runline 0.5.3 → 0.7.0

package/dist/plugins/_shared/parseSize.js

@@ -0,0 +1,26 @@
+/**
+ * Parse a `WxH` string into a { width, height } pair.
+ *
+ * Used by image-gen plugins (replicate, together, …) that take a
+ * single `size` input but talk to APIs which want explicit
+ * dimensions. Throws with the caller's plugin name in the message
+ * so error output points back at the right tool.
+ *
+ *   parseSize("1024x1024", "replicate")  // { width: 1024, height: 1024 }
+ *   parseSize(undefined, "replicate")    // defaults to 1024x1024
+ *   parseSize("1024", "replicate")       // throws
+ */
+export function parseSize(size, pluginName, defaults = { width: 1024, height: 1024 }) {
+    if (!size)
+        return { ...defaults };
+    const parts = size.split("x").map((s) => Number(s.trim()));
+    if (parts.length !== 2 ||
+        !Number.isFinite(parts[0]) ||
+        !Number.isFinite(parts[1])) {
+        throw new Error(`${pluginName}: invalid size "${size}", expected WxH`);
+    }
+    if (parts[0] <= 0 || parts[1] <= 0) {
+        throw new Error(`${pluginName}: size dimensions must be positive`);
+    }
+    return { width: parts[0], height: parts[1] };
+}

package/dist/plugins/googleImage/src/index.js

@@ -0,0 +1,79 @@
+/**
+ * Google Gemini image generation (Nano Banana / Imagen) for runline.
+ *
+ * Distinct from the rest of the googleX family — those wrap Workspace
+ * APIs over OAuth2, this one wraps Generative Language over a single
+ * API key. Kept under the `googleImage` namespace so it doesn't
+ * collide with `googleDrive`, `googleDocs`, etc.
+ *
+ *   await googleImage.image.create({ prompt: "a watercolor fox" })
+ *   await googleImage.image.create({
+ *     prompt: "edit: make the sky stormier",
+ *     model: "gemini-3-pro-image-preview",
+ *   })
+ *
+ * Nano Banana supports conversational editing — chain prompts in
+ * follow-up calls and it'll keep iterating on the last image.
+ */
+const BASE = "https://generativelanguage.googleapis.com/v1beta/models";
+export default function googleImage(rl) {
+    rl.setName("googleImage");
+    rl.setVersion("0.1.0");
+    rl.setConnectionSchema({
+        apiKey: {
+            type: "string",
+            required: true,
+            description: "Google AI API key (Gemini)",
+            env: "GOOGLE_API_KEY",
+        },
+    });
+    rl.registerAction("image.create", {
+        description: "Generate an image with Google's Gemini image models (Nano Banana / Imagen). Returns base64 bytes per candidate.",
+        inputSchema: {
+            prompt: {
+                type: "string",
+                required: true,
+                description: "Detailed description of the image",
+            },
+            model: {
+                type: "string",
+                required: false,
+                description: "gemini-2.5-flash-image (Nano Banana, default) | gemini-3-pro-image-preview | gemini-3.1-flash-image-preview",
+            },
+        },
+        async execute(input, ctx) {
+            const p = (input ?? {});
+            if (typeof p.prompt !== "string" || p.prompt.length === 0) {
+                throw new Error("googleImage: prompt is required");
+            }
+            const apiKey = ctx.connection.config.apiKey;
+            const model = p.model ?? "gemini-2.5-flash-image";
+            const url = `${BASE}/${model}:generateContent?key=${encodeURIComponent(apiKey)}`;
+            const body = {
+                contents: [{ parts: [{ text: p.prompt }] }],
+                generationConfig: { responseModalities: ["IMAGE", "TEXT"] },
+            };
+            const res = await fetch(url, {
+                method: "POST",
+                headers: { "Content-Type": "application/json" },
+                body: JSON.stringify(body),
+            });
+            if (!res.ok) {
+                throw new Error(`Google API error ${res.status}: ${await res.text()}`);
+            }
+            const data = (await res.json());
+            const images = [];
+            for (const candidate of data.candidates ?? []) {
+                for (const part of candidate.content?.parts ?? []) {
+                    if (part.inlineData?.data) {
+                        images.push({
+                            base64: part.inlineData.data,
+                            mimeType: part.inlineData.mimeType ?? "image/png",
+                        });
+                    }
+                }
+            }
+            return { provider: "googleImage", model, images };
+        },
+    });
+}