bplus_tree.h

/* Copyright (c) 2021 Xie Meiyi(xiemeiyi@hust.edu.cn) and OceanBase and/or its affiliates. All rights reserved.
miniob is licensed under Mulan PSL v2.
You can use this software according to the terms and conditions of the Mulan PSL v2.
You may obtain a copy of Mulan PSL v2 at:
         http://license.coscl.org.cn/MulanPSL2
THIS SOFTWARE IS PROVIDED ON AN "AS IS" BASIS, WITHOUT WARRANTIES OF ANY KIND,
EITHER EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO NON-INFRINGEMENT,
MERCHANTABILITY OR FIT FOR A PARTICULAR PURPOSE.
See the Mulan PSL v2 for more details. */

//
//
// Created by Xie Meiyi
// Rewritten by Longda & Wangyunlai
//
//
#ifndef __OBSERVER_STORAGE_COMMON_INDEX_MANAGER_H_
#define __OBSERVER_STORAGE_COMMON_INDEX_MANAGER_H_

#include <string.h>
#include <sstream>

#include "storage/common/record_manager.h"
#include "storage/default/disk_buffer_pool.h"
#include "sql/parser/parse_defs.h"

#define EMPTY_RID_PAGE_NUM -1
#define EMPTY_RID_SLOT_NUM -1

class AttrComparator
{
public:
  void init(AttrType type, int length)
  {
    attr_type_ = type;
    attr_length_ = length;
  }

  int attr_length() const {
    return attr_length_;
  }

  int operator()(const char *v1, const char *v2) const {
    switch (attr_type_) {
    case INTS: {
      return *(int *)v1 - *(int *)v2;
    }
      break;
    case FLOATS: {
      float result = *(float *)v1 - *(float *)v2;
      if (-1e-6 < result && result < 1e-6) {
        return 0;
      }
      return result > 0 ? 1 : -1;
    }
    case CHARS: {
      return strncmp(v1, v2, attr_length_);
    }
    default:{
      LOG_ERROR("unknown attr type. %d", attr_type_);
      abort();
    }
    }
  }
private:
  AttrType attr_type_;
  int attr_length_;
};

class KeyComparator
{
public:
  void init(AttrType type, int length)
  {
    attr_comparator_.init(type, length);
  }

  const AttrComparator &attr_comparator() const {
    return attr_comparator_;
  }

  int operator() (const char *v1, const char *v2) const {
    int result = attr_comparator_(v1, v2);
    if (result != 0) {
      return result;
    }

    const RID *rid1 = (const RID *)(v1 + attr_comparator_.attr_length());
    const RID *rid2 = (const RID *)(v2 + attr_comparator_.attr_length());
    return RID::compare(rid1, rid2);
  }

private:
  AttrComparator attr_comparator_;
};

class AttrPrinter
{
public:
  void init(AttrType type, int length)
  {
    attr_type_ = type;
    attr_length_ = length;
  }

  int attr_length() const {
    return attr_length_;
  }

  std::string operator()(const char *v) const {
    switch (attr_type_) {
    case INTS: {
      return std::to_string(*(int*)v);
    }
      break;
    case FLOATS: {
      return std::to_string(*(float*)v);
    }
    case CHARS: {
      return std::string(v, attr_length_);
    }
    default:{
      LOG_ERROR("unknown attr type. %d", attr_type_);
      abort();
    }
    }
  }
private:
  AttrType attr_type_;
  int attr_length_;
};

class KeyPrinter
{
public:
  void init(AttrType type, int length)
  {
    attr_printer_.init(type, length);
  }

  const AttrPrinter &attr_printer() const {
    return attr_printer_;
  }

  std::string operator() (const char *v) const {
    std::stringstream ss;
    ss << "{key:" << attr_printer_(v) << ",";

    const RID *rid = (const RID *)(v + attr_printer_.attr_length());
    ss << "rid:{" << rid->to_string() << "}}";
    return ss.str();
  }

private:
  AttrPrinter attr_printer_;
};

/**
 * the meta information of bplus tree
 * this is the first page of bplus tree.
 * only one field can be supported, can you extend it to multi-fields?
 */
struct IndexFileHeader {
  IndexFileHeader()
  {
    memset(this, 0, sizeof(IndexFileHeader));
    root_page = BP_INVALID_PAGE_NUM;
  }
  PageNum  root_page;
  int32_t  internal_max_size;
  int32_t  leaf_max_size;
  int32_t  attr_length;
  int32_t  key_length; // attr length + sizeof(RID)
  AttrType attr_type;

  const std::string to_string()
  {
    std::stringstream ss;

    ss << "attr_length:" << attr_length << ","
       << "key_length:" << key_length << ","
       << "attr_type:" << attr_type << ","
       << "root_page:" << root_page << ","
       << "internal_max_size:" << internal_max_size << ","
       << "leaf_max_size:" << leaf_max_size << ";";

    return ss.str();
  }
};

#define RECORD_RESERVER_PAIR_NUM 2
/**
 * the common part of page describtion of bplus tree
 * storage format:
 * | page type | item number | parent page id |
 */
struct IndexNode {
  static constexpr int HEADER_SIZE = 12;

  bool is_leaf;
  int key_num;
  PageNum parent;
};

/**
 * leaf page of bplus tree
 * storage format:
 * | common header | prev page id | next page id |
 * | key0, rid0 | key1, rid1 | ... | keyn, ridn |
 *
 * the key is in format: the key value of record and rid.
 * so the key in leaf page must be unique.
 * the value is rid.
 * can you implenment a cluster index ?
 */
struct LeafIndexNode : public IndexNode {
  static constexpr int HEADER_SIZE = IndexNode::HEADER_SIZE + 8;
  
  PageNum prev_brother;
  PageNum next_brother;
  /**
   * leaf can store order keys and rids at most
   */
  char array[0];
};

/**
 * internal page of bplus tree
 * storage format:
 * | common header | 
 * | key(0),page_id(0) | key(1), page_id(1) | ... | key(n), page_id(n) |
 *
 * the first key is ignored(key0).
 * so it will waste space, can you fix this?
 */
struct InternalIndexNode : public IndexNode {
  static constexpr int HEADER_SIZE = IndexNode::HEADER_SIZE;

  /**
   * internal node just store order -1 keys and order rids, the last rid is last rght child.
   */
  char array[0];
};

class IndexNodeHandler {
public:
  IndexNodeHandler(const IndexFileHeader &header, BPPageHandle &page_handle);

  void init_empty(bool leaf);

  bool is_leaf() const;
  int key_size() const;
  int value_size() const;
  int item_size() const;

  void increase_size(int n);
  int size() const;
  void set_parent_page_num(PageNum page_num);
  PageNum parent_page_num() const;

  PageNum page_num() const;

  bool validate() const;

  friend std::string to_string(const IndexNodeHandler &handler);

protected:
  const IndexFileHeader &header_;
  PageNum page_num_;
  IndexNode *node_;
};

class LeafIndexNodeHandler : public IndexNodeHandler {
public: 
  LeafIndexNodeHandler(const IndexFileHeader &header, BPPageHandle &page_handle);

  void init_empty();
  void set_next_page(PageNum page_num);
  void set_prev_page(PageNum page_num);
  PageNum next_page() const;
  PageNum prev_page() const;

  char *key_at(int index);
  char *value_at(int index);

  /**
   * 查找指定key的插入位置(注意不是key本身)
   * 如果key已经存在，会设置found的值
   * NOTE: 当前lookup的实现效率非常低，你是否可以优化它?
   */
  int lookup(const KeyComparator &comparator, const char *key, bool *found = nullptr) const;

  void insert(int index, const char *key, const char *value);
  void remove(int index);
  int  remove(const char *key, const KeyComparator &comparator);
  RC   move_half_to(LeafIndexNodeHandler &other, DiskBufferPool *bp, int file_id);
  RC   move_first_to_end(LeafIndexNodeHandler &other, DiskBufferPool *disk_buffer_pool, int file_id);
  RC   move_last_to_front(LeafIndexNodeHandler &other, DiskBufferPool *bp, int file_id);
  /**
   * move all items to left page
   */
  RC move_to(LeafIndexNodeHandler &other, DiskBufferPool *bp, int file_id);

  int max_size() const;
  int min_size() const;

  bool validate(const KeyComparator &comparator, DiskBufferPool *bp, int file_id) const;

  friend std::string to_string(const LeafIndexNodeHandler &handler, const KeyPrinter &printer);
private:
  char *__item_at(int index) const;
  char *__key_at(int index) const;
  char *__value_at(int index) const;

  void append(const char *item);
  void preappend(const char *item);

private:
  LeafIndexNode *leaf_node_;
};

class InternalIndexNodeHandler : public IndexNodeHandler {
public:
  InternalIndexNodeHandler(const IndexFileHeader &header, BPPageHandle &page_handle);

  void init_empty();
  void create_new_root(PageNum first_page_num, const char *key, PageNum page_num);

  void insert(const char *key, PageNum page_num, const KeyComparator &comparator);
  RC   move_half_to(LeafIndexNodeHandler &other, DiskBufferPool *bp, int file_id);
  char *key_at(int index);
  PageNum value_at(int index);

  /**
   * 返回指定子节点在当前节点中的索引
   */
  int value_index(PageNum page_num);
  void set_key_at(int index, const char *key);
  void remove(int index);

  /**
   * 与Leaf节点不同，lookup返回指定key应该属于哪个子节点，返回这个子节点在当前节点中的索引
   * 如果想要返回插入位置，就提供 `insert_position` 参数
   * NOTE: 查找效率不高，你可以优化它吗?
   */
  int lookup(const KeyComparator &comparator, const char *key,
	     bool *found = nullptr, int *insert_position = nullptr) const;
  
  int max_size() const;
  int min_size() const;

  RC move_to(InternalIndexNodeHandler &other, DiskBufferPool *disk_buffer_pool, int file_id);
  RC move_first_to_end(InternalIndexNodeHandler &other, DiskBufferPool *disk_buffer_pool, int file_id);
  RC move_last_to_front(InternalIndexNodeHandler &other, DiskBufferPool *bp, int file_id);
  RC move_half_to(InternalIndexNodeHandler &other, DiskBufferPool *bp, int file_id);

  bool validate(const KeyComparator &comparator, DiskBufferPool *bp, int file_id) const;

  friend std::string to_string(const InternalIndexNodeHandler &handler, const KeyPrinter &printer);
private:
  RC copy_from(const char *items, int num, DiskBufferPool *disk_buffer_pool, int file_id);
  RC append(const char *item, DiskBufferPool *bp, int file_id);
  RC preappend(const char *item, DiskBufferPool *bp, int file_id);

private:
  char *__item_at(int index) const;
  char *__key_at(int index) const;
  char *__value_at(int index) const;

  int value_size() const;
  int item_size() const;

private:
  InternalIndexNode *internal_node_;
};

class BplusTreeHandler {
public:
  /**
   * 此函数创建一个名为fileName的索引。
   * attrType描述被索引属性的类型，attrLength描述被索引属性的长度
   */
  RC create(const char *file_name, AttrType attr_type, int attr_length,
	    int internal_max_size = -1, int leaf_max_size = -1);

  /**
   * 打开名为fileName的索引文件。
   * 如果方法调用成功，则indexHandle为指向被打开的索引句柄的指针。
   * 索引句柄用于在索引中插入或删除索引项，也可用于索引的扫描
   */
  RC open(const char *file_name);

  /**
   * 关闭句柄indexHandle对应的索引文件
   */
  RC close();

  /**
   * 此函数向IndexHandle对应的索引中插入一个索引项。
   * 参数user_key指向要插入的属性值，参数rid标识该索引项对应的元组，
   * 即向索引中插入一个值为（user_key，rid）的键值对
   */
  RC insert_entry(const char *user_key, const RID *rid);

  /**
   * 从IndexHandle句柄对应的索引中删除一个值为（*pData，rid）的索引项
   * @return RECORD_INVALID_KEY 指定值不存在
   */
  RC delete_entry(const char *user_key, const RID *rid);

  bool is_empty() const;

  /**
   * 获取指定值的record
   * @param rid  返回值，记录记录所在的页面号和slot
   */
  RC get_entry(const char *user_key, std::list<RID> &rids);

  RC sync();

  const int get_file_id()
  {
    return file_id_;
  }

  /**
   * Check whether current B+ tree is invalid or not.
   * return true means current tree is valid, return false means current tree is invalid.
   * @return
   */
  bool validate_tree();

public:
  RC print_tree();
  RC print_leafs();

private:
  RC print_leaf(BPPageHandle &page_handle);
  RC print_internal_node_recursive(BPPageHandle &page_handle);

  bool validate_node(IndexNode *node);
  bool validate_leaf_link();
  bool validate_node_recursive(BPPageHandle &page_handle);

protected:
  RC find_leaf(const char *key, BPPageHandle &page_handle);
  RC left_most_page(BPPageHandle &page_handle);
  RC right_most_page(BPPageHandle &page_handle);
  RC find_leaf_internal(const std::function<PageNum(InternalIndexNodeHandler &)> &child_page_getter,
			BPPageHandle &page_handle);

  RC insert_into_parent(
      PageNum parent_page, BPPageHandle &left_page_handle, const char *pkey, BPPageHandle &right_page_handle);
  RC split_leaf(BPPageHandle &leaf_page_handle);

  RC delete_entry_internal(BPPageHandle &leaf_page_handle, const char *key);

  RC insert_into_new_root(BPPageHandle &left_page_handle, const char *pkey, BPPageHandle &right_page_handle);

  template <typename IndexNodeHandlerType>
  RC split(BPPageHandle &page_handle, BPPageHandle &new_page_handle);
  template <typename IndexNodeHandlerType>
  RC coalesce_or_redistribute(BPPageHandle &page_handle);
  template <typename IndexNodeHandlerType>
  RC coalesce(BPPageHandle &neighbor_page_handle, BPPageHandle &page_handle,
	      BPPageHandle &parent_page_handle, int index);
  template <typename IndexNodeHandlerType>
  RC redistribute(BPPageHandle &neighbor_page_handle, BPPageHandle &page_handle,
		  BPPageHandle &parent_page_handle, int index);

  RC insert_entry_into_parent(BPPageHandle &page_handle, BPPageHandle &new_page_handle, const char *key);
  RC insert_entry_into_leaf_node(BPPageHandle &page_handle, const char *pkey, const RID *rid);
  RC update_root_page_num();
  RC create_new_tree(const char *key, const RID *rid);

  RC adjust_root(BPPageHandle &root_page_handle);

private:
  char *make_key(const char *user_key, const RID &rid);
  void  free_key(char *key);
protected:
  DiskBufferPool *disk_buffer_pool_ = nullptr;
  int file_id_ = -1;
  bool header_dirty_ = false;
  IndexFileHeader file_header_;

  KeyComparator key_comparator_;
  KeyPrinter    key_printer_;

  common::MemPoolItem *mem_pool_item_ = nullptr;

private:
  friend class BplusTreeScanner;
  friend class BplusTreeTester;
};

class BplusTreeScanner {
public:
  BplusTreeScanner(BplusTreeHandler &tree_handler);
  ~BplusTreeScanner();

  /**
   * 扫描指定范围的数据
   * @param left_key 扫描范围的左边界，如果是null，则没有左边界
   * @param left_inclusive 左边界的值是否包含在内
   * @param right_key 扫描范围的右边界。如果是null，则没有右边界
   * @param right_inclusive 右边界的值是否包含在内
   */
  RC open(const char *left_user_key, bool left_inclusive,
	  const char *right_user_key, bool right_inclusive);

  RC next_entry(RID *rid);

  RC close();

private:
  bool inited_ = false;
  BplusTreeHandler &tree_handler_;

  /// 使用左右叶子节点和位置来表示扫描的起始位置和终止位置
  /// 起始位置和终止位置都是有效的数据
  BPPageHandle left_page_handle_;
  BPPageHandle right_page_handle_;
  int          iter_index_ = -1;
  int          end_index_ = -1; // use -1 for end of scan
};

#endif  //__OBSERVER_STORAGE_COMMON_INDEX_MANAGER_H_