tikhomirov@497: /*
tikhomirov@497:  * Copyright (c) 2012 TMate Software Ltd
tikhomirov@497:  *  
tikhomirov@497:  * This program is free software; you can redistribute it and/or modify
tikhomirov@497:  * it under the terms of the GNU General Public License as published by
tikhomirov@497:  * the Free Software Foundation; version 2 of the License.
tikhomirov@497:  *
tikhomirov@497:  * This program is distributed in the hope that it will be useful,
tikhomirov@497:  * but WITHOUT ANY WARRANTY; without even the implied warranty of
tikhomirov@497:  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
tikhomirov@497:  * GNU General Public License for more details.
tikhomirov@497:  *
tikhomirov@497:  * For information on how to redistribute this software under
tikhomirov@497:  * the terms of a license other than GNU General Public License
tikhomirov@497:  * contact TMate Software at support@hg4j.com
tikhomirov@497:  */
tikhomirov@497: package org.tmatesoft.hg.internal;
tikhomirov@497: 
tikhomirov@497: import java.io.ByteArrayInputStream;
tikhomirov@497: import java.io.ByteArrayOutputStream;
tikhomirov@497: import java.io.IOException;
tikhomirov@497: import java.io.InputStream;
tikhomirov@497: import java.io.OutputStream;
tikhomirov@497: import java.util.ArrayList;
tikhomirov@497: import java.util.Collections;
tikhomirov@497: import java.util.HashSet;
tikhomirov@497: import java.util.Iterator;
tikhomirov@497: import java.util.LinkedHashMap;
tikhomirov@497: import java.util.LinkedHashSet;
tikhomirov@497: import java.util.List;
tikhomirov@497: 
tikhomirov@497: /**
tikhomirov@497:  * Simplistic parser to allow altering configuration files without touching user modifications/formatting/comments
tikhomirov@497:  * 
tikhomirov@497:  * @author Artem Tikhomirov
tikhomirov@497:  * @author TMate Software Ltd.
tikhomirov@497:  */
tikhomirov@497: public class ConfigFileParser {
tikhomirov@497: 	private enum ParseState {Initial, Section, Entry};
tikhomirov@497: 	private ParseState state = ParseState.Initial; 
tikhomirov@497: 	private int lastNonEmptyLineEndOffset = -1;
tikhomirov@497: 	private String sectionName;
tikhomirov@497: 	private int sectionStart = -1;
tikhomirov@497: 	private String entryKey;
tikhomirov@497: 	private int entryStart = -1;
tikhomirov@497: 	private int valueStart = -1, valueEnd = -1;
tikhomirov@497: 	private ArrayList<Entry> entries;
tikhomirov@497: 	private ArrayList<Section> sections = new ArrayList<Section>();
tikhomirov@497: 	private byte[] contents;
tikhomirov@497: 	
tikhomirov@497: 	private List<String> deletions = new ArrayList<String>(5);
tikhomirov@497: 	private List<String> additions = new ArrayList<String>(5), changes = new ArrayList<String>(5);
tikhomirov@497: 
tikhomirov@497: 	
tikhomirov@497: 	public boolean exists(String section, String key) {
tikhomirov@497: 		assert contents != null;
tikhomirov@497: 		for (Section s : sections) {
tikhomirov@497: 			if (s.name.equals(section)) {
tikhomirov@497: 				for (Entry e : s.entries) {
tikhomirov@497: 					if (e.name.equals(key)) {
tikhomirov@497: 						return true;
tikhomirov@497: 					}
tikhomirov@497: 				}
tikhomirov@497: 				return false;
tikhomirov@497: 			}
tikhomirov@497: 		}
tikhomirov@497: 		return false;
tikhomirov@497: 	}
tikhomirov@497: 	
tikhomirov@497: 	public void add(String section, String key, String newValue) {
tikhomirov@497: 		additions.add(section);
tikhomirov@497: 		additions.add(key);
tikhomirov@497: 		additions.add(newValue);
tikhomirov@497: 	}
tikhomirov@497: 	
tikhomirov@497: 	public void change(String section, String key, String newValue) {
tikhomirov@497: 		changes.add(section);
tikhomirov@497: 		changes.add(key);
tikhomirov@497: 		changes.add(newValue);
tikhomirov@497: 	}
tikhomirov@497: 	
tikhomirov@497: 	public void delete(String section, String key) {
tikhomirov@497: 		deletions.add(section);
tikhomirov@497: 		deletions.add(key);
tikhomirov@497: 	}
tikhomirov@497: 
tikhomirov@497: 	public void parse(InputStream is) throws IOException {
tikhomirov@497: 		state = ParseState.Initial;
tikhomirov@497: 		sections.clear();
tikhomirov@497: 		contents = null;
tikhomirov@497: 		ByteArrayOutputStream bos = new ByteArrayOutputStream(1024);
tikhomirov@497: 		ByteArrayOutputStream line = new ByteArrayOutputStream(80);
tikhomirov@497: 		int offset = 0;
tikhomirov@497: 		int lineOffset = -1;
tikhomirov@497: 		int lineNumber = 1;
tikhomirov@497: 		boolean crDetected = false; // true when previous char was \r
tikhomirov@497: 		int b;
tikhomirov@497: 		while ( (b = is.read()) != -1) {
tikhomirov@497: 			bos.write(b);
tikhomirov@497: 			if (b == '\n' || b == '\r') {
tikhomirov@497: 				if (line.size() > 0) {
tikhomirov@497: 					processLine(lineNumber, lineOffset, line.toByteArray());
tikhomirov@497: 					line.reset();
tikhomirov@497: 					lineOffset = -1;
tikhomirov@497: 					lastNonEmptyLineEndOffset = bos.size() - 1; // offset points to EOL char
tikhomirov@497: 				} 
tikhomirov@497: 				// else: XXX does empty line closes entry???
tikhomirov@497: 				// when \n follows \r, increment line count only once
tikhomirov@497: 				if (!(b == '\n' && crDetected)) {
tikhomirov@497: 					lineNumber++;
tikhomirov@497: 				}
tikhomirov@497: 				crDetected = b == '\r';
tikhomirov@497: 			} else {
tikhomirov@497: 				crDetected = false;
tikhomirov@497: 				if (line.size() == 0) {
tikhomirov@497: 					lineOffset = offset;
tikhomirov@497: 				}
tikhomirov@497: 				line.write(b);
tikhomirov@497: 			}
tikhomirov@497: 			offset++;
tikhomirov@497: 		}
tikhomirov@497: 		// handle last line in case it's not EOL-terminated
tikhomirov@497: 		if (line.size() > 0) {
tikhomirov@497: 			processLine(lineNumber, lineOffset, line.toByteArray());
tikhomirov@497: 			// might need it for #closeSection() below
tikhomirov@497: 			lastNonEmptyLineEndOffset = bos.size();
tikhomirov@497: 		}
tikhomirov@497: 		if (state == ParseState.Entry) {
tikhomirov@497: 			closeEntry();
tikhomirov@497: 		}
tikhomirov@497: 		if (state == ParseState.Section) {
tikhomirov@497: 			closeSection();
tikhomirov@497: 		}
tikhomirov@497: 		contents = bos.toByteArray();
tikhomirov@497: 	}
tikhomirov@497: 	
tikhomirov@497: 	public void update(OutputStream out) throws IOException {
tikhomirov@497: 		if (contents == null) {
tikhomirov@497: 			throw new IOException("Shall parse first");
tikhomirov@497: 		}
tikhomirov@497: 		HashSet<String> processedSections = new HashSet<String>();
tikhomirov@497: 		int contentsOffset = 0;
tikhomirov@497: 		for (Section section : sections) {
tikhomirov@497: 			LinkedHashMap<String,String> additionsInSection = new LinkedHashMap<String,String>();
tikhomirov@497: 			LinkedHashMap<String,String> changesInSection = new LinkedHashMap<String,String>();
tikhomirov@497: 			LinkedHashSet<String> deletionsInSection = new LinkedHashSet<String>();
tikhomirov@497: 			if (!processedSections.contains(section.name)) {
tikhomirov@497: 				for (Iterator<String> it = additions.iterator(); it.hasNext();) {
tikhomirov@497: 					String s = it.next(), k = it.next(), v = it.next();
tikhomirov@497: 					if (section.name.equals(s)) {
tikhomirov@497: 						additionsInSection.put(k, v);
tikhomirov@497: 					}
tikhomirov@497: 				}
tikhomirov@497: 				for (Iterator<String> it = changes.iterator(); it.hasNext();) {
tikhomirov@497: 					String s = it.next(), k = it.next(), v = it.next();
tikhomirov@497: 					if (section.name.equals(s)) {
tikhomirov@497: 						changesInSection.put(k, v);
tikhomirov@497: 					}
tikhomirov@497: 				}
tikhomirov@497: 				for (Iterator<String> it = deletions.iterator(); it.hasNext();) {
tikhomirov@497: 					String s = it.next(), k = it.next();
tikhomirov@497: 					if (section.name.equals(s)) {
tikhomirov@497: 						deletionsInSection.add(k);
tikhomirov@497: 					}
tikhomirov@497: 				}
tikhomirov@497: 			}
tikhomirov@497: 			for (Entry e : section.entries) {
tikhomirov@497: 				if (deletionsInSection.contains(e.name)) {
tikhomirov@497: 					// write up to key start only
tikhomirov@497: 					out.write(contents, contentsOffset, e.start - contentsOffset);
tikhomirov@497: 					contentsOffset = e.valueEnd + 1;
tikhomirov@497: 				} else if (changesInSection.containsKey(e.name)) {
tikhomirov@497: 					if (e.valueStart == -1) {
tikhomirov@497: 						// e.valueEnd determines insertion point
tikhomirov@497: 						out.write(contents, contentsOffset, e.valueEnd + 1 - contentsOffset);
tikhomirov@497: 					} else {
tikhomirov@497: 						// e.valueEnd points to last character of the value 
tikhomirov@497: 						out.write(contents, contentsOffset, e.valueStart - contentsOffset);
tikhomirov@497: 					}
tikhomirov@497: 					String value = changesInSection.get(e.name);
tikhomirov@497: 					out.write(value == null ? new byte[0] : value.getBytes());
tikhomirov@497: 					contentsOffset = e.valueEnd + 1;
tikhomirov@497: 				}
tikhomirov@497: 				// else: keep contentsOffset to point to first uncopied character
tikhomirov@497: 			}
tikhomirov@497: 			if (section.entries.length == 0) {
tikhomirov@497: 				// no entries, empty or only comments, perhaps.
tikhomirov@497: 				// use end of last meaningful line (whether [section] or comment string),
tikhomirov@497: 				// which points to newline character
tikhomirov@497: 				out.write(contents, contentsOffset, section.end - contentsOffset);
tikhomirov@497: 				contentsOffset = section.end;
tikhomirov@497: 				// since it's tricky to track \n or \r\n with lastNonEmptyLineEndOffset,
tikhomirov@497: 				// we copy up to the line delimiter and insert new lines, if any, with \n prepended,
tikhomirov@497: 				// so that original EOL will be moved to the very end of the section.
tikhomirov@497: 				// Indeed, would be better to insert *after* lastNonEmptyLineEndOffset,
tikhomirov@497: 				// but I don't want to complicate #parse (if line.size() > 0 part) method.
tikhomirov@497: 				// Hope, this won't make too much trouble (if any, at all - 
tikhomirov@497: 				// if String.format translates \n to system EOL, then nobody would notice)
tikhomirov@497: 			}
tikhomirov@497: 			if (!additionsInSection.isEmpty()) {
tikhomirov@497: 				// make sure additions are written once everything else is there
tikhomirov@497: 				out.write(contents, contentsOffset, section.end - contentsOffset);
tikhomirov@497: 				contentsOffset = section.end;
tikhomirov@497: 				for (String k : additionsInSection.keySet()) {
tikhomirov@497: 					String v = additionsInSection.get(k);
tikhomirov@497: 					out.write(String.format("\n%s = %s", k, v == null ? "" : v).getBytes());
tikhomirov@497: 				}
tikhomirov@497: 			}
tikhomirov@497: 			// if section comes more than once, update only first one.
tikhomirov@497: 			processedSections.add(section.name);
tikhomirov@497: 		}
tikhomirov@497: 		out.write(contents, contentsOffset, contents.length - contentsOffset);
tikhomirov@497: 	}
tikhomirov@497: 	
tikhomirov@497: 	private void processLine(int lineNumber, int offset, byte[] line) throws IOException {
tikhomirov@497: 		int localOffset = 0, i = 0;
tikhomirov@497: 		while (i < line.length && Character.isWhitespace(line[i])) {
tikhomirov@497: 			i++;
tikhomirov@497: 		}
tikhomirov@497: 		if (i == line.length) {
tikhomirov@497: 			return;
tikhomirov@497: 		}
tikhomirov@497: 		localOffset = i;
tikhomirov@497: 		if (line[i] == '[') {
tikhomirov@497: 			if (state == ParseState.Entry) {
tikhomirov@497: 				closeEntry();
tikhomirov@497: 			}
tikhomirov@497: 			if (state == ParseState.Section) {
tikhomirov@497: 				closeSection();
tikhomirov@497: 			}
tikhomirov@497: 			
tikhomirov@497: 			while (i < line.length && line[i] != ']') {
tikhomirov@497: 				i++;
tikhomirov@497: 			}
tikhomirov@497: 			if (i == line.length) {
tikhomirov@497: 				throw new IOException(String.format("Can't find closing ']' for section name in line %d", lineNumber));
tikhomirov@497: 			}
tikhomirov@497: 			sectionName = new String(line, localOffset+1, i-localOffset-1);
tikhomirov@497: 			sectionStart = offset + localOffset;
tikhomirov@497: 			state = ParseState.Section;
tikhomirov@497: 		} else if (line[i] == '#' || line[i] == ';') {
tikhomirov@497: 			// comment line, nothing to process
tikhomirov@497: 			return;
tikhomirov@497: 		} else {
tikhomirov@497: 			// entry
tikhomirov@497: 			if (state == ParseState.Initial) {
tikhomirov@497: 				throw new IOException(String.format("Line %d doesn't belong to any section", lineNumber));
tikhomirov@497: 			}
tikhomirov@497: 			if (localOffset > 0) {
tikhomirov@497: 				if (state == ParseState.Section) {
tikhomirov@497: 					throw new IOException(String.format("Non-indented key is expected in line %d", lineNumber));
tikhomirov@497: 				}
tikhomirov@497: 				assert state == ParseState.Entry;
tikhomirov@497: 				// whitespace-indented continuation of the previous entry  
tikhomirov@497: 				if (valueStart == -1) {
tikhomirov@497: 					// value didn't start at the same line the key was found at
tikhomirov@497: 					valueStart = offset + localOffset;
tikhomirov@497: 				}
tikhomirov@497: 				// value ends with eol (assumption is trailing comments are not allowed)
tikhomirov@497: 				valueEnd = offset + line.length - 1;
tikhomirov@497: 			} else {
tikhomirov@497: 				if (state == ParseState.Entry) {
tikhomirov@497: 					closeEntry();
tikhomirov@497: 				}
tikhomirov@497: 				assert state == ParseState.Section;
tikhomirov@497: 				// it's a new entry
tikhomirov@497: 				state  = ParseState.Entry;
tikhomirov@497: 				// get name of the entry
tikhomirov@497: 				while (i < line.length && !Character.isWhitespace(line[i]) && line[i] != '=') {
tikhomirov@497: 					i++;
tikhomirov@497: 				}
tikhomirov@497: 				if (i == line.length) {
tikhomirov@497: 					throw new IOException(String.format("Can't process entry in line %d", lineNumber));
tikhomirov@497: 				}
tikhomirov@497: 				entryKey = new String(line, localOffset, i - localOffset);
tikhomirov@497: 				entryStart = offset + localOffset;
tikhomirov@497: 				// look for '=' after key name
tikhomirov@497: 				while (i < line.length && line[i] != '=') {
tikhomirov@497: 					i++;
tikhomirov@497: 				}
tikhomirov@497: 				if (i == line.length) {
tikhomirov@497: 					throw new IOException(String.format("Can't find '=' after key %s in line %d", entryKey, lineNumber));
tikhomirov@497: 				}
tikhomirov@497: 				// skip whitespaces after '='
tikhomirov@497: 				i++; // line[i] == '='
tikhomirov@497: 				while (i < line.length && Character.isWhitespace(line[i])) {
tikhomirov@497: 					i++;
tikhomirov@497: 				}
tikhomirov@497: 				// valueStart might be -1 in case no value is specified in the same line as key
tikhomirov@497: 				// but valueEnd is always initialized just in case there's no next, value continuation line
tikhomirov@497: 				if (i == line.length) {
tikhomirov@497: 					valueStart = -1;
tikhomirov@497: 				} else {
tikhomirov@497: 					valueStart = offset + i;
tikhomirov@497: 				}
tikhomirov@497: 				
tikhomirov@497: 				// if trailing comments are allowed, shall
tikhomirov@497: 				// look up comment char and set valueEnd to its position-1
tikhomirov@497: 				valueEnd = offset + line.length - 1;
tikhomirov@497: 			}
tikhomirov@497: 		}
tikhomirov@497: 	}
tikhomirov@497: 	
tikhomirov@497: 	private void closeSection() {
tikhomirov@497: 		assert state == ParseState.Section;
tikhomirov@497: 		assert sectionName != null;
tikhomirov@497: 		assert lastNonEmptyLineEndOffset != -1;
tikhomirov@497: 		Section s = new Section(sectionName, sectionStart, lastNonEmptyLineEndOffset, entries == null ? Collections.<Entry>emptyList() : entries);
tikhomirov@497: 		sections.add(s);
tikhomirov@497: 		sectionName = null;
tikhomirov@497: 		sectionStart = -1;
tikhomirov@497: 		state = ParseState.Initial;
tikhomirov@497: 		entries = null;
tikhomirov@497: 	}
tikhomirov@497: 	
tikhomirov@497: 	private void closeEntry() {
tikhomirov@497: 		assert state == ParseState.Entry;
tikhomirov@497: 		assert entryKey != null;
tikhomirov@497: 		state = ParseState.Section;
tikhomirov@497: 		Entry e = new Entry(entryKey, entryStart, valueStart, valueEnd);
tikhomirov@497: 		if (entries == null) {
tikhomirov@497: 			entries = new ArrayList<Entry>();
tikhomirov@497: 		}
tikhomirov@497: 		entries.add(e);
tikhomirov@497: 		entryKey = null;
tikhomirov@497: 		entryStart = valueStart = valueEnd -1;
tikhomirov@497: 	}
tikhomirov@497: 
tikhomirov@497: 	
tikhomirov@497: 	private static class Block {
tikhomirov@497: 		public final int start;
tikhomirov@497: 		Block(int s) {
tikhomirov@497: 			start = s;
tikhomirov@497: 		}
tikhomirov@497: 	}
tikhomirov@497: 	
tikhomirov@497: 	private static class Entry extends Block {
tikhomirov@497: 		public final int valueStart, valueEnd;
tikhomirov@497: 		public final String name;
tikhomirov@497: 		
tikhomirov@497: 		Entry(String n, int s, int vs, int ve) {
tikhomirov@497: 			super(s);
tikhomirov@497: 			name = n;
tikhomirov@497: 			valueStart = vs;
tikhomirov@497: 			valueEnd = ve;
tikhomirov@497: 		}
tikhomirov@497: 	}
tikhomirov@497: 	
tikhomirov@497: 	private static class Section extends Block {
tikhomirov@497: 		public final String name;
tikhomirov@497: 		public final Entry[] entries;
tikhomirov@497: 		public final int end;
tikhomirov@497: 
tikhomirov@497: 		Section(String n, int s, int endOffset, List<Entry> e) {
tikhomirov@497: 			super(s);
tikhomirov@497: 			name = n;
tikhomirov@497: 			end = endOffset;
tikhomirov@497: 			entries = new Entry[e.size()];
tikhomirov@497: 			e.toArray(entries);
tikhomirov@497: 		}
tikhomirov@497: 	}
tikhomirov@497: 
tikhomirov@497: 	public static void main(String[] args) throws Exception {
tikhomirov@497: 		ConfigFileParser p = new ConfigFileParser();
tikhomirov@497: 		p.parse(new ByteArrayInputStream(xx.getBytes()));
tikhomirov@497: 		System.out.println(">>>");
tikhomirov@497: 		System.out.println(xx);
tikhomirov@497: 		System.out.println("===");
tikhomirov@497: 		p.add("sect1", "key5", "x");
tikhomirov@497: 		ByteArrayOutputStream out = new ByteArrayOutputStream(xx.length());
tikhomirov@497: 		p.update(out);
tikhomirov@497: 		System.out.println(new String(out.toByteArray()));
tikhomirov@497: 		/*
tikhomirov@497: 		for (Section s : p.sections) {
tikhomirov@497: 			System.out.printf("[%s@%d]\n", s.name, s.start);
tikhomirov@497: 			for (Entry e : s.entries) {
tikhomirov@497: 				System.out.printf("%s@%d = %d..%d\n", e.name, e.start, e.valueStart, e.valueEnd);
tikhomirov@497: 			}
tikhomirov@497: 		}
tikhomirov@497: 		*/
tikhomirov@497: 	}
tikhomirov@497: 	private static final String xx = "#comment1\n [sect1]\nkey = value #not a comment2\n#comment3\nkey2=   \nkey3 =  \n  value1, #cc\n  value2\nkey4 = v1,\n  v2 \n  ,v3\n\n\n[sect2]\nx = a";
tikhomirov@497: }