java - Db4o API 在 NetBeans 中看不到动态加载的类？

Question

我认为似乎在使用 db4o 和 NetBeans 时遇到了类加载器问题。当我使用 .../jre/lib/ext 中的相同 jar 文件从终端运行完全相同的代码时，一切正常。问题是，当我对使用类加载器在运行时加载的某些类进行 native 查询时，我从数据库中得到一个空列表，我肯定应该在其中获取包含一些元素的列表(如我说过，相同的代码在命令行中工作得很好)。我觉得这可能是因为 NetBeans ClassLoader 的工作方式与 JVM ClassLoader 不同，但我不知道，我当然不是这方面的专家。这是我的主要功能的代码......

////////////////////////////////////////////////////////////////////////////////////////////
package gql;

import java.io.*;
import java.util.*;
import com.db4o.*;

public class GQL {
    ////////////////////////////////////////////////////////////////////////////
    // GLOBAL VARIABLES
    ////////////////////////////////////////////////////////////////////////////
    private static ObjectSet dbObjects;
    private static LinkedList classes = new LinkedList();
    private static String dbPath, classPath;
    private static ObjectContainer db;
    private static ClassLoader coreClassLoader =
            ClassLoader.getSystemClassLoader();
    private static ClassLoader subClassLoader =
            ClassLoader.getSystemClassLoader();

    public static void main(String[] args) {
        ////////////////////////////////////////////////////////////////////////
        // CREATE DATABASE OBJECT
        ////////////////////////////////////////////////////////////////////////
        // If no path to a database is provided on the command line, print
        // error and exit program
        if (args.length < 1) {
           System.err.println("\nError: no database path provided.\n");
           return;
        } else if (args.length > 1) {
           dbPath = args[0];
           // TODO - dubug command line classpath
           classPath = args[1];
           db = Db4o.openFile(dbPath);
        } else {     // We assume that the database Classes are stored somewhere
           dbPath = args[0];    // along the CLASSPATH, and therefore classPath
           classPath = "";      // can be left empty
           db = Db4o.openFile(dbPath);
        }

        System.out.print("GQL> ");

        // The prompt of the interpreter is within a do-while loop, which can
        // be terminated by entering "exit"
        do {
           try {

               /////////////////////////////////////////////////////////////////
               // READ IN QUERY FILE
               /////////////////////////////////////////////////////////////////
               // We create a Scanner object to read tokens from the standard in
               // stream - these will be our DLOG files provided by the user
               Scanner fileScanner = new Scanner(System.in);
               String GQLFile = fileScanner.next();

               // Break loop and exit program if user enters "exit"
               if (GQLFile.equalsIgnoreCase("exit")) {
                   break;

               // If the user input is not preceeded by "@" and teminated with
               // ";" then the input is invalid - the user is prompted again
               } else if (!(GQLFile.substring(0,1).equals("@")) ||
                           !(GQLFile.substring(GQLFile.length()-1,
                               GQLFile.length()).equals(";"))) {

                   System.out.println("\nInvalid input.\nUsage:     "
                           + " @filename;\n");
                   System.out.print("GQL> ");
                   continue;

               } else {

                   // Parse out the "@" and ";" from the user's input and send
                   // this to a file Reader object
                   GQLFile = GQLFile.substring(1,GQLFile.length()-1);
               }

               // Now we create a reader object and give it the user's parsed
               // input - in the event of a FileNotFoundException, the user is
               // prompted again
               Reader reader;
               try {
                   reader = new FileReader(GQLFile);
               } catch (FileNotFoundException e) {
                   System.out.println("\nFile " + GQLFile +
                           " does not exist.\n");
                   System.out.print("GQL> ");
                   continue;
               }

               /////////////////////////////////////////////////////////////////
               // PARSE QUERY
               /////////////////////////////////////////////////////////////////
               // The parser and Lexer objects are created in the parser.java
               // and Lexer.java files, respectively - The parser takes the
               // Lexer as an argument - the value variable generated by the
               // parse() method will return the topmost grammar construct,
               // which in this case is a Query object
               parser p = new parser(new Lexer(reader));                      

               Query query = (Query) p.parse().value;
               /////////////////////////////////////////////////////////////////
               System.out.println("\n----------------------------Input Query-----" +
                       "-----------------------\n");

               System.out.println("\n                        SUCCESSFUL PARSE   " +
                       "                       \n");

               System.out.println("--------------------------------------" +
                       "------------------------------\n");
               /////////////////////////////////////////////////////////////////

               /////////////////////////////////////////////////////////////////
               // LOAD ALL CLASSES USED IN DATABASE INTO RUNTIME
               /////////////////////////////////////////////////////////////////
               // databse Classes should be kept on the CLASSPATH, or the path
               // to these classes should be provided as a second command line
               // argument
               boolean coreClassesLoaded = loadCoreClasses(coreClassLoader,
                       classPath);
               if (!coreClassesLoaded) {
                   System.err.println("\nError: one or more of core Classes"
                           + "Node, Egge and SimplePath could not be found.\n");
                   db.close();
                   return;
               }
               //
               System.out.println("Core classes loaded.\n");
               boolean subclassesLoaded = query.loadSubclasses(subClassLoader,
                       classPath);
               if (!subclassesLoaded) {
                   System.err.println("\nError: subclasses could not be" +
                           " loaded.\n");
                   db.close();
                   return;
               }
               //
               System.out.println("Subclasses loaded.\n");


               /////////////////////////////////////////////////////////////////
               // MAKE SURE THE DATABASE ACTUALLY CONTAINS SOME OBJECTS AND,
               // IF SO, PUT AN INSTANCE OF EACH CLASS REPRESENTED INTO THE
               // LINKEDLIST CLASSLIST - SINCE WE LOADED THE DATABASE CLASSES
               // INTO THE RUNTIME ENVIRONMENT, OBJECTS RETURNED BY DATABASE
               // QUERIES WILL REMAIN TRUE TO THEIR TYPE; IF WE HADN'T DONE
               // THIS, THESE OBJECTS WOULD BE RETURNED AS TYPE GENERICOBJECT
               /////////////////////////////////////////////////////////////////
               dbObjects = db.queryByExample(Object.class);
               if (dbObjects.hasNext()) {
                  query.addClassesToList(dbObjects, classes);
               } else {
                   System.err.println("\nError: no objects in database.\n");
                   db.close();
                   return;
               }
               //
               System.out.println(classes);

               /////////////////////////////////////////////////////////////////
               // SEMANTIC CHECKS                                             //
               /////////////////////////////////////////////////////////////////
               boolean headArgsAreSet = query.setHeadArgClasses(classes);    
               if (!headArgsAreSet) {                                        
                   db.close();
                   return;                                                   
               }                                                             
               boolean typesMatch = query.checkQueryTypes(db);         
               if (!typesMatch) {                                            
                   db.close();
                   return;
               }              

               /////////////////////////////////////////////////////////////////
               // EVALUATION
               /////////////////////////////////////////////////////////////////
               query.evaluateQuery(db);

           } catch (Exception e) {
               System.out.println("\nSYNTAX ERROR\n");
               e.printStackTrace();
           }

           System.out.print("GQL> ");
        } while (true);

        System.out.println("\nExiting...\n");
        db.close();
    }

    private static boolean loadCoreClasses(ClassLoader coreClassLoader,
            String classPath) {
        try {
            coreClassLoader.loadClass("Node");
            coreClassLoader.loadClass("Edge");
            coreClassLoader.loadClass("SimplePath");
        } catch (ClassNotFoundException cnfe) {
            return false;
        }
        return true;
    }
}



////////////////////////////////////////////////////////////////////////////////////////////

奇怪的是，我需要的类肯定会加载到运行时环境中，因为我使用它们来设置一些类成员变量，例如在“语义检查”部分中。因此，就像应用程序可以看到动态加载的类，但 db4o API/数据库却不能。另外，我将 Class jar 和 db4o jar 设置为 Netbeans 库，而不仅仅是在 .../jre/lib/ext 中。这是类中的代码片段，我实际上在其中使用了 db4o native 查询，这给我带来了问题......

////////////////////////////////////////////////////////////////////////////////////////////

   public void evaluateQuery(ObjectContainer db) {
        if (this.hasPredicate) {
            ;
        } else {
            if (this.isNode) {
                List nodes = db.query(new Predicate() {
                    @Override
                    public boolean match(Node node) {
                        return (node.getName().equals("5"));
                    }
                });
                System.out.println("\n_________________RESULT__________________________");
                System.out.println("\nNode: " + nodes.get(0).getName()
                        //+ ".\n");
            }
        }
    }

////////////////////////////////////////////////////////////////////////////////////////////

...如果我执行以下操作，我仍然得到一个空列表...

////////////////////////////////////////////////////////////////////////////////////////////


   public void evaluateQuery(ObjectContainer db) {
        if (this.hasPredicate) {
            ;
        } else {
            if (this.isNode) {
                List nodes = db.queryByExample(Node.class);
                System.out.println(nodes.size());
                for (int i = 0; i < nodes.size(); i++) {
                    System.out.println("\nNode: " + nodes.get(i).getName());
                }
            }
        }
    }

////////////////////////////////////////////////////////////////////////////////////////////

Answer 1

嗯，我在这里没有看到任何动态类加载问题。您正在使用系统的类加载器，它应该对 db4o 可见。您甚至不需要加载类，db4o 将使用类加载器来完成此操作。

您确定应用程序选择相同的数据库吗？您使用相对路径吗？

顺便说一句，您可以显式设置 classloader for db4o ，像这样。

EmbeddedConfiguration configuration = Db4oEmbedded.newConfiguration();

JdkLoader classLookUp = new ClassLoaderLookup(
        Thread.currentThread().getContextClassLoader(),
        new URLClassLoader(new URL[]{new URL("file://./some/other/location")}));
configuration.common().reflectWith(new JdkReflector(classLookUp));

ObjectContainer container = Db4oEmbedded.openFile(configuration,"database.db4o");