@shop-os/modelkit 0.4.3 → 0.4.5

package/CHANGELOG.md

@@ -9,6 +9,18 @@ header in the release PR.
 
 ## [Unreleased]
 
+## [0.4.5] - Unreleased
+
+### Fixed
+- Video length maximums and combined totals forgive ffprobe's container over-read. ffprobe reads a clip a few hundredths over its nominal length (a 10 s clip reads 10.04 s), so a clip at the limit was refused. A length is now over a maximum only when it and its tenth are both over: a source that reads 10.04 s passes Gemini Omni's 10 s cap, one that reads 10.05 s or more is still refused, and a limit that is not a whole tenth (Kling O1's 10.05 s) still takes a clip at it. A combined total adds each clip's tenth, so three 5.04 s references fit a 15 s total. The rule is never stricter than the old raw comparison. It applies to the reference maximum and total and the source maximum in `videoLengthViolation`, and to the source cap in `planVideoTool`. Minimums stay raw: the over-read makes a clip that reads short shorter still, so 3.96 s is still under a 4 s minimum.
+
+## [0.4.4] - Unreleased
+
+### Added
+- `inputSpec.dimensions` on image rows: `{ field, areas, maxSide?, multipleOf? }` sends one `{ width, height }` object in `field`, computed from the request's aspect ratio and size. `areas` is the target pixel count per size tier; the long side is capped at `maxSide` and both sides round down to `multipleOf`. Nothing is sent when the ratio is missing or `adaptive`, or the size has no area, so the provider default applies (on an edit endpoint, the first reference's size). A spec with `dimensions` cannot also map `aspectRatio` or `size` in `fields`, and `field` cannot be a name another mapping already sends. This lets a fal row whose endpoint only takes `image_size` presets or `{ width, height }` (FLUX 3 Image Klein) offer 2K and 4K as data.
+
+Add a row that carries `inputSpec.dimensions` only after every app that will call it runs 0.4.4. An older kit rejects the unknown key, so the row is quarantined there.
+
 ## [0.4.3] - Unreleased
 
 ### Fixed

package/README.md

@@ -224,6 +224,7 @@ last. Video rows use `VIDEO_INPUT_SPEC_KEYS` (`prompt`, `durationSeconds`,
 with `values`); audio rows use
 `AUDIO_INPUT_SPEC_KEYS` (`prompt`, `durationSeconds`, `seed`, `negativePrompt`,
 `references`, `sourceVideo`, `instrumental`). `aspectRatio: "adaptive"` is omitted. A fal image row has no kind and builds its whole input from the spec; the optional `editEndpoint` (fal rows only) is the fal endpoint called instead of `providerModelId` when the request has references. fal rows without an `inputSpec` keep the Muse-shaped request and Ideogram keeps its own path; an `inputSpec` on an Ideogram row replaces that path. `editEndpoint` needs a `references` entry in `fields`. `inputSpec` on fal video rows is not supported.
+`dimensions` (image rows only, from 0.4.4) sends one `{ width, height }` object for a provider that takes pixel sizes instead of a ratio and a tier: `dimensions: { field: "image_size", areas: { "1K": 1048576, "2K": 4194304, "4K": 16777216 }, maxSide: 4096, multipleOf: 16 }`. Each `areas` entry is the target pixel count for that size; the long side is capped at `maxSide`, and both sides round down to `multipleOf`. With no ratio, `adaptive`, or a size with no area, the field is omitted. A spec with `dimensions` maps neither `aspectRatio` nor `size` in `fields`.
 `arrays` joins several image inputs into one provider array field, in the listed order: `arrays: { images: ["firstFrame", "referenceImages", "lastFrame"] }` (video rows: `firstFrame`, `referenceImages`, `lastFrame`; image rows: `references`). A key is mapped once, in `fields` or in one array, and the field is omitted when none of its inputs is set. A row that maps `sourceVideo` also needs `capabilities.acceptsSourceVideo`; set `capabilities.sourceVideoExcludesFrames` when the provider rejects a source video next to images.
 `validateInputSpec` is the shared contract the write schema, `dataOnlyCapable`, and
 the Replicate mapper all check. `dataOnlyCapable(row)` reports whether such a row would

package/dist/catalog.cjs

@@ -2563,16 +2563,26 @@ var ArraysSchema = import_zod2.z.record(import_zod2.z.string().min(1), import_zo
     }
   }
 });
+var DimensionsSchema = import_zod2.z.strictObject({
+  field: import_zod2.z.string().min(1),
+  areas: import_zod2.z.record(import_zod2.z.string(), import_zod2.z.number().int().positive()).refine((a) => Object.keys(a).length > 0, {
+    message: "inputSpec.dimensions.areas must name at least one size"
+  }),
+  maxSide: import_zod2.z.number().int().positive().optional(),
+  multipleOf: import_zod2.z.number().int().positive().optional()
+});
 var InputSpecSchema = import_zod2.z.strictObject({
   fields: FieldsSchema,
   values: ValuesSchema.optional(),
   static: StaticSchema.optional(),
   arrays: ArraysSchema.optional(),
-  editEndpoint: import_zod2.z.string().min(1).optional()
+  editEndpoint: import_zod2.z.string().min(1).optional(),
+  dimensions: DimensionsSchema.optional()
 }).superRefine((spec, ctx) => {
   const mappedNames = /* @__PURE__ */ new Set([
     ...Object.values(spec.fields),
-    ...Object.keys(spec.arrays ?? {})
+    ...Object.keys(spec.arrays ?? {}),
+    ...spec.dimensions ? [spec.dimensions.field] : []
   ]);
   for (const key of Object.keys(spec.static ?? {})) {
     if (mappedNames.has(key)) {
@@ -2601,7 +2611,10 @@ function validateInputSpec(raw, row) {
     return {
       ok: false,
       issues: [
-        { path: [], message: "inputSpec is only supported on replicate and fal rows" }
+        {
+          path: [],
+          message: "inputSpec is only supported on replicate and fal rows"
+        }
       ]
     };
   }
@@ -2694,6 +2707,29 @@ function validateInputSpec(raw, row) {
       }
     }
   }
+  if (spec.dimensions !== void 0) {
+    if (row.type !== "image") {
+      issues.push({
+        path: ["dimensions"],
+        message: "inputSpec.dimensions is only supported on image rows"
+      });
+    }
+    for (const key of ["aspectRatio", "size"]) {
+      if (key in spec.fields) {
+        issues.push({
+          path: ["fields", key],
+          message: `inputSpec.fields "${key}" cannot be set with inputSpec.dimensions, which sends it`
+        });
+      }
+    }
+    const owner = seenNames.get(spec.dimensions.field);
+    if (owner !== void 0 || spec.dimensions.field in (spec.arrays ?? {})) {
+      issues.push({
+        path: ["dimensions", "field"],
+        message: `inputSpec.dimensions.field "${spec.dimensions.field}" is already a mapped provider field`
+      });
+    }
+  }
   const supports = row.constraints.supports ?? [];
   for (const key of INPUT_SPEC_SUPPORTS_KEYS) {
     if (supports.includes(key)) continue;
@@ -3288,6 +3324,20 @@ function resetLiveCatalog() {
 
 // src/catalog/video-tools-spec.ts
 var import_zod6 = require("zod");
+
+// src/video-length.ts
+function tenths(seconds2) {
+  return Math.round(seconds2 * 10) / 10;
+}
+function overMax(seconds2, max) {
+  return seconds2 > max && tenths(seconds2) > max;
+}
+function overTotal(lengths, total) {
+  const sum = (values) => values.reduce((acc, len) => acc + len, 0);
+  return sum(lengths) > total && tenths(sum(lengths.map(tenths))) > total;
+}
+
+// src/catalog/video-tools-spec.ts
 var RangeSchema2 = import_zod6.z.object({ min: import_zod6.z.number(), max: import_zod6.z.number() });
 var ToolSchema = import_zod6.z.object({
   kind: import_zod6.z.enum(["upscale", "interpolate", "hdr"]),
@@ -3331,7 +3381,7 @@ function planVideoTool(row, req) {
     return fail("The source video's width, height and duration are required");
   }
   const maxSeconds = row.constraints.sourceVideoSeconds?.max;
-  if (maxSeconds !== void 0 && durationSeconds > maxSeconds) {
+  if (maxSeconds !== void 0 && overMax(durationSeconds, maxSeconds)) {
     return fail(`${row.label} takes source videos up to ${maxSeconds} seconds.`);
   }
   const base = {
@@ -3528,7 +3578,7 @@ function videoLengthViolation(model, input) {
   const lengths = (input.referenceVideos ?? []).map(knownLength);
   if (limits?.max !== void 0) {
     const max = limits.max;
-    const i = lengths.findIndex((len) => len !== void 0 && len > max);
+    const i = lengths.findIndex((len) => len !== void 0 && overMax(len, max));
     if (i >= 0)
       return `${model.label} takes reference videos up to ${String(max)} s each; video ${i + 1} is ${seconds(lengths[i], max)} s.`;
   }
@@ -3540,14 +3590,17 @@ function videoLengthViolation(model, input) {
   const source = knownLength(input.sourceVideo);
   const sourceLimits = model.constraints.sourceVideoSeconds;
   if (limits?.total !== void 0) {
-    const refSum = lengths.reduce((acc, len) => acc + (len ?? 0), 0);
-    const sum = refSum + (sourceLimits?.inReferenceTotal ? source ?? 0 : 0);
-    if (sum > limits.total)
+    const counted = [
+      ...lengths.filter((len) => len !== void 0),
+      ...sourceLimits?.inReferenceTotal && source !== void 0 ? [source] : []
+    ];
+    const sum = counted.reduce((acc, len) => acc + len, 0);
+    if (overTotal(counted, limits.total))
       return `${model.label} takes reference videos up to ${String(limits.total)} s combined; these add up to ${seconds(sum, limits.total)} s.`;
   }
   if (source === void 0) return void 0;
   const sourceMin = effectiveMin(sourceLimits, input.taskType);
-  if (sourceLimits?.max !== void 0 && source > sourceLimits.max)
+  if (sourceLimits?.max !== void 0 && overMax(source, sourceLimits.max))
     return `${model.label} takes a source video up to ${String(sourceLimits.max)} s; this one is ${seconds(source, sourceLimits.max)} s.`;
   if (sourceMin !== void 0 && source < sourceMin)
     return `${model.label} takes a source video of at least ${String(sourceMin)} s; this one is ${seconds(source, sourceMin)} s.`;