java/alt2xml-in-ini/src/cz/frantovo/alt2xml/in/ini/Reader.java
author František Kučera <franta-hg@frantovo.cz>
Sat, 06 Sep 2014 22:04:11 +0200
changeset 91 8146ad99fc67
parent 89 46c7cc4863c1
child 92 03c1c831cfcb
permissions -rw-r--r--
in-ini: no whitespace in entry keys – but do encoding
franta-hg@11
     1
/**
franta-hg@11
     2
 * Alt2XML
franta-hg@11
     3
 * Copyright © 2014 František Kučera (frantovo.cz)
franta-hg@11
     4
 *
franta-hg@11
     5
 * This program is free software: you can redistribute it and/or modify
franta-hg@11
     6
 * it under the terms of the GNU General Public License as published by
franta-hg@11
     7
 * the Free Software Foundation, either version 3 of the License, or
franta-hg@11
     8
 * (at your option) any later version.
franta-hg@11
     9
 *
franta-hg@11
    10
 * This program is distributed in the hope that it will be useful,
franta-hg@11
    11
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
franta-hg@11
    12
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
franta-hg@11
    13
 * GNU General Public License for more details.
franta-hg@11
    14
 *
franta-hg@11
    15
 * You should have received a copy of the GNU General Public License
franta-hg@11
    16
 * along with this program. If not, see <http://www.gnu.org/licenses/>.
franta-hg@11
    17
 */
franta-hg@76
    18
package cz.frantovo.alt2xml.in.ini;
franta-hg@2
    19
franta-hg@17
    20
import cz.frantovo.alt2xml.AbstractAlt2XmlReader;
franta-hg@77
    21
import cz.frantovo.alt2xml.in.Alt2ContentHandler;
franta-hg@89
    22
import cz.frantovo.alt2xml.in.Functions;
franta-hg@77
    23
import java.io.BufferedReader;
franta-hg@2
    24
import java.io.IOException;
franta-hg@77
    25
import java.io.InputStreamReader;
franta-hg@81
    26
import java.util.ArrayList;
franta-hg@81
    27
import java.util.List;
franta-hg@77
    28
import java.util.logging.Level;
franta-hg@77
    29
import java.util.logging.Logger;
franta-hg@77
    30
import java.util.regex.Matcher;
franta-hg@77
    31
import java.util.regex.Pattern;
franta-hg@2
    32
import org.xml.sax.InputSource;
franta-hg@2
    33
import org.xml.sax.SAXException;
franta-hg@2
    34
franta-hg@2
    35
/**
franta-hg@2
    36
 *
franta-hg@17
    37
 * @author Ing. František Kučera (frantovo.cz)
franta-hg@2
    38
 */
franta-hg@17
    39
public class Reader extends AbstractAlt2XmlReader {
franta-hg@13
    40
franta-hg@77
    41
	public static final String ROOT_ELEMENT = "ini";
franta-hg@77
    42
	private static final Logger log = Logger.getLogger(Reader.class.getName());
franta-hg@77
    43
franta-hg@2
    44
	@Override
franta-hg@6
    45
	public void parse(InputSource input) throws IOException, SAXException {
franta-hg@59
    46
		outputStart();
franta-hg@77
    47
franta-hg@77
    48
		try (BufferedReader br = new BufferedReader(new InputStreamReader(input.getByteStream()))) {
franta-hg@78
    49
			FileContext fc = new FileContext(contentHandler);
franta-hg@77
    50
			for (String currentLine = br.readLine(); currentLine != null; currentLine = br.readLine()) {
franta-hg@86
    51
				fc.lineNumber++;
franta-hg@86
    52
				boolean lineProcessed = false;
franta-hg@77
    53
				for (LINE_TYPE lineType : LINE_TYPE.values()) {
franta-hg@86
    54
					lineProcessed = lineType.processLine(currentLine, fc);
franta-hg@77
    55
					if (lineProcessed) {
franta-hg@77
    56
						break;
franta-hg@77
    57
					}
franta-hg@77
    58
				}
franta-hg@86
    59
				if (!lineProcessed) {
franta-hg@86
    60
					log.log(Level.SEVERE, "Invalid line in INI file: {0}", currentLine);
franta-hg@86
    61
				}
franta-hg@77
    62
			}
franta-hg@78
    63
			fc.outputEndSection(fc.lastSection);
franta-hg@78
    64
franta-hg@77
    65
		}
franta-hg@77
    66
franta-hg@59
    67
		outputEnd();
franta-hg@59
    68
	}
franta-hg@59
    69
franta-hg@59
    70
	private void outputStart() throws SAXException {
franta-hg@21
    71
		contentHandler.startDocument();
franta-hg@76
    72
		contentHandler.lineBreak();
franta-hg@77
    73
		contentHandler.startElement(null, null, ROOT_ELEMENT, null);
franta-hg@75
    74
		contentHandler.lineBreak();
franta-hg@59
    75
	}
franta-hg@59
    76
franta-hg@59
    77
	private void outputEnd() throws SAXException {
franta-hg@76
    78
		contentHandler.endElement(null, null, "ini");
franta-hg@75
    79
		contentHandler.lineBreak();
franta-hg@21
    80
		contentHandler.endDocument();
franta-hg@6
    81
	}
franta-hg@76
    82
franta-hg@77
    83
	private static class FileContext {
franta-hg@77
    84
franta-hg@77
    85
		private final Alt2ContentHandler contentHandler;
franta-hg@78
    86
		private String lastSection;
franta-hg@86
    87
		private int lineNumber;
franta-hg@77
    88
franta-hg@77
    89
		public FileContext(Alt2ContentHandler contentHandler) {
franta-hg@77
    90
			this.contentHandler = contentHandler;
franta-hg@77
    91
		}
franta-hg@78
    92
franta-hg@78
    93
		protected void outputStartSection(String name) throws SAXException {
franta-hg@78
    94
			contentHandler.indentation(1);
franta-hg@78
    95
			contentHandler.startElement(null, null, name, null);
franta-hg@78
    96
			contentHandler.lineBreak();
franta-hg@78
    97
		}
franta-hg@78
    98
franta-hg@78
    99
		protected void outputEndSection(String name) throws SAXException {
franta-hg@78
   100
			if (name != null) {
franta-hg@78
   101
				contentHandler.indentation(1);
franta-hg@78
   102
				contentHandler.endElement(null, null, name);
franta-hg@78
   103
				contentHandler.lineBreak();
franta-hg@78
   104
			}
franta-hg@78
   105
		}
franta-hg@77
   106
	}
franta-hg@77
   107
franta-hg@89
   108
	private static String encodeXmlName(String originalName, int lineNumber) {
franta-hg@89
   109
		String encodedName = Functions.encodeXmlName(originalName);
franta-hg@89
   110
		if (!encodedName.equals(originalName)) {
franta-hg@89
   111
			log.log(Level.FINE, "Line {0}: name „{1} was encoded to „{2}““", new Object[]{lineNumber, originalName, encodedName});
franta-hg@89
   112
		}
franta-hg@89
   113
		return encodedName;
franta-hg@89
   114
	}
franta-hg@89
   115
franta-hg@77
   116
	private static class LineContext {
franta-hg@77
   117
franta-hg@77
   118
		private final Matcher matcher;
franta-hg@77
   119
franta-hg@81
   120
		public LineContext(Matcher matcher) {
franta-hg@77
   121
			this.matcher = matcher;
franta-hg@77
   122
		}
franta-hg@77
   123
	}
franta-hg@77
   124
franta-hg@77
   125
	private enum LINE_TYPE {
franta-hg@77
   126
franta-hg@86
   127
		BLANK_LINE("\\s*") {
franta-hg@86
   128
					@Override
franta-hg@86
   129
					public void processLine(LineContext lc, FileContext fc) throws SAXException {
franta-hg@86
   130
						log.log(Level.FINEST, "Line {0}: skipping blank line", fc.lineNumber);
franta-hg@86
   131
					}
franta-hg@86
   132
				},
franta-hg@83
   133
		COMMENT("\\s*(;|#)\\s*(?<comment>.*)") {
franta-hg@77
   134
					@Override
franta-hg@78
   135
					public void processLine(LineContext lc, FileContext fc) throws SAXException {
franta-hg@82
   136
						// TODO: comment → LexicalHandler
franta-hg@86
   137
						log.log(Level.FINER, "Line {0}: comment: {1}", new Object[]{fc.lineNumber, lc.matcher.group("comment")});
franta-hg@77
   138
					}
franta-hg@77
   139
franta-hg@77
   140
				},
franta-hg@88
   141
		SECTION("\\s*\\[\\s*(?<name>[^\\]\\]]+)\\s*\\]\\s*") {
franta-hg@77
   142
					@Override
franta-hg@78
   143
					public void processLine(LineContext lc, FileContext fc) throws SAXException {
franta-hg@89
   144
						String name = encodeXmlName(lc.matcher.group("name"), fc.lineNumber);
franta-hg@78
   145
						fc.outputEndSection(fc.lastSection);
franta-hg@89
   146
						fc.outputStartSection(name);
franta-hg@89
   147
						fc.lastSection = name;
franta-hg@77
   148
					}
franta-hg@77
   149
franta-hg@77
   150
				},
franta-hg@81
   151
		ENTRY(
franta-hg@91
   152
				"\\s*(?<key>[^=\\s]+)\\s*=\\s*\"(?<value>[^']+)\"\\s*((;|#)\\s*(?<comment>.*)){0,1}", // quoted value → include spaces + might have comment
franta-hg@91
   153
				"\\s*(?<key>[^=\\s]+)\\s*=\\s*'(?<value>[^']+)'\\s*((;|#)\\s*(?<comment>.*)){0,1}", // apostrophed value → include spaces + might have comment
franta-hg@91
   154
				"\\s*(?<key>[^=\\s]+)\\s*=\\s*(?<value>.+)" // unquoted value → strip spaces + no comments
franta-hg@81
   155
		) {
franta-hg@77
   156
					@Override
franta-hg@78
   157
					public void processLine(LineContext lc, FileContext fc) throws SAXException {
franta-hg@89
   158
						String key = encodeXmlName(lc.matcher.group("key"), fc.lineNumber);
franta-hg@83
   159
						String value = lc.matcher.group("value");
franta-hg@78
   160
franta-hg@84
   161
						if (lc.matcher.groupCount() > 2) {
franta-hg@83
   162
							String comment = lc.matcher.group("comment");
franta-hg@82
   163
							// TODO: comment → LexicalHandler
franta-hg@86
   164
							log.log(Level.FINER, "Line {0}: comment for entry „{1}“ is: {2}", new Object[]{fc.lineNumber, key, comment});
franta-hg@82
   165
						}
franta-hg@82
   166
franta-hg@85
   167
						fc.contentHandler.indentation(fc.lastSection == null ? 1 : 2);
franta-hg@78
   168
						fc.contentHandler.textElement(value, null, null, key, null);
franta-hg@78
   169
						fc.contentHandler.lineBreak();
franta-hg@78
   170
franta-hg@77
   171
					}
franta-hg@77
   172
franta-hg@77
   173
				},;
franta-hg@77
   174
franta-hg@81
   175
		private LINE_TYPE(String... patterns) {
franta-hg@81
   176
			for (String pattern : patterns) {
franta-hg@81
   177
				this.patterns.add(Pattern.compile(pattern));
franta-hg@81
   178
			}
franta-hg@77
   179
		}
franta-hg@77
   180
franta-hg@81
   181
		private final List<Pattern> patterns = new ArrayList<>();
franta-hg@77
   182
franta-hg@86
   183
		protected boolean processLine(String currentLine, FileContext fc) throws SAXException {
franta-hg@81
   184
			for (Pattern pattern : patterns) {
franta-hg@81
   185
				Matcher m = pattern.matcher(currentLine);
franta-hg@81
   186
				if (m.matches()) {
franta-hg@86
   187
					log.log(Level.FINEST, "Line {0}: pattern „{1}“ matches „{2}“", new Object[]{fc.lineNumber, pattern, currentLine});
franta-hg@86
   188
					processLine(new LineContext(m), fc);
franta-hg@81
   189
					return true;
franta-hg@81
   190
				}
franta-hg@77
   191
			}
franta-hg@81
   192
			return false;
franta-hg@77
   193
		}
franta-hg@77
   194
franta-hg@78
   195
		public abstract void processLine(LineContext lc, FileContext fc) throws SAXException;
franta-hg@77
   196
	}
franta-hg@77
   197
franta-hg@2
   198
}