Java tutorial
/* * Copyright 2008 biaoping.yin * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ package com.frameworkset.orm.engine.model; /* * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ import java.util.List; import java.util.StringTokenizer; import org.apache.commons.lang.StringUtils; import com.frameworkset.orm.engine.EngineException; /** * A <code>NameGenerator</code> implementation for Java-esque names. * * @author <a href="mailto:dlr@finemaltcoding.com>Daniel Rall</a> * @author <a href="mailto:byron_foster@byron_foster@yahoo.com>Byron Foster</a> * @version $Id: JavaNameGenerator.java,v 1.4 2005/02/23 17:32:09 tfischer Exp $ */ public class JavaNameGenerator implements NameGenerator { /** * <code>inputs</code> should consist of two elements, the * original name of the database element and the method for * generating the name. There are currently three methods: * <code>CONV_METHOD_NOCHANGE</code> - xml names are converted * directly to java names without modification. * <code>CONV_METHOD_UNDERSCORE</code> will capitalize the first * letter, remove underscores, and capitalize each letter before * an underscore. All other letters are lowercased. "javaname" * works the same as the <code>CONV_METHOD_JAVANAME</code> method * but will not lowercase any characters. * * @param inputs list expected to contain two parameters, element * 0 contains name to convert, element 1 contains method for conversion. * @return The generated name. * @see com.frameworkset.orm.engine.model.NameGenerator */ public String generateName(List inputs) { String schemaName = (String) inputs.get(0); String method = (String) inputs.get(1); String javaName = null; if (CONV_METHOD_UNDERSCORE.equals(method)) { javaName = underscoreMethod(schemaName, false); } else if (CONV_METHOD_UNDERSCORE_OMIT_SCHEMA.equals(method)) { javaName = underscoreOmitSchemaMethod(schemaName, false); } else if (CONV_METHOD_JAVANAME.equals(method)) { javaName = javanameMethod(schemaName, false); } else if (CONV_METHOD_NOCHANGE.equals(method)) { javaName = nochangeMethod(schemaName, false); } else if (CONV_METHOD_JAVAFIELDNAME.equals(method)) { javaName = this.javanameField(schemaName, false); } else { // if for some reason nothing is defined then we default // to the traditional method. javaName = underscoreMethod(schemaName, false); } return javaName; } public String generateName(List inputs, boolean IGNORE_FIRST_TOKEN) throws EngineException { String schemaName = (String) inputs.get(0); String method = (String) inputs.get(1); String javaName = null; if (CONV_METHOD_UNDERSCORE.equals(method)) { javaName = underscoreMethod(schemaName, IGNORE_FIRST_TOKEN); } else if (CONV_METHOD_UNDERSCORE_OMIT_SCHEMA.equals(method)) { javaName = underscoreOmitSchemaMethod(schemaName, IGNORE_FIRST_TOKEN); } else if (CONV_METHOD_JAVANAME.equals(method)) { javaName = javanameMethod(schemaName, IGNORE_FIRST_TOKEN); } else if (CONV_METHOD_NOCHANGE.equals(method)) { javaName = nochangeMethod(schemaName, IGNORE_FIRST_TOKEN); } else if (CONV_METHOD_JAVAFIELDNAME.equals(method)) { javaName = this.javanameField(schemaName, IGNORE_FIRST_TOKEN); } else { // if for some reason nothing is defined then we default // to the traditional method. javaName = underscoreMethod(schemaName, IGNORE_FIRST_TOKEN); } return javaName; } /** * Converts a database schema name to java object name. Removes * <code>STD_SEPARATOR_CHAR</code> and <code>SCHEMA_SEPARATOR_CHAR</code>, * capitilizes first letter of name and each letter after the * <code>STD_SEPERATOR</code> and <code>SCHEMA_SEPARATOR_CHAR</code>, * converts the rest of the letters to lowercase. * * @param schemaName name to be converted. * @return converted name. * @see com.frameworkset.orm.engine.model.NameGenerator * @see #underscoreMethod(String) */ protected String underscoreMethod(String schemaName, boolean IGNORE_FIRST_TOKEN) { StringBuffer name = new StringBuffer(); // remove the STD_SEPARATOR_CHARs and capitalize // the tokens StringTokenizer tok = new StringTokenizer(schemaName, String.valueOf(STD_SEPARATOR_CHAR)); while (tok.hasMoreTokens()) { String namePart = ((String) tok.nextElement()).toLowerCase(); name.append(StringUtils.capitalize(namePart)); } // remove the SCHEMA_SEPARATOR_CHARs and capitalize // the tokens schemaName = name.toString(); name = new StringBuffer(); tok = new StringTokenizer(schemaName, String.valueOf(SCHEMA_SEPARATOR_CHAR)); while (tok.hasMoreTokens()) { String namePart = (String) tok.nextElement(); name.append(StringUtils.capitalize(namePart)); } return name.toString(); } /** * Converts a database schema name to java object name. * First, it removes all characters before the last occurence of * .<code>SCHEMA_SEPARATOR_CHAR</code>. Then, in a second step, removes * <code>STD_SEPARATOR_CHAR</code>, capitilizes first letter of * name and each letter after the <code>STD_SEPERATOR</code>, * and converts the rest of the letters to lowercase. * * @param schemaName name to be converted. * @return converted name. * @see com.frameworkset.orm.engine.model.NameGenerator * @see #underscoreOmitSchemaMethod(String) */ protected String underscoreOmitSchemaMethod(String schemaName, boolean IGNORE_FIRST_TOKEN) { // take only part after last dot int lastDotPos = schemaName.lastIndexOf(SCHEMA_SEPARATOR_CHAR); if (lastDotPos != -1) { schemaName = schemaName.substring(lastDotPos + 1); } StringBuffer name = new StringBuffer(); StringTokenizer tok = new StringTokenizer(schemaName, String.valueOf(STD_SEPARATOR_CHAR)); while (tok.hasMoreTokens()) { String namePart = ((String) tok.nextElement()).toLowerCase(); name.append(StringUtils.capitalize(namePart)); } return name.toString(); } /** * Converts a database schema name to java object name. Operates * same as underscoreMethod but does not convert anything to * lowercase. * * @param schemaName name to be converted. * @return converted name. * @see com.frameworkset.orm.engine.model.NameGenerator * @see #underscoreMethod(String) */ protected String javanameMethod(String schemaName, boolean IGNORE_FIRST_TOKEN) { StringBuffer name = new StringBuffer(); StringTokenizer tok = new StringTokenizer(schemaName, String.valueOf(STD_SEPARATOR_CHAR)); boolean first = true; String firstNamepart = null; while (tok.hasMoreTokens()) { if (first && IGNORE_FIRST_TOKEN) { firstNamepart = (String) tok.nextElement(); first = false; continue; } String namePart = (String) tok.nextElement(); name.append(StringUtils.capitalize(namePart)); } // remove the SCHEMA_SEPARATOR_CHARs and capitalize // the tokens if (name.length() > 0) schemaName = name.toString(); else if (IGNORE_FIRST_TOKEN && firstNamepart != null) schemaName = firstNamepart; name = new StringBuffer(); tok = new StringTokenizer(schemaName, String.valueOf(SCHEMA_SEPARATOR_CHAR)); while (tok.hasMoreTokens()) { String namePart = (String) tok.nextElement(); name.append(StringUtils.capitalize(namePart)); } return name.toString(); } /** * Converts a database schema name to java object name. Operates * same as underscoreMethod but does not convert anything to * lowercase. * * @param schemaName name to be converted. * @return converted name. * @see com.frameworkset.orm.engine.model.NameGenerator * @see #underscoreMethod(String) */ protected String javanameField(String schemaName, boolean IGNORE_FIRST_TOKEN) { StringBuffer name = new StringBuffer(); StringTokenizer tok = new StringTokenizer(schemaName, String.valueOf(STD_SEPARATOR_CHAR)); boolean first = true; while (tok.hasMoreTokens()) { String namePart = (String) tok.nextElement(); if (!first) { name.append(StringUtils.capitalize(namePart)); } else { first = false; name.append(namePart); } } // remove the SCHEMA_SEPARATOR_CHARs and capitalize // the tokens schemaName = name.toString(); name = new StringBuffer(); tok = new StringTokenizer(schemaName, String.valueOf(SCHEMA_SEPARATOR_CHAR)); first = true; while (tok.hasMoreTokens()) { String namePart = (String) tok.nextElement(); if (!first) { name.append(StringUtils.capitalize(namePart)); } else { first = false; name.append(namePart); } } return name.toString(); } /** * Converts a database schema name to java object name. In this * case no conversion is made. * * @param name name to be converted. * @return The <code>name</code> parameter, unchanged. */ protected final String nochangeMethod(String name, boolean IGNORE_FIRST_TOKEN) { return name; } }