最近在開發一個項目中，為了解決數據庫ＩＯ瓶頸，不得不把數據庫中的數據導出為文本文件。文本傳到客戶端后又要導入到數據庫。本人用Ｃ＋＋Builder嵌入PROC++寫了一個導入導出的DLL。假如對你有用深感榮幸！具體內容如下：
一、預備工作
計算機環境：Win 2000 PRO,Oracle 9i,C++ Builder 5.5
引入必要的ORACLE內部函數：要用的函數在$(ORACEL_HOME)/bin qlora9.dll鏈接庫中。為了能在C++ Builder中使用，先得生成LIB：implib sqlora9.lib sqlora9.dll
二、源文件分析
//-------------------------------------------------------------------------
//加入必要的頭文件
#include<vcl.h> #include<windows.h> #include<stdio.h> #include<stdlib.h> #include<string.h>
#include<time.h> #include<math.h> #include<fcntl.h> #include<io.h> #include<sys tat.h>
//說明DLL的輸出函數
extern "C" _declspec(dlleXPort) int _stdcall ConnectDB(const char *Username,
const char *PassWord, const char *Dbname);
extern "C" _declspec(dllexport) int _stdcall ImportTxtfile(TList *LengthArray,
String *FieldArray, const char *TableName,
const char *FileName);
extern "C" _declspec(dllexport) int _stdcall ExportTxtfile(const char *Sql,
const char *FileName);
#pragma hdrstop
//----------------------------------------------------------------------------
#define MAX_ITEMS 20 //定義最大字段數
#define MAX_VNAME_LEN 30 //定義選擇表項最大長度
#define MAX_INAME_LEN 30 //定義指示器變量名字的最大長度
EXEC SQL INCLUDE sqlca; //說明SQL通訊區
EXEC SQL INCLUDE oraca; //說明ORACLE通訊區
EXEC SQL INCLUDE sqlda; //說明SQL語句描述結構/*SQLDA結構體請查相關資料*/
EXEC ORACLE OPTION (ORACA = YES);
EXEC ORACLE OPTION (RELEASE_CURSOR = YES);
//說明ORACLE外部函數
extern "C" _declspec(dllimport) void _stdcall sqlclu(SQLDA*);
extern "C" _declspec(dllimport) void _stdcall sqlnul(short*, short*, int*);
extern "C" _declspec(dllimport) void _stdcall sqlprc(int*, int*, int*);
extern "C" _declspec(dllimport) strUCt SQLDA * _stdcall sqlald(int, unsigned int, unsigned int);
SQLDA *SelectUnit; //定義選擇項描述
SQLDA *BindUnit; //定義輸入項空間
//定義變量，以存放連接數據庫的參數
EXEC SQL BEGIN DECLARE SECTION;
char User[20];//用戶名
char Pwd[20];//密碼
char DB[20];//數據庫服務名
EXEC SQL END DECLARE SECTION;
bool bConnect = false;//是否連接標志
#pragma hdrstop
#pragma argsused
//C++ Builder DLL的主函數
BOOL WINAPI DllMain(HINSTANCE hinstDLL, DWORD fwdreason, LPVOID lpvReserved)
{
return 1;
}
/*---------------------------------------------------------------------------
連接數據庫
---------------------------------------------------------------------------*/
int _stdcall ConnectDB(const char *Username, const char *Password,
const char *Dbname)
{
strcpy(User, Username);
strcpy(Pwd, Password);
strcpy(DB, Dbname);
EXEC SQL CONNECT :User IDENTIFIED BY :Pwd USING :DB;
if (sqlca.sqlcode < 0)
return -1;
bConnect = true;
return 0;
}
/*---------------------------------------------------------------------------
導出文本函數
因為不確定SELECT語句的表及字段，所以我使用動態語句(ORACLE DYNAMIC SQL)的//第四種方式。
動態SQL方法四是在不確定SQL語句的選擇項與輸入項，且不知個數與數據類型的情況下使用的一種復雜程序設計技術。
---------------------------------------------------------------------------*/
int _stdcall ExportTxtfile(const char *Sql/*SQL選擇語句*/, const char FileName/*導出目標文本文件名*/)
{
int null_ok, precision, scale;
int handle;
if ((handle = open(FileName, O_CREATO_TEXTO_APPENDO_RDWR, S_IREADS_IWRITE)) == -1)
{
//文件打開出錯
return -1;
}
//定義變量，以存放SQL語句
EXEC SQL BEGIN DECLARE SECTION;
char sqlstr[256];
EXEC SQL END DECLARE SECTION;
//檢查是否連接數據庫
if (bConnect == false) return -2;
strcpy(sqlstr/*.arr*/, Sql);
// sqlstr.len = strlen(sql);
//給描述區分配空間
if ((SelectUnit = sqlald(MAX_ITEMS, MAX_VNAME_LEN, MAX_INAME_LEN)) == (SQLDA *)NULL)
{
//空間分配失敗
return -3;
}
if ((BindUnit = sqlald(MAX_ITEMS, MAX_VNAME_LEN, MAX_INAME_LEN)) == (SQLDA *)NULL)
{
//空間分配失敗
return -3;
}
//給查詢返回值存儲區分配空間
SelectUnit->N = MAX_ITEMS;
for (int i=0; i < MAX_ITEMS; i++)
{
BindUnit->I[i] = (short *)malloc(sizeof(short *));
BindUnit->V[i] = (char *)malloc(MAX_VNAME_LEN);
}
for (int i=0; i < MAX_ITEMS; i++)
{
SelectUnit->I[i] = (short *)malloc(sizeof(short *));
SelectUnit->V[i] = (char *)malloc(MAX_VNAME_LEN);
}
EXEC SQL WHENEVER SQLERROR GOTO sqlerr;//DO sql_error("導出出錯");
//設置ＳＱＬ語句
EXEC SQL PREPARE SQLSA FROM :sqlstr;
EXEC SQL DECLARE Cursorbase CURSOR FOR SQLSA;
//輸入描述處理
BindUnit->N = MAX_ITEMS;
EXEC SQL DESCRIBE BIND VARIABLES for SQLSA INTO BindUnit;
if (BindUnit->F < 0)
{
return -4;
//輸入項過多
}
BindUnit->N = BindUnit->F;
//打開光標
EXEC SQL OPEN Cursorbase USING DESCRipTOR BindUnit;
//選擇項處理
EXEC SQL DESCRIBE SELECT LIST for SQLSA INTO SelectUnit;
if (SelectUnit->F < 0)
{
return -4;
//選擇表項過多
}
SelectUnit->N = SelectUnit->F;
//因為所有格式，類型都是不確定的，所以要得到正確的返回值就要處理格式
for (int i=0; i < SelectUnit->F; i++)
{
sqlnul(&(SelectUnit->T[i]), &(SelectUnit->T[i]), &null_ok);
switch (SelectUnit->T[i])
{
case 1://CHAR
break;
case 2://NUMBER
sqlprc(&(SelectUnit->L[i]), &precision, &scale);
if (precision == 0)
precision = 40;
SelectUnit->L[i] = precision + 2;
break;
case 8://LONG
SelectUnit->L[i] = 240;
break;
case 11://ROWID
SelectUnit->L[i] = 18;
break;
case 12://DATE
SelectUnit->L[i] = 9;
break;
case 23://RAW
break;
case 24://LONGRAW
SelectUnit->L[i] = 240;
break;
}
SelectUnit->V[i] = (char *)realloc(SelectUnit->V[i], SelectUnit->L[i]+1);
SelectUnit->T[i] = 1;//把所有類型轉換為字符型
}
EXEC SQL WHENEVER NOT FOUND goto EndFor;
for (;;)
{
EXEC SQL FETCH Cursorbase USING DESCRIPTOR SelectUnit;
//輸出各字段
for (int i=0; i < SelectUnit->F; i++)
{
char buffer[256];
if (i != SelectUnit->F-1)
sprintf(buffer, "%s", SelectUnit->V[i]);
else sprintf(buffer, "%s/r/n", SelectUnit->V[i]);
int length = strlen(buffer);
if (write(handle, buffer, length) != length)
{
return -5;
//寫文件失敗 exit(1);
}
}
}
EndFor:
close(handle);
for (int i=0; i < MAX_ITEMS; i++)
{
if (SelectUnit->V[i] != (char *)NULL)
free(SelectUnit->V[i]);
free(SelectUnit->I[i]);
}
for (int j=0; j < MAX_ITEMS; j++)
{
if (BindUnit->V[j] != (char *)NULL)
free(BindUnit->V[j]);
free(BindUnit->I[j]);
}
sqlclu(SelectUnit);
sqlclu(BindUnit);
EXEC SQL CLOSE Cursorbase;
return 0;
sqlerr:
return -6;
}
/*----------------------------------------------------------------------------
導入文本
為了批量導入，在此我調用的sqlldr工具
首先生成SQL*Loader控制文件，后運行sqlldr
----------------------------------------------------------------------------*/
int _stdcall ImportTxtfile(TList LengthArray/*導入文本的字段長度鏈表*/,
String *FieldArray/*數據庫表的了段名數組*/, const char TableName/*導入的目標表*/, const char FileName/*導入的源文本文件*/)
{
//產生SQL*Loader控制文件
FILE *fout, *fp;
char Execommand[256];
char sqlload[] = "./ qlload.ctl";
//檢查是否連接數據庫
if (bConnect == false) return -2;
if ((fout=fopen(sqlload, "w")) == NULL)
{
//建立控制文件出錯
return -1 ;
}
fprintf(fout, "LOAD DATA/n");
fprintf(fout, "INFILE '%s'/n", FileName);
fprintf(fout, "APPEND INTO TABLE %s (/n", TableName);
int iStart = 1;
for(int i=0; i < LengthArray->Count; i++)
{
fprintf(fout, "%11s POSITION(%d:%d)", FieldArray[i], iStart, *(int*)LengthArray->Items[i]+iStart-1);
iStart += *(int*)LengthArray->Items[i];
fprintf(fout, " CHAR");
if(i < LengthArray->Count-1)
fprintf(fout, ",/n");
}
fprintf(fout, ")/n");
fclose(fout);
sprintf(Execommand, "sqlldr.exe userid=%s/%s@%s control=%s",
User, Pwd, DB, sqlload);
if (system(Execommand) == -1)
{
//SQL*Loader執行錯誤
return -1;
}
return 0 ;
}
//----------------------------------------------------------------------------
三、編譯
用ORACLE的PROC預編譯器預編后，放入C++ Builder中聯編。
聯編時需加入前面生成的sqlora9.lib。聯編時還要注重，所有PROC生成的ORACLE內部函數調用都要說明為extern "C" _declspec(dllexport) TYPE _stdcall類型。
水平有限還請見諒?。?！請多多指點。ＱＱ：5005647
