Merge pull request #1 from atrifat/add-base64-upload-support

Add base64 upload support

README.md

@@ -21,13 +21,23 @@ node src/index.mjs
 ```
 
 If you want to test the API server, you can use GUI tools like [Postman](https://www.postman.com/) or using curl.
+
+Send request by using image URL:
 ```
 curl --header "Content-Type: application/json" \
   --header "Authorization: Bearer myapitokenchangethislater" \
   --request POST \
   --data '{"url":"https://example.org/image.jpg"}' \
   http://localhost:8081/predict
 ```
+or send request by using base64 string of the image:
+```
+curl --header "Content-Type: application/json" \
+  --header "Authorization: Bearer myapitokenchangethislater" \
+  --request POST \
+  --data '{"data":"base64stringhere"}' \
+  http://localhost:8081/predict_data
+```
 The output is JSON which consists of four predicted classes (based on [NsfwSpy.js](https://github.com/NsfwSpy/NsfwSpy.js)) as follows:
 ```
 {

src/index.mjs

@@ -47,7 +47,7 @@ const ENABLE_API_TOKEN = process.env.ENABLE_API_TOKEN ? process.env.ENABLE_API_T
 const API_TOKEN = process.env.API_TOKEN || "myapitokenchangethislater";
 
 const app = express();
-app.use(bodyparser.json())
+app.use(bodyparser.json({ limit: '5mb' }));
 
 const reqLogger = function (req, _res, next) {
     console.info(`${req.method} request to "${req.url}" by ${req.hostname}`);
@@ -172,6 +172,60 @@ app.post("/predict", async (req, res) => {
     res.status(200).json({ "data": cache });
 });
 
+app.post("/predict_data", async (req, res) => {
+    let err;
+
+    const base64_data = (typeof req.body.data !== 'undefined') ? req.body.data : null;
+
+    if (base64_data === null) {
+        err = new Error("Data input is empty, please send base64 string data as input");
+        err.name = "ValidationError";
+        return res.status(400).json({ "message": err.message });
+    }
+
+    const buffer = Buffer.from(base64_data, 'base64');
+
+    const filename = sha256(base64_data);
+    let cache = await keyv.get("data" + "-" + filename);
+    // Return cache result immediately if it is exist
+    if (cache) {
+        return res.status(200).json({ "data": cache });
+    }
+
+    // Load metadata for debugging
+    const img = sharp(buffer);
+    let metadata;
+    [err, metadata] = await to.default(img.metadata());
+
+    if (err) return res.status(500).json({ "message": err.message });
+    console.debug(metadata);
+
+    console.time("Preprocess");
+    let outputInfo;
+    [err, outputInfo] = await to.default(
+        // Resize to 224 px since it is the input size of model
+        img.resize(224).jpeg().withMetadata().toFile(IMG_DOWNLOAD_PATH + filename + "_" + "final")
+    );
+    if (err) return res.status(500).json({ "message": err.message });
+    console.timeEnd("Preprocess");
+
+    console.time("Classify");
+    [err, cache] = await to.default(nsfwSpy.classifyImageFile(IMG_DOWNLOAD_PATH + filename + "_" + "final"));
+    if (err) return res.status(500).json({ "message": err.message });
+
+    // Set cache result for 1 day
+    await keyv.set("data" + "-" + filename, cache, 24 * 60 * 60 * 1000);
+
+    console.timeEnd("Classify");
+    console.debug(cache);
+
+    // Cleanup image file
+    let deleteResult;
+    [err, deleteResult] = await to.default(deleteFile(IMG_DOWNLOAD_PATH + filename + "_" + "final"));
+
+    res.status(200).json({ "data": cache });
+});
+
 app.listen(PORT, () => {
     console.log(`Listening on ${PORT} ...`);
 });