@koda-sl/baker-cli 0.270.2-dev.f69bf4bac → 0.270.4-dev.12dd573e5

package/dist/cli.js

@@ -30,6 +30,7 @@ import {
   elementMentionKeywords,
   estimateVideoCredits,
   fetchExternalBytes,
+  frameRealismDirection,
   generateCatalog,
   imageProfileFor,
   isAudioOnly,
@@ -58,7 +59,7 @@ import {
   ulid,
   validateCanvasDeep,
   ytDlpBlockSignal
-} from "./chunk-LDDTPUYV.js";
+} from "./chunk-Y2GYULGB.js";
 import {
   csvOrJson,
   daysAgoIso,
@@ -32788,6 +32789,26 @@ function classifyWatchDefects(json, windows) {
     };
   });
 }
+var FidelityReview = z29.object({
+  delivers_the_ask: z29.boolean(),
+  missing: z29.array(z29.string()).default([]),
+  contradicted: z29.array(z29.string()).default([])
+});
+function buildFidelityPrompt(brief) {
+  return `Someone asked for an advertisement in these words:
+
+"""${brief.trim()}"""
+
+Watch and LISTEN to the video, then answer only this: does it deliver what they asked for? Judge the request as it was written \u2014 not what would make a good ad, and not what you would have made. Two lists: what they asked for that is NOT in the video, and what the video does that CONTRADICTS what they asked for. Quote their own words in each entry so the answer can be checked. Empty lists when it delivers.
+Answer JSON only: {"delivers_the_ask":true|false,"missing":["..."],"contradicted":["..."]}`;
+}
+function classifyFidelityFindings(json) {
+  const parsed = FidelityReview.safeParse(json);
+  if (!parsed.success) return [];
+  const r = parsed.data;
+  if (r.delivers_the_ask) return [];
+  return [...r.missing, ...r.contradicted].map((entry) => entry.trim()).filter(Boolean).map((what) => ({ severity: "blocking", what: `Not what was asked for: ${what}.` }));
+}
 function classifyWatchFindings(json) {
   const parsed = WatchReview.safeParse(json);
   if (!parsed.success) return [];
@@ -33004,11 +33025,13 @@ async function reviewRenderedVideo(opts) {
     const unread = perFrame.filter((f) => f.json === null).length;
     if (unread === perFrame.length && reel === null) return null;
     const watch = await askVideo(opts.video, buildWatchReviewPrompt({ market: opts.market, script: opts.script }));
+    const fidelity = opts.brief?.trim() ? await askVideo(opts.video, buildFidelityPrompt(opts.brief)) : null;
     const findings = [
       ...classifyFrameFindings(perFrame, opts.windows ?? []),
       ...classifyReelFindings(reel),
       ...classifyWatchFindings(watch),
       ...classifyWatchDefects(watch, opts.windows ?? []),
+      ...classifyFidelityFindings(fidelity),
       // Measured, not judged: a model looking at frames cannot hear the mix.
       ...await measureAudio(opts.video)
     ];
@@ -33047,6 +33070,14 @@ var runCommand = defineCommand102({
   meta: { name: "run", description: "Validate and execute a canvas JSON file." },
   args: {
     file: { type: "positional", required: true, description: "Path to canvas JSON" },
+    until: {
+      type: "string",
+      description: "Stop before the expensive half and leave the run resumable. `frames` settles the pictures and stops on the doorstep of the clips, so the look can be checked for the price of a few images instead of a whole ad. Continue with `baker canvas run <file> --run-id <the same id>` \u2014 everything already done is cached and is not paid for twice."
+    },
+    brief: {
+      type: "string",
+      description: "What the person asked for, IN THEIR OWN WORDS \u2014 paste their message, do not summarise it. The review compares the finished video against this and reports what they asked for that is missing or contradicted. Without it the review can only say whether the video is well made, not whether it is the one they wanted: every other check grades the render against the spec, and a spec that misread the request agrees with itself."
+    },
     "cache-dir": { type: "string", description: "Cache root (default ./canvas/.cache)" },
     "outputs-dir": { type: "string", description: "Per-run outputs root (default ./canvas)" },
     "run-id": { type: "string", description: "Override run id (also resumes that run, re-attaching its in-flight jobs)" },
@@ -33096,6 +33127,8 @@ var runCommand = defineCommand102({
       outputsDir: args["outputs-dir"] ? String(args["outputs-dir"]) : void 0,
       runId: args["run-id"] ? String(args["run-id"]) : void 0,
       fresh: args.fresh === true,
+      brief: args.brief ? String(args.brief) : void 0,
+      until: args.until ? String(args.until) : void 0,
       cachePolicy: args["cache-policy"] ? String(args["cache-policy"]) : void 0,
       regenerate: args.regenerate !== void 0 ? String(args.regenerate) : void 0,
       // --concurrency wins; --parallel is the discoverable alias for the same bound.
@@ -33118,6 +33151,10 @@ function resolveMaxCredits(...candidates) {
   }
   return void 0;
 }
+function stopBeforeKinds(until) {
+  if (until?.trim() === "frames") return /* @__PURE__ */ new Set(["video_generate"]);
+  return null;
+}
 async function executeCanvasRun(opts) {
   const filePath = path16.resolve(opts.file);
   const raw = await readFile12(filePath, "utf8");
@@ -33306,6 +33343,8 @@ ${describeRewrites(healed.rewrites)}
   if (progress && poster) {
     poster.startKeepalive(() => progress.hasPlan() ? progress.snapshot() : null);
   }
+  const frameUrls = [];
+  const isFrameNode = (id) => parsed.nodes?.find((n) => n.id === id)?.type === "image_generate";
   try {
     const policy = opts.cachePolicy ?? "read_write";
     const result = await engine.run(parsed, {
@@ -33313,12 +33352,24 @@ ${describeRewrites(healed.rewrites)}
       signal: abort.signal,
       cache_policy: policy,
       max_credits: opts.maxCredits,
+      ...stopBeforeKinds(opts.until) ? { stop_before_kinds: stopBeforeKinds(opts.until) } : {},
       concurrency: resolveConcurrency(opts.concurrency, process.env.BAKER_CANVAS_CONCURRENCY),
       regenerate,
-      onProgress: progress && poster ? (event) => {
-        progress.apply(event);
-        if (progress.hasPlan()) poster.enqueue(progress.snapshot());
-      } : void 0
+      // Always set, because the frame collector has to run whether or not this run is
+      // reporting progress to a backend — without it a checkpoint stops with nothing to
+      // show, which is the only thing a checkpoint is for.
+      onProgress: (event) => {
+        if (event.kind === "node_settled" && isFrameNode(event.run.node_id)) {
+          for (const out of Object.values(event.outputs ?? {})) {
+            const url = out?.url;
+            if (url) frameUrls.push(url);
+          }
+        }
+        if (progress && poster) {
+          progress.apply(event);
+          if (progress.hasPlan()) poster.enqueue(progress.snapshot());
+        }
+      }
     });
     await clearRunMarker(outputsDir, filePath);
     const keepRuns = opts.keepRuns;
@@ -33327,7 +33378,11 @@ ${describeRewrites(healed.rewrites)}
 `));
     }
     const reviewable = videoPathFromOutput(result.output);
-    const review = reviewable ? await reviewRenderedVideo({ video: reviewable, windows: await sceneWindowsBeside(filePath) }) : null;
+    const review = reviewable ? await reviewRenderedVideo({
+      video: reviewable,
+      windows: await sceneWindowsBeside(filePath),
+      brief: opts.brief?.trim() || void 0
+    }) : null;
     const hints2 = reviewable ? review === null ? [RENDER_REVIEW_UNAVAILABLE] : review.length === 0 ? ["Reviewed the finished video frame by frame \u2014 nothing to report."] : [
       `Reviewed the finished video and found ${review.length} thing${review.length === 1 ? "" : "s"} to fix. Show the user what is wrong before you show them the ad.`,
       ...review.map((f) => `${f.severity === "blocking" ? "MUST FIX" : "SHOULD FIX"} \u2014 ${f.what}`),
@@ -33368,6 +33423,27 @@ ${describeRewrites(healed.rewrites)}
 `);
       process.exit(abortedBy === "SIGTERM" ? 143 : 130);
     }
+    if (e instanceof RunAbortedError && e.reason === "checkpoint") {
+      await clearRunMarker(outputsDir, filePath);
+      const frames = frameUrls;
+      process.stdout.write(
+        `${JSON.stringify(
+          {
+            ok: true,
+            data: { stopped: "frames", runId, frames },
+            hints: [
+              `${e.message}`,
+              frames.length > 0 ? `Show these ${frames.length} frames and ask whether to carry on, BEFORE the clips are paid for. They are the ad's look, at a fraction of its price.` : "No frame outputs were captured \u2014 carry on rather than blocking on nothing.",
+              `Carry on with: baker canvas run ${filePath} --run-id ${runId}`
+            ]
+          },
+          null,
+          2
+        )}
+`
+      );
+      process.exit(0);
+    }
     if (e instanceof RunAbortedError && e.reason === "cost_cap") {
       await clearRunMarker(outputsDir, filePath);
       if (poster) await poster.flush(failedPayload(e.message));
@@ -34836,12 +34912,17 @@ function voiceoverMode(spec, nativeSpeech) {
   if (spec.voiceover === false) return "none";
   return nativeSpeech ? "on_camera" : "voiceover";
 }
+function castRole(spec) {
+  if (!spec.cast) return void 0;
+  const described = spec.cast.description?.trim();
+  return described ? `${described} \u2014 a customer telling their own story, not a worker` : "a customer telling their own story, not a worker";
+}
 function adSpecToBlueprint(spec, avatarAccent) {
   const total = spec.beats.length;
   const market = marketFor(spec);
   const place = market ? ` Set in ${market}: the architecture, streets and styling are ${market}'s.` : "";
   const currency = market ? CURRENCY_BY_MARKET[market.toLowerCase()] : void 0;
-  const physics = " Everything obeys real-world physics: paper, card and screens are OPAQUE with nothing showing through from behind, every object is at believable real-world scale next to the people handling it, and every object has its real-world form and construction \u2014 a phone has ONE screen and it is on the front. NO readable text or numbers anywhere in frame \u2014 phone screens, documents and signage stay illegible or out of focus, because any figure the model invents will contradict the script. Hands are kept simple: no close-up of fingers manipulating small parts, no hand gripping the edge of an object, and each person has exactly TWO arms and TWO legs, all attached and all visible or all out of frame. Anyone working does so the way the trade actually does it: nobody stands or kneels on the equipment being installed, nothing is fitted overhanging an edge or floating unsupported, and every part rests on the structure that would really carry it." + (currency ? ` If a currency is unavoidably visible it is ${currency}.` : "");
+  const physics = frameRealismDirection({ currency, role: castRole(spec) });
   const closesOnCard = endCardWanted(spec);
   const nativeSpeech = wantsNativeSpeech(spec, closesOnCard);
   let clock = 0;
@@ -35464,7 +35545,11 @@ var scaffoldAdCommand = defineCommand106({
       ok: true,
       data: { canvas: outPath, beats: spec.data.beats.length, slug },
       hints: [
-        `Wrote ${spec.data.beats.length} beats to ${outPath}. Run it with \`baker canvas run ${outPath}\`.`,
+        // `--until frames` on anything carrying a face or a mark, because that is where
+        // getting it wrong is expensive: the pictures cost a fraction of the clips, and
+        // every defect anyone has caught in a finished ad was visible in them. The run
+        // stops, the frames are shown, and continuing pays for none of them twice.
+        handle || staged ? `Wrote ${spec.data.beats.length} beats to ${outPath}. Run it with \`baker canvas run ${outPath} --until frames --brief "<what they asked for, their words>"\`, SHOW the frames it returns and ask whether to carry on \u2014 then continue with the \`--run-id\` it gives you. This ad carries ${handle ? "a person's face" : "the client's mark"}, and a wrong look costs a whole ad to discover after the clips.` : `Wrote ${spec.data.beats.length} beats to ${outPath}. Run it with \`baker canvas run ${outPath} --brief "<what they asked for, their words>"\`.`,
         ...autoCast ? [
           `Cast \`${autoCast}\` \u2014 the spec described this person in words, and they are one of this company's avatars. Every shot now grounds on their identity sheet and they speak in their own pinned voice. Write \`"cast": { "avatar": "` + autoCast + '" }` yourself next time; a written description is for someone the company does not have.'
         ] : [],