/*******************************************************************************
 * Copyright (c) 2000, 2013 IBM Corporation and others.
 *
 * This program and the accompanying materials
 * are made available under the terms of the Eclipse Public License 2.0
 * which accompanies this distribution, and is available at
 * https://www.eclipse.org/legal/epl-2.0/
 *
 * SPDX-License-Identifier: EPL-2.0
 *
 * Contributors:
 *     IBM Corporation - initial API and implementation
 *     Christian Walther (Indel AG) - Bug 402009: Disallow "whole word" together with regex
 *******************************************************************************/
package org.eclipse.search.internal.core.text;

import java.util.regex.Pattern;
import java.util.regex.PatternSyntaxException;

import org.eclipse.core.runtime.Assert;

import org.eclipse.jface.text.FindReplaceDocumentAdapter;

import org.eclipse.search.internal.ui.SearchMessages;

/**
 *
 */
public class PatternConstructor {


	private PatternConstructor() {
		// don't instantiate
	}

	public static Pattern createPattern(String pattern, boolean isCaseSensitive, boolean isRegex) throws PatternSyntaxException {
		return createPattern(pattern, isRegex, true, isCaseSensitive, false);
	}

	/**
	 * Creates a pattern element from the pattern string which is either a reg-ex expression or in
	 * our old 'StringMatcher' format.
	 *
	 * @param pattern The search pattern
	 * @param isRegex <code>true</code> if the passed string already is a reg-ex pattern
	 * @param isStringMatcher <code>true</code> if the passed string is in the StringMatcher format.
	 * @param isCaseSensitive Set to <code>true</code> to create a case insensitive pattern
	 * @param isWholeWord <code>true</code> to create a pattern that requires a word boundary at the
	 *            beginning and the end.
	 * @return The created pattern
	 * @throws PatternSyntaxException if "\R" is at an illegal position
	 */
	public static Pattern createPattern(String pattern, boolean isRegex, boolean isStringMatcher, boolean isCaseSensitive, boolean isWholeWord) throws PatternSyntaxException {
		if (isRegex) {
			pattern= substituteLinebreak(pattern);
			Assert.isTrue(!isWholeWord, "isWholeWord unsupported together with isRegex"); //$NON-NLS-1$
		} else {
			int len= pattern.length();
			StringBuilder buffer= new StringBuilder(len + 10);
			// don't add a word boundary if the search text does not start with
			// a word char. (this works around a user input error).
			if (isWholeWord && len > 0 && isWordChar(pattern.charAt(0))) {
				buffer.append("\\b"); //$NON-NLS-1$
			}
			appendAsRegEx(isStringMatcher, pattern, buffer);
			if (isWholeWord && len > 0 && isWordChar(pattern.charAt(len - 1))) {
				buffer.append("\\b"); //$NON-NLS-1$
			}
			pattern= buffer.toString();
		}

		int regexOptions= Pattern.MULTILINE;
		if (!isCaseSensitive) {
			regexOptions|= Pattern.CASE_INSENSITIVE | Pattern.UNICODE_CASE;
		}
		return Pattern.compile(pattern, regexOptions);
	}

	/**
	 * Copied from {@link org.eclipse.jface.text.FindReplaceDocumentAdapter}' to support '\R'
	 *
	 * @param findString the string to substitute
	 * @return the new string
	 * @throws PatternSyntaxException if "\R" is at an illegal position
	 */
	private static String substituteLinebreak(String findString) throws PatternSyntaxException {
		int length= findString.length();
		StringBuilder buf= new StringBuilder(length);

		int inCharGroup= 0;
		int inBraces= 0;
		boolean inQuote= false;
		for (int i= 0; i < length; i++) {
			char ch= findString.charAt(i);
			switch (ch) {
				case '[':
					buf.append(ch);
					if (! inQuote)
						inCharGroup++;
					break;

				case ']':
					buf.append(ch);
					if (! inQuote)
						inCharGroup--;
					break;

				case '{':
					buf.append(ch);
					if (! inQuote && inCharGroup == 0)
						inBraces++;
					break;

				case '}':
					buf.append(ch);
					if (! inQuote && inCharGroup == 0)
						inBraces--;
					break;

				case '\\':
					if (i + 1 < length) {
						char ch1= findString.charAt(i + 1);
						if (inQuote) {
							if (ch1 == 'E')
								inQuote= false;
							buf.append(ch).append(ch1);
							i++;

						} else if (ch1 == 'R') {
							if (inCharGroup > 0 || inBraces > 0) {
								String msg= SearchMessages.PatternConstructor_error_line_delim_position;
								throw new PatternSyntaxException(msg, findString, i);
							}
							buf.append("(?>\\r\\n?|\\n)"); //$NON-NLS-1$
							i++;

						} else {
							if (ch1 == 'Q') {
								inQuote= true;
							}
							buf.append(ch).append(ch1);
							i++;
						}
					} else {
						buf.append(ch);
					}
					break;

				default:
					buf.append(ch);
					break;
			}

		}
		return buf.toString();
	}


	private static boolean isWordChar(char c) {
		return Character.isLetterOrDigit(c);
	}

	/**
	 * Creates a pattern element from an array of patterns in the old 'StringMatcher' format.
	 *
	 * @param patterns The search patterns
	 * @param isCaseSensitive Set to <code>true</code> to create a case insensitive pattern
	 * @return The created pattern
	 * @throws PatternSyntaxException if "\R" is at an illegal position
	 */
	public static Pattern createPattern(String[] patterns, boolean isCaseSensitive) throws PatternSyntaxException {
		StringBuilder pattern= new StringBuilder();
		for (int i= 0; i < patterns.length; i++) {
			if (i > 0) {
				// note that this works only as we know that the operands of the
				// or expression will be simple and need no brackets.
				pattern.append('|');
			}
			appendAsRegEx(true, patterns[i], pattern);
		}
		return createPattern(pattern.toString(), true, true, isCaseSensitive, false);
	}


	public static StringBuilder appendAsRegEx(boolean isStringMatcher, String pattern, StringBuilder buffer) {
		boolean isEscaped= false;
		for (int i = 0; i < pattern.length(); i++) {
			char c = pattern.charAt(i);
			switch(c) {
			// the backslash
			case '\\':
				// the backslash is escape char in string matcher
				if (isStringMatcher && !isEscaped) {
					isEscaped= true;
				}
				else {
					buffer.append("\\\\");  //$NON-NLS-1$
					isEscaped= false;
				}
				break;
			// characters that need to be escaped in the regex.
			case '(':
			case ')':
			case '{':
			case '}':
			case '.':
			case '[':
			case ']':
			case '$':
			case '^':
			case '+':
			case '|':
				if (isEscaped) {
					buffer.append("\\\\");  //$NON-NLS-1$
					isEscaped= false;
				}
				buffer.append('\\');
				buffer.append(c);
				break;
			case '?':
				if (isStringMatcher && !isEscaped) {
					buffer.append('.');
				}
				else {
					buffer.append('\\');
					buffer.append(c);
					isEscaped= false;
				}
				break;
			case '*':
				if (isStringMatcher && !isEscaped) {
					buffer.append(".*"); //$NON-NLS-1$
				}
				else {
					buffer.append('\\');
					buffer.append(c);
					isEscaped= false;
				}
				break;
			default:
				if (isEscaped) {
					buffer.append("\\\\");  //$NON-NLS-1$
					isEscaped= false;
				}
				buffer.append(c);
				break;
			}
		}
		if (isEscaped) {
			buffer.append("\\\\");  //$NON-NLS-1$
			isEscaped= false;
		}
		return buffer;
	}

	/**
	 * Interprets escaped characters in the given replace pattern.
	 *
	 * @param replaceText the replace pattern
	 * @param foundText the found pattern to be replaced
	 * @param lineDelim the line delimiter to use for \R
	 * @return a replace pattern with escaped characters substituted by the respective characters
	 * @since 3.4
	 */
	public static String interpretReplaceEscapes(String replaceText, String foundText, String lineDelim) {
		return new ReplaceStringConstructor(lineDelim).interpretReplaceEscapes(replaceText, foundText);
	}

	/**
	 * Copied from {@link FindReplaceDocumentAdapter}}
	 *
	 * FindReplaceDocumentAdapter with contributions from:
	 * Cagatay Calli <ccalli@gmail.com> - [find/replace] retain caps when replacing - https://bugs.eclipse.org/bugs/show_bug.cgi?id=28949
	 * Cagatay Calli <ccalli@gmail.com> - [find/replace] define & fix behavior of retain caps with other escapes and text before \C - https://bugs.eclipse.org/bugs/show_bug.cgi?id=217061
	 */
	private static class ReplaceStringConstructor {

		private static final int RC_MIXED= 0;
		private static final int RC_UPPER= 1;
		private static final int RC_LOWER= 2;
		private static final int RC_FIRSTUPPER= 3;


		private int fRetainCaseMode;
		private final String fLineDelim;

		public ReplaceStringConstructor(String lineDelim) {
			fLineDelim= lineDelim;

		}

		/**
		 * Interprets escaped characters in the given replace pattern.
		 *
		 * @param replaceText the replace pattern
		 * @param foundText the found pattern to be replaced
		 * @return a replace pattern with escaped characters substituted by the respective characters
		 * @since 3.4
		 */
		private String interpretReplaceEscapes(String replaceText, String foundText) {
			int length= replaceText.length();
			boolean inEscape= false;
			StringBuilder buf= new StringBuilder(length);

			/* every string we did not check looks mixed at first
			 * so initialize retain case mode with RC_MIXED
			 */
			fRetainCaseMode= RC_MIXED;

			for (int i= 0; i < length; i++) {
				final char ch= replaceText.charAt(i);
				if (inEscape) {
					i= interpretReplaceEscape(ch, i, buf, replaceText, foundText);
					inEscape= false;

				} else if (ch == '\\') {
					inEscape= true;

				} else if (ch == '$') {
					buf.append(ch);

					/*
					 * Feature in java.util.regex.Matcher#replaceFirst(String):
					 * $00, $000, etc. are interpreted as $0 and
					 * $01, $001, etc. are interpreted as $1, etc. .
					 * If we support \0 as replacement pattern for capturing group 0,
					 * it would not be possible any more to write a replacement pattern
					 * that appends 0 to a capturing group (like $0\0).
					 * The fix is to interpret \00 and $00 as $0\0, and
					 * \01 and $01 as $0\1, etc.
					 */
					if (i + 2 < length) {
						char ch1= replaceText.charAt(i + 1);
						char ch2= replaceText.charAt(i + 2);
						if (ch1 == '0' && '0' <= ch2 && ch2 <= '9') {
							buf.append("0\\"); //$NON-NLS-1$
							i++; // consume the 0
						}
					}
				} else {
					interpretRetainCase(buf, ch);
				}
			}

			if (inEscape) {
				// '\' as last character is invalid, but we still add it to get an error message
				buf.append('\\');
			}
			return buf.toString();
		}

		/**
		 * Interprets the escaped character <code>ch</code> at offset <code>i</code>
		 * of the <code>replaceText</code> and appends the interpretation to <code>buf</code>.
		 *
		 * @param ch the escaped character
		 * @param i the offset
		 * @param buf the output buffer
		 * @param replaceText the original replace pattern
		 * @param foundText the found pattern to be replaced
		 * @return the new offset
		 * @since 3.4
		 */
		private int interpretReplaceEscape(final char ch, int i, StringBuilder buf, String replaceText, String foundText) {
			int length= replaceText.length();
			switch (ch) {
				case 'r':
					buf.append('\r');
					break;
				case 'n':
					buf.append('\n');
					break;
				case 't':
					buf.append('\t');
					break;
				case 'f':
					buf.append('\f');
					break;
				case 'a':
					buf.append('\u0007');
					break;
				case 'e':
					buf.append('\u001B');
					break;
				case 'R': //see http://www.unicode.org/unicode/reports/tr18/#Line_Boundaries
					buf.append(fLineDelim);
					break;
				/*
				 * \0 for octal is not supported in replace string, since it
				 * would conflict with capturing group \0, etc.
				 */
				case '0':
					buf.append('$').append(ch);
					/*
					 * See explanation in "Feature in java.util.regex.Matcher#replaceFirst(String)"
					 * in interpretReplaceEscape(String) above.
					 */
					if (i + 1 < length) {
						char ch1= replaceText.charAt(i + 1);
						if ('0' <= ch1 && ch1 <= '9') {
							buf.append('\\');
						}
					}
					break;

				case '1':
				case '2':
				case '3':
				case '4':
				case '5':
				case '6':
				case '7':
				case '8':
				case '9':
					buf.append('$').append(ch);
					break;

				case 'c':
					if (i + 1 < length) {
						char ch1= replaceText.charAt(i + 1);
						interpretRetainCase(buf, (char)(ch1 ^ 64));
						i++;
					} else {
						String msg= SearchMessages.PatternConstructor_error_escape_sequence;
						throw new PatternSyntaxException(msg, replaceText, i);
					}
					break;

				case 'x':
					if (i + 2 < length) {
						int parsedInt;
						try {
							parsedInt= Integer.parseInt(replaceText.substring(i + 1, i + 3), 16);
							if (parsedInt < 0)
								throw new NumberFormatException();
						} catch (NumberFormatException e) {
							String msg= SearchMessages.PatternConstructor_error_hex_escape_sequence;
							throw new PatternSyntaxException(msg, replaceText, i);
						}
						interpretRetainCase(buf, (char) parsedInt);
						i+= 2;
					} else {
						String msg= SearchMessages.PatternConstructor_error_hex_escape_sequence;
						throw new PatternSyntaxException(msg, replaceText, i);
					}
					break;

				case 'u':
					if (i + 4 < length) {
						int parsedInt;
						try {
							parsedInt= Integer.parseInt(replaceText.substring(i + 1, i + 5), 16);
							if (parsedInt < 0)
								throw new NumberFormatException();
						} catch (NumberFormatException e) {
							String msg= SearchMessages.PatternConstructor_error_unicode_escape_sequence;
							throw new PatternSyntaxException(msg, replaceText, i);
						}
						interpretRetainCase(buf, (char) parsedInt);
						i+= 4;
					} else {
						String msg= SearchMessages.PatternConstructor_error_unicode_escape_sequence;
						throw new PatternSyntaxException(msg, replaceText, i);
					}
					break;

				case 'C':
					if(foundText.toUpperCase().equals(foundText)) // is whole match upper-case?
						fRetainCaseMode= RC_UPPER;
					else if (foundText.toLowerCase().equals(foundText)) // is whole match lower-case?
						fRetainCaseMode= RC_LOWER;
					else if(Character.isUpperCase(foundText.charAt(0))) // is first character upper-case?
						fRetainCaseMode= RC_FIRSTUPPER;
					else
						fRetainCaseMode= RC_MIXED;
					break;

				default:
					// unknown escape k: append uninterpreted \k
					buf.append('\\').append(ch);
					break;
			}
			return i;
		}

		/**
		 * Interprets current Retain Case mode (all upper-case,all lower-case,capitalized or mixed)
		 * and appends the character <code>ch</code> to <code>buf</code> after processing.
		 *
		 * @param buf the output buffer
		 * @param ch the character to process
		 * @since 3.4
		 */
		private void interpretRetainCase(StringBuilder buf, char ch) {
			if (fRetainCaseMode == RC_UPPER)
				buf.append(String.valueOf(ch).toUpperCase());
			else if (fRetainCaseMode == RC_LOWER)
				buf.append(String.valueOf(ch).toLowerCase());
			else if (fRetainCaseMode == RC_FIRSTUPPER) {
				buf.append(String.valueOf(ch).toUpperCase());
				fRetainCaseMode= RC_MIXED;
			} else
				buf.append(ch);
		}

	}
}