              or json-ld:<path> such as json-ld:Product.offers.price
            --capture <id>    check the text of this capture, not the current text
            --user-provided   as for read
+  embed    --budget <time>   the longest time (default 10m); no embedding starts after it
+           --max-docs <n>    how many texts get vectors at most
+           --kind <kind>     only this kind of document
+  venue    --name <text>     the name of the venue: the result tells if the home page names it
+           --require-name    with --name: read no hours from a site that does not name the venue
+           --address <text>  the street address, for a page that lists more than one location
+           --lat <n> --lon <n>  the position of the venue, for the same use (within 150 m);
+                             write a value below 0 after =, such as --lon=-20.5
+           --osm <value>     an OpenStreetMap opening_hours value, read as one more reading;
+                             no request goes to OpenStreetMap
+           --pages <n>       the pages after the home page: 0 to 8 (default 3)
+           --compare <file>  compare the week of the site with a kept week (JSON), day by day
+  venue, product  --render, --max-age, --archive-fallback, --archive-despite-robots  as for fetch
+           --budget <time>   no request starts after this time, such as 2m (default: no limit)
+           --on-behalf-of <agent>  the robots.txt rules of this agent apply too, such as
+                             Claude-User (repeat it)
+           --user-provided   as for read
 
 Exit codes:
   0  done
   1  not done: a page not fetched, no match, a quote not found, a document unknown or
      held back, a watch check that failed or that is still due, a field value that extract
-     did not find or that is not clear, or an error
+     did not find or that is not clear, embed that stopped before each page had vectors,
+     or an error
      (fetch: a page with an archived copy counts as done)
   2  the command line is wrong
+  3  venue, product: nothing read: no opening hours of the site or no Product data, a page
+     held back, or a page that did not come
+  4  venue, product: the site refused the crawler (a wall): a 401, a 402 or a 403, a
+     challenge, a bot check, robots.txt or a stopped host. The crawler does not try a
+     different identity
 `;
 /**
 * This function does the command `export`, and gives the exit code. `readLine` checked the whole
 * command line first. The export only reads the store: it makes no store, and it sends no request.
 */
 const exportCommand = async (plan, dir, json) => {
-	const result = plan.format === "wacz" ? await (await import("./wacz-sfOLT8e9.mjs").then((n) => n.n)).exportWacz(dir, plan.options) : await (await import("./warc-VIPscWxY.mjs").then((n) => n.a)).exportWarc(dir, plan.options);
+	const result = plan.format === "wacz" ? await (await import("./wacz-QyzLp4mC.mjs").then((n) => n.n)).exportWacz(dir, plan.options) : await (await import("./warc-BHcaRKHw.mjs").then((n) => n.a)).exportWarc(dir, plan.options);
 	process.stdout.write(`${json ? JSON.stringify(result, null, 2) : describeExport(result, plan.options.includeRefusals)}\n`);
 	return 0;
 };
@@ -861,7 +1441,7 @@ const runWatchCommand = async (crawler, line, values, limits, json, out) => {
 };
 /** This function gives the text of a diff for a person: the two captures, then the text of `formatDiff`. */
 const describeDiff = async (diff) => {
-	const { formatDiff } = await import("./diff-Dk2eSvi-.mjs").then((n) => n.i);
+	const { formatDiff } = await import("./diff-BUM91Rev.mjs").then((n) => n.i);
 	return [
 		`${diff.doc}  ${diff.from.capture} (${diff.from.fetched_at}) -> ${diff.to.capture} (${diff.to.fetched_at})`,
 		formatDiff(diff),
@@ -871,13 +1451,54 @@ const describeDiff = async (diff) => {
 };
 /** This function checks the fields against a stored document, and prints a line for each field. The exit code is 0 when each field is verified or normalized. */
 const runExtract = async (crawler, doc, fields, options, json, out) => {
-	const { extractDocument } = await import("./extract-DcwNrTK7.mjs").then((n) => n.i);
+	const { extractDocument } = await import("./extract-CefB5Vfn.mjs").then((n) => n.i);
 	const claims = fields;
 	const result = crawler.extract === void 0 ? await extractDocument(crawler.runtime, doc, claims, options) : await crawler.extract(doc, claims, options);
 	if (json) out(JSON.stringify(result, null, 2));
 	else for (const line of describeExtract(result)) out(line);
 	return extractCode(result);
 };
+/**
+* This function makes the embedder of the command line. It loads no model: the model loads at the
+* first text. The transformers embedder keeps the files of its model in `<store>/models`.
+*/
+const embedderFor = async (choice, dir) => {
+	try {
+		if (choice.kind === "ollama") {
+			const { ollamaEmbedder } = await import("./ollama-BJxkjny4.mjs").then((n) => n.a);
+			return ollamaEmbedder({
+				...choice.model === void 0 ? {} : { model: choice.model },
+				...choice.url === void 0 ? {} : { baseUrl: choice.url },
+				allowRemote: choice.allowRemote
+			});
+		}
+		const { transformersEmbedder } = await import("./transformers-BQMlaGj4.mjs").then((n) => n.i);
+		return transformersEmbedder({
+			...choice.model === void 0 ? {} : { model: choice.model },
+			cacheDir: join(dir, "models")
+		});
+	} catch (error) {
+		throw new UsageError(`--embedder ${choice.kind}: ${error.message}`, { cause: error });
+	}
+};
+/** This function gives vectors to the stored pages within the budget. The exit code is 0 when each page has vectors. */
+const runEmbed = async (crawler, limits, options, json, out) => {
+	const controller = new AbortController();
+	const stop = () => controller.abort(/* @__PURE__ */ new Error("interrupted"));
+	process.once("SIGINT", stop);
+	try {
+		const result = await crawler.embed({
+			budget: limits.budgetMs,
+			signal: controller.signal,
+			...options.maxDocs === void 0 ? {} : { maxDocs: options.maxDocs },
+			...options.kind === void 0 ? {} : { kind: options.kind }
+		});
+		out(json ? JSON.stringify(result, null, 2) : describeEmbed(result));
+		return result.stopped === "done" ? 0 : 1;
+	} finally {
+		process.removeListener("SIGINT", stop);
+	}
+};
 /** This function does the command of the command line, and gives the exit code. */
 const main = async (args) => {
 	const out = (text) => void process.stdout.write(`${text}\n`);
@@ -891,6 +1512,7 @@ const main = async (args) => {
 		return 0;
 	}
 	const { command, rest, values, render, rank, limit, from, length, maxAge, crawlLimits, at, diffFrom, diffTo, watchLine, extractFields, dir, signKey, signatureAgent } = line;
+	const embedder = line.embedder === void 0 ? void 0 : await embedderFor(line.embedder, dir);
 	if (command === "key-directory") return keyDirectoryCommand(keyFile(signKey), signatureAgent);
 	let signer;
 	if (signKey !== void 0 && signatureAgent !== void 0) {
@@ -911,7 +1533,8 @@ const main = async (args) => {
 		contact: line.contact,
 		...values["no-browser"] === true ? { renderer: false } : {},
 		...line.allowPrivate ? { egress: { allowPrivate: true } } : {},
-		...signer === void 0 ? {} : { signer }
+		...signer === void 0 ? {} : { signer },
+		...embedder === void 0 ? {} : { embedder }
 	};
 	if (command === "mcp") {
 		const { startMcpServer } = await import("./mcp.mjs");
@@ -921,11 +1544,22 @@ const main = async (args) => {
 		});
 		return 0;
 	}
+	if (command === "daemon") {
+		const { daemonCommand } = await import("./cli-daemon-ZYRdy6EW.mjs");
+		return daemonCommand(line.daemonLine, options, {
+			allowUserProvided: line.allowUserProvided,
+			json: values.json === true,
+			out
+		});
+	}
 	if (line.exportPlan !== void 0) return exportCommand(line.exportPlan, dir, values.json === true);
-	const { openCrawler } = await import("./open-DnrWgkC0.mjs").then((n) => n.a);
+	const recipeLine = line.recipeLine;
+	const recipes = recipeLine === void 0 ? void 0 : await import("./cli-recipes-Bci6HREA.mjs");
+	const recipeInput = recipes === void 0 || recipeLine === void 0 ? void 0 : recipes.prepareRecipe(recipeLine);
+	const { openCrawler } = await import("./open-CbpSCEZc.mjs").then((n) => n.a);
 	let renderOff;
 	let renderer = options.renderer;
-	if (!(command === "fetch" || command === "crawl" || watchLine?.action === "run") || render === "never") renderer = false;
+	if (!(command === "fetch" || command === "crawl" || recipeLine !== void 0 || watchLine?.action === "run") || render === "never") renderer = false;
 	else if (renderer === void 0) {
 		const { browserMissing } = await import("./browser-CKRNbIDT.mjs").then((n) => n.n);
 		renderOff = await browserMissing();
@@ -1005,6 +1639,19 @@ const main = async (args) => {
 				userProvided: values["user-provided"] === true,
 				at
 			}, values.json === true, out);
+			case "embed": return await runEmbed(crawler, crawlLimits, {
+				maxDocs: line.maxDocs,
+				kind: values.kind
+			}, values.json === true, out);
+			case "venue":
+			case "product": {
+				const io = {
+					json: values.json === true,
+					out,
+					renderOff
+				};
+				return command === "venue" ? await recipes.venueCommand(crawler, recipeLine, recipeInput, io) : await recipes.productCommand(crawler, recipeLine, io);
+			}
 			default: {
 				const counts = await crawler.reindex();
 				print(counts, () => describeReindex(counts));
@@ -1024,6 +1671,6 @@ main(process.argv.slice(2)).then((code) => {
 	process.exitCode = usage ? 2 : 1;
 });
 //#endregion
-export {};
+export { keptWeekOf as a, describeVenue as i, describeDaemonStatus as n, productCode as o, describeProduct as r, venueCode as s, UsageError as t };
 
 //# sourceMappingURL=cli.mjs.map