From 177cee9f913a72eb402be9fb2485038127dc843c Mon Sep 17 00:00:00 2001
From: YuhangQ <i@yuhangq.com>
Date: Tue, 26 Oct 2021 12:29:59 +0800
Subject: [PATCH] btree insert finished!

---
 .gitignore                   |   1 +
 invodb/btree/btree_node.cpp  |  84 ++++++++++++++++++++--
 invodb/btree/btree_node.h    | 103 ++++-----------------------
 invodb/btree/btree_uuid.cpp  | 131 ++++++++++++++++++++++-------------
 invodb/btree/btree_uuid.h    |   7 +-
 invodb/file/page_manager.cpp |   1 +
 invodb/file/storage_page.h   |   2 +-
 invodb/main.cpp              |   4 +-
 invodb/main.h                |   2 +
 invodb/models/collection.cpp |   1 -
 10 files changed, 188 insertions(+), 148 deletions(-)
 create mode 100644 .gitignore

diff --git a/.gitignore b/.gitignore
new file mode 100644
index 0000000..0d28bdd
--- /dev/null
+++ b/.gitignore
@@ -0,0 +1 @@
+test.invodb
\ No newline at end of file
diff --git a/invodb/btree/btree_node.cpp b/invodb/btree/btree_node.cpp
index aca49a8..2fea8c2 100644
--- a/invodb/btree/btree_node.cpp
+++ b/invodb/btree/btree_node.cpp
@@ -7,12 +7,86 @@
 std::map<int, BTreeNodeUUID*> BTreeNodeUUID::map;
 
 BTreeNodeUUID *BTreeNodeUUID::getNode(const int &address) {
-    if(address == 0) {
-        printf("???");
-        exit(0);
-    }
-    if(map.count(address) == 0) {
+    if(true || map.count(address) == 0) {
         map[address] = new BTreeNodeUUID(address);
     }
     return map[address];
 }
+
+BTreeNodeUUID::BTreeNodeUUID(const int& address):address(address) {
+    clear();
+    StoragePage page = PageManager::Instance().getPage(address);
+    int p = 0;
+    size = page.getIntStartFrom(p);
+    p += 4;
+    parent = page.getIntStartFrom(p);
+    p += 4;
+    left = page.getIntStartFrom(p);
+    p += 4;
+    right = page.getIntStartFrom(p);
+    p += 4;
+    leaf = !page.getIntStartFrom(p);
+    p += 4;
+    for(int i=0; i<m; i++) {
+        for(int j=0; j<32; j++) {
+            key[i] += page[p++];
+        }
+    }
+    for(int i=0; i<m+1; i++) {
+        val[i] = page.getIntStartFrom(p);
+        p += 4;
+    }
+}
+
+int BTreeNodeUUID::insert(const std::string uuid) {
+    int pos = 0;
+    while(pos < size && uuid > key[pos]) pos++;
+    val[size + 1] = val[size];
+    for(int i=size; i>pos; i--) {
+        val[i] = val[i - 1];
+        key[i] = key[i - 1];
+    }
+    key[pos] = uuid;
+    size++;
+    return pos;
+}
+
+void BTreeNodeUUID::print() {
+    printf("---------BTreeNode---------\n");
+    for(int i=0; i<size; i++) {
+        printf("%s %d\n", key[i].c_str(), val[i]);
+    }
+}
+
+void BTreeNodeUUID::clear() {
+    for(int i=0; i<m+1; i++) key[i].clear(), val[i] = 0;
+    size = 0;
+    leaf = false;
+    parent = 0;
+}
+
+void BTreeNodeUUID::save() {
+    StoragePage page(address);
+
+    int p = 0;
+    page.setIntStartFrom(p, size);
+    p += 4;
+    page.setIntStartFrom(p, parent);
+    p += 4;
+    page.setIntStartFrom(p, left);
+    p += 4;
+    page.setIntStartFrom(p, right);
+    p += 4;
+    page.setIntStartFrom(p, !leaf);
+    p += 4;
+    for(int i=0; i<m; i++) {
+        for(int j=0; j<32; j++) {
+            page[p++] = key[i][j];
+        }
+    }
+    for(int i=0; i<m+1; i++) {
+        page.setIntStartFrom(p, val[i]);
+        p += 4;
+    }
+    page.save();
+}
diff --git a/invodb/btree/btree_node.h b/invodb/btree/btree_node.h
index 171b3fb..3d2c439 100644
--- a/invodb/btree/btree_node.h
+++ b/invodb/btree/btree_node.h
@@ -12,7 +12,7 @@
 #include "file/page_manager.h"
 
 /**
- * m = 28
+ * m = 27
  * value string max
  * (32 + 4)*28 + 5 = 1013
  */
@@ -21,96 +21,23 @@
 class BTreeNodeUUID {
 public:
     static BTreeNodeUUID* getNode(const int& address);
-    BTreeNodeUUID(const int& address):address(address) {
-        clear();
-        StoragePage page = PageManager::Instance().getPage(address);
-        for(int i=0; i<28; i++) {
-            for(int j=0; j<32; j++) {
-                key[i] += page[i * 32 + j];
-            }
-        }
-        int p = 28 * 32;
-        for(int i=0; i<m+1; i++) {
-            val[i] = page.getIntStartFrom(p);
-            p += 4;
-        }
-        leaf = !page[1013];
-        parent = page.getIntStartFrom(1014);
-    }
-    bool isLeaf() {
-        return leaf;
-    }
-
-    void setLeaf(const bool& value) {
-        leaf = value;
-    }
-    void updateParent() {
-        if(isLeaf()) return;
-        for(int i=0; i<=size(); i++) {
-            BTreeNodeUUID child = getNode(val[i])->parent;
-            child.parent = address;
-            child.save();
-        }
-    }
-    int size() {
-        return n_size;
-    }
-    int insert(const std::string uuid) {
-
-//        static int cnt = 0;
-//        if(!isLeaf() && size() >= 27) {
-//            if(cnt) throw "fuck";
-//            cnt++;
-//        }
-        int pos = 0;
-        while(pos < n_size && uuid > key[pos]) pos++;
-        val[n_size + 1] = val[n_size];
-        for(int i=n_size; i>pos; i--) {
-            val[i] = val[i - 1];
-            key[i] = key[i - 1];
-        }
-        key[pos] = uuid;
-        n_size++;
-        return pos;
-    }
-    void print() {
-        printf("---------BTreeNode---------\n");
-        for(int i=0; i<size(); i++) {
-            printf("%s %d\n", key[i].c_str(), val[i]);
-        }
-    }
-    void clear() {
-        for(int i=0; i<m; i++) key[i].clear(), val[i] = 0;
-        n_size = 0;
-        leaf = false;
-        parent = 0;
-    }
-    int getAddress() { return address; }
-    void save() {
-        StoragePage page(address);
-        for(int i=0; i<28; i++) {
-            for(int j=0; j<32; j++) {
-                page[i * 32 + j] = key[i][j];
-            }
-        }
-        int p = 28 * 32;
-        for(int i=0; i<m+1; i++) {
-            page.setIntStartFrom(p, val[i]);
-            p += 4;
-        }
-        page[1013] = !leaf;
-        page.setIntStartFrom(1014, parent);
-        page.save();
-    }
-    static const int m = 28;
-    std::string key[m];
+    int insert(const std::string uuid);
+    void print();
+    void clear();
+    void save();
+    static const int m = 27;
+    std::string key[m+1];
     int val[m+1];
     int parent;
-private:
-    static std::map<int, BTreeNodeUUID*> map;
-    int address;
-    int n_size;
+    int left;
+    int right;
     bool leaf;
+    int size;
+    int address;
+private:
+    BTreeNodeUUID(const int& address);
+    static std::map<int, BTreeNodeUUID*> map;
+
 };
 
 #endif //INVODB_BTREE_NODE_H
diff --git a/invodb/btree/btree_uuid.cpp b/invodb/btree/btree_uuid.cpp
index c47a65a..357ae27 100644
--- a/invodb/btree/btree_uuid.cpp
+++ b/invodb/btree/btree_uuid.cpp
@@ -5,41 +5,46 @@
 #include "btree_uuid.h"
 
 BTreeUUID::BTreeUUID(const int& address) {
-    root = BTreeNodeUUID::getNode(address);
+    root = address;
 }
 
 void BTreeUUID::insert(const char *uuid, int address) {
 
-    BTreeNodeUUID* cur = root;
+    BTreeNodeUUID* cur = BTreeNodeUUID::getNode(root);
     BTreeNodeUUID* parent = nullptr;
 
-    while(!cur->isLeaf()) {
+    while(!cur->leaf) {
         parent = cur;
-        for(int i=0; i<cur->size(); i++) {
+        for(int i=0; i<cur->size; i++) {
             if(uuid < cur->key[i]) {
                 cur = BTreeNodeUUID::getNode(cur->val[i]);
                 break;
             }
-            if(i == cur->size() - 1) {
+            if(i == cur->size - 1) {
                 cur = BTreeNodeUUID::getNode(cur->val[i + 1]);
                 break;
             }
         }
-        cur->parent = parent->getAddress();
+        cur->parent = parent->address;
+        cur->save();
     }
 
     // insert directly
-    if(cur->size() < cur->m - 1) {
+    if(cur->size < cur->m - 1) {
         cur->val[cur->insert(uuid)] = address;
         cur->save();
         return;
     }
 
     // split
-    split(uuid, address, parent, cur);
+    if(parent == nullptr) split(uuid, address, 0, cur->address);
+    else split(uuid, address, parent->address, cur->address);
 }
 
-void BTreeUUID::split(std::string uuid, int address, BTreeNodeUUID *parent, BTreeNodeUUID *cur) {
+void BTreeUUID::split(std::string uuid, int address, int parentAddr, int curAddr) {
+
+    BTreeNodeUUID* parent = BTreeNodeUUID::getNode(parentAddr);
+    BTreeNodeUUID* cur = BTreeNodeUUID::getNode(curAddr);
 
     cur->val[cur->insert(uuid)] = address;
 
@@ -48,30 +53,42 @@ void BTreeUUID::split(std::string uuid, int address, BTreeNodeUUID *parent, BTre
 
     int mid = (cur->m / 2);
     for(int i=0; i<mid; i++) lLeaf->val[lLeaf->insert(cur->key[i])] = cur->val[i];
+    lLeaf->right = rLeaf->address;
+    lLeaf->left = cur->left;
     for(int i=mid; i<cur->m; i++) rLeaf->val[rLeaf->insert(cur->key[i])] = cur->val[i];
+    rLeaf->left = rLeaf->address;
+    rLeaf->right = cur->right;
 
-    if(cur == root) {
+    if(cur->address == root) {
         BTreeNodeUUID* newRoot = BTreeNodeUUID::getNode(PageManager::Instance().allocate());
         newRoot->insert(rLeaf->key[0]);
-        newRoot->val[0] = lLeaf->getAddress();
-        newRoot->val[1] = rLeaf->getAddress();
-        newRoot->setLeaf(false);
-        root = newRoot;
-        lLeaf->parent = rLeaf->parent = root->getAddress();
+        newRoot->val[0] = lLeaf->address;
+        newRoot->val[1] = rLeaf->address;
+        newRoot->leaf = false;
+        root = newRoot->address;
+        lLeaf->parent = rLeaf->parent = root;
 
         newRoot->save();
         lLeaf->save();
         rLeaf->save();
-    } else insertInternal(rLeaf->key[0], parent, lLeaf, rLeaf);
+    } else {
+        lLeaf->save();
+        rLeaf->save();
+        insertInternal(rLeaf->key[0], parent->address, lLeaf->address, rLeaf->address);
+    }
 }
 
-void BTreeUUID::insertInternal(std::string uuid, BTreeNodeUUID *cur, BTreeNodeUUID *lLeaf, BTreeNodeUUID *rLeaf) {
+void BTreeUUID::insertInternal(std::string uuid, int curAddr, int lLeafAddr, int rLeafAddr) {
 
-    if(cur->size() < cur->m - 1) {
+    BTreeNodeUUID *cur = BTreeNodeUUID::getNode(curAddr);
+    BTreeNodeUUID *lLeaf = BTreeNodeUUID::getNode(lLeafAddr);
+    BTreeNodeUUID *rLeaf = BTreeNodeUUID::getNode(rLeafAddr);
+
+    if(cur->size < cur->m - 1) {
         int pos = cur->insert(uuid);
-        cur->val[pos] = lLeaf->getAddress();
-        cur->val[pos+1] = rLeaf->getAddress();
-        lLeaf->parent = rLeaf->parent = root->getAddress();
+        cur->val[pos] = lLeaf->address;
+        cur->val[pos+1] = rLeaf->address;
+        lLeaf->parent = rLeaf->parent = root;
         cur->save();
         lLeaf->save();
         rLeaf->save();
@@ -80,19 +97,14 @@ void BTreeUUID::insertInternal(std::string uuid, BTreeNodeUUID *cur, BTreeNodeUU
 
     BTreeNodeUUID* newLChild = BTreeNodeUUID::getNode(PageManager::Instance().allocate());
     BTreeNodeUUID* newRChild = BTreeNodeUUID::getNode(PageManager::Instance().allocate());
-    newLChild->setLeaf(false);
-    newRChild->setLeaf(false);
-
-    if(cur->size() != 27) {
-        printf("%d %d %d p:%d cur:%d\n", cur->size(), cur->isLeaf(), cur==root, cur->parent, cur->getAddress());
-        exit(0);
-    }
+    newLChild->leaf = false;
+    newRChild->leaf = false;
 
     int pos = cur->insert(uuid);
-    cur->val[pos] = lLeaf->getAddress();
-    cur->val[pos+1] = rLeaf->getAddress();
+    cur->val[pos] = lLeaf->address;
+    cur->val[pos+1] = rLeaf->address;
 
-    int mid = cur->size() / 2;
+    int mid = cur->size / 2;
 
     for(int i=0; i<mid; i++) newLChild->insert(cur->key[i]);
     for(int i=0; i<=mid; i++) newLChild->val[i] = cur->val[i];
@@ -100,46 +112,70 @@ void BTreeUUID::insertInternal(std::string uuid, BTreeNodeUUID *cur, BTreeNodeUU
     for(int i=mid+1; i<cur->m; i++) newRChild->insert(cur->key[i]);
     for(int i=mid+1; i<=cur->m; i++) newRChild->val[i-mid-1] = cur->val[i];
 
-    if(cur == root) {
+    lLeaf->save();
+    rLeaf->save();
+
+    if(cur->address == root) {
         BTreeNodeUUID* newRoot = BTreeNodeUUID::getNode(PageManager::Instance().allocate());
         newRoot->insert(cur->key[mid]);
-        newRoot->val[0] = newLChild->getAddress();
-        newRoot->val[1] = newRChild->getAddress();
-        newRoot->setLeaf(false);
-        root = newRoot;
-        newLChild->parent = newRChild->parent = root->getAddress();
+        newRoot->val[0] = newLChild->address;
+        newRoot->val[1] = newRChild->address;
+        newRoot->leaf = false;
+        root = newRoot->address;
+        newLChild->parent = newRChild->parent = root;
 
         newRoot->save();
         newLChild->save();
         newRChild->save();
     } else {
-        insertInternal(cur->key[mid], BTreeNodeUUID::getNode(cur->parent), newLChild, newRChild);
+        newLChild->save();
+        newRChild->save();
+        if(cur->parent == 0) throw "fuck";
+        insertInternal(cur->key[mid], cur->parent, newLChild->address, newRChild->address);
     }
-
 }
 
 void BTreeUUID::print() {
-    innerPrint(root);
+    innerPrint(BTreeNodeUUID::getNode(root));
 }
 
 void BTreeUUID::innerPrint(BTreeNodeUUID *cur) {
-    if(cur == root) {
+    if(cur->address == root) {
         cnt = 0;
     }
-    if(cur->isLeaf()) cnt += cur->size();
-    printf("---------%d(%d)count=%d&sum=%d--------\n", cur->getAddress(), cur->isLeaf(), cur->size(), cnt);
-    for(int i=0; i<cur->size(); i++) {
+    if(cur->leaf) cnt += cur->size;
+    printf("---------%d(%d)count=%d&sum=%d---l:%d,r:%d-----\n", cur->address, cur->leaf, cur->size, cnt, cur->left, cur->right);
+    for(int i=0; i<cur->size; i++) {
         printf("%d:%s ", i, cur->key[i].substr(0, 4).c_str());
     }
     printf("\n");
-    for(int i=0; i<=cur->size(); i++) {
+    for(int i=0; i<=cur->size; i++) {
         printf("%d:%d ", i, cur->val[i]);
     }
     printf("\n");
-    if(cur->isLeaf()) return;
-    for(int i=0; i<=cur->size(); i++) {
+
+
+    if(cur->leaf) return;
+    for(int i=0; i<=cur->size; i++) {
         innerPrint(BTreeNodeUUID::getNode(cur->val[i]));
     }
+
+
+    /*
+    BTreeNodeUUID test = *cur;
+    test.address = PageManager::Instance().allocate();
+    test.print();
+
+
+    test.save();
+
+    BTreeNodeUUID test2 = *BTreeNodeUUID::getNode(test.address);
+
+    printf("test: size:%d l:%d r:%d\n", test2.size, test2.left, test2.right);
+    test2.print();
+     */
+
+
 }
 /*
 void BTreeUUID::innerInsert(BTreeNodeUUID* &p, BTreeNodeUUID* f, const char *uuid, int address) {
@@ -161,3 +197,4 @@ void BTreeUUID::innerInsert(BTreeNodeUUID* &p, BTreeNodeUUID* f, const char *uui
 }
 */
 
+
diff --git a/invodb/btree/btree_uuid.h b/invodb/btree/btree_uuid.h
index 0b44d13..d489557 100644
--- a/invodb/btree/btree_uuid.h
+++ b/invodb/btree/btree_uuid.h
@@ -10,14 +10,13 @@
 class BTreeUUID {
 public:
     BTreeUUID(const int& address);
-    ~BTreeUUID() { delete root; };
     void insert(const char* uuid, int address);
     void print();
 private:
     void innerPrint(BTreeNodeUUID* cur);
-    void split(std::string uuid, int address, BTreeNodeUUID* parent, BTreeNodeUUID* cur);
-    void insertInternal(std::string uuid, BTreeNodeUUID* cur, BTreeNodeUUID* lLeaf, BTreeNodeUUID* rLeaf);
-    BTreeNodeUUID *root;
+    void split(std::string uuid, int address, int parentAddr, int curAddr);
+    void insertInternal(std::string uuid, int curAddr, int lLeafAddr, int rLeafAddr);
+    int root;
     int cnt;
 };
 
diff --git a/invodb/file/page_manager.cpp b/invodb/file/page_manager.cpp
index 4f5face..1f839bb 100644
--- a/invodb/file/page_manager.cpp
+++ b/invodb/file/page_manager.cpp
@@ -22,6 +22,7 @@ void PageManager::setPage(const int &index, const StoragePage &page) {
     stream.clear();
     stream.seekg(index * 1024);
     stream.write(page, 1024);
+    stream.flush();
 }
 
 int PageManager::allocate() {
diff --git a/invodb/file/storage_page.h b/invodb/file/storage_page.h
index 71c5e00..90b902a 100644
--- a/invodb/file/storage_page.h
+++ b/invodb/file/storage_page.h
@@ -24,7 +24,7 @@ public:
     void setStringStartFrom(const int &index, const char *str);
     int *intArray();
     StoragePage(const int& id) { memset(page, 0, sizeof(page)); this->address = id; }
-    char& operator[] (int index) { return this->page[index]; }
+    char& operator[] (int index) { if(index>=1024 || index < 0) throw "overflow"; else return this->page[index]; }
     operator const char *() const { return this->page; }
     operator char *() { return this->page; }
     int getAddress();
diff --git a/invodb/main.cpp b/invodb/main.cpp
index 63ced14..911e34b 100644
--- a/invodb/main.cpp
+++ b/invodb/main.cpp
@@ -14,6 +14,7 @@ int main() {
 
     PageManager& manager = PageManager::Instance();
 
+
     Collection *col;
     try {
        col = &Collection::getCollection("hello");
@@ -24,10 +25,9 @@ int main() {
     JSON json("{\"hello\": 1}");
     col->insert(json);
 
-
     BTreeUUID *btree = new BTreeUUID(PageManager::Instance().allocate());
     char uuid[32];
-    for(int i=0; i<100000; i++) {
+    for(int i=0; i<1000000; i++) {
         generateUUID(uuid);
         btree->insert(uuid, PageManager::Instance().allocate());
     }
diff --git a/invodb/main.h b/invodb/main.h
index b4d5c0e..38d6d77 100644
--- a/invodb/main.h
+++ b/invodb/main.h
@@ -6,6 +6,8 @@
 #define INVODB_MAIN_H
 
 #include <iostream>
+#include <cstdlib>
+#include <ctime>
 #include <models/collection.h>
 
 
diff --git a/invodb/models/collection.cpp b/invodb/models/collection.cpp
index dff3bbf..7aa8a7d 100644
--- a/invodb/models/collection.cpp
+++ b/invodb/models/collection.cpp
@@ -9,7 +9,6 @@ Collection::Collection(const std::string &name, const int &firstPage) {
 }
 
 void Collection::insert(JSON &json) {
-
     if(!json.HasMember("__Invo_ID__")) {
         char uuid[32];
         generateUUID(uuid);