Add XGBoost test files (#1452)

Author: lutzroeder

source/gguf.js

@@ -191,7 +191,8 @@ gguf.Reader = class {
     static open(context) {
         const stream = context.stream;
         if (stream && stream.length > 4) {
-            const signature = String.fromCharCode.apply(null, stream.peek(4));
+            const buffer = stream.peek(4);
+            const signature = String.fromCharCode.apply(null, buffer);
             if (signature === 'GGUF') {
                 return new gguf.Reader(context);
             }

source/mlir.js

@@ -40,7 +40,7 @@ mlir.ModelFactory = class {
                 const reader = await context.read('binary');
                 const parser = new mlir.BytecodeReader(reader);
                 parser.read();
-                throw new mlir.Error('Invalid file content. File contains MLIR bytecode data.');
+                throw new mlir.Error('File contains unsupported MLIR bytecode data.');
             }
             default: {
                 throw new mlir.Error(`Unsupported MLIR format '${context.type}'.`);

source/view.js

@@ -6151,6 +6151,7 @@ view.ModelFactoryService = class {
         this.register('./weka', ['.model']);
         this.register('./qnn', ['.json', '.bin', '.serialized', '.dlc']);
         this.register('./kann', ['.kann', '.bin', '.kgraph'], [], [/^....KaNN/]);
+        this.register('./xgboost', ['.xgb', '.xgboost', '.json', '.model', '.bin', '.txt'], [], [/^{L\x00\x00/, /^binf/, /^bs64/, /^\s*booster\[0\]:/]);
         this.register('', ['.cambricon', '.vnnmodel', '.nnc']);
         /* eslint-enable no-control-regex */
     }
@@ -6695,7 +6696,6 @@ view.ModelFactoryService = class {
                 { name: 'Cambricon model', value: /^\x7fMEF/ },
                 { name: 'Cambricon model', value: /^cambricon_offline/ },
                 { name: 'VNN model', value: /^\x2F\x4E\x00\x00.\x00\x00\x00/, identifier: /.vnnmodel$/ },
-                { name: 'XGBoost model', value: /^(binf|bs64)/ }, // https://github.com/dmlc/xgboost/blob/master/src/learner.cc
                 { name: 'SQLite data', value: /^SQLite format/ },
                 { name: 'Optimium model', value: /^EZMODEL/ }, // https://github.com/EZ-Optimium/Optimium,
                 { name: 'undocumented NNC data', value: /^(\xC0|\xBC)\x0F\x00\x00ENNC/ },

source/xgboost.js

@@ -0,0 +1,54 @@
+
+// Experimental
+
+const xgboost = {};
+
+xgboost.ModelFactory = class {
+
+    async match(context) {
+        const obj = await context.peek('json');
+        if (obj && obj.learner && obj.version) {
+            return context.set('xgboost.json', obj);
+        }
+        const stream = context.stream;
+        if (stream && stream.length > 4) {
+            const buffer = stream.peek(4);
+            if (buffer[0] === 0x7B && buffer[1] === 0x4C && buffer[2] === 0x00 && buffer[3] === 0x00) {
+                return context.set('xgboost.ubj', stream);
+            }
+            const signature = String.fromCharCode.apply(null, buffer);
+            if (signature.startsWith('binf')) {
+                return context.set('xgboost.binf', stream);
+            }
+            if (signature.startsWith('bs64')) {
+                return context.set('xgboost.bs64', stream);
+            }
+            const reader = await context.read('text', 0x100);
+            const line = reader.read('\n');
+            if (line !== undefined && line.trim() === 'booster[0]:') {
+                return context.set('xgboost.text', stream);
+            }
+        }
+        return null;
+    }
+
+    async open(context) {
+        if (context.type === 'xgboost.json') {
+            throw new xgboost.Error('File contains unsupported XGBoost JSON data.');
+        }
+        if (context.type === 'xgboost.text') {
+            throw new xgboost.Error('File contains unsupported XGBoost text data.');
+        }
+        throw new xgboost.Error('File contains unsupported XGBoost data.');
+    }
+};
+
+xgboost.Error = class extends Error {
+
+    constructor(message) {
+        super(message);
+        this.name = 'Error loading XGBoost model.';
+    }
+};
+
+export const ModelFactory = xgboost.ModelFactory;

test/models.json

@@ -3312,7 +3312,7 @@
     "target":   "model.mlirbc",
     "source":   "https://github.com/user-attachments/files/17179955/model.mlirbc.zip[model.mlirbc]",
     "format":   "MLIR",
-    "error":    "Invalid file content. File contains MLIR bytecode data.",
+    "error":    "File contains unsupported MLIR bytecode data.",
     "link":     "https://github.com/lutzroeder/netron/issues/1044"
   },
   {
@@ -3363,7 +3363,7 @@
     "target":   "versioned-op-2.0.mlirbc",
     "source":   "https://github.com/user-attachments/files/17174958/versioned-op-2.0.mlirbc.zip[versioned-op-2.0.mlirbc]",
     "format":   "MLIR",
-    "error":    "Invalid file content. File contains MLIR bytecode data.",
+    "error":    "File contains unsupported MLIR bytecode data.",
     "link":     "https://github.com/lutzroeder/netron/issues/1044"
   },
   {
@@ -8451,6 +8451,34 @@
     "source":   "https://raw.githubusercontent.com/PTaati/wekaTree2python/master/j48model.model",
     "error":    "Unsupported type 'weka.classifiers.trees.J48'."
   },
+  {
+    "type":     "xgboost",
+    "target":   "xgb_classifier.json",
+    "source":   "https://github.com/user-attachments/files/20028544/xgb_classifier.zip[xgb_classifier.json]",
+    "error":    "File contains unsupported XGBoost JSON data.",
+    "link":     "https://github.com/lutzroeder/netron/issues/1452"
+  },
+  {
+    "type":     "xgboost",
+    "target":   "xgb_classifier.model",
+    "source":   "https://github.com/user-attachments/files/20028544/xgb_classifier.zip[xgb_classifier.model]",
+    "error":    "File contains unsupported XGBoost data.",
+    "link":     "https://github.com/lutzroeder/netron/issues/1452"
+  },
+  {
+    "type":     "xgboost",
+    "target":   "xgb_classifier.pkl",
+    "source":   "https://github.com/user-attachments/files/20028544/xgb_classifier.zip[xgb_classifier.pkl]",
+    "format":   "scikit-learn",
+    "link":     "https://github.com/lutzroeder/netron/issues/1452"
+  },
+  {
+    "type":     "xgboost",
+    "target":   "xgb_classifier.txt",
+    "source":   "https://github.com/user-attachments/files/20028544/xgb_classifier.zip[xgb_classifier.txt]",
+    "error":    "File contains unsupported XGBoost text data.",
+    "link":     "https://github.com/lutzroeder/netron/issues/1452"
+  },
   {
     "type":     "xmodel",
     "target":   "face-quality_pt.xmodel",