Scala正则和抽取器：解析方法参数

　　在《正则表达式基础知识》中概括了正则表达式的基础知识， 本文讲解如何使用正则表达式解析方法参数，从而可以根据 DAO 自动生成 Service.

　　在做 Java 项目时，常常要根据 DAO 生成 Service , 而 Service 有时是简单的调用 DAO 方法。比如根据 public CreativeDO findByCreativeId(Long creativeId)  生成如下代码：

public CreativeDO findByCreativeId(Long creativeId) {
      return creativeDAO.findByCreativeId(creativeId);
}

实际上就是将分号替换成带左右大括号的部分， 而 return 语句里可变的部分是方法名称 (findByCreativeId)、方法列表 (creativeId) ， 因此，需要从方法签名中提取出方法名称和参数列表。正则表达式尤其适合做这件事。

　　编写正则表达式的技巧：

      1.  从简单的表达式写起， 反复试验，逐步逼近；

      2.  分而治之：将要匹配的字符串分解成有意义的小组，分别写出匹配小组的子正则表达式，然后组合成完整的正则表达式；　  

      3.  为了兼容处理输入含前导后缀空格的情况，通常在正则表达式前后添加 \s*Regex\s* 。

　

从最简单的开始　

　　编写正则表达式，从最简单的开始。先处理只有一个参数的方法签名，可以拆解为方法名称及方法参数列表两个部分：

val methodNameRegexStr = "\\s*(?:\\w+\\s+)?\\w+<?\\w+>?\\s+(\\w+)"
val singleParamRegexStr = "[^,]*\\w+<?\\w+>?\\s+(\\w+)\\s*"
val simpleMethodSignRexStr = methodNameRegexStr + "\\(" + singleParamRegexStr + "\\)\\s*;\\s*"

　　

　　这里小小地使用到了"分而治之"的技巧。其中：

　　1.  带访问修饰符或不带访问修饰符：  CreativeDO findByCreativeId(Long creativeId) 或  public CreativeDO findByCreativeId(Long creativeId) ;  这里使用了  (?:\\w+\\s+)? 来匹配带 public 或不带 public 的情况， ? 表示可有可无； (?:regex) 表示匹配 regex 的字符串但是并不捕获该分组，这是由于只要使用了小括号的正则都会被捕获，可以在后续引用匹配结果，但是这里 public 并不是我们感兴趣的目标，忽略掉；

2.  参数可能是集合类型：  \\w+<?\\w+>? ， 比如 List<String> ，　这里不能匹配嵌套的结构， 比如 List<List<String>> 实际中在 DAO 层也很少出现；

　　3.  方法名称使用了 (\\w+) 来捕获；

　　4.  方法参数中使用  \\w+<?\\w+>?  来匹配参数类型， 比如 Long 或 List<Long> ;  使用  [^,]* 来匹配参数类型前面的部分，比如  @Param(\"orderNo\") ;  [^charset] 使用了字符集排除组，排除匹配 charset 指定的任何字符；   [^,]*\\w+<?\\w+>?\\s+(\\w+)\\s* 可以匹配　Long creativeId　或 @Param(\"creativeId\")  Long creativeId  ;

5. 合并起来就是  simpleMethodSignRexStr 用来匹配单参数的方法签名。

　  使用：　在字符串上调用方法 r 即可获取对应的正则表达式用来匹配； 使用 regex(values) = text 可以从 text 中抽取匹配 regex 中的分组的值 values 。　　

val methodSign = "  int insert(@Param(\"kdtId\") BuyerAddressDO buyerAddressDO); "
val simpleMethodSignRex = simpleMethodSignRexStr.r
val simpleMethodSignRex(methodName, arg) = methodSign
println(methodName + " " + arg)

处理双参数

　  有了匹配单参数方法签名的正则表达式， 又有匹配单参数的正则表达式， 编写处理双参数的方法签名就简单多了：　　

    val twoParamMethodSignRegStr = methodNameRegexStr + "\\(" + singleParamRegexStr + "," + singleParamRegexStr + "\\);\\s*"

　  这里感受到了正则表达式复用的滋味了吧！ ^_^

使用：　

val twoParamMethodSign = "OrderExpressDO getById(@Param(\"id\") int id, @Param(\"kdtId\") int kdtId); "
val twoParamMethodSignRex = twoParamMethodSignRegStr.r
val twoParamMethodSignRex(methodName2, arg1, arg2) = twoParamMethodSign
println(List(methodName2, arg1 + ", " + arg2))

　　

处理任意多个参数

　 通常扩展下双参数情况就可以了。　

    //val generalParamMethodSignRegStr = methodNameRegexStr + "\\((" + singleParamRegexStr + "(?:," + singleParamRegexStr + ")*)\\);\\s*"

不过，事实却没有那么美好。即使这样能够匹配整个方法签名，却只能捕获第一个参数和最后一个参数，中间的参数会被忽略。怎么办了？ 查看 Scala 正则表达式的部分寻找答案，发现抽取器似乎能够解决整个问题。

　　抽取器

　　抽取器并不是什么特别的事物，只是 Scala 增加的一个语法糖，用于从对象值中抽取出感兴趣的值或值集合；就好比 Java 里实现了迭代器方法的对象都可以使用 foreach 遍历一样。实际上， val twoParamMethodSignRex(methodName2, arg1, arg2) = twoParamMethodSign 已经使用到了抽取器的语法，这里正则表达式实现了抽取器的功能。只要对象实现了 unapply 或 unapplySeq 就可以使用抽取器语法。 对于解析方法签名这个例子，可以先将参数列表使用正则表达式抽取出来，然后用逗号分隔，最后使用单参数正则表达式(又一次复用！) 抽取，实现如下：

val generalParamMethodSignRegStr = methodNameRegexStr + "\\((.*)\\);\\s*"

object Method {

        def unapplySeq(methodSign:String): Option[(String, Seq[String])] = {
            try {
                val generalParamMethodSignReg = generalParamMethodSignRegStr.r
                val generalParamMethodSignReg(methodName, args) = methodSign
                val params = args.split(',').toList
                val argNames = params.map(extractArgName(_))
                println("parsed: " + Some(methodName, argNames))
                Some(methodName, argNames)
            } catch {
                case _ => Some("", List())
            }

        }

        def extractArgName(singleParam: String): String = {
            val singleParamRegex = singleParamRegexStr.r
            val singleParamRegex(argName) = singleParam
            return argName
        }

    }

　　然后可以这样使用：

　　 generalParamMethodSign match {
            case Method(methodName, firstArg, restArgs @ _*) =>
                println("Case Match Way: " + methodName + "(" + firstArg + ", " + restArgs.mkString(", ") + ")")
            case _ => println("Not matched")
        }

        val Method(methodName5, firstArg, restArgs @ _*) = generalParamMethodSign
        println("Extractor Way: " + methodName5 + "(" + firstArg + ", " + restArgs.mkString(", ") + ")")

　　其中： firstArg 是抽取列表的第一个元素， restArgs 是抽取列表的剩余元素。如果写成  val Method(methodName, args) = generalParamMethodSign 是不行的，估计是因为列表是通过链表的形式实现的： 列表总是第一个元素链接到剩余元素的列表。

　　　　

　　完整代码：

package scalastudy.basic

import java.io.PrintWriter

import scalastudy.utils.DefaultFileUtil._
import scalastudy.utils.PathConstants

/**
 * Created by shuqin on 16/4/22.
 */
object AutoGenerateJavaCodes extends App {

    val methodNameRegexStr = "\\s*(?:\\w+\\s+)?\\w+<?\\w+>?\\s+(\\w+)"
    val singleParamRegexStr = "[^,]*\\w+<?\\w+>?\\s+(\\w+)\\s*"
    val simpleMethodSignRexStr = methodNameRegexStr + "\\(" + singleParamRegexStr + "\\)\\s*;\\s*"
    val twoParamMethodSignRegStr = methodNameRegexStr + "\\(" + singleParamRegexStr + "," + singleParamRegexStr + "\\);\\s*"
    //val generalParamMethodSignRegStr = methodNameRegexStr + "\\((" + singleParamRegexStr + "(?:," + singleParamRegexStr + ")*)\\);\\s*"
    val generalParamMethodSignRegStr = methodNameRegexStr + "\\((.*)\\);\\s*"

    val apiPackageName = "cc.lovesqcc"

    launch()

    object Method {

        def unapplySeq(methodSign:String): Option[(String, Seq[String])] = {
            try {
                val generalParamMethodSignReg = generalParamMethodSignRegStr.r
                val generalParamMethodSignReg(methodName, args) = methodSign
                val params = args.split(',').toList
                val argNames = params.map(extractArgName(_))
                println("parsed: " + Some(methodName, argNames))
                Some(methodName, argNames)
            } catch {
                case _ => Some("", List())
            }

        }

        def extractArgName(singleParam: String): String = {
            val singleParamRegex = singleParamRegexStr.r
            val singleParamRegex(argName) = singleParam
            return argName
        }

    }

    def generateJavaFile(filepath: String): List[Any] = {

        val basePath = "/tmp/"
        val biz = List("order", "payment")

        var writePath = basePath
        var bizType = ""
        biz.foreach { e =>
            if (filepath.contains(e)) {
                writePath = writePath + "/" + e + "/"
                bizType = e
            }
        }
        val daoFileName = extraFilename(filepath)
        val daoClassName = daoFileName.substring(0, daoFileName.indexOf('.'))
        val writeFilename = writePath + daoFileName.replaceAll("DAO", "ServiceImpl")
        val serviceClassName = daoClassName.replace("DAO", "ServiceImpl")
        val daoRefName = firstLower(daoClassName)
        val lines = readFileLines(filepath)
        var fileContents = ""
        var daoFlag = false
        lines.foreach { line =>
            if (daoFlag) {
                fileContents += "\n\t@Resource\n"
                fileContents += "\tprivate " + daoClassName + " " + daoRefName + ";\n\n"
                daoFlag = false
            }
            else if (line.contains("interface")) {
                fileContents += "@Service\npublic class " + serviceClassName + " implements " + daoClassName.replace("DAO", "Service") + " { \n"
                daoFlag = true
            }
            else if (line.contains(";")) {
                if (!line.contains("import") && !line.contains("package")) {
                    val parsed = parseMethod(line)
                    val replaceStr = " {\n\t\treturn " + daoRefName + "." + parsed(0) + "(" + parsed(1) + ");\n\t}\n"
                    var assertQualifier = ""
                    if (!line.contains("public")) {
                        assertQualifier = "public "
                    }
                    fileContents += "\t" + assertQualifier + " " + line.trim.replace(";", replaceStr)
                }
                else if (line.contains("package")) {
                    fileContents += line.replace("dao", "service") + "\n\n"
                    var bizTypeStr = "."
                    if (bizType.length > 0) {
                        bizTypeStr = "." + bizType + "."
                    }
                    fileContents += "import " + apiPackageName + bizTypeStr + "service." + daoClassName.replace("DAO", "Service") + ";\n"
                    fileContents += "import javax.annotation.Resource;\n"
                    fileContents += "import org.springframework.stereotype.Service;\n"
                }
                else {
                    fileContents += line + "\n"
                }
            }
            else {
                fileContents += line + "\n"
            }
        }

        mkdir(writePath)
        val writeFile = new PrintWriter(writeFilename)
        writeFile.println(fileContents)
        writeFile.close

        return List[Any]();
    }

    def daoRefName(daoClassName: String): String = {
        return firstLower(daoClassName)
    }

    def firstLower(str: String): String = {
        return str.charAt(0).toLower + str.substring(1)
    }

    def parseMethod(methodSign: String): List[String] = {

        val simpleMethodSignRex = simpleMethodSignRexStr.r
        try {
            val Method(methodName, firstArg, restArgs @ _*) = methodSign
            if (restArgs.size == 0) {
                return List(methodName, firstArg)
            }
            else {
                return List(methodName, firstArg + ", " + restArgs.mkString(", "))
            }

        } catch {
            case _ => return List("", "")
        }

    }

    def debug(): Unit = {

        // simple catch regex groups
        val methodSign = "  int insert(@Param(\"kdtId\") BuyerAddressDO buyerAddressDO); "
        val simpleMethodSignRex = simpleMethodSignRexStr.r
        val simpleMethodSignRex(methodName, arg) = methodSign
        println(methodName + " " + arg)

        val twoParamMethodSign = "OrderExpressDO getById(@Param(\"id\") int id, @Param(\"kdtId\") int kdtId); "
        val twoParamMethodSignRex = twoParamMethodSignRegStr.r
        val twoParamMethodSignRex(methodName2, arg1, arg2) = twoParamMethodSign
        println(List(methodName2, arg1 + ", " + arg2))

        val text = "Good query(@Param(\"goodId\") goodId, int kdtId)"
        val regex = "\\s*(?:\\w+\\s+)?\\w+<?\\w+>?\\s+(\\w+)\\((.*)\\)".r
        val regex(methodName3, wholearg1) = text
        println(methodName3 + " " + wholearg1);

        val generalParamMethodSign = " OrderDO queryOrder(@Param(\"orderNos\") List<String> orderNos, @Param(\"page\") Integer page, @Param(\"pageSize\") Integer pageSize, boolean flag); "
        val regex2 = generalParamMethodSignRegStr.r
        val regex2(methodName4, wholearg2) = generalParamMethodSign
        println(methodName4 + " : " + wholearg2);

        generalParamMethodSign match {
            case Method(methodName, firstArg, restArgs @ _*) =>
                println("Case Match Way: " + methodName + "(" + firstArg + ", " + restArgs.mkString(", ") + ")")
            case _ => println("Not matched")
        }

        val Method(methodName5, firstArg, restArgs @ _*) = generalParamMethodSign
        println("Extractor Way: " + methodName5 + "(" + firstArg + ", " + restArgs.mkString(", ") + ")")
    }

    def testParseMethod(): Unit = {
        val testMethods = Map(
            " List<OrderDO> queryOrder(int kdtId); " -> List("queryOrder", "kdtId"),
            " List<OrderDO> queryOrder( int kdtId ); " -> List("queryOrder", "kdtId"),
            " OrderDO queryOrder(@Param(\"kdtId\") int kdtId); " -> List("queryOrder", "kdtId"),
            " List<OrderDO> queryOrder(List<String> orderNos); " -> List("queryOrder", "orderNos"),
            " List<OrderDO> queryOrder(@Param(\"orderNos\") List<String> orderNos); " -> List("queryOrder", "orderNos"),
            " OrderDO queryOrder(String orderNo, Integer kdtId); " -> List("queryOrder", "orderNo, kdtId"),
            " OrderDO queryOrder(String orderNo, @Param(\"kdtId\") Integer kdtId); " -> List("queryOrder", "orderNo, kdtId"),
            " OrderDO queryOrder(@Param(\"orderNo\") String orderNo, Integer kdtId); " -> List("queryOrder", "orderNo, kdtId"),
            " OrderDO queryOrder(@Param(\"orderNo\") String orderNo, @Param(\"kdtId\") Integer kdtId); " -> List("queryOrder", "orderNo, kdtId"),
            " OrderDO queryOrder(List<String> orderNos, Integer kdtId); \n" -> List("queryOrder", "orderNos, kdtId"),
            " OrderDO queryOrder(@Param(\"orderNos\") List<String> orderNos, Integer kdtId); " -> List("queryOrder", "orderNos, kdtId"),
            " OrderDO queryOrder(List<String> orderNos, @Param(\"kdtId\") Integer kdtId); " -> List("queryOrder", "orderNos, kdtId"),
            " OrderDO queryOrder(@Param(\"orderNos\") List<String> orderNos, @Param(\"kdtId\") Integer kdtId); " -> List("queryOrder", "orderNos, kdtId"),
            " OrderDO queryOrder(@Param(\"orderNos\") List<String> orderNos, @Param(\"page\") Integer page, @Param(\"pageSize\") Integer pageSize); " -> List("queryOrder", "orderNos, page, pageSize"))
        testMethods.foreach { testMethod =>
            println("test method: " + testMethod._1)
            val parsed = parseMethod(testMethod._1)
            println(parsed)
            assert(parseMethod(testMethod._1) == testMethod._2)
        }
        println("test ParseMethod passed.")
    }

    def launch(): Unit = {

        debug
        testParseMethod

        val dirpath = "/tmp/";
        handleFiles(fetchAllFiles)((file: String) => file.endsWith("DAO.java"))(List(generateJavaFile(_)))((liststr: List[Any]) => "")(dirpath);
    }

}