franta-hg@11: /**
franta-hg@11:  * Alt2XML
franta-hg@11:  * Copyright © 2014 František Kučera (frantovo.cz)
franta-hg@11:  *
franta-hg@11:  * This program is free software: you can redistribute it and/or modify
franta-hg@11:  * it under the terms of the GNU General Public License as published by
franta-hg@111:  * the Free Software Foundation, version 3 of the License.
franta-hg@11:  *
franta-hg@11:  * This program is distributed in the hope that it will be useful,
franta-hg@11:  * but WITHOUT ANY WARRANTY; without even the implied warranty of
franta-hg@11:  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
franta-hg@11:  * GNU General Public License for more details.
franta-hg@11:  *
franta-hg@11:  * You should have received a copy of the GNU General Public License
franta-hg@11:  * along with this program. If not, see <http://www.gnu.org/licenses/>.
franta-hg@11:  */
franta-hg@76: package cz.frantovo.alt2xml.in.ini;
franta-hg@2: 
franta-hg@17: import cz.frantovo.alt2xml.AbstractAlt2XmlReader;
franta-hg@77: import cz.frantovo.alt2xml.in.Alt2ContentHandler;
franta-hg@89: import cz.frantovo.alt2xml.in.Functions;
franta-hg@77: import java.io.BufferedReader;
franta-hg@2: import java.io.IOException;
franta-hg@77: import java.io.InputStreamReader;
franta-hg@81: import java.util.ArrayList;
franta-hg@81: import java.util.List;
franta-hg@77: import java.util.logging.Level;
franta-hg@77: import java.util.logging.Logger;
franta-hg@77: import java.util.regex.Matcher;
franta-hg@77: import java.util.regex.Pattern;
franta-hg@2: import org.xml.sax.InputSource;
franta-hg@2: import org.xml.sax.SAXException;
franta-hg@92: import org.xml.sax.helpers.AttributesImpl;
franta-hg@2: 
franta-hg@2: /**
franta-hg@95:  * Reads INI files with sections and entries.
franta-hg@95:  * Example:
franta-hg@95:  * <pre>; this is comment
franta-hg@95:  *random=value outside of any groups
franta-hg@95:  *
franta-hg@95:  *[some_section]
franta-hg@95:  *
franta-hg@95:  *; simple entry:
franta-hg@95:  *key=value
franta-hg@95:  *
franta-hg@95:  *; entry starting/ending with whitespace
franta-hg@95:  *white="  spaces everywhere  " ; might have comment
franta-hg@95:  *alternative='  spaces everywhere  ' ; same
franta-hg@95:  *
franta-hg@95:  *; entries with subkeys:
franta-hg@95:  *key[subkey_a]=value
franta-hg@95:  *key[subkey_b]=value
franta-hg@95:  *
franta-hg@95:  *# alternative way to comment
franta-hg@95:  *
franta-hg@95:  *[another secion]
franta-hg@95:  *yes=there might be spaces in names
franta-hg@95:  *because=they are encoded before putting into XML element names
franta-hg@95:  * </pre>
franta-hg@2:  *
franta-hg@17:  * @author Ing. František Kučera (frantovo.cz)
franta-hg@2:  */
franta-hg@17: public class Reader extends AbstractAlt2XmlReader {
franta-hg@13: 
franta-hg@77: 	public static final String ROOT_ELEMENT = "ini";
franta-hg@77: 	private static final Logger log = Logger.getLogger(Reader.class.getName());
franta-hg@77: 
franta-hg@2: 	@Override
franta-hg@6: 	public void parse(InputSource input) throws IOException, SAXException {
franta-hg@59: 		outputStart();
franta-hg@77: 
franta-hg@77: 		try (BufferedReader br = new BufferedReader(new InputStreamReader(input.getByteStream()))) {
franta-hg@78: 			FileContext fc = new FileContext(contentHandler);
franta-hg@77: 			for (String currentLine = br.readLine(); currentLine != null; currentLine = br.readLine()) {
franta-hg@86: 				fc.lineNumber++;
franta-hg@86: 				boolean lineProcessed = false;
franta-hg@77: 				for (LINE_TYPE lineType : LINE_TYPE.values()) {
franta-hg@86: 					lineProcessed = lineType.processLine(currentLine, fc);
franta-hg@77: 					if (lineProcessed) {
franta-hg@77: 						break;
franta-hg@77: 					}
franta-hg@77: 				}
franta-hg@86: 				if (!lineProcessed) {
franta-hg@86: 					log.log(Level.SEVERE, "Invalid line in INI file: {0}", currentLine);
franta-hg@86: 				}
franta-hg@77: 			}
franta-hg@78: 			fc.outputEndSection(fc.lastSection);
franta-hg@78: 
franta-hg@77: 		}
franta-hg@77: 
franta-hg@59: 		outputEnd();
franta-hg@59: 	}
franta-hg@59: 
franta-hg@59: 	private void outputStart() throws SAXException {
franta-hg@21: 		contentHandler.startDocument();
franta-hg@76: 		contentHandler.lineBreak();
franta-hg@77: 		contentHandler.startElement(null, null, ROOT_ELEMENT, null);
franta-hg@75: 		contentHandler.lineBreak();
franta-hg@59: 	}
franta-hg@59: 
franta-hg@59: 	private void outputEnd() throws SAXException {
franta-hg@102: 		contentHandler.endElement(null, null, ROOT_ELEMENT);
franta-hg@75: 		contentHandler.lineBreak();
franta-hg@21: 		contentHandler.endDocument();
franta-hg@6: 	}
franta-hg@76: 
franta-hg@77: 	private static class FileContext {
franta-hg@77: 
franta-hg@77: 		private final Alt2ContentHandler contentHandler;
franta-hg@78: 		private String lastSection;
franta-hg@86: 		private int lineNumber;
franta-hg@77: 
franta-hg@77: 		public FileContext(Alt2ContentHandler contentHandler) {
franta-hg@77: 			this.contentHandler = contentHandler;
franta-hg@77: 		}
franta-hg@78: 
franta-hg@78: 		protected void outputStartSection(String name) throws SAXException {
franta-hg@78: 			contentHandler.indentation(1);
franta-hg@78: 			contentHandler.startElement(null, null, name, null);
franta-hg@78: 			contentHandler.lineBreak();
franta-hg@78: 		}
franta-hg@78: 
franta-hg@78: 		protected void outputEndSection(String name) throws SAXException {
franta-hg@78: 			if (name != null) {
franta-hg@78: 				contentHandler.indentation(1);
franta-hg@78: 				contentHandler.endElement(null, null, name);
franta-hg@78: 				contentHandler.lineBreak();
franta-hg@78: 			}
franta-hg@78: 		}
franta-hg@77: 	}
franta-hg@77: 
franta-hg@89: 	private static String encodeXmlName(String originalName, int lineNumber) {
franta-hg@89: 		String encodedName = Functions.encodeXmlName(originalName);
franta-hg@89: 		if (!encodedName.equals(originalName)) {
franta-hg@89: 			log.log(Level.FINE, "Line {0}: name „{1} was encoded to „{2}““", new Object[]{lineNumber, originalName, encodedName});
franta-hg@89: 		}
franta-hg@89: 		return encodedName;
franta-hg@89: 	}
franta-hg@89: 
franta-hg@77: 	private static class LineContext {
franta-hg@77: 
franta-hg@77: 		private final Matcher matcher;
franta-hg@77: 
franta-hg@81: 		public LineContext(Matcher matcher) {
franta-hg@77: 			this.matcher = matcher;
franta-hg@77: 		}
franta-hg@77: 	}
franta-hg@77: 
franta-hg@77: 	private enum LINE_TYPE {
franta-hg@77: 
franta-hg@86: 		BLANK_LINE("\\s*") {
franta-hg@86: 					@Override
franta-hg@86: 					public void processLine(LineContext lc, FileContext fc) throws SAXException {
franta-hg@86: 						log.log(Level.FINEST, "Line {0}: skipping blank line", fc.lineNumber);
franta-hg@86: 					}
franta-hg@86: 				},
franta-hg@83: 		COMMENT("\\s*(;|#)\\s*(?<comment>.*)") {
franta-hg@77: 					@Override
franta-hg@78: 					public void processLine(LineContext lc, FileContext fc) throws SAXException {
franta-hg@82: 						// TODO: comment → LexicalHandler
franta-hg@86: 						log.log(Level.FINER, "Line {0}: comment: {1}", new Object[]{fc.lineNumber, lc.matcher.group("comment")});
franta-hg@77: 					}
franta-hg@77: 
franta-hg@77: 				},
franta-hg@95: 		SECTION("\\s*\\[\\s*(?<name>[^\\]]+)\\s*\\]\\s*") {
franta-hg@77: 					@Override
franta-hg@78: 					public void processLine(LineContext lc, FileContext fc) throws SAXException {
franta-hg@89: 						String name = encodeXmlName(lc.matcher.group("name"), fc.lineNumber);
franta-hg@78: 						fc.outputEndSection(fc.lastSection);
franta-hg@89: 						fc.outputStartSection(name);
franta-hg@89: 						fc.lastSection = name;
franta-hg@77: 					}
franta-hg@77: 
franta-hg@77: 				},
franta-hg@81: 		ENTRY(
franta-hg@94: 				"\\s*(?<key>[^=\\]]+?[^=\\s\\]]*)(\\[(?<subkey>[^\\]]+)\\])?\\s*=\\s*\"(?<value>[^']+)\"\\s*((;|#)\\s*(?<comment>.*))?", // quoted value → include spaces + might have comment
franta-hg@94: 				"\\s*(?<key>[^=\\]]+?[^=\\s\\]]*)(\\[(?<subkey>[^\\]]+)\\])?\\s*=\\s*'(?<value>[^']+)'\\s*((;|#)\\s*(?<comment>.*))?", // apostrophed value → include spaces + might have comment
franta-hg@94: 				"\\s*(?<key>[^=\\]]+?[^=\\s\\]]*)(\\[(?<subkey>[^\\]]+)\\])?\\s*=\\s*(?<value>.+)" // unquoted value → strip spaces + no comments
franta-hg@81: 		) {
franta-hg@77: 					@Override
franta-hg@78: 					public void processLine(LineContext lc, FileContext fc) throws SAXException {
franta-hg@89: 						String key = encodeXmlName(lc.matcher.group("key"), fc.lineNumber);
franta-hg@83: 						String value = lc.matcher.group("value");
franta-hg@78: 
franta-hg@92: 						if (lc.matcher.groupCount() > 4) {
franta-hg@83: 							String comment = lc.matcher.group("comment");
franta-hg@82: 							// TODO: comment → LexicalHandler
franta-hg@86: 							log.log(Level.FINER, "Line {0}: comment for entry „{1}“ is: {2}", new Object[]{fc.lineNumber, key, comment});
franta-hg@82: 						}
franta-hg@82: 
franta-hg@92: 						AttributesImpl attributes = null;
franta-hg@92: 						String subkey = lc.matcher.group("subkey");
franta-hg@92: 						if (subkey != null) {
franta-hg@92: 							attributes = new AttributesImpl();
franta-hg@92: 							attributes.addAttribute(null, "sub", "sub", "xs:string", subkey);
franta-hg@92: 						}
franta-hg@92: 
franta-hg@85: 						fc.contentHandler.indentation(fc.lastSection == null ? 1 : 2);
franta-hg@92: 						fc.contentHandler.textElement(value, null, null, key, attributes);
franta-hg@78: 						fc.contentHandler.lineBreak();
franta-hg@78: 
franta-hg@77: 					}
franta-hg@77: 
franta-hg@77: 				},;
franta-hg@77: 
franta-hg@95: 		/**
franta-hg@95: 		 * @param patterns regular expression (or expressions) that describes this line type
franta-hg@95: 		 */
franta-hg@81: 		private LINE_TYPE(String... patterns) {
franta-hg@81: 			for (String pattern : patterns) {
franta-hg@81: 				this.patterns.add(Pattern.compile(pattern));
franta-hg@81: 			}
franta-hg@77: 		}
franta-hg@77: 
franta-hg@81: 		private final List<Pattern> patterns = new ArrayList<>();
franta-hg@77: 
franta-hg@95: 		/**
franta-hg@95: 		 *
franta-hg@95: 		 * @param currentLine input line to be parsed
franta-hg@95: 		 * @param fc
franta-hg@95: 		 * @return whether line matches and was thus processed
franta-hg@95: 		 * @throws SAXException
franta-hg@95: 		 */
franta-hg@86: 		protected boolean processLine(String currentLine, FileContext fc) throws SAXException {
franta-hg@81: 			for (Pattern pattern : patterns) {
franta-hg@81: 				Matcher m = pattern.matcher(currentLine);
franta-hg@81: 				if (m.matches()) {
franta-hg@86: 					log.log(Level.FINEST, "Line {0}: pattern „{1}“ matches „{2}“", new Object[]{fc.lineNumber, pattern, currentLine});
franta-hg@86: 					processLine(new LineContext(m), fc);
franta-hg@81: 					return true;
franta-hg@81: 				}
franta-hg@77: 			}
franta-hg@81: 			return false;
franta-hg@77: 		}
franta-hg@77: 
franta-hg@78: 		public abstract void processLine(LineContext lc, FileContext fc) throws SAXException;
franta-hg@77: 	}
franta-hg@2: }