chore: import upstream snapshot with attribution

tools/train/source/data/RandomSampler.cpp

@@ -0,0 +1,75 @@
+//
+//  RandomSampler.cpp
+//  MNN
+//
+//  Created by MNN on 2019/11/14.
+//  Copyright © 2018, Alibaba Group Holding Limited
+//
+
+#include "RandomSampler.hpp"
+#include <algorithm>
+#include <iterator>
+#include <random>
+#include "Distributions.hpp"
+#include "RandomGenerator.hpp"
+using namespace MNN::Express;
+
+namespace MNN {
+namespace Train {
+
+RandomSampler::RandomSampler(size_t size, bool shuffle) {
+    mIndices.reserve(size);
+    for (int i = 0; i < size; i++) {
+        mIndices.emplace_back(i);
+    }
+
+    mShuffle = shuffle;
+    if (mShuffle) {
+        std::shuffle(mIndices.begin(), mIndices.end(), RandomGenerator::generator());
+    }
+}
+
+void RandomSampler::reset(size_t size) {
+    mIndices.clear();
+    mIndices.reserve(size);
+    for (int i = 0; i < size; i++) {
+        mIndices.emplace_back(i);
+    }
+
+    if (mShuffle) {
+        std::shuffle(mIndices.begin(), mIndices.end(), RandomGenerator::generator());
+    }
+
+    mIndex = 0;
+}
+
+size_t RandomSampler::size() {
+    return mIndices.size();
+}
+
+const std::vector<size_t> RandomSampler::indices() {
+    return mIndices;
+}
+
+size_t RandomSampler::index() {
+    return mIndex;
+}
+
+std::vector<size_t> RandomSampler::next(size_t batchSize) {
+    MNN_ASSERT(mIndex <= mIndices.size());
+
+    auto remainIndices = mIndices.size() - mIndex;
+    if (remainIndices == 0) {
+        return {};
+    }
+
+    std::vector<size_t> batchIndex(std::min(batchSize, remainIndices));
+    std::copy(mIndices.begin() + mIndex, mIndices.begin() + mIndex + batchIndex.size(), batchIndex.begin());
+
+    mIndex += batchIndex.size();
+
+    return batchIndex;
+}
+
+} // namespace Train
+} // namespace MNN