feat: add Phase 4 AI tools with Python bridge and 6 new tools

Add Python bridge (packages/ai/src/bridge.ts) that calls Python scripts
via child_process with venv-first fallback to system python3. Implements
6 AI-powered tools:

- Remove Background: rembg-based with U2-Net/IS-Net models
- Image Upscaling: Real-ESRGAN with Lanczos fallback
- OCR/Text Extraction: Tesseract + PaddleOCR engines
- Face/PII Blur: MediaPipe face detection with configurable blur
- Object Eraser: LaMa inpainting with mask-based input
- Smart Crop: Sharp attention-based entropy cropping (no Python needed)

Each tool includes: Python script, TypeScript wrapper, API route,
and React settings component. All Python scripts handle ImportError
gracefully with clear installation messages.
This commit is contained in:
Siddharth Kumar Sah
2026-03-22 04:31:49 +08:00
parent a8cc611eb2
commit 5524939b6f
30 changed files with 1880 additions and 2 deletions
+86
View File
@@ -0,0 +1,86 @@
import type { FastifyInstance, FastifyRequest, FastifyReply } from "fastify";
import { randomUUID } from "node:crypto";
import { writeFile } from "node:fs/promises";
import { join, basename } from "node:path";
import { blurFaces } from "@stirling-image/ai";
import { createWorkspace } from "../../lib/workspace.js";
/**
* Face detection and blurring route.
* Uses MediaPipe for detection, PIL for blurring.
*/
export function registerBlurFaces(app: FastifyInstance) {
app.post(
"/api/v1/tools/blur-faces",
async (request: FastifyRequest, reply: FastifyReply) => {
let fileBuffer: Buffer | null = null;
let filename = "image";
let settingsRaw: string | null = null;
try {
const parts = request.parts();
for await (const part of parts) {
if (part.type === "file") {
const chunks: Buffer[] = [];
for await (const chunk of part.file) {
chunks.push(chunk);
}
fileBuffer = Buffer.concat(chunks);
filename = basename(part.filename ?? "image");
} else if (part.fieldname === "settings") {
settingsRaw = part.value as string;
}
}
} catch (err) {
return reply.status(400).send({
error: "Failed to parse multipart request",
details: err instanceof Error ? err.message : String(err),
});
}
if (!fileBuffer || fileBuffer.length === 0) {
return reply.status(400).send({ error: "No image file provided" });
}
try {
const settings = settingsRaw ? JSON.parse(settingsRaw) : {};
const jobId = randomUUID();
const workspacePath = await createWorkspace(jobId);
// Save input
const inputPath = join(workspacePath, "input", filename);
await writeFile(inputPath, fileBuffer);
// Process
const result = await blurFaces(
fileBuffer,
join(workspacePath, "output"),
{
blurRadius: settings.blurRadius ?? 30,
sensitivity: settings.sensitivity ?? 0.5,
},
);
// Save output
const outputFilename =
filename.replace(/\.[^.]+$/, "") + "_blurred.png";
const outputPath = join(workspacePath, "output", outputFilename);
await writeFile(outputPath, result.buffer);
return reply.send({
jobId,
downloadUrl: `/api/v1/download/${jobId}/${encodeURIComponent(outputFilename)}`,
originalSize: fileBuffer.length,
processedSize: result.buffer.length,
facesDetected: result.facesDetected,
faces: result.faces,
});
} catch (err) {
return reply.status(422).send({
error: "Face blur failed",
details: err instanceof Error ? err.message : "Unknown error",
});
}
},
);
}
+88
View File
@@ -0,0 +1,88 @@
import type { FastifyInstance, FastifyRequest, FastifyReply } from "fastify";
import { randomUUID } from "node:crypto";
import { writeFile } from "node:fs/promises";
import { join, basename } from "node:path";
import { inpaint } from "@stirling-image/ai";
import { createWorkspace } from "../../lib/workspace.js";
/**
* Object eraser / inpainting route.
* Accepts an image and a mask image, erases masked areas.
*/
export function registerEraseObject(app: FastifyInstance) {
app.post(
"/api/v1/tools/erase-object",
async (request: FastifyRequest, reply: FastifyReply) => {
let imageBuffer: Buffer | null = null;
let maskBuffer: Buffer | null = null;
let filename = "image";
try {
const parts = request.parts();
for await (const part of parts) {
if (part.type === "file") {
const chunks: Buffer[] = [];
for await (const chunk of part.file) {
chunks.push(chunk);
}
const buf = Buffer.concat(chunks);
if (part.fieldname === "mask") {
maskBuffer = buf;
} else {
imageBuffer = buf;
filename = basename(part.filename ?? "image");
}
}
}
} catch (err) {
return reply.status(400).send({
error: "Failed to parse multipart request",
details: err instanceof Error ? err.message : String(err),
});
}
if (!imageBuffer || imageBuffer.length === 0) {
return reply.status(400).send({ error: "No image file provided" });
}
if (!maskBuffer || maskBuffer.length === 0) {
return reply
.status(400)
.send({ error: "No mask image provided. Upload a mask as a second file with fieldname 'mask'" });
}
try {
const jobId = randomUUID();
const workspacePath = await createWorkspace(jobId);
// Save input
const inputPath = join(workspacePath, "input", filename);
await writeFile(inputPath, imageBuffer);
// Process
const resultBuffer = await inpaint(
imageBuffer,
maskBuffer,
join(workspacePath, "output"),
);
// Save output
const outputFilename =
filename.replace(/\.[^.]+$/, "") + "_erased.png";
const outputPath = join(workspacePath, "output", outputFilename);
await writeFile(outputPath, resultBuffer);
return reply.send({
jobId,
downloadUrl: `/api/v1/download/${jobId}/${encodeURIComponent(outputFilename)}`,
originalSize: imageBuffer.length,
processedSize: resultBuffer.length,
});
} catch (err) {
return reply.status(422).send({
error: "Object erasing failed",
details: err instanceof Error ? err.message : "Unknown error",
});
}
},
);
}
+16 -1
View File
@@ -32,6 +32,13 @@ import { registerFavicon } from "./favicon.js";
import { registerImageToPdf } from "./image-to-pdf.js";
// Phase 3: Adjustments extra
import { registerReplaceColor } from "./replace-color.js";
// Phase 4: AI Tools
import { registerRemoveBackground } from "./remove-background.js";
import { registerUpscale } from "./upscale.js";
import { registerOcr } from "./ocr.js";
import { registerBlurFaces } from "./blur-faces.js";
import { registerEraseObject } from "./erase-object.js";
import { registerSmartCrop } from "./smart-crop.js";
/**
* Registry that imports and registers all tool routes.
@@ -79,5 +86,13 @@ export async function registerToolRoutes(app: FastifyInstance): Promise<void> {
// Phase 3: Adjustments extra
registerReplaceColor(app);
app.log.info("Tool routes registered (26 tools, 29 endpoints)");
// Phase 4: AI Tools
registerRemoveBackground(app);
registerUpscale(app);
registerOcr(app);
registerBlurFaces(app);
registerEraseObject(app);
registerSmartCrop(app);
app.log.info("Tool routes registered (32 tools, 35 endpoints)");
}
+68
View File
@@ -0,0 +1,68 @@
import type { FastifyInstance, FastifyRequest, FastifyReply } from "fastify";
import { randomUUID } from "node:crypto";
import { basename } from "node:path";
import { extractText } from "@stirling-image/ai";
import { createWorkspace } from "../../lib/workspace.js";
/**
* OCR / text extraction route.
* Returns JSON with extracted text rather than an image.
*/
export function registerOcr(app: FastifyInstance) {
app.post(
"/api/v1/tools/ocr",
async (request: FastifyRequest, reply: FastifyReply) => {
let fileBuffer: Buffer | null = null;
let filename = "image";
let settingsRaw: string | null = null;
try {
const parts = request.parts();
for await (const part of parts) {
if (part.type === "file") {
const chunks: Buffer[] = [];
for await (const chunk of part.file) {
chunks.push(chunk);
}
fileBuffer = Buffer.concat(chunks);
filename = basename(part.filename ?? "image");
} else if (part.fieldname === "settings") {
settingsRaw = part.value as string;
}
}
} catch (err) {
return reply.status(400).send({
error: "Failed to parse multipart request",
details: err instanceof Error ? err.message : String(err),
});
}
if (!fileBuffer || fileBuffer.length === 0) {
return reply.status(400).send({ error: "No image file provided" });
}
try {
const settings = settingsRaw ? JSON.parse(settingsRaw) : {};
const jobId = randomUUID();
const workspacePath = await createWorkspace(jobId);
const result = await extractText(fileBuffer, workspacePath, {
engine: settings.engine,
language: settings.language,
});
return reply.send({
jobId,
filename,
text: result.text,
engine: result.engine,
});
} catch (err) {
return reply.status(422).send({
error: "OCR failed",
details: err instanceof Error ? err.message : "Unknown error",
});
}
},
);
}
@@ -0,0 +1,80 @@
import type { FastifyInstance, FastifyRequest, FastifyReply } from "fastify";
import { randomUUID } from "node:crypto";
import { writeFile } from "node:fs/promises";
import { join, basename } from "node:path";
import { removeBackground } from "@stirling-image/ai";
import { createWorkspace } from "../../lib/workspace.js";
/**
* AI background removal route.
* Uses Python + rembg under the hood.
*/
export function registerRemoveBackground(app: FastifyInstance) {
app.post(
"/api/v1/tools/remove-background",
async (request: FastifyRequest, reply: FastifyReply) => {
let fileBuffer: Buffer | null = null;
let filename = "image";
let settingsRaw: string | null = null;
try {
const parts = request.parts();
for await (const part of parts) {
if (part.type === "file") {
const chunks: Buffer[] = [];
for await (const chunk of part.file) {
chunks.push(chunk);
}
fileBuffer = Buffer.concat(chunks);
filename = basename(part.filename ?? "image");
} else if (part.fieldname === "settings") {
settingsRaw = part.value as string;
}
}
} catch (err) {
return reply.status(400).send({
error: "Failed to parse multipart request",
details: err instanceof Error ? err.message : String(err),
});
}
if (!fileBuffer || fileBuffer.length === 0) {
return reply.status(400).send({ error: "No image file provided" });
}
try {
const settings = settingsRaw ? JSON.parse(settingsRaw) : {};
const jobId = randomUUID();
const workspacePath = await createWorkspace(jobId);
// Save input
const inputPath = join(workspacePath, "input", filename);
await writeFile(inputPath, fileBuffer);
// Process
const resultBuffer = await removeBackground(
fileBuffer,
join(workspacePath, "output"),
{ model: settings.model },
);
// Save output
const outputFilename = filename.replace(/\.[^.]+$/, "") + "_nobg.png";
const outputPath = join(workspacePath, "output", outputFilename);
await writeFile(outputPath, resultBuffer);
return reply.send({
jobId,
downloadUrl: `/api/v1/download/${jobId}/${encodeURIComponent(outputFilename)}`,
originalSize: fileBuffer.length,
processedSize: resultBuffer.length,
});
} catch (err) {
return reply.status(422).send({
error: "Background removal failed",
details: err instanceof Error ? err.message : "Unknown error",
});
}
},
);
}
+33
View File
@@ -0,0 +1,33 @@
import { z } from "zod";
import sharp from "sharp";
import type { FastifyInstance } from "fastify";
import { createToolRoute } from "../tool-factory.js";
const settingsSchema = z.object({
width: z.number().int().positive(),
height: z.number().int().positive(),
});
/**
* Smart crop using Sharp's attention-based strategy.
* Uses entropy/saliency detection to find the most interesting region.
* No Python needed.
*/
export function registerSmartCrop(app: FastifyInstance) {
createToolRoute(app, {
toolId: "smart-crop",
settingsSchema,
process: async (inputBuffer, settings, filename) => {
const result = await sharp(inputBuffer)
.resize(settings.width, settings.height, {
fit: "cover",
position: sharp.strategy.attention,
})
.png()
.toBuffer();
const outputFilename = filename.replace(/\.[^.]+$/, "") + "_smartcrop.png";
return { buffer: result, filename: outputFilename, contentType: "image/png" };
},
});
}
+86
View File
@@ -0,0 +1,86 @@
import type { FastifyInstance, FastifyRequest, FastifyReply } from "fastify";
import { randomUUID } from "node:crypto";
import { writeFile } from "node:fs/promises";
import { join, basename } from "node:path";
import { upscale } from "@stirling-image/ai";
import { createWorkspace } from "../../lib/workspace.js";
/**
* AI image upscaling route.
* Uses Real-ESRGAN when available, falls back to Lanczos.
*/
export function registerUpscale(app: FastifyInstance) {
app.post(
"/api/v1/tools/upscale",
async (request: FastifyRequest, reply: FastifyReply) => {
let fileBuffer: Buffer | null = null;
let filename = "image";
let settingsRaw: string | null = null;
try {
const parts = request.parts();
for await (const part of parts) {
if (part.type === "file") {
const chunks: Buffer[] = [];
for await (const chunk of part.file) {
chunks.push(chunk);
}
fileBuffer = Buffer.concat(chunks);
filename = basename(part.filename ?? "image");
} else if (part.fieldname === "settings") {
settingsRaw = part.value as string;
}
}
} catch (err) {
return reply.status(400).send({
error: "Failed to parse multipart request",
details: err instanceof Error ? err.message : String(err),
});
}
if (!fileBuffer || fileBuffer.length === 0) {
return reply.status(400).send({ error: "No image file provided" });
}
try {
const settings = settingsRaw ? JSON.parse(settingsRaw) : {};
const scale = Number(settings.scale) || 2;
const jobId = randomUUID();
const workspacePath = await createWorkspace(jobId);
// Save input
const inputPath = join(workspacePath, "input", filename);
await writeFile(inputPath, fileBuffer);
// Process
const result = await upscale(
fileBuffer,
join(workspacePath, "output"),
{ scale },
);
// Save output
const outputFilename =
filename.replace(/\.[^.]+$/, "") + `_${scale}x.png`;
const outputPath = join(workspacePath, "output", outputFilename);
await writeFile(outputPath, result.buffer);
return reply.send({
jobId,
downloadUrl: `/api/v1/download/${jobId}/${encodeURIComponent(outputFilename)}`,
originalSize: fileBuffer.length,
processedSize: result.buffer.length,
width: result.width,
height: result.height,
method: result.method,
});
} catch (err) {
return reply.status(422).send({
error: "Upscaling failed",
details: err instanceof Error ? err.message : "Unknown error",
});
}
},
);
}