try to add lowmemory load

debug/final_load.cc

+/*1. store data as column files, and restore data as plaintext insert query
+* 2. plaintext insert query should be able to recover directly
+* 3. should be able to used exsisting data to reduce the computation overhead(to be implemented)
+*/
+#include <stdlib.h>
+#include <stdio.h>
+#include <string>
+#include <map>
+#include <iostream>
+#include <fstream>
+#include <sstream>
+#include <vector>
+#include <algorithm>
+#include "wrapper/reuse.hh"
+#include "wrapper/common.hh"
+#include "wrapper/insert_lib.hh"
+using std::cout;
+using std::cin;
+using std::endl;
+using std::vector;
+using std::string;
+using std::to_string;
+static std::string embeddedDir="/t/cryt/shadow";
+char * globalEsp=NULL;
+int num_of_pipe = 4;
+//global map, for each client, we have one WrapperState which contains ProxyState.
+static std::map<std::string, WrapperState*> clients;
+//This connection mimics the behaviour of MySQL-Proxy
+Connect  *globalConn;
+/*for each field, convert the format to FieldMeta_Wrapper*/
+static void init(){
+    std::string client="192.168.1.1:1234";
+    //one Wrapper per user.
+    clients[client] = new WrapperState();    
+    //Connect phase
+    ConnectionInfo ci("localhost", "root", "letmein",3306);
+    const std::string master_key = "113341234";
+    char *buffer;
+    if((buffer = getcwd(NULL, 0)) == NULL){  
+        perror("getcwd error");  
+    }
+    embeddedDir = std::string(buffer)+"/shadow";
+    SharedProxyState *shared_ps = 
+			new SharedProxyState(ci, embeddedDir , master_key, 
+                                            determineSecurityRating());
+    assert(0 == mysql_thread_init());
+    //we init embedded database here.
+    clients[client]->ps = std::unique_ptr<ProxyState>(new ProxyState(*shared_ps));
+    clients[client]->ps->safeCreateEmbeddedTHD();
+    //Connect end!!
+    globalConn = new Connect(ci.server, ci.user, ci.passwd, ci.port);
+}
+
+//========================================================================================//
+
+
+fullBackUp gfb;
+
+struct batch{
+    vector<string> field_names;
+    vector<int> field_types;
+    vector<int> field_lengths;
+};
+
+batch ggbt;
+
+/*should choose the right decryption onion*/
+static
+std::shared_ptr<ReturnMeta> getReturnMeta(std::vector<FieldMeta*> fms,
+                                      std::vector<FieldMetaTrans> &tfds){
+    assert(fms.size()==tfds.size());
+    std::shared_ptr<ReturnMeta> myReturnMeta = std::make_shared<ReturnMeta>();
+    int pos=0;
+    //construct OLK
+    for(auto i=0u;i<tfds.size();i++){
+        //the order is DET,OPE,ASHE,AGG. other onions are not decryptable!!
+        int index = getDecryptionOnionIndex(tfds[i]);
+        if(index==-1) assert(0);
+
+        onion o = tfds[i].getChoosenOnionO()[index];
+        SECLEVEL l = tfds[i].getOriginalFieldMeta()->getOnionMeta(o)->getSecLevel();
+        FieldMeta *k = tfds[i].getOriginalFieldMeta();
+        OLK curOLK(o,l,k);
+        bool use_salt = false;
+        if(needsSalt(curOLK))
+            use_salt = true;
+	addToReturn(myReturnMeta.get(),pos++,curOLK,use_salt,k->getFieldName());
+
+        if(use_salt)
+            addSaltToReturn(myReturnMeta.get(),pos++);
+
+        ggbt.field_types.push_back(tfds[i].getChoosenFieldTypes()[index]);
+        ggbt.field_names.push_back(tfds[i].getChoosenOnionName()[index]);
+        ggbt.field_lengths.push_back(tfds[i].getChoosenFieldLengths()[index]);
+
+        if(use_salt){
+            ggbt.field_types.push_back(tfds[i].getSaltType());
+            ggbt.field_names.push_back(tfds[i].getSaltName());
+            ggbt.field_lengths.push_back(tfds[i].getSaltLength());
+        }
+    }
+    return myReturnMeta;
+}
+
+/*init global full backup. */
+static
+void initGfb(std::vector<FieldMetaTrans> &res,std::string db,std::string table){
+    vector<string> field_names;
+    vector<int> field_types;
+    vector<int> field_lengths;
+    /*choosen onions should all be included in gfb. salt is also included
+      it's hard to decide whether a FieldMetaTrans has salt because the senmantic is different from that of FieldMeta.
+    */
+    for(auto &item:res){
+        for(auto i:item.getChoosenOnionName()){
+            field_names.push_back(i);
+        }
+        for(auto i:item.getChoosenFieldTypes()){
+            field_types.push_back(i);
+        }
+        for(auto i:item.getChoosenFieldLengths()){
+            field_lengths.push_back(i);
+        }
+        if(item.getHasSalt()){
+            field_names.push_back(item.getSaltName());
+            field_types.push_back(item.getSaltType());
+            field_lengths.push_back(item.getSaltLength());
+        }
+    }
+    gfb.field_names = field_names;
+    gfb.field_types = field_types;
+    gfb.field_lengths = field_lengths;
+    //then we should read the vector
+    std::string prefix = std::string("data/")+db+"/"+table+"/";
+    for(unsigned int i=0u; i<gfb.field_names.size(); i++) {
+        std::string filename = prefix + gfb.field_names[i];
+        std::vector<std::string> column;
+        if(IS_NUM(gfb.field_types[i])){
+            load_num_file(filename,column);
+        }else{
+            load_string_file(filename,column,gfb.field_lengths[i]);
+        }
+        gfb.annoOnionNameToFileVector[gfb.field_names[i]] = std::move(column);
+    }
+    //init another map
+    for(unsigned int i=0;i<gfb.field_names.size();i++){
+        gfb.annoOnionNameToType[gfb.field_names[i]] = gfb.field_types[i];
+    }
+}
+
+/*load file, decrypt, and then return data plain fields in the type ResType*/
+static ResType load_files(std::string db, std::string table){
+    std::unique_ptr<SchemaInfo> schema =  myLoadSchemaInfo(embeddedDir);
+    //get all the fields in the tables.
+    std::vector<FieldMeta*> fms = getFieldMeta(*schema,db,table);
+    TableMetaTrans res_meta = loadTableMetaTrans(db,table);
+    std::vector<FieldMetaTrans> res = res_meta.getFts();
+    for(unsigned int i=0;i<fms.size();i++){
+        res[i].trans(fms[i]);
+    }
+    //then we should load all the fields available
+    initGfb(res,db,table);   
+
+    std::shared_ptr<ReturnMeta> rm = getReturnMeta(fms,res);
+
+    vector<string> field_names = ggbt.field_names;
+    vector<int> field_types = ggbt.field_types;
+    vector<int> field_lengths = ggbt.field_lengths;
+
+    //why do we need this??
+    create_embedded_thd(0);
+    rawMySQLReturnValue resraw;
+//    vector<vector<string>> resss_field = loadTableFieldsForDecryption(db,
+//                                         table,field_names, field_types, field_lengths);
+    vector<vector<string>> res_field;   
+    for(auto item:field_names){
+        res_field.push_back(gfb.annoOnionNameToFileVector[item]);
+    }
+    //then transform it to ress_fields
+    unsigned int length = res_field[0].size();
+
+    vector<vector<string>> ress_field;
+    for(unsigned int i=0u;i<length;i++){
+        vector<string> row;
+        for(unsigned int j=0u;j<res_field.size();j++){
+            row.push_back(res_field[j][i]);
+        }
+        ress_field.push_back(row);
+    }
+
+    resraw.rowValues = ress_field;
+    resraw.fieldNames = field_names;
+    for(unsigned int i=0;i<field_types.size();++i){
+	resraw.fieldTypes.push_back(static_cast<enum_field_types>(field_types[i]));
+    }
+    ResType rawtorestype = rawMySQLReturnValue_to_ResType(false, &resraw);
+    auto finalresults = decryptResults(rawtorestype,*rm);
+    return finalresults;
+}
+
+static
+void local_wrapper(const Item &i, const FieldMeta &fm, Analysis &a,
+                           List<Item> *const append_list){
+    //append_list->push_back(&(const_cast<Item&>(i)));
+    //do not use the plain strategy 
+    std::vector<Item *> l;
+    const uint64_t salt = fm.getHasSalt() ? randomValue() : 0;
+    uint64_t IV = salt;
+    for (auto it : fm.orderedOnionMetas()) {
+        const onion o = it.first->getValue();
+        OnionMeta * const om = it.second;
+        std::string annoOnionName = om->getAnonOnionName();
+        if(gfb.annoOnionNameToFileVector.find(annoOnionName)!=gfb.annoOnionNameToFileVector.end()){
+            enum_field_types type = static_cast<enum_field_types>(gfb.annoOnionNameToType[annoOnionName]);
+            std::vector<std::string> &tempFileVector = gfb.annoOnionNameToFileVector[annoOnionName];
+            std::string in = tempFileVector.back();            
+            if(IS_NUM(type)){
+                //std::string in("11");
+                unsigned int len = annoOnionName.size();
+                if(len>4u&&annoOnionName.substr(len-4)=="ASHE"){
+                    l.push_back(MySQLFieldTypeToItem(type,in));
+                }else{
+                    l.push_back( new (current_thd->mem_root)
+                                Item_int(static_cast<ulonglong>(valFromStr(in))) );
+                }
+            }else{
+                //std::string in("hehe");
+                l.push_back(MySQLFieldTypeToItem(type,in));
+            }
+            tempFileVector.pop_back();
+            //l.push_back(&(const_cast<Item&>(i)));
+        }else{
+            l.push_back(my_encrypt_item_layers(i, o, *om, a, IV));
+        }
+    }
+    std::string saltName = fm.getSaltName();
+    if (fm.getHasSalt()) {
+        if(gfb.annoOnionNameToFileVector.find(saltName)!=gfb.annoOnionNameToFileVector.end()){
+            std::vector<std::string> &tempFileVector = gfb.annoOnionNameToFileVector[saltName];
+            std::string in = tempFileVector.back();
+//            enum_field_types type = static_cast<enum_field_types>(gfb.annoOnionNameToType[saltName]);
+//            l.push_back(MySQLFieldTypeToItem(type,in));
+            l.push_back( new (current_thd->mem_root)
+                                Item_int(static_cast<ulonglong>(valFromStr(in)))
+             );
+            tempFileVector.pop_back();
+        }else{
+            l.push_back(new Item_int(static_cast<ulonglong>(salt)));
+        }
+    }
+
+    for (auto it : l) {
+        append_list->push_back(it);
+    }
+}
+
+
+int
+main(int argc, char* argv[]){
+    init();
+    create_embedded_thd(0);
+    std::string db="tdb",table="student";
+    std::string ip="localhost";
+    if(argc==4){
+        ip = std::string(argv[1]);
+        db = std::string(argv[2]);
+        table = std::string(argv[3]);
+    }
+
+    std::unique_ptr<SchemaInfo> schema =  myLoadSchemaInfo(embeddedDir);
+    schema.get();
+    const std::unique_ptr<AES_KEY> &TK = std::unique_ptr<AES_KEY>(getKey(std::string("113341234")));
+    Analysis analysis(db, *schema, TK, SECURITY_RATING::SENSITIVE);
+
+    /*choose decryption onion, load and decrypt to plain text*/
+    ResType res =  load_files(db,table);
+    std::string annoTableName = analysis.getTableMeta(db,table).getAnonTableName();
+
+    const std::string head = std::string("INSERT INTO `")+db+"`.`"+annoTableName+"` ";
+
+    /*reencryption to get the encrypted insert!!!*/
+    for(auto &row:res.rows){
+        List<List_item> newList;
+        List<Item> *const newList0 = new List<Item>();
+        for(auto i=0u;i<res.names.size();i++){
+            std::string field_name = res.names[i];
+            FieldMeta & fm = analysis.getFieldMeta(db,table,field_name);
+            local_wrapper(*row[i],fm,analysis,newList0);
+        }
+        newList.push_back(newList0);
+        std::ostringstream o;
+        insertManyValues(o,newList);
+        std::cout<<(head+o.str())<<std::endl;
+    }
+    return 0;
+}
+

debug/final_load_low_memory.cc

@@ -132,6 +132,8 @@ void initGfb(std::vector<FieldMetaTrans> &res,std::string db,std::string table){
     gfb.field_lengths = field_lengths;
     //then we should read the vector
     std::string prefix = std::string("data/")+db+"/"+table+"/";
+
+    unsigned long tupleNum=0;
     for(unsigned int i=0u; i<gfb.field_names.size(); i++) {
         std::string filename = prefix + gfb.field_names[i];
         std::vector<std::string> column;
@@ -140,12 +142,17 @@ void initGfb(std::vector<FieldMetaTrans> &res,std::string db,std::string table){
         }else{
             load_string_file(filename,column,gfb.field_lengths[i]);
         }
+        tupleNum = column.size();
         gfb.annoOnionNameToFileVector[gfb.field_names[i]] = std::move(column);
     }
     //init another map
     for(unsigned int i=0;i<gfb.field_names.size();i++){
         gfb.annoOnionNameToType[gfb.field_names[i]] = gfb.field_types[i];
     }
+    //extra transformation. transform rows to item*
+    for(unsigned int i=0;i<gfb.field_names.size();i++){
+        gfb.annoOnionNameToItemVector[gfb.field_names[i]] = itemNullVector(tupleNum);
+    }
 }
 
 /*load file, decrypt, and then return data plain fields in the type ResType*/
@@ -192,7 +199,7 @@ static ResType load_files_low_memory(std::string db, std::string table){
     for(unsigned int i=0;i<field_types.size();++i){
 	resraw.fieldTypes.push_back(static_cast<enum_field_types>(field_types[i]));
     }
-
+    //mem 91%
     ResType rawtorestype = rawMySQLReturnValue_to_ResType(false, &resraw);
     auto finalresults = decryptResults(rawtorestype,*rm);
     return std::move(finalresults);
@@ -259,10 +266,12 @@ int
 main(int argc, char* argv[]){
     init();
     create_embedded_thd(0);
+    std::string ip = "localhost";
     std::string db="tdb",table="student";
     if(argc==3){
-        db = std::string(argv[1]);
-        table = std::string(argv[2]);
+        ip = std::string(argv[1]);
+        db = std::string(argv[2]);
+        table = std::string(argv[3]);
     }
 
     std::unique_ptr<SchemaInfo> schema =  myLoadSchemaInfo(embeddedDir);

debug/final_store.cc

@@ -121,11 +121,10 @@ main(int argc, char* argv[]){
     std::string db="tdb",table="student";
     std::string ip="127.0.0.1";
     int port=3306;
-    if(argc==5){
+    if(argc==4){
         ip = std::string(argv[1]);
-        port = std::stoi(std::string(argv[2]));
-        db = std::string(argv[3]);
-        table = std::string(argv[4]);
+        db = std::string(argv[2]);
+        table = std::string(argv[3]);
     }
     init(ip,port);
     store(db,table);

debug/micro_init_str.cc

@@ -4,18 +4,19 @@ using std::string;
 
 int
 main(int argc,char ** argv) {
-    if(argc<2){
-        std::cout<<"expect 1 or 2 arguments"<<std::endl;
+    if(argc!=3){
+        std::cout<<"expect ip and length"<<std::endl;
         return 0;
     }
-    string length(argv[1]);
+    string ip(argv[1]);
+    string length(argv[2]);
 
     std::vector<string> create{
         "create database micro_db;",
         "use micro_db;",
         string("create table str_table(name varchar(")+length+"));"
     };
-    big_proxy b;  
+    big_proxy b("tdb",ip,"root","letmein",3306);
     for(auto item:create){
         b.go(item);
     }

debug/micro_insert_int.cc

+#include "main/big_proxy.hh"
+#include <vector>
+using std::string;
+
+int
+main(int argc,char ** argv) {
+    std::vector<string> create{
+        "use micro_db;",
+        "insert into int_table values(1),(2),(3),(4),(5);",
+    };
+    if(argc!=2){
+        std::cout<<"1:ip"<<std::endl;
+        return 0;
+    }
+    std::string ip = std::string(argv[1]);
+    big_proxy b("tdb",ip,"root","letmein",3306);
+    for(auto item:create){
+        b.go(item);
+    }
+    return 0;
+}

debug/micro_insert_str.cc

+#include "main/big_proxy.hh"
+#include <vector>
+using std::string;
+
+int
+main(int argc,char ** argv) {
+    if(argc!=2){
+        std::cout<<"expect ip"<<std::endl;
+        return 0;
+    }
+    string ip(argv[1]);
+
+    std::vector<string> create{
+        "use micro_db;",
+        "insert into str_table values('a'),('b'),('c'),('d'),('e');"
+    };
+    big_proxy b("tdb",ip,"root","letmein",3306);
+    for(auto item:create){
+        b.go(item);
+    }
+    return 0;
+}
+

debug/micro_select_int.cc

+#include "main/big_proxy.hh"
+#include <vector>
+using std::string;
+
+int
+main(int argc,char ** argv) {
+    std::vector<string> create{
+        "use micro_db;",
+        "select * from int_table;",
+    };
+    if(argc!=2){
+        std::cout<<"1:ip"<<std::endl;
+        return 0;
+    }
+    std::string ip = std::string(argv[1]);
+    big_proxy b("tdb",ip,"root","letmein",3306);
+    for(auto item:create){
+        b.go(item);
+    }
+    return 0;
+}

debug/micro_select_str.cc

+#include "main/big_proxy.hh"
+#include <vector>
+using std::string;
+
+int
+main(int argc,char ** argv) {
+    if(argc!=2){
+        std::cout<<"expect ip and length"<<std::endl;
+        return 0;
+    }
+    string ip(argv[1]);
+
+    std::vector<string> create{
+        "use micro_db;",
+        "select * from str_table;"
+    };
+    big_proxy b("tdb",ip,"root","letmein",3306);
+    for(auto item:create){
+        b.go(item);
+    }
+    return 0;
+}
+

wrapper/reuse.cc

@@ -438,6 +438,16 @@ void load_num_file(std::string filename,std::vector<std::string> &res){
     }
     infile.close();
 }
+/*
+void load_num_file(std::string filename,std::vector<Item> &res,enum_field_types intype){
+    std::ifstream infile(filename);
+    std::string line;     
+    while(std::getline(infile,line)){
+        res.pusb_back((void*)MySQLFieldTypeToItem(intype,line));
+    }
+    infile.close();
+}*/
+
 
 
 void load_string_file(std::string filename, std::vector<std::string> &res,unsigned long length){
@@ -450,6 +460,17 @@ void load_string_file(std::string filename, std::vector<std::string> &res,unsign
     close(fd);
 }
 
+/*
+void load_string_file(std::string filename,std::vector<void*> &res,unsigned long length,enum_field_types intype){
+    char *buf = new char[length];
+    int fd = open(filename.c_str(),O_RDONLY);
+    if(fd==-1) assert(0);//reading from -1 may cause errors
+    while(read(fd,buf,length)!=0){
+        res.push_back((void*)MySQLFieldTypeToItem(intype,std::string(buf,length)));
+    }
+    close(fd);
+}
+*/
 
 std::ostream&
 insertManyValues(std::ostream &out,List<List_item> &newList){

wrapper/reuse.hh

@@ -167,6 +167,7 @@ struct fullBackUp{
     std::vector<int> field_types;
     std::vector<int> field_lengths;
     std::map<std::string,std::vector<std::string>> annoOnionNameToFileVector;//field name to vector of string
+    std::map<std::string,std::vector<Item*>> annoOnionNameToItemVector;
     std::map<std::string,int> annoOnionNameToType;
 };
 
@@ -184,3 +185,11 @@ getIntItem(int i);
 std::unique_ptr<Item>
 getStringItem(std::string s);
 
+
+//Item* do not work
+//void load_num_file(std::string filename,std::vector<Item> &res,enum_field_types intype);
+
+//void load_string_file(std::string filename,std::vector<Item> &res,unsigned long length,enum_field_types intype);
+
+
+