我的github：codetoys，所有代码都将会位于ctfc库中。已经放入库中我会指出在库中的位置。
这些代码大部分以Linux为目标但部分代码是纯C++的，可以在任何平台上使用。
专题：一个自制代码生成器（嵌入式脚本语言）之总述-CSDN博客
专题：一个自制代码生成器（嵌入式脚本语言）之对象模型-CSDN博客
专题：一个自制代码生成器（嵌入式脚本语言）之堆栈结构和总入口-CSDN博客
专题：一个自制代码生成器（嵌入式脚本语言）之核心逻辑-CSDN博客
专题：一个自制代码生成器（嵌入式脚本语言）之辅助逻辑-CSDN博客
专题：一个自制代码生成器（嵌入式脚本语言）之应用实例-CSDN博客
专题：一个自制代码生成器（嵌入式脚本语言）之模型开发-CSDN博客（本篇）
专题：一个自制代码生成器（嵌入式脚本语言）之代码模板详解-CSDN博客
这是对数据库操作的模型化实现，原则上与代码生成器无关，对同样的模型也可以直接写函数操作数据库，就像普通的动态方法一样。不过使用了代码生成器生成静态代码性能会比较高，也比较方便检查错误。
目录
一、回顾一下对象模型CCTObject
代码生成器接受的模型用CCTObject来定义，和一般脚本语言定义对象差不多，通过级联实现任意复杂的对象树，主要结构如下：
struct CCTObject
{
enum TYPE { OBJECT, POINTER, PROPERTY, ARRAY };
TYPE m_type;//根据type决定哪个成员有效
CCTObject* m_Pointer;//指针
string m_Property;//属性（值）
map<string, CCTObject > m_Object;//对象
vector<CCTObject > m_Array;//数组
......
}整个模型最终生成到一个CCTObject对象里面。
二、各种要素的定义
2.1 列 字段 member
对应数据库的列，最关键的属性是名字和类型。
//成员变量定义
struct member
{
//group为空则为列定义，使用下面一组成员
string member_name;
string member_comment;//单行
string member_type;//long time double string
string member_default;//默认值
string member_show_type;//显示方式
//group不为空则为列组访问，使用下面一组成员
string group;//组名，
string var_name;//变量名
//生成数据库类型
string makeDBType()const
{
stringstream ss;
if ("string" == member_type)ss << "COLUMN_TYPE_STRING_POOL";
else if ("long" == member_type)ss << "COLUMN_TYPE_LONG";
else if ("time" == member_type)ss << "COLUMN_TYPE_LONG";
else if ("double" == member_type)ss << "COLUMN_TYPE_DOUBLE";
else ss << "错误的类型 " << member_type;
return ss.str();
}
//生成参数类型
string makeParamType()const
{
stringstream ss;
if ("string" == member_type)ss << "char const *";
else if ("long" == member_type)ss << "long const";
else if ("time" == member_type)ss << "long const";
else if ("double" == member_type)ss << "double const";
else ss << "错误的类型 " << member_type;
return ss.str();
}
//生成参数值
string makeParamValue()const
{
stringstream ss;
if ("string" == member_type)ss << member_name << ".c_str()";
else if ("long" == member_type)ss << member_name;
else if ("time" == member_type)ss << member_name;
else if ("double" == member_type)ss << member_name;
else ss << "错误的类型 " << member_type;
return ss.str();
}
//生成printf类型
string makePrintfType()const
{
stringstream ss;
if ("string" == member_type)ss << "%s";
else if ("long" == member_type)ss << "%ld";
else if ("time" == member_type)ss << "%ld";
//else if ("double" == member_type)ss << "%f";
else ss << "错误的类型 " << member_type;
return ss.str();
}
//生成HTML输出类型
string makeHtmlType()const
{
stringstream ss;
if ("string" == member_type)ss << "CHtmlDoc::CHtmlDoc_DATACLASS_LEFT";
else if ("time" == member_type)ss << "CHtmlDoc::CHtmlDoc_DATACLASS_TIME";
else ss << "CHtmlDoc::CHtmlDoc_DATACLASS_RIGHT";
return ss.str();
}
//输出到CCTObject
CCTObject toCCTObject()
{
CCTObject O;
if (0 == group.size())
{
O.SetObjectAddProperty("name", member_name);
O.SetObjectAddProperty("GetName", member_name);
O.SetObjectAddProperty("GroupIndexVar", "");
O.SetObjectAddProperty("comment", member_comment);
O.SetObjectAddProperty("type", ("time" == member_type ? "long" : member_type));
O.SetObjectAddProperty("default", member_default);
O.SetObjectAddProperty("show_type", member_show_type);
O.SetObjectAddProperty("DBType", makeDBType());
O.SetObjectAddProperty("ParamType", makeParamType());
O.SetObjectAddProperty("ParamName", member_name);
O.SetObjectAddProperty("PrintfType", makePrintfType());
O.SetObjectAddProperty("ParamValue", makeParamValue());
O.SetObjectAddProperty("HtmlType", makeHtmlType());
}
else
{
O.SetObjectAddProperty("name", var_name);
O.SetObjectAddProperty("GetName", "\" + GetGroup" + group + "ColName(" + var_name + ") + \"");
O.SetObjectAddProperty("GroupIndexVar", " " + var_name + ",");
O.SetObjectAddProperty("ParamType", "long " + var_name + ", long");
O.SetObjectAddProperty("ParamName", var_name);
O.SetObjectAddProperty("PrintfType", "%ld");
}
return O;
}
};
成员变量的功能是比较显而易见的，另外包含了一组把成员变量的值变成对象模型的属性的方法，最后由toCCTObject()方法变成数据模型对象。
2.2 索引 index
索引比较简单，因为只是在创建表的时候用一下：
//索引定义
struct index
{
string index_name;
string index_comment;
bool unique{ false };
string index_members;
//输出到CCTObject
CCTObject toCCTObject()
{
CCTObject O;
O.SetObjectAddProperty("name", index_name);
O.SetObjectAddProperty("comment", index_comment);
O.SetObjectAddProperty("unique", (unique ? "true" : "false"));
O.SetObjectAddProperty("members", index_members);
return O;
}
};
2.3 列组 group
这个只是为了简化经常使用的一组列，并不对应数据库概念：
//列组定义（可以用下标标识的多个列）
struct group
{
string group_name;
string group_comment;
vector<member > group_members;
//输出到CCTObject
CCTObject toCCTObject()
{
CCTObject O;
O.SetObjectAddProperty("name", group_name);
O.SetObjectAddProperty("comment", group_comment);
for (vector<member >::iterator it = group_members.begin(); it != group_members.end(); ++it)
{
O.SetObjectArrayPushBack("members", it->toCCTObject());
}
return O;
}
};
2.4 增删改查 DML
所有SQL都用函数包装起来，这样客户代码不会直接使用SQL，等于所有对数据库的操作都集中在一起，以后想找出来哪些表被用到、哪些没有用到就很容易了，只要包装函数没有被调用自然就没有被用到。
//DML定义
struct dml
{
string dml_name;
string dml_type;//select insert update delete
string dml_comment;
vector<member > op_members;//操作的列
bool hasWhere;//是否有where，当where_members和other_where均为空则不添加此属性
vector<member > where_members;//条件列，仅支持 and =
string other_where;//直接写死的条件部分
//输出到CCTObject
CCTObject toCCTObject()
{
hasWhere = (where_members.size() != 0 || other_where.size() != 0);
CCTObject O;
O.SetObjectAddProperty("name", dml_name);
O.SetObjectAddProperty("type", dml_type);
O.SetObjectAddProperty("comment", dml_comment);
for (vector<member >::iterator it = op_members.begin(); it != op_members.end(); ++it)
{
O.SetObjectArrayPushBack("op_members", it->toCCTObject());
}
if (hasWhere)O.SetObjectAddProperty("hasWhere", hasWhere);
for (vector<member >::iterator it = where_members.begin(); it != where_members.end(); ++it)
{
O.SetObjectArrayPushBack("where_members", it->toCCTObject());
}
if(other_where.size() != 0)O.SetObjectAddProperty("other_where", other_where);
return O;
}
};
这个模型不允许直接传入SQL的片段，因为这样会破坏封装，实际数据库操作无法预期。
2.5 表 table
表对象包含以上所有部分，对应一张实体表。
class table
{
public:
string table_name;
string table_comment;
vector<member > table_members;
string PK_cols;//主键的列
vector<index > table_indexs;//主键名称为PK
vector<group > table_groups;//可以用下标访问的列组
vector<dml > table_dmls;//数据操作，对应一个sql
member * _getMember(string const & name)
{
for (size_t i = 0; i < table_members.size(); ++i)
{
if (name == table_members[i].member_name)return &table_members[i];
}
return NULL;
}
bool SetTable(char const * name, char const * comment)
{
table_name = name;
table_comment = comment;
return true;
}
bool AddMember(char const * name, char const * type, char const * comment, char const * _default = "", char const * _show_type="")
{
member m;
m.member_name = name;
m.member_type = type;
m.member_comment = comment;
m.member_default = _default;
m.member_show_type = _show_type;
if ("string" == m.member_type)m.member_default = "\"" + m.member_default + "\"";
else
{
if (0 == m.member_default.size())m.member_default = "0";
}
table_members.push_back(m);
return true;
}
//设置主键
bool SetPK(char const * PK)
{
PK_cols = PK;
StringTokenizer st(PK_cols, ",");
for (size_t i = 0; i < st.size(); ++i)
{
member * p = _getMember(st[i].c_str());
if (NULL == p)
{
thelog << "主键指定的列不存在 " << st[i] << ende;
exit(0);
}
}
return true;
}
bool AddIndex(char const * name, bool unique,char const * members, char const * comment)
{
StringTokenizer st(members, ",");
index tmp;
tmp.index_name = name;
tmp.unique = unique;
tmp.index_comment = comment;
tmp.index_members = members;
//检查
for (size_t i = 0; i < st.size(); ++i)
{
member* p = _getMember(st[i].c_str());
if (NULL == p)
{
thelog << "索引指定的列不存在 " << st[i] << ende;
exit(0);
}
}
table_indexs.push_back(tmp);
return true;
}
bool AddGroup(char const * name, char const * members, char const * comment)
{
StringTokenizer st(members, ",");
group tmp;
tmp.group_name = name;
tmp.group_comment = comment;
for (size_t i = 0; i < st.size(); ++i)
{
member * p = _getMember(st[i].c_str());
if (NULL == p)
{
thelog << "列组指定的列不存在 " << st[i] << ende;
exit(0);
}
tmp.group_members.push_back(*p);
}
table_groups.push_back(tmp);
return true;
}
bool AddDML(char const * name, char const * type, char const * op_members, char const * where_members, char const * comment, char const * other_where = "")
{
dml tmp;
tmp.dml_name = name;
tmp.dml_type = type;
tmp.dml_comment = comment;
tmp.other_where = other_where;
if (0 != strlen(op_members))
{
StringTokenizer st(op_members, ",");
for (size_t i = 0; i < st.size(); ++i)
{
if ("*" == st[i])
{//全部
for (vector<member >::const_iterator it = table_members.begin(); it != table_members.end(); ++it)
{
tmp.op_members.push_back(*it);
}
}
else
{
StringTokenizer st2(st[i], ":");
if (2 == st2.size())
{//列组
bool isGroupFound = false;
for (vector<group >::const_iterator it = table_groups.begin(); it != table_groups.end(); ++it)
{
if (it->group_name == st2[0])
{
isGroupFound = true;
}
}
if (!isGroupFound)
{
thelog << "op_members指定的列组不存在 " << st2[0] << ende;
exit(0);
}
member tmpmember;
tmpmember.group = st2[0];
tmpmember.var_name = st2[1];
tmp.op_members.push_back(tmpmember);
}
else
{//简单列
member * p = _getMember(st[i].c_str());
if (NULL == p)
{
thelog << "op_members指定的列不存在 " << st[i] << ende;
exit(0);
}
tmp.op_members.push_back(*p);
}
}
}
}
if (0 != strlen(where_members))
{
StringTokenizer st(where_members, ",");
for (size_t i = 0; i < st.size(); ++i)
{
member * p = _getMember(st[i].c_str());
if (NULL == p)
{
thelog << "where_members指定的列不存在 " << st[i] << ende;
exit(0);
}
tmp.where_members.push_back(*p);
}
}
table_dmls.push_back(tmp);
return true;
}
//输出到CCTObject
CCTObject toCCTObject()
{
CCTObject O;
O.SetObjectAddProperty("name", table_name);
O.SetObjectAddProperty("comment", table_comment);
O.SetObjectAddProperty("PK_cols", PK_cols);
for (vector<member >::iterator it = table_members.begin(); it != table_members.end(); ++it)
{
O.SetObjectArrayPushBack("members", it->toCCTObject());
}
for (vector<index >::iterator it = table_indexs.begin(); it != table_indexs.end(); ++it)
{
O.SetObjectArrayPushBack("indexs", it->toCCTObject());
}
for (vector<group >::iterator it = table_groups.begin(); it != table_groups.end(); ++it)
{
O.SetObjectArrayPushBack("groups", it->toCCTObject());
}
for (vector<dml >::iterator it = table_dmls.begin(); it != table_dmls.end(); ++it)
{
O.SetObjectArrayPushBack("dmls", it->toCCTObject());
}
return O;
}
};
2.6 序列 sequence
序列是个比较头疼的东西，因为不同数据库的支持很不一样，Oracle的序列是独立的数据库对象，Sql Server则是列可以被指定为自增列。
class sequence
{
public:
string sequence_name;
string sequence_init;
string sequence_comment;
//输出到CCTObject
CCTObject toCCTObject()
{
CCTObject O;
O.SetObjectAddProperty("name", sequence_name);
O.SetObjectAddProperty("comment", sequence_comment);
O.SetObjectAddProperty("init", sequence_init);
return O;
}
};
三、骨干代码
class CCTModel_UniversalDB
{
private:
vector<table > m_tables;
vector<sequence > m_sequences;
set<string > _table_names;//用来检查是否已经存在
set<string > _sequence_names;//用来检查是否已经存在
public:
CCTObject toCCTObject()
{
CCTObject O;
O.SetObjectAddProperty("name", "CodeTemplate");
for (vector<table >::iterator it = m_tables.begin(); it != m_tables.end(); ++it)
{
O.SetObjectArrayPushBack("tables", it->toCCTObject());
}
for (vector<sequence >::iterator it = m_sequences.begin(); it != m_sequences.end(); ++it)
{
O.SetObjectArrayPushBack("sequences", it->toCCTObject());
}
return O;
}
sequence * AddNewSequence(char const * sequencename, char const * comment, long init)
{
if (_sequence_names.end() != _sequence_names.find(sequencename))
{
thelog << sequencename << " 已经存在" << ende;
return NULL;
}
else
{
_sequence_names.insert(sequencename);
}
m_sequences.resize(m_sequences.size() + 1);
sequence * p = &m_sequences[m_sequences.size() - 1];
p->sequence_name = sequencename;
p->sequence_comment = comment;
char buf[256];
sprintf(buf, "%ld", init);
p->sequence_init = buf;
return p;
}
table * AddNewTable(char const * tablename, char const * comment)
{
if (_table_names.end() != _table_names.find(tablename))
{
thelog << tablename << " 已经存在" << ende;
return NULL;
}
else
{
_table_names.insert(tablename);
}
m_tables.resize(m_tables.size() + 1);
table * p = &m_tables[m_tables.size() - 1];
p->SetTable(tablename, comment);
return p;
}
public:
//_ns 名字空间，sys 总类的名称的一部分
bool CreateCode(char const * _ns, char const * sys, char const * output_dir)
{
CCTObject O;
O = toCCTObject();
O.SetObjectAddProperty("NAMESPACE", _ns);
O.SetObjectAddProperty("SYS", sys);
CCTStack S;
CCodeTemplate ct;
string str;
//thelog << endl << O.toString(str) << endi;
S.clear();
if (!ct.ProcessFile("_t_UniversalDB.h.ct", (string(output_dir) + "/_cc_" + sys + ".h").c_str(), O, S))return false;
vector<CCTObject > * p;
if (NULL != O.FindObject("tables"))
{
p = &O.FindObject("tables")->m_Array;
for (vector<CCTObject >::iterator it = p->begin(); it != p->end(); ++it)
{
S.clear();
S.Push();
S.Add("table", *it);
//thelog << endl << S.toString(str) << endi;
if (!ct.ProcessFile("_t_UniversalDB_table.h.ct", (string(output_dir) + "/_cc_" + sys + "_" + it->GetDefaultValue() + ".h").c_str(), O, S))return false;
}
}
if (NULL != O.FindObject("sequences"))
{
p = &O.FindObject("sequences")->m_Array;
for (vector<CCTObject >::iterator it = p->begin(); it != p->end(); ++it)
{
S.clear();
S.Push();
S.Add("sequence", *it);
thelog << endl << S.toString(str) << endi;
if (!ct.ProcessFile("_t_UniversalDB_sequence.h.ct", (string(output_dir) + "/_cc_" + sys + "_" + it->GetDefaultValue() + ".h").c_str(), O, S))return false;
}
}
return true;
}
};
代码很容易懂。
四、使用模型
专题：一个自制代码生成器（嵌入式脚本语言）之应用实例-CSDN博客
五、代码模板详解
下一篇：专题：一个自制代码生成器（嵌入式脚本语言）之代码模板详解-CSDN博客
（这里是结束但不是整个系列的结束）
3184
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
