java/alt2xml-in-ini/src/cz/frantovo/alt2xml/in/ini/Reader.java
author František Kučera <franta-hg@frantovo.cz>
Sat, 06 Sep 2014 23:34:53 +0200
changeset 93 0d8e8903d638
parent 92 03c1c831cfcb
child 94 f5b287fa69b6
permissions -rw-r--r--
in-ini: re-eble whitespace in entry keys – but trim spaces before „=“ to support:
„ a b c = d“ → „<a-b-c>…“ instead of „<a-b-c->…“
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@92
    34
import org.xml.sax.helpers.AttributesImpl;
franta-hg@2
    35
franta-hg@2
    36
/**
franta-hg@2
    37
 *
franta-hg@17
    38
 * @author Ing. František Kučera (frantovo.cz)
franta-hg@2
    39
 */
franta-hg@17
    40
public class Reader extends AbstractAlt2XmlReader {
franta-hg@13
    41
franta-hg@77
    42
	public static final String ROOT_ELEMENT = "ini";
franta-hg@77
    43
	private static final Logger log = Logger.getLogger(Reader.class.getName());
franta-hg@77
    44
franta-hg@2
    45
	@Override
franta-hg@6
    46
	public void parse(InputSource input) throws IOException, SAXException {
franta-hg@59
    47
		outputStart();
franta-hg@77
    48
franta-hg@77
    49
		try (BufferedReader br = new BufferedReader(new InputStreamReader(input.getByteStream()))) {
franta-hg@78
    50
			FileContext fc = new FileContext(contentHandler);
franta-hg@77
    51
			for (String currentLine = br.readLine(); currentLine != null; currentLine = br.readLine()) {
franta-hg@86
    52
				fc.lineNumber++;
franta-hg@86
    53
				boolean lineProcessed = false;
franta-hg@77
    54
				for (LINE_TYPE lineType : LINE_TYPE.values()) {
franta-hg@86
    55
					lineProcessed = lineType.processLine(currentLine, fc);
franta-hg@77
    56
					if (lineProcessed) {
franta-hg@77
    57
						break;
franta-hg@77
    58
					}
franta-hg@77
    59
				}
franta-hg@86
    60
				if (!lineProcessed) {
franta-hg@86
    61
					log.log(Level.SEVERE, "Invalid line in INI file: {0}", currentLine);
franta-hg@86
    62
				}
franta-hg@77
    63
			}
franta-hg@78
    64
			fc.outputEndSection(fc.lastSection);
franta-hg@78
    65
franta-hg@77
    66
		}
franta-hg@77
    67
franta-hg@59
    68
		outputEnd();
franta-hg@59
    69
	}
franta-hg@59
    70
franta-hg@59
    71
	private void outputStart() throws SAXException {
franta-hg@21
    72
		contentHandler.startDocument();
franta-hg@76
    73
		contentHandler.lineBreak();
franta-hg@77
    74
		contentHandler.startElement(null, null, ROOT_ELEMENT, null);
franta-hg@75
    75
		contentHandler.lineBreak();
franta-hg@59
    76
	}
franta-hg@59
    77
franta-hg@59
    78
	private void outputEnd() throws SAXException {
franta-hg@76
    79
		contentHandler.endElement(null, null, "ini");
franta-hg@75
    80
		contentHandler.lineBreak();
franta-hg@21
    81
		contentHandler.endDocument();
franta-hg@6
    82
	}
franta-hg@76
    83
franta-hg@77
    84
	private static class FileContext {
franta-hg@77
    85
franta-hg@77
    86
		private final Alt2ContentHandler contentHandler;
franta-hg@78
    87
		private String lastSection;
franta-hg@86
    88
		private int lineNumber;
franta-hg@77
    89
franta-hg@77
    90
		public FileContext(Alt2ContentHandler contentHandler) {
franta-hg@77
    91
			this.contentHandler = contentHandler;
franta-hg@77
    92
		}
franta-hg@78
    93
franta-hg@78
    94
		protected void outputStartSection(String name) throws SAXException {
franta-hg@78
    95
			contentHandler.indentation(1);
franta-hg@78
    96
			contentHandler.startElement(null, null, name, null);
franta-hg@78
    97
			contentHandler.lineBreak();
franta-hg@78
    98
		}
franta-hg@78
    99
franta-hg@78
   100
		protected void outputEndSection(String name) throws SAXException {
franta-hg@78
   101
			if (name != null) {
franta-hg@78
   102
				contentHandler.indentation(1);
franta-hg@78
   103
				contentHandler.endElement(null, null, name);
franta-hg@78
   104
				contentHandler.lineBreak();
franta-hg@78
   105
			}
franta-hg@78
   106
		}
franta-hg@77
   107
	}
franta-hg@77
   108
franta-hg@89
   109
	private static String encodeXmlName(String originalName, int lineNumber) {
franta-hg@89
   110
		String encodedName = Functions.encodeXmlName(originalName);
franta-hg@89
   111
		if (!encodedName.equals(originalName)) {
franta-hg@89
   112
			log.log(Level.FINE, "Line {0}: name „{1} was encoded to „{2}““", new Object[]{lineNumber, originalName, encodedName});
franta-hg@89
   113
		}
franta-hg@89
   114
		return encodedName;
franta-hg@89
   115
	}
franta-hg@89
   116
franta-hg@77
   117
	private static class LineContext {
franta-hg@77
   118
franta-hg@77
   119
		private final Matcher matcher;
franta-hg@77
   120
franta-hg@81
   121
		public LineContext(Matcher matcher) {
franta-hg@77
   122
			this.matcher = matcher;
franta-hg@77
   123
		}
franta-hg@77
   124
	}
franta-hg@77
   125
franta-hg@77
   126
	private enum LINE_TYPE {
franta-hg@77
   127
franta-hg@86
   128
		BLANK_LINE("\\s*") {
franta-hg@86
   129
					@Override
franta-hg@86
   130
					public void processLine(LineContext lc, FileContext fc) throws SAXException {
franta-hg@86
   131
						log.log(Level.FINEST, "Line {0}: skipping blank line", fc.lineNumber);
franta-hg@86
   132
					}
franta-hg@86
   133
				},
franta-hg@83
   134
		COMMENT("\\s*(;|#)\\s*(?<comment>.*)") {
franta-hg@77
   135
					@Override
franta-hg@78
   136
					public void processLine(LineContext lc, FileContext fc) throws SAXException {
franta-hg@82
   137
						// TODO: comment → LexicalHandler
franta-hg@86
   138
						log.log(Level.FINER, "Line {0}: comment: {1}", new Object[]{fc.lineNumber, lc.matcher.group("comment")});
franta-hg@77
   139
					}
franta-hg@77
   140
franta-hg@77
   141
				},
franta-hg@88
   142
		SECTION("\\s*\\[\\s*(?<name>[^\\]\\]]+)\\s*\\]\\s*") {
franta-hg@77
   143
					@Override
franta-hg@78
   144
					public void processLine(LineContext lc, FileContext fc) throws SAXException {
franta-hg@89
   145
						String name = encodeXmlName(lc.matcher.group("name"), fc.lineNumber);
franta-hg@78
   146
						fc.outputEndSection(fc.lastSection);
franta-hg@89
   147
						fc.outputStartSection(name);
franta-hg@89
   148
						fc.lastSection = name;
franta-hg@77
   149
					}
franta-hg@77
   150
franta-hg@77
   151
				},
franta-hg@81
   152
		ENTRY(
franta-hg@93
   153
				"\\s*(?<key>[^=\\]]+?[^=\\s\\]]*)(\\[(?<subkey>[^\\]]+)\\]){0,1}\\s*=\\s*\"(?<value>[^']+)\"\\s*((;|#)\\s*(?<comment>.*)){0,1}", // quoted value → include spaces + might have comment
franta-hg@93
   154
				"\\s*(?<key>[^=\\]]+?[^=\\s\\]]*)(\\[(?<subkey>[^\\]]+)\\]){0,1}\\s*=\\s*'(?<value>[^']+)'\\s*((;|#)\\s*(?<comment>.*)){0,1}", // apostrophed value → include spaces + might have comment
franta-hg@93
   155
				"\\s*(?<key>[^=\\]]+?[^=\\s\\]]*)(\\[(?<subkey>[^\\]]+)\\]){0,1}\\s*=\\s*(?<value>.+)" // unquoted value → strip spaces + no comments
franta-hg@81
   156
		) {
franta-hg@77
   157
					@Override
franta-hg@78
   158
					public void processLine(LineContext lc, FileContext fc) throws SAXException {
franta-hg@89
   159
						String key = encodeXmlName(lc.matcher.group("key"), fc.lineNumber);
franta-hg@83
   160
						String value = lc.matcher.group("value");
franta-hg@78
   161
franta-hg@92
   162
						if (lc.matcher.groupCount() > 4) {
franta-hg@83
   163
							String comment = lc.matcher.group("comment");
franta-hg@82
   164
							// TODO: comment → LexicalHandler
franta-hg@86
   165
							log.log(Level.FINER, "Line {0}: comment for entry „{1}“ is: {2}", new Object[]{fc.lineNumber, key, comment});
franta-hg@82
   166
						}
franta-hg@82
   167
franta-hg@92
   168
						AttributesImpl attributes = null;
franta-hg@92
   169
						String subkey = lc.matcher.group("subkey");
franta-hg@92
   170
						if (subkey != null) {
franta-hg@92
   171
							attributes = new AttributesImpl();
franta-hg@92
   172
							attributes.addAttribute(null, "sub", "sub", "xs:string", subkey);
franta-hg@92
   173
						}
franta-hg@92
   174
franta-hg@85
   175
						fc.contentHandler.indentation(fc.lastSection == null ? 1 : 2);
franta-hg@92
   176
						fc.contentHandler.textElement(value, null, null, key, attributes);
franta-hg@78
   177
						fc.contentHandler.lineBreak();
franta-hg@78
   178
franta-hg@77
   179
					}
franta-hg@77
   180
franta-hg@77
   181
				},;
franta-hg@77
   182
franta-hg@81
   183
		private LINE_TYPE(String... patterns) {
franta-hg@81
   184
			for (String pattern : patterns) {
franta-hg@81
   185
				this.patterns.add(Pattern.compile(pattern));
franta-hg@81
   186
			}
franta-hg@77
   187
		}
franta-hg@77
   188
franta-hg@81
   189
		private final List<Pattern> patterns = new ArrayList<>();
franta-hg@77
   190
franta-hg@86
   191
		protected boolean processLine(String currentLine, FileContext fc) throws SAXException {
franta-hg@81
   192
			for (Pattern pattern : patterns) {
franta-hg@81
   193
				Matcher m = pattern.matcher(currentLine);
franta-hg@81
   194
				if (m.matches()) {
franta-hg@86
   195
					log.log(Level.FINEST, "Line {0}: pattern „{1}“ matches „{2}“", new Object[]{fc.lineNumber, pattern, currentLine});
franta-hg@86
   196
					processLine(new LineContext(m), fc);
franta-hg@81
   197
					return true;
franta-hg@81
   198
				}
franta-hg@77
   199
			}
franta-hg@81
   200
			return false;
franta-hg@77
   201
		}
franta-hg@77
   202
franta-hg@78
   203
		public abstract void processLine(LineContext lc, FileContext fc) throws SAXException;
franta-hg@77
   204
	}
franta-hg@77
   205
franta-hg@2
   206
}