zoukankan      html  css  js  c++  java
  • JSP编译成Servlet(五)JDT Compiler编译器

    本文转载自  http://blog.csdn.net/wangyangzhizhou/article/details/50990798

    点击订购作者《Tomcat内核设计剖析》

    通过JSP编译器编译后生成了对应的java文件,接下去要把Java文件编译成class文件。对于这部分完全没有必要重新造轮子,常见的优秀编译工具有Eclipse JDT Java编译器和Ant编译器。Tomcat其实是同时支持两个编译器的,通过配置可以选择,而默认是使用Eclipse JDT编译器。

    通过调用这些现成的编译器的API就可以方便地实现对java文件的编译,由于两个编译器功能基本一样,我们就挑默认编辑器看下它是如何进行编译的,下面仅看如何用Eclipse JDT编译器编译java文件。

    Eclipse JDT提供了Compiler类用于编译,它的构造函数比较复杂,如下所示,其实就是实现自定义构造函数包含的参数即基本完成了编译工作。

    public Compiler(

      INameEnvironment environment,

      IErrorHandlingPolicy policy,

      CompilerOptions options,

      final ICompilerRequestor requestor,

      IProblemFactory problemFactory) {

    }

    为了说明方便直接上一个简单的编译实现,如下:

    public class JDTCompile {

        private static final File WORKDIR = new File("D:\Program Files\tomcat7\work\Catalina\localhost\test");

        public static void main(String[] args) {

            INameEnvironment nameEnvironment = new INameEnvironment() {

                public NameEnvironmentAnswer findType(final char[][] compoundTypeName) {

                    return findType(join(compoundTypeName));

                }

                public NameEnvironmentAnswer findType(final char[] typeName, final char[][] packageName) {

                    return findType(join(packageName) + "." + new String(typeName));

                }

                private NameEnvironmentAnswer findType(final String name) {

                    File file = new File(WORKDIR, name.replace('.', '/') + ".java");

                    if (file.isFile()) {

                        return new NameEnvironmentAnswer(new CompilationUnit(file), null);

                    }

                    try {

                        InputStream input =

                                this.getClass().getClassLoader().getResourceAsStream(name.replace(".", "/") + ".class");

                        if (input != null) {

                            byte[] bytes = IOUtils.toByteArray(input);

                            if (bytes != null) {

                                ClassFileReader classFileReader = new ClassFileReader(bytes, name.toCharArray(), true);

                                return new NameEnvironmentAnswer(classFileReader, null);

                            }

                        }

                    } catch (ClassFormatException e) {

                        throw new RuntimeException(e);

                    } catch (IOException e) {

                        throw new RuntimeException(e);

                    }

                    return null;

                }

                public boolean isPackage(char[][] parentPackageName, char[] packageName) {

                    String name = new String(packageName);

                    if (parentPackageName != null) {

                        name = join(parentPackageName) + "." + name;

                    }

                    File target = new File(WORKDIR, name.replace('.', '/'));

                    return !target.isFile();

                }

                public void cleanup() {}

            };

            ICompilerRequestor compilerRequestor = new ICompilerRequestor() {

                public void acceptResult(CompilationResult result) {

                    if (result.hasErrors()) {

                        for (IProblem problem : result.getErrors()) {

                            String className = new String(problem.getOriginatingFileName()).replace("/", ".");

                            className = className.substring(0, className.length() - 5);

                            String message = problem.getMessage();

                            if (problem.getID() == IProblem.CannotImportPackage) {

                                message = problem.getArguments()[0] + " cannot be resolved";

                            }

                            throw new RuntimeException(className + ":" + message);

                        }

                    }

                    ClassFile[] clazzFiles = result.getClassFiles();

                    for (int i = 0; i < clazzFiles.length; i++) {

                        String clazzName = join(clazzFiles[i].getCompoundName());

                        File target = new File(WORKDIR, clazzName.replace(".", "/") + ".class");

                        try {

                            FileUtils.writeByteArrayToFile(target, clazzFiles[i].getBytes());

                        } catch (IOException e) {

                            throw new RuntimeException(e);

                        }

                    }

                }

            };

            IProblemFactory problemFactory = new DefaultProblemFactory(Locale.ENGLISH);

            IErrorHandlingPolicy policy = DefaultErrorHandlingPolicies.exitOnFirstError();

            org.eclipse.jdt.internal.compiler.Compiler jdtCompiler =

                    new org.eclipse.jdt.internal.compiler.Compiler(nameEnvironment, policy, getCompilerOptions(),

                            compilerRequestor, problemFactory);

            jdtCompiler

                    .compile(new ICompilationUnit[] {new CompilationUnit(new File(WORKDIR, "org\apache\jsp\HelloWorld_jsp.java"))});

        }

        public static CompilerOptions getCompilerOptions() {

            Map settings = new HashMap();

            String javaVersion = CompilerOptions.VERSION_1_7;

            settings.put(CompilerOptions.OPTION_Source, javaVersion);

            settings.put(CompilerOptions.OPTION_Compliance, javaVersion);

            return new CompilerOptions(settings);

        }

        private static class CompilationUnit implements ICompilationUnit {

            private File file;

            public CompilationUnit(File file) {

                this.file = file;

            }

            public char[] getContents() {

                try {

                    return FileUtils.readFileToString(file).toCharArray();

                } catch (IOException e) {

                    throw new RuntimeException(e);

                }

            }

            public char[] getMainTypeName() {

                return file.getName().replace(".java", "").toCharArray();

            }

            public char[][] getPackageName() {

                String fullPkgName = this.file.getParentFile().getAbsolutePath().replace(WORKDIR.getAbsolutePath(), "");

                fullPkgName = fullPkgName.replace("/", ".").replace("\", ".");

                if (fullPkgName.startsWith("."))

                    fullPkgName = fullPkgName.substring(1);

                String[] items = fullPkgName.split("[.]");

                char[][] pkgName = new char[items.length][];

                for (int i = 0; i < items.length; i++) {

                    pkgName[i] = items[i].toCharArray();

                }

                return pkgName;

            }

            public boolean ignoreOptionalProblems() {

                return false;

            }

            public char[] getFileName() {

                return this.file.getName().toCharArray();

            }

        }

        private static String join(char[][] chars) {

            StringBuilder sb = new StringBuilder();

            for (char[] item : chars) {

                if (sb.length() > 0) {

                    sb.append(".");

                }

                sb.append(item);

            }

            return sb.toString();

        }

    }

    为了更好理解,我们根据构造函数的参数分别看看,

    INameEnvironment接口,主要需要实现的方法是findTypeisPackagefindType是帮助JDT找到相应的java源文件或者class字节码,根据传进来的包名和类名去寻找。例如传入“java.lang.String”或“org.apache.jsp.HelloWorld_jsp”则分别要找到JDK自带的String字节码及tomcat中编译的HelloWorld_jsp.java文件。接着按要求封装这些对象返回JDT规定的NameEnvironmentAnswer对象。而isPackage则提供是否是包的判断。

    IErrorHandlingPolicy接口,用于描述错误策略,可直接使用DefaultErrorHandlingPolicies.exitOnFirstError(),表示第一个错误就退出编译。

    CompilerOptions对象,指定编译时的一些参数,例如这里指定编译的Java版本为1.7

    ICompilerRequestor接口,它只有一个acceptResult方法,这个方法用于处理编译后的结果,如果包含了错误信息则抛异常,否则则把编译成功的字节码写到指定路径的HelloWorld_jsp.class文件中,即生成字节码。

    IProblemFactory接口,主要是控制编译错误信息的格式。

    所有Compiler构造函数需要的参数对象都已经具备,传入这些参数后创建一个Compiler对象,然后调用compile方法即可对指定的java文件进行编译。这里完成了HelloWorld_jsp.java的编译,结果生成了HelloWorld_jsp.class字节码。实际的tomcat中基本也是类似这样使用JDT实现servlet的编译,但它使用的某些策略可能不相同,例如使用DefaultErrorHandlingPolicies.proceedWithAllProblems()作为错误策略。

    通过这两章节“从JSPServlet”及“从Servletclass字节码”,我们已经清楚tomcatJSP编译处理的整个过程了,先根据JSP语法解析生成类似xxxx.javaServlet,然后再通过Eclipse JDTxxxx.java编译,最后生成了JVM能识别的class字节码。

  • 相关阅读:
    CF1439E
    CF1446
    CSP2020 游记
    CF1442
    CF1444E
    CF1444
    CF850F Rainbow Balls
    A
    uoj266[清华集训2016]Alice和Bob又在玩游戏(SG函数)
    loj536「LibreOJ Round #6」花札(二分图博弈)
  • 原文地址:https://www.cnblogs.com/abc8023/p/7770544.html
Copyright © 2011-2022 走看看