DiagramSourceCompressor.java
/* ========================================================================
* PlantUML : a free UML diagram generator
* ========================================================================
*
* (C) Copyright 2009-2024, Arnaud Roques
*
* Project Info: https://plantuml.com
*
* If you like this project or if you find it useful, you can support us at:
*
* https://plantuml.com/patreon (only 1$ per month!)
* https://plantuml.com/paypal
*
* This file is part of PlantUML.
*
* PlantUML is free software; you can redistribute it and/or modify it
* under the terms of the GNU General Public License as published by
* the Free Software Foundation, either version 3 of the License, or
* (at your option) any later version.
*
* PlantUML distributed in the hope that it will be useful, but
* WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
* or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public
* License for more details.
*
* You should have received a copy of the GNU General Public
* License along with this library; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301,
* USA.
*
*
* Original Author: Arnaud Roques
*
*
*/
package net.sourceforge.plantuml.code;
import java.io.IOException;
import java.io.StringReader;
import net.sourceforge.plantuml.StringUtils;
import net.sourceforge.plantuml.preproc.ReadLine;
import net.sourceforge.plantuml.preproc.ReadLineReader;
import net.sourceforge.plantuml.preproc.UncommentReadLine;
import net.sourceforge.plantuml.text.StringLocated;
public class DiagramSourceCompressor implements StringCompressor {
private static final String START_UML = "@startuml";
private static final String END_UML = "@enduml";
public String compress(final String data) throws IOException {
final ReadLine r = new UncommentReadLine(ReadLineReader.create(new StringReader(data), "COMPRESS"));
final StringBuilder sb = new StringBuilder();
final StringBuilder full = new StringBuilder();
StringLocated s = null;
boolean startDone = false;
while ((s = r.readLine()) != null) {
append(full, s);
if (s.getString().startsWith("@startuml")) {
startDone = true;
} else if (s.getString().startsWith("@enduml")) {
return sb.toString();
} else if (startDone) {
append(sb, s);
}
}
if (startDone == false) {
return compressOld(full.toString());
}
return sb.toString();
}
private void append(final StringBuilder sb, StringLocated s) {
if (sb.length() > 0) {
sb.append('\n');
}
sb.append(s.getString());
}
// Package-private, as the other cleaning methods below, so that they can be unit tested
String compressOld(String s) throws IOException {
return clean(stripStartEnd(s, true));
}
public String decompress(String s) {
String result = clean(s);
if (result.startsWith("@start")) {
return result;
}
result = "@startuml\n" + result;
if (result.endsWith("\n") == false) {
result += "\n";
}
result += "@enduml";
return result;
}
String clean(String s) {
// s = s.replace("\0", "");
s = StringUtils.trin(s);
s = clean1(s);
s = removeFromToken(s, END_UML);
s = removeFromToken(s, START_UML);
s = StringUtils.trin(s);
return s;
}
/**
* Same as {@code s.replaceAll(token + "[^\\n\\r]*", "")}, with the token taken
* literally: removes every occurrence of the token, and the rest of its
* line (up to, and not including, the next {@code \n} or {@code \r}).
* <p>
* Plain loop, like {@link #stripStartEnd}, so that the cost does not depend
* on the regex engine of the TeaVM build when the source is big (issue
* #2934).
*/
static String removeFromToken(String s, String token) {
int pos = s.indexOf(token);
if (pos == -1)
return s;
final int len = s.length();
final StringBuilder sb = new StringBuilder(len);
int from = 0;
while (pos != -1) {
sb.append(s, from, pos);
from = pos + token.length();
while (from < len && s.charAt(from) != '\n' && s.charAt(from) != '\r')
from++;
pos = s.indexOf(token, from);
}
sb.append(s, from, len);
return sb.toString();
}
private String clean1(String s) {
return stripStartEnd(s, false);
}
/**
* What {@code [%s]} stands for in a PlantUML regex: {@code \s} and the
* non-breaking space. This is deliberately not {@link StringUtils#isTrimmable},
* which is wider (every char up to the space, control chars included).
*/
static boolean isBlank(char c) {
return c == ' ' || c == '\t' || c == '\n' || c == '\u000B' || c == '\f' || c == '\r' || c == '\u00A0';
}
/**
* The line terminators, besides {@code \n} and {@code \r} which are blanks, that
* a regex {@code $} accepts just before the end of the input.
*/
private static boolean isOtherLineTerminator(char c) {
return c == '\u0085' || c == '\u2028' || c == '\u2029';
}
/**
* Returns what group 2 captures in the regex
* <pre>(?s)^[%s]*(@startuml[^\n\r]*)?[%s]*(.*?)[%s]*(@enduml)?[%s]*$</pre>
* that is, the text without its leading blanks, without a first
* {@code @startuml} line, without its trailing blanks and without a final
* {@code @enduml}.
* <p>
* It is computed with plain loops because that regex, applied to a whole
* diagram source, overflows the JavaScript call stack in the TeaVM build as
* soon as the source is a few KB long (issue #2934): the lazy {@code (.*?)}
* costs one stack frame per character there.
*
* @param dollarBeforeFinalTerminator {@code true} for {@code Matcher.find()},
* where {@code $} also matches before a
* final U+0085, U+2028 or U+2029;
* {@code false} for {@code Matcher.matches()},
* where the match must reach the very end
*/
static String stripStartEnd(String s, boolean dollarBeforeFinalTerminator) {
final int len = s.length();
int start = 0;
while (start < len && isBlank(s.charAt(start)))
start++;
if (s.startsWith(START_UML, start)) {
start += START_UML.length();
while (start < len && s.charAt(start) != '\n' && s.charAt(start) != '\r')
start++;
while (start < len && isBlank(s.charAt(start)))
start++;
}
int end = len;
if (dollarBeforeFinalTerminator && end > start && isOtherLineTerminator(s.charAt(end - 1)))
end--;
while (end > start && isBlank(s.charAt(end - 1)))
end--;
final int endUml = end - END_UML.length();
if (endUml >= start && s.startsWith(END_UML, endUml)) {
end = endUml;
while (end > start && isBlank(s.charAt(end - 1)))
end--;
}
return s.substring(start, end);
}
}