org.fife.ui.rsyntaxtextarea.folding.CurlyFoldParser Maven / Gradle / Ivy
/*
* 10/08/2011
*
* CurlyFoldParser.java - Fold parser for languages with C-style syntax.
*
* This library is distributed under a modified BSD license. See the included
* LICENSE file for details.
*/
package org.fife.ui.rsyntaxtextarea.folding;
import java.util.ArrayList;
import java.util.List;
import javax.swing.text.BadLocationException;
import org.fife.ui.rsyntaxtextarea.RSyntaxTextArea;
import org.fife.ui.rsyntaxtextarea.Token;
/**
* A basic fold parser that can be used for languages such as C, that use
* curly braces to denote code blocks. This parser searches for curly brace
* pairs and creates code folds out of them. It can also optionally find
* C-style multi-line comments ("/* ... */") and make them
* foldable as well.
*
* This parser knows nothing about language semantics; it uses
* RSyntaxTextArea's syntax highlighting tokens to identify
* curly braces. By default, it looks for single-char tokens of type
* {@link Token#SEPARATOR}, with lexemes '{' or '}'.
* If your {@link org.fife.ui.rsyntaxtextarea.TokenMaker} uses a different token
* type for curly braces, you should override the {@link #isLeftCurly(Token)} and
* {@link #isRightCurly(Token)} methods with your own definitions. In theory,
* you could extend this fold parser to parse languages that use completely
* different tokens than curly braces to denote foldable regions by overriding
* those two methods.
*
* Note also that this class may impose somewhat of a performance penalty on
* large source files, since it re-parses the entire document each time folds
* are reevaluated.
*
* @author Robert Futrell
* @version 1.0
*/
public class CurlyFoldParser implements FoldParser {
/**
* Whether to scan for C-style multi-line comments and make them foldable.
*/
private boolean foldableMultiLineComments;
/**
* Whether this parser is folding Java.
*/
private final boolean java;
/**
* Used to find import statements when folding Java code.
*/
private static final char[] KEYWORD_IMPORT = "import".toCharArray();
/**
* Ending of a multi-line comment in C, C++, Java, etc.
*/
protected static final char[] C_MLC_END = "*/".toCharArray();
/**
* Creates a fold parser that identifies foldable regions via curly braces
* as well as C-style multi-line comments.
*/
public CurlyFoldParser() {
this(true, false);
}
/**
* Constructor.
*
* @param cStyleMultiLineComments Whether to scan for C-style multi-line
* comments and make them foldable.
* @param java Whether this parser is folding Java. This adds extra
* parsing rules, such as grouping all import statements into a
* fold section.
*/
public CurlyFoldParser(boolean cStyleMultiLineComments, boolean java) {
this.foldableMultiLineComments = cStyleMultiLineComments;
this.java = java;
}
/**
* Returns whether multi-line comments are foldable with this parser.
*
* @return Whether multi-line comments are foldable.
* @see #setFoldableMultiLineComments(boolean)
*/
public boolean getFoldableMultiLineComments() {
return foldableMultiLineComments;
}
/**
* {@inheritDoc}
*/
@Override
public List getFolds(RSyntaxTextArea textArea) {
List folds = new ArrayList<>();
Fold currentFold = null;
int lineCount = textArea.getLineCount();
boolean inMLC = false;
int mlcStart = 0;
int importStartLine = -1;
int lastSeenImportLine = -1;
int importGroupStartOffs = -1;
int importGroupEndOffs = -1;
int lastRightCurlyLine = -1;
Fold prevFold = null;
try {
for (int line=0; line-1) {
if (lastSeenImportLine>importStartLine) {
Fold fold = null;
// Any imports found *should* be a top-level fold,
// but we're extra lenient here and allow groups
// of them anywhere to keep our parser better-behaved
// if they have random "imports" throughout code.
if (currentFold==null) {
fold = new Fold(FoldType.IMPORTS,
textArea, importGroupStartOffs);
folds.add(fold);
}
else {
fold = currentFold.createChild(FoldType.IMPORTS,
importGroupStartOffs);
}
fold.setEndOffset(importGroupEndOffs);
}
importStartLine = lastSeenImportLine =
importGroupStartOffs = importGroupEndOffs = -1;
}
}
if (inMLC) {
// If we found the end of an MLC that started
// on a previous line...
if (t.endsWith(C_MLC_END)) {
int mlcEnd = t.getEndOffset() - 1;
if (currentFold==null) {
currentFold = new Fold(FoldType.COMMENT, textArea, mlcStart);
currentFold.setEndOffset(mlcEnd);
folds.add(currentFold);
currentFold = null;
}
else {
currentFold = currentFold.createChild(FoldType.COMMENT, mlcStart);
currentFold.setEndOffset(mlcEnd);
currentFold = currentFold.getParent();
}
//System.out.println("Ending MLC at: " + mlcEnd + ", parent==" + currentFold);
inMLC = false;
mlcStart = 0;
}
// Otherwise, this MLC is continuing on to yet
// another line.
}
else {
// If we're an MLC that ends on a later line...
if (t.getType()!=Token.COMMENT_EOL && !t.endsWith(C_MLC_END)) {
//System.out.println("Starting MLC at: " + t.offset);
inMLC = true;
mlcStart = t.getOffset();
}
}
}
else if (isLeftCurly(t)) {
// Java-specific stuff
if (java) {
if (importStartLine>-1) {
if (lastSeenImportLine>importStartLine) {
Fold fold = null;
// Any imports found *should* be a top-level fold,
// but we're extra lenient here and allow groups
// of them anywhere to keep our parser better-behaved
// if they have random "imports" throughout code.
if (currentFold==null) {
fold = new Fold(FoldType.IMPORTS,
textArea, importGroupStartOffs);
folds.add(fold);
}
else {
fold = currentFold.createChild(FoldType.IMPORTS,
importGroupStartOffs);
}
fold.setEndOffset(importGroupEndOffs);
}
importStartLine = lastSeenImportLine =
importGroupStartOffs = importGroupEndOffs = -1;
}
}
// If a new fold block starts on the same line as the
// previous one ends, we treat it as one big block
// (e.g. K&R-style "} else {")
if (prevFold != null && line == lastRightCurlyLine) {
currentFold = prevFold;
// Keep currentFold.endOffset where it was, so that
// unclosed folds at end of the file work as well
// as possible
prevFold = null;
lastRightCurlyLine = -1;
}
else if (currentFold==null) { // A top-level fold
currentFold = new Fold(FoldType.CODE, textArea, t.getOffset());
folds.add(currentFold);
}
else { // A nested fold
currentFold = currentFold.createChild(FoldType.CODE, t.getOffset());
}
}
else if (isRightCurly(t)) {
if (currentFold!=null) {
currentFold.setEndOffset(t.getOffset());
Fold parentFold = currentFold.getParent();
//System.out.println("... Adding regular fold at " + t.offset + ", parent==" + parentFold);
// Don't add fold markers for single-line blocks
if (currentFold.isOnSingleLine()) {
if (!currentFold.removeFromParent()) {
folds.remove(folds.size()-1);
}
}
else {
// Remember the end of the last completed fold,
// in case it needs to get merged with the next
// one (e.g. K&R "} else {" style)
lastRightCurlyLine = line;
prevFold = currentFold;
}
currentFold = parentFold;
}
}
// Java-specific folding rules
else if (java) {
if (t.is(Token.RESERVED_WORD, KEYWORD_IMPORT)) {
if (importStartLine==-1) {
importStartLine = line;
importGroupStartOffs = t.getOffset();
importGroupEndOffs = t.getOffset();
}
lastSeenImportLine = line;
}
else if (importStartLine>-1 &&
t.isIdentifier() &&//SEPARATOR &&
t.isSingleChar(';')) {
importGroupEndOffs = t.getOffset();
}
}
t = t.getNextToken();
}
}
} catch (BadLocationException ble) { // Should never happen
ble.printStackTrace();
}
return folds;
}
/**
* Returns whether the token is a left curly brace. This method exists
* so subclasses can provide their own curly brace definition.
*
* @param t The token.
* @return Whether it is a left curly brace.
* @see #isRightCurly(Token)
*/
public boolean isLeftCurly(Token t) {
return t.isLeftCurly();
}
/**
* Returns whether the token is a right curly brace. This method exists
* so subclasses can provide their own curly brace definition.
*
* @param t The token.
* @return Whether it is a right curly brace.
* @see #isLeftCurly(Token)
*/
public boolean isRightCurly(Token t) {
return t.isRightCurly();
}
/**
* Sets whether multi-line comments are foldable with this parser.
*
* @param foldable Whether multi-line comments are foldable.
* @see #getFoldableMultiLineComments()
*/
public void setFoldableMultiLineComments(boolean foldable) {
this.foldableMultiLineComments = foldable;
}
}