@koda-sl/baker-cli 0.466.0 → 0.467.0

package/dist/chunk-IK2S7QGF.js

@@ -0,0 +1,363 @@
+import {
+  MAX_VIDEO_INGEST_BYTES,
+  MAX_VIDEO_INGEST_SECONDS,
+  directDedupeKey,
+  downloadTimeoutMs,
+  isDirectMediaUrl,
+  isDrmProtected,
+  isRegionBlocked,
+  isToolingOrBlockFailure,
+  refuseByDuration,
+  resolveIngestSource,
+  worthDownloadingInstead
+} from "./chunk-FKOECEVC.js";
+import {
+  YtDlpError,
+  canonicalUrl,
+  durationSeconds,
+  isAudioOnly,
+  isVimeoAuthFailure,
+  probeYtDlp,
+  runYtDlp,
+  ytDlpBlockSignal
+} from "./chunk-GT4UFORX.js";
+import {
+  videosIngestRequestSchema
+} from "./chunk-WXC3I4CT.js";
+import {
+  ApiError,
+  apiPost
+} from "./chunk-KB5XJHQX.js";
+import {
+  captureVimeoCookie,
+  isProxyFailure
+} from "./chunk-QXDEZKAR.js";
+import {
+  registerSchema,
+  writeJson
+} from "./chunk-4DM6OAQ6.js";
+
+// src/commands/videos/ingest.ts
+import { mkdtemp, rm, stat } from "fs/promises";
+import { tmpdir } from "os";
+import path from "path";
+import { defineCommand } from "citty";
+
+// src/lib/streamUpload.ts
+import { createHash } from "crypto";
+import { createReadStream } from "fs";
+import { Readable } from "stream";
+import { pipeline } from "stream/promises";
+async function hashFile(filePath) {
+  const hash = createHash("sha256");
+  await pipeline(createReadStream(filePath), hash);
+  return hash.digest("hex");
+}
+async function putFileStreamed(args) {
+  const body = Readable.toWeb(createReadStream(args.filePath));
+  return await fetch(args.putUrl, {
+    method: "PUT",
+    body,
+    headers: {
+      "Content-Type": args.contentType,
+      "Content-Length": String(args.sizeBytes)
+    },
+    duplex: "half"
+  });
+}
+
+// src/commands/videos/ingest.ts
+registerSchema({
+  command: "videos.ingest",
+  description: "Add a video to the library from a URL \u2014 a direct file, or a YouTube/TikTok/Vimeo page.",
+  args: {
+    url: { type: "string", description: "Video URL (direct file or a watch/post page)", required: true },
+    source: { type: "string", description: "Provenance: url | youtube", required: false },
+    "external-id": { type: "string", description: "Provider asset id (deduped on)", required: false },
+    "external-url": { type: "string", description: "Canonical page URL", required: false },
+    download: {
+      type: "boolean",
+      description: "Force the download path instead of letting Baker fetch the URL directly",
+      required: false,
+      default: false
+    },
+    country: {
+      type: "string",
+      description: "Two-letter country to fetch from (e.g. US, DE, JP) \u2014 use when a video is blocked in your region",
+      required: false
+    },
+    context: {
+      type: "string",
+      description: "What you know that the video cannot show \u2014 the campaign, whether the person on camera is a real customer, what it is for. Highest-leverage flag here: it is what makes the clip findable later.",
+      required: false
+    },
+    "dry-run": { type: "boolean", description: "Preview the operation without executing", required: false }
+  }
+});
+var ingestCommand = defineCommand({
+  meta: {
+    name: "ingest",
+    description: "Add a video to the library from a URL. A direct file URL is handed straight to Baker, which fetches it. A page URL (YouTube, TikTok, Vimeo, Instagram) is downloaded here first, then uploaded \u2014 and a direct URL that Baker cannot fetch falls back to that same path automatically.\n\nExample: baker videos ingest https://www.youtube.com/watch?v=abc123"
+  },
+  args: {
+    url: { type: "positional", description: "Video URL", required: false },
+    source: { type: "string", description: "Provenance: url | youtube", required: false },
+    "external-id": { type: "string", description: "Provider asset id", required: false },
+    "external-url": { type: "string", description: "Canonical page URL", required: false },
+    download: { type: "boolean", description: "Force the download path", required: false, default: false },
+    country: { type: "string", description: "Two-letter country to fetch from, e.g. US", required: false },
+    context: {
+      type: "string",
+      description: "Context for the analysis \u2014 campaign, who is on camera, intended use",
+      required: false
+    },
+    "dry-run": { type: "boolean", description: "Preview without executing", required: false, default: false }
+  },
+  run: ({ args }) => runVideoIngest({
+    url: args.url,
+    source: args.source,
+    externalId: args["external-id"],
+    externalUrl: args["external-url"],
+    forceDownload: args.download === true,
+    country: args.country,
+    context: args.context,
+    dryRun: args["dry-run"] === true
+  })
+});
+async function runVideoIngest(opts) {
+  const url = opts.url;
+  if (!url) {
+    writeJson({ ok: false, error: { code: "VALIDATION_ERROR", message: "URL is required" } });
+    process.exit(1);
+  }
+  const direct = !opts.forceDownload && isDirectMediaUrl(url);
+  const country = opts.country?.trim() || void 0;
+  const args = {
+    url,
+    source: resolveIngestSource(opts.source, url),
+    externalId: opts.externalId,
+    // Deliberately NOT pre-filled with the pasted URL. Doing so shadowed the
+    // canonical URL the probe resolves (`?? canonicalUrl(...)` below could never
+    // run), so the same video filed under a different `externalUrl` depending on
+    // whether the link was copied from a share sheet, a shorts page or the
+    // address bar. `downloadThenIngest` falls back to `args.url` when the probe
+    // resolves nothing, which is the value this used to hard-code.
+    externalUrl: opts.externalUrl,
+    descriptionContext: opts.context?.trim() || void 0
+  };
+  const validation = videosIngestRequestSchema.safeParse({ ...args, url });
+  if (!validation.success) {
+    const problem = validation.error.issues[0];
+    writeJson({
+      ok: false,
+      error: {
+        code: "VALIDATION_ERROR",
+        message: problem ? `${problem.path.join(".") || "request"}: ${problem.message}` : "Invalid request"
+      }
+    });
+    process.exit(1);
+  }
+  if (opts.dryRun) {
+    reportDryRun(args, direct, country);
+    return;
+  }
+  try {
+    const data = await ingestByRoute(args, direct, country);
+    writeJson({ ok: true, data, hints: ingestHints(data) });
+  } catch (err) {
+    reportIngestFailure(err);
+  }
+}
+function ingestHints(data) {
+  if (data.reducedQuality !== true) return void 0;
+  return [
+    "Added, but at standard quality (360p): the site refused to release the full-quality file and this was the best it would hand over. Tell the user \u2014 the clip is in the library and usable, it is just not high-resolution. If they need the crisp version, ask them for the original file and use `baker videos upload <file>`."
+  ];
+}
+function reportDryRun(args, direct, country) {
+  writeJson({
+    ok: true,
+    dryRun: true,
+    operation: "videos.ingest",
+    params: {
+      url: args.url,
+      source: args.source,
+      // Known up front on the direct route, because it is derived from the URL
+      // alone. The download route's key comes from the probe, which a dry run
+      // does not perform, so there it stays null rather than becoming a guess.
+      externalId: args.externalId ?? (direct ? directDedupeKey(args.source, args.url) : null),
+      // A page URL's canonical form is only known after the probe, which a dry
+      // run does not perform — so this is null rather than a guess.
+      externalUrl: args.externalUrl ?? null,
+      direct,
+      country: country ?? null,
+      descriptionContext: args.descriptionContext ?? null
+    }
+  });
+}
+async function ingestByRoute(args, direct, country) {
+  if (!direct) {
+    return { ...await downloadThenIngest(args, country), via: "download" };
+  }
+  try {
+    const externalId = args.externalId ?? directDedupeKey(args.source, args.url);
+    return { ...await ingestUrl({ ...args, externalId }), via: "direct" };
+  } catch (err) {
+    if (!(err instanceof ApiError) || !worthDownloadingInstead(err.code)) throw err;
+    return { ...await downloadThenIngest(args, country), via: "download", fallbackFrom: err.message };
+  }
+}
+function reportYtDlpFailure(err) {
+  const detail = `${err.stderrTail} ${err.message}`;
+  if (isProxyFailure(ytDlpBlockSignal(detail))) {
+    fail(
+      "Couldn't download that video: our connection to the internet was refused.",
+      "This is Baker's egress proxy, not the link \u2014 its credentials look wrong or expired. The link is probably fine. Report it; retrying won't help until the proxy is fixed."
+    );
+  }
+  if (isDrmProtected(detail)) {
+    fail(
+      "That video is copy-protected, so it can't be added to the library.",
+      "The publisher encrypted this one against downloading. Nothing will change that \u2014 not retrying, not a different region. Ask whoever owns the video for the original file and add it with `baker videos upload <file>`."
+    );
+  }
+  if (isVimeoAuthFailure(detail)) {
+    fail(
+      "Vimeo wouldn't hand over that video \u2014 Baker isn't signed in to Vimeo.",
+      "Vimeo now refuses anonymous downloads entirely, so this needs a signed-in session and the one Baker holds has either expired or was never set. Nothing is wrong with the link. Report it so the Vimeo cookie can be refreshed; meanwhile download the file and use `baker videos upload <file>`."
+    );
+  }
+  if (isRegionBlocked(detail)) {
+    fail(
+      `That video isn't available in the region we're fetching from. ${err.stderrTail || err.message}`,
+      "Re-run with `--country <two-letter code>` for a country where the video plays \u2014 e.g. `--country US`. Pick the one the video belongs to (the uploader's country is the usual answer); each attempt costs a paid connection, so choose rather than work through the list."
+    );
+  }
+  fail(
+    `Couldn't download that video. ${err.stderrTail || err.message}`,
+    isToolingOrBlockFailure(detail) ? "The site wouldn't hand over the video \u2014 usually the downloader being out of date against a site that changed, or the request being blocked. Retrying won't help, and the link is probably fine. Report it so the tool can be updated; meanwhile download the file and use `baker videos upload <file>`." : "Check the link is public and playable. Private or age-restricted videos can't be downloaded \u2014 download the file yourself and use `baker videos upload <file>`."
+  );
+}
+function fail(message, fix, code = "INGEST_FAILED") {
+  writeJson({ ok: false, error: { code, message, fix } });
+  process.exit(1);
+}
+function reportIngestFailure(err) {
+  if (err instanceof ApiError) {
+    writeJson({ ok: false, error: { code: err.code, message: err.message } });
+    process.exit(1);
+  }
+  if (err instanceof YtDlpError) reportYtDlpFailure(err);
+  const detail = err instanceof Error ? err.message : String(err);
+  if (detail.includes("timed out after")) {
+    fail(
+      "That video took too long to download and was stopped part-way.",
+      "Long or slow-serving videos can outrun the download window. Try a shorter clip, or download this one yourself and use `baker videos upload <file>`.",
+      "INTERNAL_ERROR"
+    );
+  }
+  fail(
+    `Couldn't add that video. ${detail}`,
+    "This is a fault on Baker's side, not a problem with the link. Report it.",
+    "INTERNAL_ERROR"
+  );
+}
+function dedupeKey(args, info) {
+  const id = typeof info.id === "string" ? info.id : void 0;
+  if (!id) return void 0;
+  if (info.extractor === "generic") return `${args.source}:${canonicalUrl(info) ?? args.url}`;
+  return `${args.source}:${id}`;
+}
+function ingestUrl(args) {
+  const body = {
+    url: args.url,
+    source: args.source
+  };
+  if (args.externalId) body.externalId = args.externalId;
+  if (args.externalUrl) body.externalUrl = args.externalUrl;
+  if (args.descriptionContext) body.descriptionContext = args.descriptionContext;
+  return apiPost("/api/videos/ingest", body);
+}
+async function downloadThenIngest(args, country) {
+  const vimeoCookie = captureVimeoCookie();
+  const workDir = await mkdtemp(path.join(tmpdir(), "videos-ingest-"));
+  try {
+    const probe = await probeYtDlp({ url: args.url, country, vimeoCookie, cookieDir: workDir });
+    if (isAudioOnly(probe.info)) {
+      throw new ApiError(
+        "VALIDATION_ERROR",
+        "That link is audio, not video, so it can't go in the video library. Music and podcast links have no picture to store."
+      );
+    }
+    const tooLong = refuseByDuration(durationSeconds(probe.info), MAX_VIDEO_INGEST_SECONDS);
+    if (tooLong) {
+      throw new ApiError("VALIDATION_ERROR", tooLong);
+    }
+    const externalId = args.externalId ?? dedupeKey(args, probe.info);
+    const { filePath, tier, reducedQuality } = await runYtDlp({
+      url: args.url,
+      audioOnly: false,
+      workDir,
+      // Deliberately uncapped: take the best the site will release and let Mux
+      // decide what to serve. The 1080p ceiling this replaces was reasoning
+      // about the wrong cost — it saved download bytes by throwing away source
+      // pixels *before* the one system that can make renditions from them ever
+      // saw them, so a 4K master became a 1080p master permanently, for
+      // everything the library is ever asked to do with that clip afterwards.
+      // Downscaling is Mux's job and it is cheap; re-acquiring a master is not.
+      country,
+      vimeoCookie,
+      // Budgeted from the probed length, so the duration we accept and the time
+      // we allow to fetch it cannot drift apart.
+      timeoutMs: downloadTimeoutMs(durationSeconds(probe.info))
+    });
+    const stats = await stat(filePath);
+    if (stats.size > MAX_VIDEO_INGEST_BYTES) {
+      throw new ApiError(
+        "VALIDATION_ERROR",
+        `That video is ${(stats.size / 1024 ** 3).toFixed(1)} GB, over the ${Math.round(MAX_VIDEO_INGEST_BYTES / 1024 ** 3)} GB limit for this route.`
+      );
+    }
+    const publicUrl = await uploadToAssetStore(filePath, stats.size);
+    return {
+      ...await ingestUrl({
+        ...args,
+        url: publicUrl,
+        externalId,
+        externalUrl: args.externalUrl ?? canonicalUrl(probe.info) ?? args.url
+      }),
+      egress: tier,
+      // Named in the result for the same reason `egress` is: a geo-targeted exit
+      // is a deliberate, billable choice and must not be invisible afterwards.
+      ...country ? { country } : {},
+      // Third member of the same family. The site refused the good formats and
+      // the download quietly took the one it would release, which is the right
+      // call — but a clip that is 360p because of that must say so here, or the
+      // only way to find out is to watch it.
+      ...reducedQuality ? { reducedQuality: true } : {}
+    };
+  } finally {
+    await rm(workDir, { recursive: true, force: true }).catch(() => {
+    });
+  }
+}
+async function uploadToAssetStore(filePath, sizeBytes) {
+  const sha256 = await hashFile(filePath);
+  const { putUrl, publicUrl } = await apiPost("/api/canvas/assets/presign", {
+    sha256,
+    mime: "video/mp4",
+    purpose: "source"
+  });
+  const putRes = await putFileStreamed({ putUrl, filePath, contentType: "video/mp4", sizeBytes });
+  if (!putRes.ok) {
+    throw new ApiError("INTERNAL_ERROR", `Storing the downloaded video failed: HTTP ${putRes.status}`);
+  }
+  return publicUrl;
+}
+
+export {
+  ingestCommand,
+  runVideoIngest,
+  ingestHints
+};
+//# sourceMappingURL=chunk-IK2S7QGF.js.map

package/dist/chunk-INI7U2SZ.js

@@ -0,0 +1,188 @@
+import {
+  analyticsErrorEnvelope
+} from "./chunk-JHLGXB6K.js";
+import {
+  PEOPLE_SEARCH_MAX_TERMS,
+  parsePeopleQuery,
+  peopleLookupRequestSchema
+} from "./chunk-SSJK7FEV.js";
+import {
+  writeJsonEnvelope
+} from "./chunk-LHNL6JM2.js";
+import {
+  apiPost
+} from "./chunk-KB5XJHQX.js";
+import {
+  registerSchema
+} from "./chunk-4DM6OAQ6.js";
+
+// src/commands/analytics/lookup.ts
+import { readFileSync } from "fs";
+import { defineCommand } from "citty";
+function peopleArgs(args2) {
+  const parts = peopleParts(args2);
+  if (parts.length === 0) return {};
+  const people = parsePeopleQuery(parts).terms.map((term) => term.input);
+  return people.length === 0 ? {} : { people };
+}
+function peopleParts(args2) {
+  return [...args2.person ? [String(args2.person)] : [], ...args2["people-file"] ? [readPeopleFile(args2)] : []];
+}
+function readPeopleFile(args2) {
+  try {
+    return readFileSync(String(args2["people-file"]), "utf8");
+  } catch {
+    return "";
+  }
+}
+function unreadablePeopleFile(args2) {
+  if (!args2["people-file"]) return void 0;
+  const path = String(args2["people-file"]);
+  try {
+    readFileSync(path, "utf8");
+    return void 0;
+  } catch {
+    return `Couldn't read --people-file ${path}. Check the path, or pass the people inline with --person.`;
+  }
+}
+function peopleLimitError(args2) {
+  const unreadable = unreadablePeopleFile(args2);
+  if (unreadable !== void 0) return unreadable;
+  const { terms, dropped } = parsePeopleQuery(peopleParts(args2));
+  if (dropped === 0) return void 0;
+  return `${terms.length + dropped} people named; one search takes up to ${PEOPLE_SEARCH_MAX_TERMS}. Split the list and run each part.`;
+}
+function compactRow(row) {
+  const clicks = [...row.googleClick ? ["google"] : [], ...row.metaClick ? ["meta"] : []];
+  const candidates = {
+    input: row.input,
+    readAs: row.kind,
+    found: row.found,
+    problem: row.problem,
+    contacts: row.contacts,
+    events: row.events,
+    firstSeen: row.firstSeen,
+    lastSeen: row.lastSeen,
+    pageViews: row.pageViews,
+    landings: row.landings,
+    visits: row.visits,
+    formSubmits: row.formSubmits,
+    firstSource: row.firstSource,
+    systemEvents: row.systemEvents,
+    pageEvents: row.pageEvents,
+    uploads: row.uploads,
+    uploadedTo: row.uploadedTo,
+    clicks,
+    knownBy: row.knownBy
+  };
+  return Object.fromEntries(
+    Object.entries(candidates).filter(
+      ([key, value]) => key === "found" || !(value === 0 || value === "" || Array.isArray(value) && value.length === 0)
+    )
+  );
+}
+function lookupHints(data) {
+  const hints = [];
+  if (data.dropped > 0) {
+    hints.push(
+      `${data.dropped} lines past the limit of ${PEOPLE_SEARCH_MAX_TERMS} were not looked up. Run the rest separately.`
+    );
+  }
+  const unseen = data.people.filter((row) => !row.found && row.problem === "");
+  if (unseen.length > 0) {
+    hints.push(
+      `Not seen by Baker: ${unseen.map((row) => row.input).join(", ")}. Baker knows a person by what its pages, Forms and connected systems recorded \u2014 a lead that came in through a form or site Baker does not measure, or a source not connected, is not here.`
+    );
+  }
+  for (const row of data.people.filter((person) => person.problem !== "")) {
+    hints.push(`${row.input}: ${row.problem}`);
+  }
+  const found = data.people.filter((row) => row.found);
+  const neverVisited = found.filter((row) => row.pageViews === 0);
+  if (neverVisited.length > 0) {
+    hints.push(
+      `${neverVisited.length} of ${found.length} found never visited a page Baker publishes: Baker knows them only from what their systems reported, so no landing can be credited with them.`
+    );
+  }
+  if (found.length > 0) {
+    hints.push(
+      `\`baker analytics feed --person '${found[0]?.input}' --days 90\` lists everything one of them did, in order \u2014 add more, comma-separated, to follow several.`
+    );
+  }
+  return hints;
+}
+function lookupReport(data, full = false) {
+  const found = data.people.filter((row) => row.found);
+  return {
+    summary: {
+      searched: data.people.length,
+      found: found.length,
+      notFound: data.people.length - found.length,
+      visitedABakerPage: found.filter((row) => row.pageViews > 0).length
+    },
+    people: full ? data.people : data.people.map(compactRow),
+    hints: lookupHints(data)
+  };
+}
+var NO_PEOPLE = "Name at least one person: --person 'ana@empresa.com, deal:40001234567' or --people-file leads.txt.";
+var args = {
+  person: {
+    type: "string",
+    description: "The people, comma-separated: an email, a phone (+34\u2026), a HubSpot contact id, deal:<HubSpot deal id>, a gclid/fbclid, a browser id. Prefix to be exact: email:, phone:, hubspot:, deal:, gclid:, fbclid:, browser:",
+    required: false
+  },
+  "people-file": {
+    type: "string",
+    description: "A file with one person per line \u2014 a column pasted from a CRM export",
+    required: false
+  },
+  full: {
+    type: "boolean",
+    description: "Every field of every line, zeros included",
+    required: false
+  }
+};
+registerSchema({
+  command: "analytics.lookup",
+  description: "What Baker saw of each person in a list, by email, phone, CRM id, deal or click id",
+  args
+});
+var lookupCommand = defineCommand({
+  meta: {
+    name: "lookup",
+    description: `What Baker saw of each person in a list \u2014 all time, one line each \u2014 by whatever the client holds about them. The reconciliation question: "HubSpot says 26 leads yesterday, which of them does Baker have, and where did they come from".
+
+Start here:
+  baker analytics lookup --person 'ana@empresa.com, +34600111222, 90001234567'
+  baker analytics lookup --person 'deal:40001234567, deal:40001234568'   \u2014 HubSpot deal ids, looked up through the deal's contacts
+  baker analytics lookup --people-file leads.txt                           \u2014 one per line, up to ${PEOPLE_SEARCH_MAX_TERMS}
+
+Each line says whether Baker has them, when it first and last saw them, the pages and Forms they touched, what their systems reported (systemEvents: CRM stages, orders), what Baker uploaded for them to ad platforms, which ad click they carried, and every kind of key they are linked by. A bare number is read as a HubSpot contact id or a phone; prefix it to be exact. Emails and phones are hashed by Baker \u2014 say them to the client as they gave them, never as a hash. Follow one with \`baker analytics feed --person <same value>\`.`
+  },
+  args,
+  run: async ({ args: raw }) => {
+    const { people } = peopleArgs(raw);
+    const parsed = peopleLookupRequestSchema.safeParse({ people: people ?? [] });
+    const invalid = peopleLimitError(raw) ?? (parsed.success ? void 0 : NO_PEOPLE);
+    if (invalid !== void 0 || !parsed.success) {
+      writeJsonEnvelope({ ok: false, error: { code: "VALIDATION_ERROR", message: invalid ?? NO_PEOPLE } });
+      process.exit(1);
+    }
+    try {
+      const response = await apiPost("/api/analytics/people", parsed.data);
+      const { hints, ...data } = lookupReport(response.data, raw.full === true);
+      writeJsonEnvelope({ ok: true, data, ...hints.length > 0 ? { hints } : {} });
+    } catch (error) {
+      writeJsonEnvelope({ ok: false, error: analyticsErrorEnvelope(error) });
+      process.exit(1);
+    }
+  }
+});
+
+export {
+  peopleArgs,
+  peopleLimitError,
+  lookupReport,
+  lookupCommand
+};
+//# sourceMappingURL=chunk-INI7U2SZ.js.map