tabs.java

来自「java操作excel的类」· Java 代码 · 共 613 行 · 第 1/2 页

	 * Read form the input stream, changing the tabs at the beginning of each line
	 * to the specified number of spaces or the other way around, write the result
	 * to the output stream.
	 *
	 * The current system's line separator is used.
	 *
	 * @param in stream that contains the text which needs line number conversion.
	 * @param out stream where converted text is written.
	 * @param inputTabWidth number of spaces used instead of a tab in the input.
	 * @param outputTabWidth TABS if tabs should be used, otherwise, number of spaces to use.
	 * @return true if the output was modified from the input, false if it is exactly the same
	 * @throws BinaryDataException if non-text data is encountered.
	 * @throws IOException if an input or output error occurs.
	 *
	 * @since ostermillerutils 1.00.00
	 */
	public static boolean convert(InputStream in, OutputStream out, int inputTabWidth, int outputTabWidth) throws IOException {
		return convert(in, out, inputTabWidth, outputTabWidth, DEFAULT_MODIFY_BINARY);
	}

	/**
	 * Read form the input stream, changing the tabs at the beginning of each line
	 * to the specified number of spaces or the other way around, write the result
	 * to the output stream.
	 *
	 * The current system's line separator is used.
	 *
	 * @param in stream that contains the text which needs line number conversion.
	 * @param out stream where converted text is written.
	 * @param inputTabWidth number of spaces used instead of a tab in the input.
	 * @param outputTabWidth TABS if tabs should be used, otherwise, number of spaces to use.
	 * @param binaryException throw an exception and abort the operation if binary data is encountered and binaryExcepion is false.
	 * @return true if the output was modified from the input, false if it is exactly the same.
	 * @throws BinaryDataException if non-text data is encountered.
	 * @throws IOException if an input or output error occurs.
	 *
	 * @since ostermillerutils 1.00.00
	 */
	public static boolean convert(InputStream in, OutputStream out, int inputTabWidth, int outputTabWidth, boolean binaryException) throws IOException {
		if ((inputTabWidth < 1 || inputTabWidth > 20) && inputTabWidth != TABS){
			throw new IllegalArgumentException(labels.getString("widtherror"));
		}
		if ((outputTabWidth < 1 || outputTabWidth > 20) && outputTabWidth != TABS){
			throw new IllegalArgumentException(labels.getString("widtherror"));
		}
		int state = STATE_INIT;
		int spaces = 0;
		int tabs = 0;
		int tabStops = 0;
		int extraSpaces = 0;
		boolean modified = false;

		byte[] buffer = new byte[BUFFER_SIZE];
		int read;
		while((read = in.read(buffer)) != -1){
			for (int i=0; i<read; i++){
				byte b = buffer[i];
				if(binaryException && b!='\r' && b!='\n' && b!='\t' && b!='\f' && (b & 0xff)<32){
					throw new BinaryDataException(labels.getString("binaryexcepion"));
				}
				switch (b){
					case ' ': {
						if (state == STATE_INIT) {
							spaces++;
							extraSpaces++;
							if (extraSpaces == inputTabWidth){
								tabStops++;
								extraSpaces = 0;
							}
						} else {
							out.write(b);
						}
					} break;
					case '\t': {
						if (state == STATE_INIT) {
							if (spaces > 0){
								// put tabs before spaces
								modified = true;
							}
							tabs++;
							tabStops++;
							extraSpaces = 0;
						} else {
							out.write(b);
						}
					} break;
					case '\r': case '\n': {
						out.write(b);
						spaces = 0;
						tabs = 0;
						tabStops = 0;
						extraSpaces = 0;
						state = STATE_INIT;
					} break;
					default: {
						if (state == STATE_INIT){
							if (outputTabWidth == TABS){
								for (int j=0; j<tabStops; j++){
									out.write((byte)'\t');
								}
							} else {
								extraSpaces += tabStops * outputTabWidth;
								tabStops = 0;
							}
							for (int j=0; j<extraSpaces; j++){
								out.write((byte)' ');
							}
							if (extraSpaces != spaces || tabStops != tabs) modified = true;
						}
						out.write(b);
						state = STATE_SOMETHING;
					} break;
				}
			}
		}
		return modified;
	}

	/**
	 * Change the tabs at the beginning of each line of the file to four spaces.
	 * Guess the tab width of the input file.
	 *
	 * @param f File to be converted.
	 * @return true if the file was modified, false if it was already in the correct format
	 * @throws BinaryDataException if non-text data is encountered.
	 * @throws IOException if an input or output error occurs.
	 *
	 * @since ostermillerutils 1.00.00
	 */
	public static boolean convert(File f) throws IOException {
		return convert(f, DEFAULT_INPUT_FILE_TAB_WIDTH, DEFAULT_OUTPUT_TAB_WIDTH, DEFAULT_MODIFY_BINARY);
	}

	/**
	 * Change the tabs at the beginning of each line of the file
	 * to the specified number of spaces.
	 *
	 * @param f File to be converted.
	 * @param inputTabWidth number of spaces used instead of a tab in the input, or TAB to guess.
	 * @return true if the output was modified from the input, false if it is exactly the same
	 * @throws BinaryDataException if non-text data is encountered.
	 * @throws IOException if an input or output error occurs.
	 * @throws IllegalArgumentException if tab widths are not between 1 and 20 or TABS.
	 *
	 * @since ostermillerutils 1.00.00
	 */
	public static boolean convert(File f, int inputTabWidth) throws IOException {
		return convert(f, inputTabWidth, DEFAULT_OUTPUT_TAB_WIDTH, DEFAULT_MODIFY_BINARY);
	}

	/**
	 * Change the tabs at the beginning of each line of the file
	 * to the specified number of spaces or the other way around.
	 *
	 * @param f File to be converted.
	 * @param inputTabWidth number of spaces used instead of a tab in the input, or TAB to guess.
	 * @param outputTabWidth true if tabs should be used, false if spaces should be used.
	 * @return true if the output was modified from the input, false if it is exactly the same
	 * @throws BinaryDataException if non-text data is encountered.
	 * @throws IOException if an input or output error occurs.
	 *
	 * @since ostermillerutils 1.00.00
	 */
	public static boolean convert(File f, int inputTabWidth, int outputTabWidth) throws IOException {
		return convert(f, inputTabWidth, outputTabWidth, DEFAULT_MODIFY_BINARY);
	}

	/**
	 * Change the tabs at the beginning of each line of the file
	 * to the specified number of spaces or the other way around.
	 *
	 * @param f File to be converted.
	 * @param inputTabWidth number of spaces used instead of a tab in the input, or TABS to guess.
	 * @param outputTabWidth true if tabs should be used, false if spaces should be used.
	 * @param binaryException throw an exception and abort the operation if binary data is encountered and binaryExcepion is false.
	 * @return true if the file was modified, false if it was already in the correct format
	 * @throws BinaryDataException if non-text data is encountered.
	 * @throws IOException if an input or output error occurs.
	 * @throws IllegalArgumentException if tab widths are not between 1 and 20 or TABS.
	 *
	 * @since ostermillerutils 1.00.00
	 */
	public static boolean convert(File f, int inputTabWidth, int outputTabWidth, boolean binaryException) throws IOException {
		File temp = null;
		InputStream in = null;
		OutputStream out = null;
		boolean modified = false;
		try {
			if (inputTabWidth == TABS){
				inputTabWidth = guessTabWidth(new FileInputStream(f));
			}
			in = new FileInputStream(f);
			temp = File.createTempFile("LineEnds", null, null);
			out = new FileOutputStream(temp);
			modified = convert(in, out, inputTabWidth, outputTabWidth, binaryException);
			in.close();
			in = null;
			out.flush();
			out.close();
			out = null;
			if (modified){
				FileHelper.move(temp, f, true);
			} else {
				if (!temp.delete()){
					throw new IOException(
						MessageFormat.format(
							labels.getString("tempdeleteerror"),
							new String[] {temp.toString()}
						)
					);
				}
			}
		} finally {
			if (in != null){
				in.close();
				in = null;
			}
			if (out != null){
				out.flush();
				out.close();
				out = null;
			}
		}
		return modified;
	}

	/**
	 * Buffer size when reading from input stream.
	 *
	 * @since ostermillerutils 1.00.00
	 */
	private final static int BUFFER_SIZE = 1024;
	private final static int STATE_INIT = 0;
	private final static int STATE_SOMETHING = 1;

	final private static int MAX_SPACES = 128;
	final private static int MAX_TABS = 16;
	final private static int MAX_COMBINED = 256;
	/**
	 * Guess the number of spaces per tab at the beginning of each line.
	 *
	 * @return the least value (two or greater) which has some line that starts with n times spaces for n zero to max spaces starting a line.
	 *
	 * @since ostermillerutils 1.00.00
	 */
	public static int guessTabWidth(InputStream in) throws IOException {
		byte[] buffer = new byte[BUFFER_SIZE];
		int[][] data = new int[MAX_SPACES][MAX_TABS];
		int[] spaceData = new int[MAX_SPACES*MAX_TABS];
		int read;
		int state = STATE_INIT;
		int tabs = 0;
		int spaces = 0;
		int mostTabs = 0;
		int mostSpaces = 0;
		boolean spaceUsed = false;
		while((read = in.read(buffer)) != -1){
			for (int i=0; i<read; i++){
				byte b = buffer[i];
				switch (b){
					case ' ': {
						if (state == STATE_INIT) spaces++;
					} break;
					case '\t': {
						if (state == STATE_INIT) tabs++;
					} break;
					case '\r': case '\n': {
						state = STATE_INIT;
						if (spaces < MAX_SPACES && tabs < MAX_TABS){
							data[spaces][tabs]++;
							if (tabs > mostTabs) mostTabs = tabs;
							if (spaces > mostSpaces) mostSpaces = spaces;
							spaces = 0;
							tabs = 0;
						}
					} break;
					default: {
						state = STATE_SOMETHING;
					} break;
				}
			}
		}
		for (int tabWidth=2; tabWidth<=20; tabWidth++){
			int mostCombined=0;
			for (int tabInd=0; tabInd <= mostTabs; tabInd++){
				for (int spaceInd=0; spaceInd <= mostSpaces; spaceInd++){
					int totInd = spaceInd + (tabInd * tabWidth);
					if (totInd < MAX_COMBINED){
						int numLines = data[spaceInd][tabInd];
						if (numLines > 0){
							if (mostCombined < totInd) mostCombined = totInd;
							spaceData[totInd] += numLines;
						}
					}
				}
			}
			boolean found = true;
			for(int combInd=0; found && combInd < mostCombined; combInd+=tabWidth){
				found = spaceData[combInd] > 0;
			}
			if (found) return tabWidth;
		}
		return 2;
	}
}