@koda-sl/baker-cli 0.462.0 → 0.464.0

package/dist/chunk-UX7I6S3B.js

@@ -0,0 +1,106 @@
+import {
+  apiGet
+} from "./chunk-64RONDRZ.js";
+
+// src/commands/testimonials/emptyCorpusHints.ts
+function resolveEmptyReason({
+  resultCount,
+  activeFilters,
+  sourceCount
+}) {
+  if (resultCount > 0) {
+    return null;
+  }
+  if (activeFilters.length > 0) {
+    return "filtered_out";
+  }
+  return sourceCount === 0 ? "no_corpus_connected" : "corpus_empty";
+}
+var NO_CORPUS_EXIT = "Testimonials are ingested from the client's public review pages \u2014 an agent cannot write one \u2014 so no rewording will find quotes. If the user named the client's own review page (Google Maps, Trustpilot, Yelp, Tripadvisor, G2, Capterra, Amazon, App Store, Google Play), add it with `baker testimonials sources add <link>` \u2014 its reviews arrive over the next minutes, not in this turn. Stop searching and source proof another way: a result or case study the client already published, a named-customer logo (`baker images logo`), or a founder claim attributed as a claim. Otherwise ship without a testimonial block and file the gap: `baker actions create --name \"Collect customer quotes for <page>\" --priority medium --tags <slug>`. Never invent, paraphrase-into-existence, or pad with unrelated praise.";
+function emptyReasonDetail(reason, sourceCount, activeFilters) {
+  if (reason === "filtered_out") {
+    return `No rows matched, and ${activeFilters.join(", ")} narrowed this read. The corpus was not checked \u2014 drop every filter to see whether the library itself is empty.`;
+  }
+  if (reason === "no_corpus_connected") {
+    return "This company has no review source connected \u2014 checked against its source list, not inferred from the empty result.";
+  }
+  return `${sourceCount} review source${sourceCount === 1 ? " is" : "s are"} connected but hold no testimonials yet \u2014 the source is there, the rows are not. Ingestion may still be running.`;
+}
+function morePagesHint(resultCount, cursor) {
+  const found = resultCount === 0 ? "This page held no matches" : `This page held ${resultCount}`;
+  return `${found}, and the library continues past it \u2014 so this is one page, not the client's whole set. Do not report it as the corpus and do not conclude anything is missing from it. Read the next page with \`baker testimonials list --cursor ${cursor}\` (same filters), or go straight at the rows you want with \`baker testimonials search <query>\`.`;
+}
+function endOfWalkHint(command) {
+  const noun = command === "search" ? "results" : "rows";
+  return `That was the end of the library \u2014 no more ${noun} past the page before this one. Nothing here says the corpus is empty (the earlier pages held it), so answer from what those pages returned rather than re-reading.`;
+}
+async function measureEmptyCorpus({
+  command,
+  resultCount,
+  activeFilters,
+  nextCursor = null,
+  continued = false
+}) {
+  if (nextCursor) {
+    return { hints: [morePagesHint(resultCount, nextCursor)] };
+  }
+  if (resultCount > 0) {
+    return { hints: [] };
+  }
+  if (continued) {
+    return { hints: [endOfWalkHint(command)] };
+  }
+  if (activeFilters.length > 0) {
+    const reason2 = "filtered_out";
+    return {
+      hints: emptyCorpusHints({ command, reason: reason2, sourceCount: 0, activeFilters }),
+      empty: { reason: reason2, detail: emptyReasonDetail(reason2, 0, activeFilters) }
+    };
+  }
+  let sourceCount;
+  try {
+    const sources = await apiGet("/api/testimonials/sources");
+    sourceCount = Array.isArray(sources) ? sources.length : 0;
+  } catch {
+    return {
+      hints: [
+        "No rows, and the check for connected review sources failed \u2014 so this is NOT evidence that the client has no reviews. Do not record an absence you could not verify. Re-run once; if it fails again, say the library could not be read rather than that it is empty."
+      ]
+    };
+  }
+  const reason = resolveEmptyReason({ resultCount, activeFilters, sourceCount });
+  if (reason === null) {
+    return { hints: [] };
+  }
+  return {
+    hints: emptyCorpusHints({ command, reason, sourceCount, activeFilters }),
+    empty: { reason, detail: emptyReasonDetail(reason, sourceCount, activeFilters) }
+  };
+}
+function emptyCorpusHints({
+  command,
+  reason,
+  sourceCount,
+  activeFilters
+}) {
+  if (reason === null) {
+    return [];
+  }
+  if (reason === "filtered_out") {
+    return [
+      `No rows \u2014 ${activeFilters.join(", ")} narrowed this. Drop every filter and re-run once before touching the query; a filter combination and an empty library look identical from here.`
+    ];
+  }
+  const lead = command === "search" ? "No hits" : "No rows";
+  if (reason === "no_corpus_connected") {
+    return [`${lead}, and this company's source list is empty: there is no review corpus to search. ${NO_CORPUS_EXIT}`];
+  }
+  return [
+    `${lead}. ${sourceCount} review source${sourceCount === 1 ? " is" : "s are"} connected but no testimonials have landed yet \u2014 so this is a timing or ingestion gap, not a missing connection, and re-querying will not change it. Ingestion may still be running: say so rather than reporting the client has no reviews. ${NO_CORPUS_EXIT}`
+  ];
+}
+
+export {
+  measureEmptyCorpus
+};
+//# sourceMappingURL=chunk-UX7I6S3B.js.map

package/dist/chunk-V5YMTW6L.js

@@ -0,0 +1,296 @@
+import {
+  DIRECT_ROUTE,
+  captureProxyCredentials,
+  childEnvWith,
+  plannedRoutes,
+  shouldEscalate
+} from "./chunk-S25EINEI.js";
+
+// src/lib/vimeoCookie.ts
+import { writeFile } from "fs/promises";
+import path from "path";
+var COOKIE_NAME = "vimeo";
+async function writeVimeoCookieFile(cookie, workDir) {
+  const value = cookie?.trim();
+  if (!value) return void 0;
+  const expires = Math.floor(Date.now() / 1e3) + 365 * 24 * 60 * 60;
+  const jar = [
+    "# Netscape HTTP Cookie File",
+    [".vimeo.com", "TRUE", "/", "TRUE", String(expires), COOKIE_NAME, value].join("	"),
+    ""
+  ].join("\n");
+  const filePath = path.join(workDir, "vimeo-cookies.txt");
+  await writeFile(filePath, jar, { mode: 384 });
+  return filePath;
+}
+var VIMEO_AUTH_SIGNATURES = ["oauth token", "only works when logged-in", "unable to fetch new oauth tokens"];
+function isVimeoAuthFailure(detail) {
+  const haystack = detail.toLowerCase();
+  if (!haystack.includes("[vimeo]")) return false;
+  return VIMEO_AUTH_SIGNATURES.some((signature) => haystack.includes(signature));
+}
+
+// src/lib/ytDlpSignal.ts
+var HTTP_ERROR = /HTTP Error (\d{3})/i;
+var CHALLENGE_PHRASES = [
+  "sign in to confirm you're not a bot",
+  "sign in to confirm you\u2019re not a bot",
+  "confirm you are not a bot",
+  "unusual traffic",
+  "captcha"
+];
+var NOT_A_BLOCK_PHRASES = ["private video", "video unavailable", "members-only", "this video is available to"];
+var PROXY_TUNNEL_FAILURE = /tunnel connection failed: 407|proxy authentication required|http error 407/i;
+function ytDlpBlockSignal(stderr) {
+  const text = stderr.toLowerCase();
+  if (PROXY_TUNNEL_FAILURE.test(stderr)) return { status: 407 };
+  if (NOT_A_BLOCK_PHRASES.some((phrase) => text.includes(phrase))) return {};
+  const status = HTTP_ERROR.exec(stderr);
+  if (status) return { status: Number(status[1]) };
+  if (CHALLENGE_PHRASES.some((phrase) => text.includes(phrase))) return { challenge: true };
+  return {};
+}
+
+// src/lib/ytDlp.ts
+import { spawn } from "child_process";
+import { readdir, readFile } from "fs/promises";
+import path2 from "path";
+
+// src/lib/ytDlpFormat.ts
+function videoFormatArgs(maxResolution) {
+  const sort = maxResolution ? ["-S", `res:${maxResolution}`] : [];
+  return ["--format", "bv*+ba/b", ...sort, "--merge-output-format", "mp4"];
+}
+
+// src/lib/ytDlpPlayerClient.ts
+var YOUTUBE_FALLBACK_CLIENT_ARGS = ["--extractor-args", "youtube:player_client=android"];
+var MEDIA_FETCH_REFUSED = "unable to download video data";
+var YOUTUBE_HOSTS = ["youtube.com", "youtu.be"];
+function isYouTube(url) {
+  let host;
+  try {
+    host = new URL(url).hostname.toLowerCase();
+  } catch {
+    return false;
+  }
+  return YOUTUBE_HOSTS.some((known) => host === known || host.endsWith(`.${known}`));
+}
+function needsPlayerClientRetry(url, stderr) {
+  return isYouTube(url) && stderr.toLowerCase().includes(MEDIA_FETCH_REFUSED);
+}
+
+// src/lib/ytDlp.ts
+var YT_DLP_BIN = "yt-dlp";
+var YT_DLP_TIMEOUT_MS = 10 * 60 * 1e3;
+var YT_DLP_PROBE_TIMEOUT_MS = 60 * 1e3;
+var TRUST_SYSTEM_CERTS = ["--compat-options", "no-certifi"];
+var YtDlpError = class extends Error {
+  constructor(message, stderrTail) {
+    super(message);
+    this.stderrTail = stderrTail;
+    this.name = "YtDlpError";
+  }
+};
+function tail(text, maxLines) {
+  return text.split("\n").slice(-maxLines).join("\n");
+}
+async function runYtDlp(args) {
+  const outTemplate = path2.join(args.workDir, "out.%(ext)s");
+  const infoPath = path2.join(args.workDir, "out.info.json");
+  const argv = [
+    args.url,
+    "--no-playlist",
+    "--no-progress",
+    "--no-warnings",
+    "--quiet",
+    "--write-info-json",
+    ...TRUST_SYSTEM_CERTS,
+    "-o",
+    outTemplate
+  ];
+  if (args.audioOnly) {
+    argv.push("-x", "--audio-format", "m4a");
+  } else {
+    argv.push(...videoFormatArgs(args.maxHeight));
+  }
+  const cookieFile = await writeVimeoCookieFile(args.vimeoCookie, args.workDir);
+  if (cookieFile) argv.push("--cookies", cookieFile);
+  args.onLog?.(`yt-dlp: downloading ${args.audioOnly ? "audio" : "video"}`);
+  const timeoutMs = args.timeoutMs ?? YT_DLP_TIMEOUT_MS;
+  const { code, stderr, tier, reducedQuality } = await downloadWithClientFallback({
+    url: args.url,
+    argv,
+    onLog: args.onLog,
+    ladder: (attemptArgv) => runViaLadder({ url: args.url, argv: attemptArgv, timeoutMs, onLog: args.onLog, country: args.country }),
+    repeat: (attemptArgv, route) => spawnAndWait(YT_DLP_BIN, attemptArgv, timeoutMs, { extraEnv: proxyEnvFor(route) })
+  });
+  if (code !== 0) {
+    throw new YtDlpError(`yt-dlp exited ${code}`, tail(stderr, 40));
+  }
+  const files = await readdir(args.workDir);
+  const wanted = args.audioOnly ? ".m4a" : ".mp4";
+  const downloaded = files.find((f) => f.startsWith("out.") && f.endsWith(wanted));
+  if (!downloaded) {
+    throw new YtDlpError(`yt-dlp finished but produced no ${wanted} file (saw: ${files.join(", ")})`, "");
+  }
+  let info = {};
+  try {
+    info = JSON.parse(await readFile(infoPath, "utf-8"));
+  } catch {
+  }
+  return { filePath: path2.join(args.workDir, downloaded), info, tier, reducedQuality };
+}
+async function probeYtDlp(args) {
+  const argv = [
+    args.url,
+    "--no-playlist",
+    "--no-warnings",
+    "--dump-single-json",
+    "--skip-download",
+    // Without this the probe fails wherever the *download* would fail, which
+    // defeats the point of probing: yt-dlp still runs format selection for
+    // `--dump-single-json`, so a binary too old to read a site's format list
+    // exits "Requested format is not available" and we learn nothing — measured
+    // against a stale build, which returned full metadata the moment this flag
+    // was added. Duration and id come from the page, not from a format.
+    "--ignore-no-formats-error",
+    ...TRUST_SYSTEM_CERTS
+  ];
+  if (args.cookieDir) {
+    const cookieFile = await writeVimeoCookieFile(args.vimeoCookie, args.cookieDir);
+    if (cookieFile) argv.push("--cookies", cookieFile);
+  }
+  let stdout = "";
+  const { code, stderr, tier } = await walkYtDlpLadder({
+    url: args.url,
+    credentials: captureProxyCredentials(),
+    attempt: async (route) => {
+      const result = await spawnAndWait(YT_DLP_BIN, argv, args.timeoutMs ?? YT_DLP_PROBE_TIMEOUT_MS, {
+        extraEnv: proxyEnvFor(route),
+        captureStdout: true
+      });
+      stdout = result.stdout ?? "";
+      return result;
+    },
+    onLog: args.onLog,
+    country: args.country
+  });
+  if (code !== 0) {
+    throw new YtDlpError(`yt-dlp exited ${code}`, tail(stderr, 40));
+  }
+  let parsed;
+  try {
+    parsed = JSON.parse(stdout);
+  } catch {
+    throw new YtDlpError("yt-dlp returned metadata that could not be read", tail(stdout, 5));
+  }
+  if (parsed === null || typeof parsed !== "object") {
+    throw new YtDlpError("yt-dlp returned no metadata for that link", tail(stderr, 40));
+  }
+  return { info: parsed, tier };
+}
+function isAudioOnly(info) {
+  return info.vcodec === "none";
+}
+function durationSeconds(info) {
+  const raw = info.duration;
+  return typeof raw === "number" && Number.isFinite(raw) && raw > 0 ? raw : void 0;
+}
+function canonicalUrl(info) {
+  const raw = info.webpage_url;
+  return typeof raw === "string" && raw.length > 0 ? raw : void 0;
+}
+async function walkYtDlpLadder(args) {
+  const routes = plannedRoutes(args.url, args.credentials, args.country ? { country: args.country } : void 0);
+  let last = { code: -1, stderr: "", tier: "direct", route: DIRECT_ROUTE };
+  for (const route of routes) {
+    const tier = route.kind === "proxy" ? route.tier : "direct";
+    if (route.kind === "proxy") args.onLog?.(`yt-dlp: retrying via ${tier}`);
+    last = { ...await args.attempt(route), tier, route };
+    if (last.code === 0) return last;
+    if (!shouldEscalate(ytDlpBlockSignal(last.stderr))) return last;
+  }
+  return last;
+}
+async function downloadWithClientFallback(args) {
+  const first = await args.ladder(args.argv);
+  if (first.code === 0 || !needsPlayerClientRetry(args.url, first.stderr)) {
+    return { ...first, reducedQuality: false };
+  }
+  args.onLog?.("yt-dlp: retrying with the fallback player client");
+  const again = await args.repeat([...args.argv, ...YOUTUBE_FALLBACK_CLIENT_ARGS], first.route);
+  return { ...again, tier: first.tier, route: first.route, reducedQuality: again.code === 0 };
+}
+function runViaLadder(args) {
+  return walkYtDlpLadder({
+    url: args.url,
+    credentials: captureProxyCredentials(),
+    attempt: (route) => spawnAndWait(YT_DLP_BIN, args.argv, args.timeoutMs, { extraEnv: proxyEnvFor(route) }),
+    onLog: args.onLog,
+    country: args.country
+  });
+}
+function proxyEnvFor(route) {
+  if (route.kind !== "proxy") return void 0;
+  const uri = new URL(route.server);
+  uri.username = encodeURIComponent(route.username);
+  uri.password = encodeURIComponent(route.password);
+  return {
+    HTTP_PROXY: uri.toString(),
+    HTTPS_PROXY: uri.toString(),
+    http_proxy: uri.toString(),
+    https_proxy: uri.toString()
+  };
+}
+function spawnAndWait(bin, argv, timeoutMs, options = {}) {
+  return new Promise((resolve, reject) => {
+    const child = spawn(bin, argv, {
+      stdio: ["ignore", "pipe", "pipe"],
+      // `childEnvWith` because `spawn`'s `env` REPLACES the environment rather
+      // than extending it — passing only the additions would strip PATH.
+      ...options.extraEnv ? { env: childEnvWith(options.extraEnv) } : {}
+    });
+    const stderrChunks = [];
+    const stdoutChunks = [];
+    let stderrLen = 0;
+    const MAX_STDERR = 256 * 1024;
+    child.stdout.on("data", (chunk) => {
+      if (options.captureStdout) stdoutChunks.push(chunk);
+    });
+    child.stderr.on("data", (chunk) => {
+      if (stderrLen < MAX_STDERR) {
+        stderrChunks.push(chunk);
+        stderrLen += chunk.length;
+      }
+    });
+    const timer = setTimeout(() => {
+      child.kill("SIGKILL");
+      reject(new Error(`${bin} timed out after ${timeoutMs}ms`));
+    }, timeoutMs);
+    child.on("error", (err) => {
+      clearTimeout(timer);
+      reject(new Error(`failed to spawn "${bin}" \u2014 ${err.message}`));
+    });
+    child.on("close", (code) => {
+      clearTimeout(timer);
+      resolve({
+        code: code ?? -1,
+        stderr: Buffer.concat(stderrChunks).toString("utf-8"),
+        ...options.captureStdout ? { stdout: Buffer.concat(stdoutChunks).toString("utf-8") } : {}
+      });
+    });
+  });
+}
+
+export {
+  isVimeoAuthFailure,
+  ytDlpBlockSignal,
+  YT_DLP_BIN,
+  YtDlpError,
+  runYtDlp,
+  probeYtDlp,
+  isAudioOnly,
+  durationSeconds,
+  canonicalUrl
+};
+//# sourceMappingURL=chunk-V5YMTW6L.js.map