feat: accept a source file on generation assets

paths/generate.yaml

@@ -65,8 +65,8 @@
               schema:
                 $ref: "../schemas/responses/generationerrorresponse.yaml#/GenerationErrorResponse"
       description: |
-        Generate a single image, video or audio asset from a text prompt without
-        rendering a full edit. Submit a prompt-bearing asset; the response is
+        Generate a single image, video or audio asset from a prompt or, for some models, a source file, without
+        rendering a full edit. Submit an asset with a prompt or source file; the response is
         immediate when an identical asset has been generated before (results are
         cached by prompt, model and options), otherwise the job is queued and can
         be polled via the status endpoint.

paths/models.yaml

@@ -1,7 +1,7 @@
     get:
       description: |
         List the generation models available for `prompt`-bearing image, video and audio
-        assets, with the options each accepts. Use `POST /generate/quote` for a credit estimate.
+        assets generated from a prompt or, for some models, a source file, with the options each accepts. Use `POST /generate/quote` for a credit estimate.
 
         Use this to populate a model picker and render its option fields, rather than
         hard coding a model list. A newly launched model appears here without any change

schemas/generationasset.yaml

@@ -1,5 +1,7 @@
     GenerationAsset:
-      description: An image, video or audio asset to generate from a text prompt.
+      description: >-
+        An asset to generate, from a text `prompt` or, for models that take one, a source
+        `src`.
       type: object
       properties:
         type:
@@ -16,6 +18,12 @@
           maxLength: 4000
           pattern: '\S'
           example: A lighthouse on a rocky coast at sunset, cinematic lighting
+        src:
+          description: >-
+            The source file for models that take one instead of a prompt, such as
+            `shotstack-music-analysis`. A public HTTPS URL.
+          type: string
+          format: uri
         model:
           description: >-
             The generation model. Defaults to `nano-banana-2` for images,
@@ -37,4 +45,3 @@
       additionalProperties: false
       required:
         - type
-        - prompt

schemas/generationmodel.yaml

@@ -1,6 +1,6 @@
     GenerationModel:
       description: >-
-        A generation model available to `prompt`-bearing image, video and audio assets,
+        A generation model available to image, video and audio assets generated from a prompt or, for some models, a source file,
         with the options it accepts. Use `POST /generate/quote` for a credit estimate. Render a model picker and its
         option fields from this rather than hard coding a model list, so a newly launched
         model is available without a client release.

scripts/fix-discriminator.cjs

@@ -366,6 +366,7 @@ function addSrcOrPromptRefine(code, zPrefix) {
     'imageassetImageAssetSchema',
     'videoassetVideoAssetSchema',
     'audioassetAudioAssetSchema',
+    'generationassetGenerationAssetSchema',
   ];
   const strictMarker = '.strict()';
   const superRefine = strictMarker + '.superRefine((data, ctx) => {\n' +

tests/smoke.cjs

@@ -120,6 +120,12 @@ async function run() {
     assert.strictEqual(zodCjs.generationmodelpricingGenerationModelPricingSchema, undefined);
   });
 
+  check("Generation assets accept a source file or a prompt", () => {
+    const analysis = { body: { asset: { type: "audio", model: "shotstack-music-analysis", src: "https://example.com/a.mp3" } } };
+    assert(zodCjs.postGenerateRequest.safeParse(analysis).success, "a src-input asset parses");
+    assert(!zodCjs.postGenerateRequest.safeParse({ body: { asset: { type: "audio" } } }).success, "an asset needs src or prompt");
+  });
+
   check("Parse valid rich-text asset", () => {
     const result = zodCjs.richTextAssetSchema.parse({
       type: "rich-text",