leveldb 0.0.1

data/ext/leveldb/table/merger.cc

@@ -0,0 +1,197 @@
+// Copyright (c) 2011 The LevelDB Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file. See the AUTHORS file for names of contributors.
+
+#include "table/merger.h"
+
+#include "leveldb/comparator.h"
+#include "leveldb/iterator.h"
+#include "table/iterator_wrapper.h"
+
+namespace leveldb {
+
+namespace {
+class MergingIterator : public Iterator {
+ public:
+  MergingIterator(const Comparator* comparator, Iterator** children, int n)
+      : comparator_(comparator),
+        children_(new IteratorWrapper[n]),
+        n_(n),
+        current_(NULL),
+        direction_(kForward) {
+    for (int i = 0; i < n; i++) {
+      children_[i].Set(children[i]);
+    }
+  }
+
+  virtual ~MergingIterator() {
+    delete[] children_;
+  }
+
+  virtual bool Valid() const {
+    return (current_ != NULL);
+  }
+
+  virtual void SeekToFirst() {
+    for (int i = 0; i < n_; i++) {
+      children_[i].SeekToFirst();
+    }
+    FindSmallest();
+    direction_ = kForward;
+  }
+
+  virtual void SeekToLast() {
+    for (int i = 0; i < n_; i++) {
+      children_[i].SeekToLast();
+    }
+    FindLargest();
+    direction_ = kReverse;
+  }
+
+  virtual void Seek(const Slice& target) {
+    for (int i = 0; i < n_; i++) {
+      children_[i].Seek(target);
+    }
+    FindSmallest();
+    direction_ = kForward;
+  }
+
+  virtual void Next() {
+    assert(Valid());
+
+    // Ensure that all children are positioned after key().
+    // If we are moving in the forward direction, it is already
+    // true for all of the non-current_ children since current_ is
+    // the smallest child and key() == current_->key().  Otherwise,
+    // we explicitly position the non-current_ children.
+    if (direction_ != kForward) {
+      for (int i = 0; i < n_; i++) {
+        IteratorWrapper* child = &children_[i];
+        if (child != current_) {
+          child->Seek(key());
+          if (child->Valid() &&
+              comparator_->Compare(key(), child->key()) == 0) {
+            child->Next();
+          }
+        }
+      }
+      direction_ = kForward;
+    }
+
+    current_->Next();
+    FindSmallest();
+  }
+
+  virtual void Prev() {
+    assert(Valid());
+
+    // Ensure that all children are positioned before key().
+    // If we are moving in the reverse direction, it is already
+    // true for all of the non-current_ children since current_ is
+    // the largest child and key() == current_->key().  Otherwise,
+    // we explicitly position the non-current_ children.
+    if (direction_ != kReverse) {
+      for (int i = 0; i < n_; i++) {
+        IteratorWrapper* child = &children_[i];
+        if (child != current_) {
+          child->Seek(key());
+          if (child->Valid()) {
+            // Child is at first entry >= key().  Step back one to be < key()
+            child->Prev();
+          } else {
+            // Child has no entries >= key().  Position at last entry.
+            child->SeekToLast();
+          }
+        }
+      }
+      direction_ = kReverse;
+    }
+
+    current_->Prev();
+    FindLargest();
+  }
+
+  virtual Slice key() const {
+    assert(Valid());
+    return current_->key();
+  }
+
+  virtual Slice value() const {
+    assert(Valid());
+    return current_->value();
+  }
+
+  virtual Status status() const {
+    Status status;
+    for (int i = 0; i < n_; i++) {
+      status = children_[i].status();
+      if (!status.ok()) {
+        break;
+      }
+    }
+    return status;
+  }
+
+ private:
+  void FindSmallest();
+  void FindLargest();
+
+  // We might want to use a heap in case there are lots of children.
+  // For now we use a simple array since we expect a very small number
+  // of children in leveldb.
+  const Comparator* comparator_;
+  IteratorWrapper* children_;
+  int n_;
+  IteratorWrapper* current_;
+
+  // Which direction is the iterator moving?
+  enum Direction {
+    kForward,
+    kReverse
+  };
+  Direction direction_;
+};
+
+void MergingIterator::FindSmallest() {
+  IteratorWrapper* smallest = NULL;
+  for (int i = 0; i < n_; i++) {
+    IteratorWrapper* child = &children_[i];
+    if (child->Valid()) {
+      if (smallest == NULL) {
+        smallest = child;
+      } else if (comparator_->Compare(child->key(), smallest->key()) < 0) {
+        smallest = child;
+      }
+    }
+  }
+  current_ = smallest;
+}
+
+void MergingIterator::FindLargest() {
+  IteratorWrapper* largest = NULL;
+  for (int i = n_-1; i >= 0; i--) {
+    IteratorWrapper* child = &children_[i];
+    if (child->Valid()) {
+      if (largest == NULL) {
+        largest = child;
+      } else if (comparator_->Compare(child->key(), largest->key()) > 0) {
+        largest = child;
+      }
+    }
+  }
+  current_ = largest;
+}
+}  // namespace
+
+Iterator* NewMergingIterator(const Comparator* cmp, Iterator** list, int n) {
+  assert(n >= 0);
+  if (n == 0) {
+    return NewEmptyIterator();
+  } else if (n == 1) {
+    return list[0];
+  } else {
+    return new MergingIterator(cmp, list, n);
+  }
+}
+
+}  // namespace leveldb

data/ext/leveldb/table/merger.h

@@ -0,0 +1,26 @@
+// Copyright (c) 2011 The LevelDB Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file. See the AUTHORS file for names of contributors.
+
+#ifndef STORAGE_LEVELDB_TABLE_MERGER_H_
+#define STORAGE_LEVELDB_TABLE_MERGER_H_
+
+namespace leveldb {
+
+class Comparator;
+class Iterator;
+
+// Return an iterator that provided the union of the data in
+// children[0,n-1].  Takes ownership of the child iterators and
+// will delete them when the result iterator is deleted.
+//
+// The result does no duplicate suppression.  I.e., if a particular
+// key is present in K child iterators, it will be yielded K times.
+//
+// REQUIRES: n >= 0
+extern Iterator* NewMergingIterator(
+    const Comparator* comparator, Iterator** children, int n);
+
+}  // namespace leveldb
+
+#endif  // STORAGE_LEVELDB_TABLE_MERGER_H_

data/ext/leveldb/table/table.cc

@@ -0,0 +1,275 @@
+// Copyright (c) 2011 The LevelDB Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file. See the AUTHORS file for names of contributors.
+
+#include "leveldb/table.h"
+
+#include "leveldb/cache.h"
+#include "leveldb/comparator.h"
+#include "leveldb/env.h"
+#include "leveldb/filter_policy.h"
+#include "leveldb/options.h"
+#include "table/block.h"
+#include "table/filter_block.h"
+#include "table/format.h"
+#include "table/two_level_iterator.h"
+#include "util/coding.h"
+
+namespace leveldb {
+
+struct Table::Rep {
+  ~Rep() {
+    delete filter;
+    delete [] filter_data;
+    delete index_block;
+  }
+
+  Options options;
+  Status status;
+  RandomAccessFile* file;
+  uint64_t cache_id;
+  FilterBlockReader* filter;
+  const char* filter_data;
+
+  BlockHandle metaindex_handle;  // Handle to metaindex_block: saved from footer
+  Block* index_block;
+};
+
+Status Table::Open(const Options& options,
+                   RandomAccessFile* file,
+                   uint64_t size,
+                   Table** table) {
+  *table = NULL;
+  if (size < Footer::kEncodedLength) {
+    return Status::InvalidArgument("file is too short to be an sstable");
+  }
+
+  char footer_space[Footer::kEncodedLength];
+  Slice footer_input;
+  Status s = file->Read(size - Footer::kEncodedLength, Footer::kEncodedLength,
+                        &footer_input, footer_space);
+  if (!s.ok()) return s;
+
+  Footer footer;
+  s = footer.DecodeFrom(&footer_input);
+  if (!s.ok()) return s;
+
+  // Read the index block
+  BlockContents contents;
+  Block* index_block = NULL;
+  if (s.ok()) {
+    s = ReadBlock(file, ReadOptions(), footer.index_handle(), &contents);
+    if (s.ok()) {
+      index_block = new Block(contents);
+    }
+  }
+
+  if (s.ok()) {
+    // We've successfully read the footer and the index block: we're
+    // ready to serve requests.
+    Rep* rep = new Table::Rep;
+    rep->options = options;
+    rep->file = file;
+    rep->metaindex_handle = footer.metaindex_handle();
+    rep->index_block = index_block;
+    rep->cache_id = (options.block_cache ? options.block_cache->NewId() : 0);
+    rep->filter_data = NULL;
+    rep->filter = NULL;
+    *table = new Table(rep);
+    (*table)->ReadMeta(footer);
+  } else {
+    if (index_block) delete index_block;
+  }
+
+  return s;
+}
+
+void Table::ReadMeta(const Footer& footer) {
+  if (rep_->options.filter_policy == NULL) {
+    return;  // Do not need any metadata
+  }
+
+  // TODO(sanjay): Skip this if footer.metaindex_handle() size indicates
+  // it is an empty block.
+  ReadOptions opt;
+  BlockContents contents;
+  if (!ReadBlock(rep_->file, opt, footer.metaindex_handle(), &contents).ok()) {
+    // Do not propagate errors since meta info is not needed for operation
+    return;
+  }
+  Block* meta = new Block(contents);
+
+  Iterator* iter = meta->NewIterator(BytewiseComparator());
+  std::string key = "filter.";
+  key.append(rep_->options.filter_policy->Name());
+  iter->Seek(key);
+  if (iter->Valid() && iter->key() == Slice(key)) {
+    ReadFilter(iter->value());
+  }
+  delete iter;
+  delete meta;
+}
+
+void Table::ReadFilter(const Slice& filter_handle_value) {
+  Slice v = filter_handle_value;
+  BlockHandle filter_handle;
+  if (!filter_handle.DecodeFrom(&v).ok()) {
+    return;
+  }
+
+  // We might want to unify with ReadBlock() if we start
+  // requiring checksum verification in Table::Open.
+  ReadOptions opt;
+  BlockContents block;
+  if (!ReadBlock(rep_->file, opt, filter_handle, &block).ok()) {
+    return;
+  }
+  if (block.heap_allocated) {
+    rep_->filter_data = block.data.data();     // Will need to delete later
+  }
+  rep_->filter = new FilterBlockReader(rep_->options.filter_policy, block.data);
+}
+
+Table::~Table() {
+  delete rep_;
+}
+
+static void DeleteBlock(void* arg, void* ignored) {
+  delete reinterpret_cast<Block*>(arg);
+}
+
+static void DeleteCachedBlock(const Slice& key, void* value) {
+  Block* block = reinterpret_cast<Block*>(value);
+  delete block;
+}
+
+static void ReleaseBlock(void* arg, void* h) {
+  Cache* cache = reinterpret_cast<Cache*>(arg);
+  Cache::Handle* handle = reinterpret_cast<Cache::Handle*>(h);
+  cache->Release(handle);
+}
+
+// Convert an index iterator value (i.e., an encoded BlockHandle)
+// into an iterator over the contents of the corresponding block.
+Iterator* Table::BlockReader(void* arg,
+                             const ReadOptions& options,
+                             const Slice& index_value) {
+  Table* table = reinterpret_cast<Table*>(arg);
+  Cache* block_cache = table->rep_->options.block_cache;
+  Block* block = NULL;
+  Cache::Handle* cache_handle = NULL;
+
+  BlockHandle handle;
+  Slice input = index_value;
+  Status s = handle.DecodeFrom(&input);
+  // We intentionally allow extra stuff in index_value so that we
+  // can add more features in the future.
+
+  if (s.ok()) {
+    BlockContents contents;
+    if (block_cache != NULL) {
+      char cache_key_buffer[16];
+      EncodeFixed64(cache_key_buffer, table->rep_->cache_id);
+      EncodeFixed64(cache_key_buffer+8, handle.offset());
+      Slice key(cache_key_buffer, sizeof(cache_key_buffer));
+      cache_handle = block_cache->Lookup(key);
+      if (cache_handle != NULL) {
+        block = reinterpret_cast<Block*>(block_cache->Value(cache_handle));
+      } else {
+        s = ReadBlock(table->rep_->file, options, handle, &contents);
+        if (s.ok()) {
+          block = new Block(contents);
+          if (contents.cachable && options.fill_cache) {
+            cache_handle = block_cache->Insert(
+                key, block, block->size(), &DeleteCachedBlock);
+          }
+        }
+      }
+    } else {
+      s = ReadBlock(table->rep_->file, options, handle, &contents);
+      if (s.ok()) {
+        block = new Block(contents);
+      }
+    }
+  }
+
+  Iterator* iter;
+  if (block != NULL) {
+    iter = block->NewIterator(table->rep_->options.comparator);
+    if (cache_handle == NULL) {
+      iter->RegisterCleanup(&DeleteBlock, block, NULL);
+    } else {
+      iter->RegisterCleanup(&ReleaseBlock, block_cache, cache_handle);
+    }
+  } else {
+    iter = NewErrorIterator(s);
+  }
+  return iter;
+}
+
+Iterator* Table::NewIterator(const ReadOptions& options) const {
+  return NewTwoLevelIterator(
+      rep_->index_block->NewIterator(rep_->options.comparator),
+      &Table::BlockReader, const_cast<Table*>(this), options);
+}
+
+Status Table::InternalGet(const ReadOptions& options, const Slice& k,
+                          void* arg,
+                          void (*saver)(void*, const Slice&, const Slice&)) {
+  Status s;
+  Iterator* iiter = rep_->index_block->NewIterator(rep_->options.comparator);
+  iiter->Seek(k);
+  if (iiter->Valid()) {
+    Slice handle_value = iiter->value();
+    FilterBlockReader* filter = rep_->filter;
+    BlockHandle handle;
+    if (filter != NULL &&
+        handle.DecodeFrom(&handle_value).ok() &&
+        !filter->KeyMayMatch(handle.offset(), k)) {
+      // Not found
+    } else {
+      Iterator* block_iter = BlockReader(this, options, iiter->value());
+      block_iter->Seek(k);
+      if (block_iter->Valid()) {
+        (*saver)(arg, block_iter->key(), block_iter->value());
+      }
+      s = block_iter->status();
+      delete block_iter;
+    }
+  }
+  if (s.ok()) {
+    s = iiter->status();
+  }
+  delete iiter;
+  return s;
+}
+
+
+uint64_t Table::ApproximateOffsetOf(const Slice& key) const {
+  Iterator* index_iter =
+      rep_->index_block->NewIterator(rep_->options.comparator);
+  index_iter->Seek(key);
+  uint64_t result;
+  if (index_iter->Valid()) {
+    BlockHandle handle;
+    Slice input = index_iter->value();
+    Status s = handle.DecodeFrom(&input);
+    if (s.ok()) {
+      result = handle.offset();
+    } else {
+      // Strange: we can't decode the block handle in the index block.
+      // We'll just return the offset of the metaindex block, which is
+      // close to the whole file size for this case.
+      result = rep_->metaindex_handle.offset();
+    }
+  } else {
+    // key is past the last key in the file.  Approximate the offset
+    // by returning the offset of the metaindex block (which is
+    // right near the end of the file).
+    result = rep_->metaindex_handle.offset();
+  }
+  delete index_iter;
+  return result;
+}
+
+}  // namespace leveldb