/**
* This function composes the perl regular expression to perform pattern matching with negative
* look ahead. This becomes handy when a replacement string includes some string that is also a
* pattern.
*
* <p>i.e.) pattern - & replacement - ", &
*
* @param pattern - pattern string to match
* @param replacements - an array of replacement strings
* @return returns the format pattern for perl5 negative look-ahead
* @exception throws a FrameworkException when the pattern/replacement is null.
*/
private static String makePerl5MatchPatternNegativeLookAhead(
String pattern, String replacements[]) throws FrameworkException {
if (pattern == null || replacements.length < 0) {
throw new FrameworkException(
"RegexUtils: makePerl5MatchPatternNegativeLookAhead: pattern or replacement is null."
+ "\npattern = "
+ pattern
+ "\nreplacements.length = "
+ replacements.length);
}
String result = null;
StringBuffer patternBuffer = new StringBuffer();
Perl5Util util = new Perl5Util();
String formatPattern = makePerl5MatchPattern(pattern);
for (int i = 0; i < replacements.length; i++) {
if (util.match(formatPattern, replacements[i])) {
if (replacements[i].startsWith(pattern)) {
result = util.postMatch();
// very first one
if (i == 0) {
patternBuffer.append(pattern);
patternBuffer.append("(?!");
}
patternBuffer.append(result);
// the last one
if (i == (replacements.length - 1)) {
patternBuffer.append(")");
} else // not the last one, cat the perl5 separater
{
patternBuffer.append("|");
}
} else {
throw new FrameworkException(
"ERROR: RegexUtils: makePerl5MatchPatternNegativeLookAhead: The pattern in the "
+ "replacement string should be at the beginning.");
}
} else // no match found meaning invalid use of this function
{
throw new FrameworkException(
"ERROR: RegexUtils: makePerl5MatchPatternNegativeLookAhead: "
+ "Invalid use of the function.");
}
}
if (Debug.isLevelEnabled(Debug.MSG_STATUS)) {
Debug.log(
Debug.MSG_STATUS,
"RegexUtils: makePerl5MatchPatternNegativeLookAhead: patternBuffer.toString() = "
+ patternBuffer.toString());
}
return patternBuffer.toString();
}
/**
* This function composes the perl regular expression to perform pattern matching with negative
* look ahead. This becomes handy when a replacement string includes some string that is also a
* pattern. i.e.) pattern - ", & replacement - ", & original input - I didn't &0 to say
* "GOO".
*
* <p>After the first substitution with " and ": processed input - I didn't &0 to say
* "GOO".
*
* <p>When we do the second substitution with '&', we do not want to perform the pattern matching
* on the '&' of """ since it is already a replacement. If we give the pattern like this with
* perl's negative look-ahead functionality:
*
* <p>not-negative look-ahead pattern - & negative look-ahead pattern - &(?!quot)
*
* <p>After the second substitution with & and & with non-negative look-ahead: processed input
* - I didn't &0 to say &quotGOO&quot.
*
* <p>After the second substitution with & and & with negative look-ahead: processed input - I
* didn't &0 to say "GOO".
*
* @param pattern - pattern string to match
* @param replacement - replacement
* @return returns the format pattern for perl5 negative look-ahead
* @exception throws a FrameworkException when the input/pattern/replacement is null.
*/
private static String makePerl5MatchPatternNegativeLookAhead(String pattern, String replacement)
throws FrameworkException {
if (pattern == null || replacement == null) {
throw new FrameworkException(
"RegexUtils: makePerl5MatchPatternNegativeLookAhead: pattern or replacement is null."
+ "\npattern = "
+ pattern
+ "\nreplacement = "
+ replacement);
}
String result = pattern;
StringBuffer patternBuffer = new StringBuffer();
Perl5Util util = new Perl5Util();
String formatPattern = makePerl5MatchPattern(pattern);
// check if the pattern string is a part of the replacement string
if (util.match(formatPattern, replacement)) {
if (replacement.startsWith(pattern)) {
result = util.postMatch();
patternBuffer.append(pattern);
patternBuffer.append("(?!");
patternBuffer.append(result);
patternBuffer.append(")");
result = patternBuffer.toString();
} else {
throw new FrameworkException(
"ERROR: RegexUtils: makePerl5MatchPatternNegativeLookAhead: The pattern in the "
+ "replacement string should be at the beginning.");
}
}
return result;
}
/**
* Makes WikiText from a Collection.
*
* @param links Collection to make into WikiText.
* @param separator Separator string to use.
* @param numItems How many items to show.
* @return The WikiText
*/
protected String wikitizeCollection(Collection links, String separator, int numItems) {
if (links == null || links.isEmpty()) return "";
StringBuffer output = new StringBuffer();
Iterator it = links.iterator();
int count = 0;
//
// The output will be B Item[1] A S B Item[2] A S B Item[3] A
//
while (it.hasNext() && ((count < numItems) || (numItems == ALL_ITEMS))) {
String value = (String) it.next();
if (count > 0) {
output.append(m_after);
output.append(m_separator);
}
output.append(m_before);
// Make a Wiki markup link. See TranslatorReader.
output.append("[" + m_engine.beautifyTitle(value) + "|" + value + "]");
count++;
}
//
// Output final item - if there have been none, no "after" is printed
//
if (count > 0) output.append(m_after);
return output.toString();
}
// if numChars is 0, this means match as many as you want
private int __parseUnsignedInteger(int radix, int minDigits, int maxDigits)
throws MalformedPatternException {
int num, digits = 0;
StringBuffer buf;
// We don't expect huge numbers, so an initial buffer of 4 is fine.
buf = new StringBuffer(4);
while (Character.digit(__lookahead, radix) != -1 && digits < maxDigits) {
buf.append((char) __lookahead);
__match(__lookahead);
++digits;
}
if (digits < minDigits || digits > maxDigits)
throw new MalformedPatternException(
"Parse error: unexpected number of digits at position " + __bytesRead);
try {
num = Integer.parseInt(buf.toString(), radix);
} catch (NumberFormatException e) {
throw new MalformedPatternException(
"Parse error: numeric value at " + "position " + __bytesRead + " is invalid");
}
return num;
}
/**
* This function replace only the first pattern matched with the replacement.
*
* @param pattern pattern string to match
* @param input input string
* @param replacement replacement string
* @return returns the processed string
* @exception throws a FrameworkException when the pattern/replacement is null.
*/
public static String replaceLast(String pattern, String input, String replacement)
throws FrameworkException {
if ((pattern == null) || (input == null) || (replacement == null)) {
throw new FrameworkException(
"RegexUtil: replaceLast(): pattern or input cannot be null. "
+ "pattern = "
+ pattern
+ "input = "
+ input
+ "replacement = "
+ replacement);
}
Perl5Util util = new Perl5Util();
MatchResult matchResult = null;
String result = null;
String pre = null;
String post = null;
StringBuffer resultBuffer = new StringBuffer();
int length = input.length();
String regex = makePerl5SubstitutionPattern(pattern, replacement);
pattern = makePerl5MatchPattern(pattern);
// counts the number of match and grab the last one
while (util.match(pattern, input)) {
// getting the string before match
pre = util.preMatch();
resultBuffer.append(pre);
// get the matched string
matchResult = util.getMatch();
resultBuffer.append(matchResult.toString());
// get the post string after the match
post = util.postMatch();
// the post becomes the new input
input = post;
}
// get the last match found
matchResult = util.getMatch();
// do the string replacement on the pattern found
result = util.substitute(regex, matchResult.toString());
resultBuffer.append(result);
resultBuffer.append(post);
return resultBuffer.toString();
}
/**
* This function makes a perl5 match pattern from non-format pattern for the substitution. This
* function also checks for the case of special character. It the patter passed in is such a
* chracter, the escape character back slash is attached in the front. The format is:
*
* <p>s/pattern/replacement/
*
* <p>i.e.) input: pattern - " replacement - "
*
* <p>output: s/\"/"/
*
* @param pattern pattern string to match
* @param replacement replacement string
* @return returns the perl5 format for substitution
* @exception throws a FrameworkException when the pattern/replacement is null.
*/
private static String makePerl5SubstitutionPattern(String pattern, String replacement)
throws FrameworkException {
if (replacement == null || pattern == null) {
throw new FrameworkException(
"RegexUtils: makePerl5SubstitutionPattern: pattern or replacement is null."
+ "pattern = "
+ pattern
+ " replacement = "
+ replacement);
}
StringBuffer regex = new StringBuffer();
String formatPattern = pattern;
// check for the special characters which needs to be preceeed it by a backslash
if (pattern.equals("|")
|| pattern.equals(")")
|| pattern.equals("$")
|| pattern.equals("*")
|| pattern.equals("^")
|| pattern.equals("/")
|| pattern.equals("+")
|| pattern.equals(".")
|| pattern.equals("[")
|| pattern.equals("?")
|| pattern.equals("(")
|| pattern.equals("]")) {
// add the escape character
formatPattern = "\\" + pattern;
}
// the case the replacement has an option
if (checkOption(replacement)) {
regex.append("s/");
regex.append(formatPattern);
regex.append("/");
regex.append(replacement);
regex.append("g");
} else {
regex.append("s/");
regex.append(formatPattern);
regex.append("/");
regex.append(replacement);
regex.append("/g");
}
return regex.toString();
}
/**
* This function makes a perl5 match pattern from non-format pattern. This function also checks
* for the case of special character. It the patter passed in is such a chracter, the escape
* character back slash is attached in the front.
*
* <p>i.e.) input "&" output "/&/"
*
* @param pattern pattern string to match
* @return returns the perl5 format pattern
* @exception throws a FrameworkException when the pattern is null.
*/
private static String makePerl5MatchPattern(String pattern) throws FrameworkException {
if (pattern == null) {
throw new FrameworkException("RegexUtil: makePerl5MatchPattern(): the pattern is null.");
}
StringBuffer perl5Pattern = new StringBuffer();
String formatPattern = pattern;
// check for the special characters which needs to be preceeed it by a backslash
if (pattern.equals("|")
|| pattern.equals(")")
|| pattern.equals("$")
|| pattern.equals("*")
|| pattern.equals("^")
|| pattern.equals("/")
|| pattern.equals("+")
|| pattern.equals(".")
|| pattern.equals("[")
|| pattern.equals("?")
|| pattern.equals("(")
|| pattern.equals("]")) {
// add the escape character
formatPattern = "\\" + pattern;
}
/* check if the pattern has the option. user can specify the pattern with the
* perl5 option: pattern/[i][m][s][x]
* checkOption returns true if the pattern has index of the one of
* those: /i, /m, /s, /x
*/
if (checkOption(formatPattern)) {
perl5Pattern.append("/");
perl5Pattern.append(formatPattern);
} else {
perl5Pattern.append("/");
perl5Pattern.append(formatPattern);
perl5Pattern.append("/");
}
return perl5Pattern.toString();
}
/**
* The function checks and handles the situation where another begin token is found in between the
* first begin token and the first end token. Such first beginToken is ignored. The current input
* to be passed in to the replacement function will be the string between the LAST begin token and
* the FIRST end token found.
*
* @param beginToken the pattern to look for
* @param current input string
* @param resultBuffer the buffer to hold the result
* @return returns a Vector containing the substrings of the input that occur
*/
private static final String skipOrphanedBeginToken(
String beginToken, String current, StringBuffer resultBuffer) {
Perl5Util util = new Perl5Util();
boolean nextBeginTokenMatch = true;
MatchResult matchResult = null;
String remainderStr = null;
String subCurrent = null;
StringBuffer currentBuffer = new StringBuffer();
String pre = null;
// subCurrent is copy of the current to check if there is any next beginTokens
subCurrent = current;
while (nextBeginTokenMatch) {
nextBeginTokenMatch = util.match(beginToken, subCurrent);
if (nextBeginTokenMatch == true) {
// pre is the string before the beginToken
pre = util.preMatch();
currentBuffer.append(pre);
// get the matched beginToken
matchResult = util.getMatch();
currentBuffer.append(matchResult.toString());
// get the remaining string after the beginToken
remainderStr = util.postMatch();
subCurrent = remainderStr;
} else // there is no match
{
// appending the string before the last begin token to the result
resultBuffer.append(currentBuffer.toString());
// finally get the string to perform the substitution on
current = subCurrent;
break;
}
} // while
return current;
}
/**
* This function substitues ALL the occurence of the pattern in the input with the replacement
* passed in. The substitution is done only in the range bound by the begin token and the end
* token.
*
* @param pattern pattern string to match
* @param input input string
* @param replacement replacement string to replace with
* @param beginToken begin boundary token
* @param endToken end boundary token
* @param beginTokens[] an array of beginTokens - this is used for negative look ahead where there
* are other begin tokens to be recognized.
* @return returns processed string if both beginToken and the endToken are successfully found,
* returns the unproessed original input otherwise.
* @exception throws a FrameworkException when either pattern/input/replacement is null.
*/
public static String replaceAll(
String pattern,
String input,
String replacement,
String beginToken,
String endToken,
String beginTokens[])
throws FrameworkException {
if ((pattern == null) || (replacement == null) || (input == null)) {
Debug.log(
Debug.ALL_ERRORS,
"RegexUtils: replaceAll(): pattern or replacement or input is null. "
+ "\npattern = "
+ pattern
+ "\ninput = "
+ input
+ "\nreplacement = "
+ replacement);
throw new FrameworkException(
"RegexUtils: replaceAll(): pattern or replacement or input is null.");
}
Perl5Util util = new Perl5Util();
StringBuffer resultBuffer = new StringBuffer();
if ((beginToken == null) || (endToken == null)) {
// either beginToken or endToken cannot be null, however both can be null.
throw new FrameworkException(
"RegexUtils: replaceAll(): Either begin or end token is null. BeginToken = "
+ beginToken
+ ", "
+ "endToken = "
+ endToken);
} else // do pattern match
{
// making the Perl5 regular expression format pattern
String begin = makePerl5MatchPattern(beginToken);
String end = makePerl5MatchPattern(endToken);
boolean beginTokenMatch = true;
// if begin token found
while (beginTokenMatch) {
// check the input for each iteration. when there is no input, break out of the loop.
if (input == null) break;
beginTokenMatch = util.match(begin, input);
if (beginTokenMatch) {
String negativeLookAheadPattern =
makePerl5MatchPatternNegativeLookAhead(pattern, replacement);
int beginOffsetForBeginToken = util.beginOffset(0);
int endOffsetForBeginToken = util.endOffset(0);
// the input after the begin token
String subInput = input.substring(endOffsetForBeginToken);
// if end token passed in was an empty string
if (endToken.equals("")) {
String result = null;
try {
result = replaceAllWithBeginToken(pattern, input, replacement, beginToken);
} catch (FrameworkException e) {
Debug.log(Debug.ALL_ERRORS, "RegexUtils: replaceAll() failed." + e.getMessage());
}
return result;
} else if (util.match(end, subInput)) // endToken found
{
// begin offset for the end token relative to the input not subInput
int beginOffsetForEndToken = endOffsetForBeginToken + util.beginOffset(0);
int endOffsetForEndToken = endOffsetForBeginToken + util.endOffset(0);
// pre is the string before the beginToken and the beginToken
String pre = input.substring(0, endOffsetForBeginToken);
resultBuffer.append(pre);
// current is the string between the beginToken and the endToken
String current = input.substring(endOffsetForBeginToken, beginOffsetForEndToken);
// theRest is the rest of the input string after the endToken
String theRest = input.substring(endOffsetForEndToken);
// current is the string between begin token and the endtoken
current = skipOrphanedBeginToken(begin, current, resultBuffer);
if (isOtherBeginTokenThere(beginToken, current, beginTokens) == false) {
current = replaceAll(pattern, current, replacement);
} // isOtherBeginTokenThere
resultBuffer.append(current);
resultBuffer.append(input.substring(beginOffsetForEndToken, endOffsetForEndToken));
input = theRest;
} else // endToken not found
{
resultBuffer.append(input);
break;
}
} // if beginToken found
else // beginToken not found
{
resultBuffer.append(input);
break;
}
} // while begin token found
} // else do pattern match
return resultBuffer.toString();
}
@NotNull
private Pattern getPattern(String pattern) {
if (!Comparing.strEqual(pattern, myPattern)) {
myCompiledPattern = null;
myPattern = pattern;
}
if (myCompiledPattern == null) {
boolean allowToLower = true;
final int eol = pattern.indexOf('\n');
if (eol != -1) {
pattern = pattern.substring(0, eol);
}
if (pattern.length() >= 80) {
pattern = pattern.substring(0, 80);
}
final @NonNls StringBuffer buffer = new StringBuffer();
if (containsOnlyUppercaseLetters(pattern)) {
allowToLower = false;
}
if (allowToLower) {
buffer.append(".*");
}
boolean firstIdentifierLetter = true;
for (int i = 0; i < pattern.length(); i++) {
final char c = pattern.charAt(i);
if (Character.isLetterOrDigit(c)) {
// This logic allows to use uppercase letters only to catch the name like PDM for
// PsiDocumentManager
if (Character.isUpperCase(c) || Character.isDigit(c)) {
if (!firstIdentifierLetter) {
buffer.append("[^A-Z]*");
}
buffer.append("[");
buffer.append(c);
if (allowToLower || i == 0) {
buffer.append('|');
buffer.append(Character.toLowerCase(c));
}
buffer.append("]");
} else if (Character.isLowerCase(c)) {
buffer.append('[');
buffer.append(c);
buffer.append('|');
buffer.append(Character.toUpperCase(c));
buffer.append(']');
} else {
buffer.append(c);
}
firstIdentifierLetter = false;
} else if (c == '*') {
buffer.append(".*");
firstIdentifierLetter = true;
} else if (c == '.') {
buffer.append("\\.");
firstIdentifierLetter = true;
} else if (c == ' ') {
buffer.append("[^A-Z]*\\ ");
firstIdentifierLetter = true;
} else {
firstIdentifierLetter = true;
// for standard RegExp engine
// buffer.append("\\u");
// buffer.append(Integer.toHexString(c + 0x20000).substring(1));
// for OROMATCHER RegExp engine
buffer.append("\\x");
buffer.append(Integer.toHexString(c + 0x20000).substring(3));
}
}
buffer.append(".*");
try {
myCompiledPattern = new Perl5Compiler().compile(buffer.toString());
} catch (MalformedPatternException e) {
// do nothing
}
}
return myCompiledPattern;
}