Support Chunking in StreamData (facebookincubator#248)

Summary:
Pull Request resolved: facebookincubator#248

Another feature in the new chunking policy described in this [doc](https://fburl.com/gdoc/gkdwwju1) is the ability to split large streams above a specified limit into smaller chunks. In this diff, we implement a `popChunk` method in each `StreamData` class to handle this functionality. With this feature we are not forced to encode extremely large streams into a single chunk.

Integration will happen in the next diff.

Differential Revision: D81824143

Author: macvincent

dwio/nimble/common/Vector.h

@@ -182,6 +182,16 @@ class Vector {
     }
   }
 
+  // Removes the first |count| elements from the vector.
+  // Does NOT shrink capacity to fit the new size.
+  void trimLeft(uint64_t count) {
+    NIMBLE_DASSERT(
+        count <= size_, "Cannot trim more elements than exist in vector");
+    const uint64_t remainingSize = size_ - count;
+    std::move(begin() + count, end(), begin());
+    resize(remainingSize);
+  }
+
   // Add |copies| copies of |value| to the end of the vector.
   void extend(uint64_t copies, T value) {
     reserve(size_ + copies);

dwio/nimble/velox/StreamChunker.cpp

@@ -0,0 +1,76 @@
+// (c) Meta Platforms, Inc. and affiliates. Confidential and proprietary.
+
+#include "dwio/nimble/velox/StreamChunker.h"
+
+namespace facebook::nimble {
+template <typename T>
+std::unique_ptr<StreamChunker> encodeStreamTyped(
+    StreamData& streamData,
+    uint64_t maxChunkSize,
+    uint64_t minChunkSize,
+    bool lastChunk) {
+  const auto& streamDataType = streamData.type();
+  if (streamDataType == ContentStreamData<T>::TYPE_NAME) {
+    return std::make_unique<ContentStreamChunker<T>>(
+        static_cast<ContentStreamData<T>&>(streamData),
+        maxChunkSize,
+        minChunkSize,
+        lastChunk);
+  } else if (streamDataType == NullsStreamData::TYPE_NAME) {
+    return std::make_unique<NullsStreamChunker>(
+        static_cast<NullsStreamData&>(streamData),
+        maxChunkSize,
+        minChunkSize,
+        lastChunk);
+  } else if (streamDataType == NullableContentStreamData<T>::TYPE_NAME) {
+    return std::make_unique<NullableContentStreamChunker<T>>(
+        static_cast<NullableContentStreamData<T>&>(streamData),
+        maxChunkSize,
+        minChunkSize,
+        lastChunk);
+  }
+  NIMBLE_UNREACHABLE(
+      fmt::format("Unsupported streamData type {}", streamDataType))
+}
+
+std::unique_ptr<StreamChunker> getStreamChunker(
+    StreamData& streamData,
+    uint64_t maxChunkSize,
+    uint64_t minChunkSize,
+    bool lastChunk) {
+  const auto scalarKind = streamData.descriptor().scalarKind();
+  switch (scalarKind) {
+    case ScalarKind::Bool:
+      return encodeStreamTyped<bool>(
+          streamData, maxChunkSize, minChunkSize, lastChunk);
+    case ScalarKind::Int8:
+      return encodeStreamTyped<int8_t>(
+          streamData, maxChunkSize, minChunkSize, lastChunk);
+    case ScalarKind::Int16:
+      return encodeStreamTyped<int16_t>(
+          streamData, maxChunkSize, minChunkSize, lastChunk);
+    case ScalarKind::Int32:
+      return encodeStreamTyped<int32_t>(
+          streamData, maxChunkSize, minChunkSize, lastChunk);
+    case ScalarKind::UInt32:
+      return encodeStreamTyped<uint32_t>(
+          streamData, maxChunkSize, minChunkSize, lastChunk);
+    case ScalarKind::Int64:
+      return encodeStreamTyped<int64_t>(
+          streamData, maxChunkSize, minChunkSize, lastChunk);
+    case ScalarKind::Float:
+      return encodeStreamTyped<float>(
+          streamData, maxChunkSize, minChunkSize, lastChunk);
+    case ScalarKind::Double:
+      return encodeStreamTyped<double>(
+          streamData, maxChunkSize, minChunkSize, lastChunk);
+    case ScalarKind::String:
+    case ScalarKind::Binary:
+      return encodeStreamTyped<std::string_view>(
+          streamData, maxChunkSize, minChunkSize, lastChunk);
+    default:
+      NIMBLE_UNREACHABLE(
+          fmt::format("Unsupported scalar kind {}", toString(scalarKind)));
+  }
+}
+} // namespace facebook::nimble