文本处理学习笔记1

环境情况如下：
python-2.5.2
ibatis-2.3.4.726
jdk1.6.0_07
Oracle9i Enterprise Edition 9.2.0.1.0

一个简单的ibatis配置文件生成工具：
测试文件如下(test.sql)：

-----------------------------------------------------
-- Export file for user PS                         --
-- Created by administrator on 2010/1/26, 10:56:19 --
-----------------------------------------------------

spool test.log

prompt
prompt Creating table TAB_BSC
prompt =====================
prompt
create table PS.TAB_BSC
(
    TB_ID NUMBER(10, 0),
    TB_NAME VARCHAR2(50),
    TB_TIME DATE,
    TB_VALUE NUMBER(10, 5)
)
tablespace PSTBS
  pctfree 10
  initrans 1
  maxtrans 255
  storage
  (
    initial 64K
    minextents 1
    maxextents unlimited
  );

prompt
prompt Creating table TAB_CELL
prompt ======================
prompt
create table PS.TAB_CELL
(
    TC_ID NUMBER(10, 0),
    TC_NAME VARCHAR2(50),
    TC_TIME DATE,
    TC_VALUE NUMBER(10, 5)
)
tablespace PSTBS
  pctfree 10
  initrans 1
  maxtrans 255
  storage
  (
    initial 64K
    minextents 1
    maxextents unlimited
  );

spool off

生成工具的Python代码如下(ibatisGen.py)：

# coding:gbk

import os
import re

def ibatisxml(rawsqls):
    tabs = re.findall("(create table PS.(\w+).+?;)", rawsqls, re.DOTALL)

    for i in tabs:
        tabU = i[1]
        cols = re.findall("  ([_A-Z0-9]+)\s+([_A-Z0-9]+)", i[0])

        txml = ["""<?xml version="1.0" encoding="UTF-8" ?>
<!DOCTYPE sqlMap PUBLIC "-//ibatis.apache.org//DTD SQL Map 2.0//EN" "http://ibatis.apache.org/dtd/sql-map-2.dtd">"""]
        columnStr = ",\n".join(["""  %s"""%j[0] for j in cols])
        valueStr = ",\n".join(["""  #%s#"""%j[0] for j in cols])

        txml.append("""<sqlMap namespace="%(tabU)s">
  <typeAlias alias="%(tabU)s" type="com.test.beans.%(tabU)s"/>
  <select id="selectSource%(tabU)s" parameterClass="%(tabU)s" resultClass="%(tabU)s">
    <![CDATA[
      select
%(columnStr)s
      from %(tabU)s
    ]]>
  </select>
  <insert id="insertTarget%(tabU)s" parameterClass="%(tabU)s">
    <![CDATA[
      insert into %(tabU)s (
%(columnStr)s
      ) values (
%(valueStr)s
      )
    ]]>
  </insert>
  <delete id="deleteTarget%(tabU)s" parameterClass="%(tabU)s">
    <![CDATA[
      delete from %(tabU)s where 1=1
    ]]>
  </delete>
</sqlMap>
"""%locals())

        if not os.path.exists("ibatisxml"):
            os.makedirs("ibatisxml")
        ofile = file("ibatisxml/%s.xml"%i[1], "w")
        ofile.write("\n".join(txml))
        ofile.close()

def javaclass(rawsqls):
    tabs = re.findall("(create table PS.(\w+).+?;)", rawsqls, re.DOTALL)

    for vv in tabs:
        tabU = vv[1]

        tclass = []
        cols = re.findall("  ([_A-Z0-9]+)\s+([_A-Z0-9]+)", vv[0])
        tclass.append("""package com.test.beans;

import java.io.Serializable;
import java.math.BigDecimal;
import java.sql.Timestamp;

public class %(tabU)s implements Serializable {

    private static final long serialVersionUID = 1L;
"""%locals())

        for iv, cv in enumerate(cols):
            colU = cv[0]
            colL = cv[0].capitalize()
            colT = "String"
            if cv[1]=="TIMESTAMP" or cv[1]=="DATE":
                colT = "Timestamp"
            elif cv[1]=="INTEGER" or cv[1]=="FLOAT" or cv[1]=="NUMBER":
                colT = "BigDecimal"
            tclass.append("""    public %(colT)s get%(colU)s() {
        return %(colU)s;
    }

    public void set%(colU)s(%(colT)s %(colL)s) {
        %(colU)s = %(colL)s;
    }
"""%locals())

        for iv, cv in enumerate(cols):
            colU = cv[0]
            colL = cv[0].capitalize()
            colT = "String"
            if cv[1]=="TIMESTAMP" or cv[1]=="DATE":
                colT = "Timestamp"
            elif cv[1]=="INTEGER" or cv[1]=="FLOAT" or cv[1]=="NUMBER":
                colT = "BigDecimal"
            tclass.append("""    public %(colT)s %(colU)s;"""%locals())

        tclass.append("}")

        if not os.path.exists("javaclass"):
            os.makedirs("javaclass")
        ofile = file("javaclass/%s.java"%tabU, "w")
        ofile.write("\n".join(tclass))
        ofile.close()

ifile = file("test.sql", "r")
rsql = ifile.read()
ifile.close()
ibatisxml(rsql)
javaclass(rsql)

来看看测试结果吧，
ibatisxml文件夹中生成两个文件：

(TAB_BSC.xml)：

<?xml version="1.0" encoding="UTF-8" ?>
<!DOCTYPE sqlMap PUBLIC "-//ibatis.apache.org//DTD SQL Map 2.0//EN" "http://ibatis.apache.org/dtd/sql-map-2.dtd">
<sqlMap namespace="TAB_BSC">
  <typeAlias alias="TAB_BSC" type="com.test.beans.TAB_BSC"/>
  <select id="selectSourceTAB_BSC" parameterClass="TAB_BSC" resultClass="TAB_BSC">
    <![CDATA[
      select
  TB_ID,
  TB_NAME,
  TB_TIME,
  TB_VALUE
      from TAB_BSC
    ]]>
  </select>
  <insert id="insertTargetTAB_BSC" parameterClass="TAB_BSC">
    <![CDATA[
      insert into TAB_BSC (
  TB_ID,
  TB_NAME,
  TB_TIME,
  TB_VALUE
      ) values (
  #TB_ID#,
  #TB_NAME#,
  #TB_TIME#,
  #TB_VALUE#
      )
    ]]>
  </insert>
  <delete id="deleteTargetTAB_BSC" parameterClass="TAB_BSC">
    <![CDATA[
      delete from TAB_BSC where 1=1
    ]]>
  </delete>
</sqlMap>

(TAB_CELL.xml)：

<?xml version="1.0" encoding="UTF-8" ?>
<!DOCTYPE sqlMap PUBLIC "-//ibatis.apache.org//DTD SQL Map 2.0//EN" "http://ibatis.apache.org/dtd/sql-map-2.dtd">
<sqlMap namespace="TAB_CELL">
  <typeAlias alias="TAB_CELL" type="com.test.beans.TAB_CELL"/>
  <select id="selectSourceTAB_CELL" parameterClass="TAB_CELL" resultClass="TAB_CELL">
    <![CDATA[
      select
  TC_ID,
  TC_NAME,
  TC_TIME,
  TC_VALUE
      from TAB_CELL
    ]]>
  </select>
  <insert id="insertTargetTAB_CELL" parameterClass="TAB_CELL">
    <![CDATA[
      insert into TAB_CELL (
  TC_ID,
  TC_NAME,
  TC_TIME,
  TC_VALUE
      ) values (
  #TC_ID#,
  #TC_NAME#,
  #TC_TIME#,
  #TC_VALUE#
      )
    ]]>
  </insert>
  <delete id="deleteTargetTAB_CELL" parameterClass="TAB_CELL">
    <![CDATA[
      delete from TAB_CELL where 1=1
    ]]>
  </delete>
</sqlMap>

javaclass文件夹中也生成了两个文件：

(TAB_BSC.java)：

package com.test.beans;

import java.io.Serializable;
import java.math.BigDecimal;
import java.sql.Timestamp;

public class TAB_BSC implements Serializable {

    private static final long serialVersionUID = 1L;

    public BigDecimal getTB_ID() {
        return TB_ID;
    }

    public void setTB_ID(BigDecimal Tb_id) {
        TB_ID = Tb_id;
    }

    public String getTB_NAME() {
        return TB_NAME;
    }

    public void setTB_NAME(String Tb_name) {
        TB_NAME = Tb_name;
    }

    public Timestamp getTB_TIME() {
        return TB_TIME;
    }

    public void setTB_TIME(Timestamp Tb_time) {
        TB_TIME = Tb_time;
    }

    public BigDecimal getTB_VALUE() {
        return TB_VALUE;
    }

    public void setTB_VALUE(BigDecimal Tb_value) {
        TB_VALUE = Tb_value;
    }

    public BigDecimal TB_ID;
    public String TB_NAME;
    public Timestamp TB_TIME;
    public BigDecimal TB_VALUE;
}

(TAB_CELL.java)：

package com.test.beans;

import java.io.Serializable;
import java.math.BigDecimal;
import java.sql.Timestamp;

public class TAB_CELL implements Serializable {

    private static final long serialVersionUID = 1L;

    public BigDecimal getTC_ID() {
        return TC_ID;
    }

    public void setTC_ID(BigDecimal Tc_id) {
        TC_ID = Tc_id;
    }

    public String getTC_NAME() {
        return TC_NAME;
    }

    public void setTC_NAME(String Tc_name) {
        TC_NAME = Tc_name;
    }

    public Timestamp getTC_TIME() {
        return TC_TIME;
    }

    public void setTC_TIME(Timestamp Tc_time) {
        TC_TIME = Tc_time;
    }

    public BigDecimal getTC_VALUE() {
        return TC_VALUE;
    }

    public void setTC_VALUE(BigDecimal Tc_value) {
        TC_VALUE = Tc_value;
    }

    public BigDecimal TC_ID;
    public String TC_NAME;
    public Timestamp TC_TIME;
    public BigDecimal TC_VALUE;
}

JavaBean的生成稍微偷了下懒，
使用 Timestamp 接所有的时间，
使用 BigDecimal 接所有的数字，
幸好 java支持这样做，jdbc支持得也不错。
以后用ibatis时可以先导一份表结构，
根据表结构可以生成配置文件，生成JavaBean，
方便了。