History

My first initial commit with all of mine experimental changes.

Dmitry Kabak (userdima committed 13 years ago

Commit d37b88d1 parent 3f5b46618352 files · +2598093 −-8690Browse files

Showing the changes to java/pkg/sun/invoke/util/BytecodeName.java (1 of 18352 files). Show all

Unified · Split

java/pkg/sun/invoke/util/BytecodeName.java+627−0
@@ -0,0 +1,627 @@
1+/*
2+ * Copyright (c) 2007, 2011, Oracle and/or its affiliates. All rights reserved.
3+ * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
4+ *
5+ * This code is free software; you can redistribute it and/or modify it
6+ * under the terms of the GNU General Public License version 2 only, as
7+ * published by the Free Software Foundation. Oracle designates this
8+ * particular file as subject to the "Classpath" exception as provided
9+ * by Oracle in the LICENSE file that accompanied this code.
10+ *
11+ * This code is distributed in the hope that it will be useful, but WITHOUT
12+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
13+ * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
14+ * version 2 for more details (a copy is included in the LICENSE file that
15+ * accompanied this code).
16+ *
17+ * You should have received a copy of the GNU General Public License version
18+ * 2 along with this work; if not, write to the Free Software Foundation,
19+ * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
20+ *
21+ * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
22+ * or visit www.oracle.com if you need additional information or have any
23+ * questions.
24+ */
25+
26+package sun.invoke.util;
27+
28+/**
29+ * Utility routines for dealing with bytecode-level names.
30+ * Includes universal mangling rules for the JVM.
31+ *
32+ * <h3>Avoiding Dangerous Characters </h3>
33+ *
34+ * <p>
35+ * The JVM defines a very small set of characters which are illegal
36+ * in name spellings. We will slightly extend and regularize this set
37+ * into a group of <cite>dangerous characters</cite>.
38+ * These characters will then be replaced, in mangled names, by escape sequences.
39+ * In addition, accidental escape sequences must be further escaped.
40+ * Finally, a special prefix will be applied if and only if
41+ * the mangling would otherwise fail to begin with the escape character.
42+ * This happens to cover the corner case of the null string,
43+ * and also clearly marks symbols which need demangling.
44+ * </p>
45+ * <p>
46+ * Dangerous characters are the union of all characters forbidden
47+ * or otherwise restricted by the JVM specification,
48+ * plus their mates, if they are brackets
49+ * (<code><big><b>[</b></big></code> and <code><big><b>]</b></big></code>,
50+ * <code><big><b>&lt;</b></big></code> and <code><big><b>&gt;</b></big></code>),
51+ * plus, arbitrarily, the colon character <code><big><b>:</b></big></code>.
52+ * There is no distinction between type, method, and field names.
53+ * This makes it easier to convert between mangled names of different
54+ * types, since they do not need to be decoded (demangled).
55+ * </p>
56+ * <p>
57+ * The escape character is backslash <code><big><b>\</b></big></code>
58+ * (also known as reverse solidus).
59+ * This character is, until now, unheard of in bytecode names,
60+ * but traditional in the proposed role.
61+ *
62+ * </p>
63+ * <h3> Replacement Characters </h3>
64+ *
65+ *
66+ * <p>
67+ * Every escape sequence is two characters
68+ * (in fact, two UTF8 bytes) beginning with
69+ * the escape character and followed by a
70+ * <cite>replacement character</cite>.
71+ * (Since the replacement character is never a backslash,
72+ * iterated manglings do not double in size.)
73+ * </p>
74+ * <p>
75+ * Each dangerous character has some rough visual similarity
76+ * to its corresponding replacement character.
77+ * This makes mangled symbols easier to recognize by sight.
78+ * </p>
79+ * <p>
80+ * The dangerous characters are
81+ * <code><big><b>/</b></big></code> (forward slash, used to delimit package components),
82+ * <code><big><b>.</b></big></code> (dot, also a package delimiter),
83+ * <code><big><b>;</b></big></code> (semicolon, used in signatures),
84+ * <code><big><b>$</b></big></code> (dollar, used in inner classes and synthetic members),
85+ * <code><big><b>&lt;</b></big></code> (left angle),
86+ * <code><big><b>&gt;</b></big></code> (right angle),
87+ * <code><big><b>[</b></big></code> (left square bracket, used in array types),
88+ * <code><big><b>]</b></big></code> (right square bracket, reserved in this scheme for language use),
89+ * and <code><big><b>:</b></big></code> (colon, reserved in this scheme for language use).
90+ * Their replacements are, respectively,
91+ * <code><big><b>|</b></big></code> (vertical bar),
92+ * <code><big><b>,</b></big></code> (comma),
93+ * <code><big><b>?</b></big></code> (question mark),
94+ * <code><big><b>%</b></big></code> (percent),
95+ * <code><big><b>^</b></big></code> (caret),
96+ * <code><big><b>_</b></big></code> (underscore), and
97+ * <code><big><b>{</b></big></code> (left curly bracket),
98+ * <code><big><b>}</b></big></code> (right curly bracket),
99+ * <code><big><b>!</b></big></code> (exclamation mark).
100+ * In addition, the replacement character for the escape character itself is
101+ * <code><big><b>-</b></big></code> (hyphen),
102+ * and the replacement character for the null prefix is
103+ * <code><big><b>=</b></big></code> (equal sign).
104+ * </p>
105+ * <p>
106+ * An escape character <code><big><b>\</b></big></code>
107+ * followed by any of these replacement characters
108+ * is an escape sequence, and there are no other escape sequences.
109+ * An equal sign is only part of an escape sequence
110+ * if it is the second character in the whole string, following a backslash.
111+ * Two consecutive backslashes do <em>not</em> form an escape sequence.
112+ * </p>
113+ * <p>
114+ * Each escape sequence replaces a so-called <cite>original character</cite>
115+ * which is either one of the dangerous characters or the escape character.
116+ * A null prefix replaces an initial null string, not a character.
117+ * </p>
118+ * <p>
119+ * All this implies that escape sequences cannot overlap and may be
120+ * determined all at once for a whole string. Note that a spelling
121+ * string can contain <cite>accidental escapes</cite>, apparent escape
122+ * sequences which must not be interpreted as manglings.
123+ * These are disabled by replacing their leading backslash with an
124+ * escape sequence (<code><big><b>\-</b></big></code>). To mangle a string, three logical steps
125+ * are required, though they may be carried out in one pass:
126+ * </p>
127+ * <ol>
128+ * <li>In each accidental escape, replace the backslash with an escape sequence
129+ * (<code><big><b>\-</b></big></code>).</li>
130+ * <li>Replace each dangerous character with an escape sequence
131+ * (<code><big><b>\|</b></big></code> for <code><big><b>/</b></big></code>, etc.).</li>
132+ * <li>If the first two steps introduced any change, <em>and</em>
133+ * if the string does not already begin with a backslash, prepend a null prefix (<code><big><b>\=</b></big></code>).</li>
134+ * </ol>
135+ *
136+ * To demangle a mangled string that begins with an escape,
137+ * remove any null prefix, and then replace (in parallel)
138+ * each escape sequence by its original character.
139+ * <p>Spelling strings which contain accidental
140+ * escapes <em>must</em> have them replaced, even if those
141+ * strings do not contain dangerous characters.
142+ * This restriction means that mangling a string always
143+ * requires a scan of the string for escapes.
144+ * But then, a scan would be required anyway,
145+ * to check for dangerous characters.
146+ *
147+ * </p>
148+ * <h3> Nice Properties </h3>
149+ *
150+ * <p>
151+ * If a bytecode name does not contain any escape sequence,
152+ * demangling is a no-op: The string demangles to itself.
153+ * Such a string is called <cite>self-mangling</cite>.
154+ * Almost all strings are self-mangling.
155+ * In practice, to demangle almost any name &ldquo;found in nature&rdquo;,
156+ * simply verify that it does not begin with a backslash.
157+ * </p>
158+ * <p>
159+ * Mangling is a one-to-one function, while demangling
160+ * is a many-to-one function.
161+ * A mangled string is defined as <cite>validly mangled</cite> if
162+ * it is in fact the unique mangling of its spelling string.
163+ * Three examples of invalidly mangled strings are <code><big><b>\=foo</b></big></code>,
164+ * <code><big><b>\-bar</b></big></code>, and <code><big><b>baz\!</b></big></code>, which demangle to <code><big><b>foo</b></big></code>, <code><big><b>\bar</b></big></code>, and
165+ * <code><big><b>baz\!</b></big></code>, but then remangle to <code><big><b>foo</b></big></code>, <code><big><b>\bar</b></big></code>, and <code><big><b>\=baz\-!</b></big></code>.
166+ * If a language back-end or runtime is using mangled names,
167+ * it should never present an invalidly mangled bytecode
168+ * name to the JVM. If the runtime encounters one,
169+ * it should also report an error, since such an occurrence
170+ * probably indicates a bug in name encoding which
171+ * will lead to errors in linkage.
172+ * However, this note does not propose that the JVM verifier
173+ * detect invalidly mangled names.
174+ * </p>
175+ * <p>
176+ * As a result of these rules, it is a simple matter to
177+ * compute validly mangled substrings and concatenations
178+ * of validly mangled strings, and (with a little care)
179+ * these correspond to corresponding operations on their
180+ * spelling strings.
181+ * </p>
182+ * <ul>
183+ * <li>Any prefix of a validly mangled string is also validly mangled,
184+ * although a null prefix may need to be removed.</li>
185+ * <li>Any suffix of a validly mangled string is also validly mangled,
186+ * although a null prefix may need to be added.</li>
187+ * <li>Two validly mangled strings, when concatenated,
188+ * are also validly mangled, although any null prefix
189+ * must be removed from the second string,
190+ * and a trailing backslash on the first string may need escaping,
191+ * if it would participate in an accidental escape when followed
192+ * by the first character of the second string.</li>
193+ * </ul>
194+ * <p>If languages that include non-Java symbol spellings use this
195+ * mangling convention, they will enjoy the following advantages:
196+ * </p>
197+ * <ul>
198+ * <li>They can interoperate via symbols they share in common.</li>
199+ * <li>Low-level tools, such as backtrace printers, will have readable displays.</li>
200+ * <li>Future JVM and language extensions can safely use the dangerous characters
201+ * for structuring symbols, but will never interfere with valid spellings.</li>
202+ * <li>Runtimes and compilers can use standard libraries for mangling and demangling.</li>
203+ * <li>Occasional transliterations and name composition will be simple and regular,
204+ * for classes, methods, and fields.</li>
205+ * <li>Bytecode names will continue to be compact.
206+ * When mangled, spellings will at most double in length, either in
207+ * UTF8 or UTF16 format, and most will not change at all.</li>
208+ * </ul>
209+ *
210+ *
211+ * <h3> Suggestions for Human Readable Presentations </h3>
212+ *
213+ *
214+ * <p>
215+ * For human readable displays of symbols,
216+ * it will be better to present a string-like quoted
217+ * representation of the spelling, because JVM users
218+ * are generally familiar with such tokens.
219+ * We suggest using single or double quotes before and after
220+ * mangled symbols which are not valid Java identifiers,
221+ * with quotes, backslashes, and non-printing characters
222+ * escaped as if for literals in the Java language.
223+ * </p>
224+ * <p>
225+ * For example, an HTML-like spelling
226+ * <code><big><b>&lt;pre&gt;</b></big></code> mangles to
227+ * <code><big><b>\^pre\_</b></big></code> and could
228+ * display more cleanly as
229+ * <code><big><b>'&lt;pre&gt;'</b></big></code>,
230+ * with the quotes included.
231+ * Such string-like conventions are <em>not</em> suitable
232+ * for mangled bytecode names, in part because
233+ * dangerous characters must be eliminated, rather
234+ * than just quoted. Otherwise internally structured
235+ * strings like package prefixes and method signatures
236+ * could not be reliably parsed.
237+ * </p>
238+ * <p>
239+ * In such human-readable displays, invalidly mangled
240+ * names should <em>not</em> be demangled and quoted,
241+ * for this would be misleading. Likewise, JVM symbols
242+ * which contain dangerous characters (like dots in field
243+ * names or brackets in method names) should not be
244+ * simply quoted. The bytecode names
245+ * <code><big><b>\=phase\,1</b></big></code> and
246+ * <code><big><b>phase.1</b></big></code> are distinct,
247+ * and in demangled displays they should be presented as
248+ * <code><big><b>'phase.1'</b></big></code> and something like
249+ * <code><big><b>'phase'.1</b></big></code>, respectively.
250+ * </p>
251+ *
252+ * @author John Rose
253+ * @version 1.2, 02/06/2008
254+ * @see http://blogs.sun.com/jrose/entry/symbolic_freedom_in_the_vm
255+ */
256+public class BytecodeName {
257+ private BytecodeName() { } // static only class
258+
259+ /** Given a source name, produce the corresponding bytecode name.
260+ * The source name should not be qualified, because any syntactic
261+ * markers (dots, slashes, dollar signs, colons, etc.) will be mangled.
262+ * @param s the source name
263+ * @return a valid bytecode name which represents the source name
264+ */
265+ public static String toBytecodeName(String s) {
266+ String bn = mangle(s);
267+ assert((Object)bn == s || looksMangled(bn)) : bn;
268+ assert(s.equals(toSourceName(bn))) : s;
269+ return bn;
270+ }
271+
272+ /** Given an unqualified bytecode name, produce the corresponding source name.
273+ * The bytecode name must not contain dangerous characters.
274+ * In particular, it must not be qualified or segmented by colon {@code ':'}.
275+ * @param s the bytecode name
276+ * @return the source name, which may possibly have unsafe characters
277+ * @throws IllegalArgumentException if the bytecode name is not {@link #isSafeBytecodeName safe}
278+ * @see #isSafeBytecodeName(java.lang.String)
279+ */
280+ public static String toSourceName(String s) {
281+ checkSafeBytecodeName(s);
282+ String sn = s;
283+ if (looksMangled(s)) {
284+ sn = demangle(s);
285+ assert(s.equals(mangle(sn))) : s+" => "+sn+" => "+mangle(sn);
286+ }
287+ return sn;
288+ }
289+
290+ /**
291+ * Given a bytecode name from a classfile, separate it into
292+ * components delimited by dangerous characters.
293+ * Each resulting array element will be either a dangerous character,
294+ * or else a safe bytecode name.
295+ * (The safe name might possibly be mangled to hide further dangerous characters.)
296+ * For example, the qualified class name {@code java/lang/String}
297+ * will be parsed into the array {@code {"java", '/', "lang", '/', "String"}}.
298+ * The name {@code &lt;init&gt;} will be parsed into { '&lt;', "init", '&gt;'}}
299+ * The name {@code foo/bar$:baz} will be parsed into
300+ * {@code {"foo", '/', "bar", '$', ':', "baz"}}.
301+ * The name {@code ::\=:foo:\=bar\!baz} will be parsed into
302+ * {@code {':', ':', "", ':', "foo", ':', "bar:baz"}}.
303+ */
304+ public static Object[] parseBytecodeName(String s) {
305+ int slen = s.length();
306+ Object[] res = null;
307+ for (int pass = 0; pass <= 1; pass++) {
308+ int fillp = 0;
309+ int lasti = 0;
310+ for (int i = 0; i <= slen; i++) {
311+ int whichDC = -1;
312+ if (i < slen) {
313+ whichDC = DANGEROUS_CHARS.indexOf(s.charAt(i));
314+ if (whichDC < DANGEROUS_CHAR_FIRST_INDEX) continue;
315+ }
316+ // got to end of string or next dangerous char
317+ if (lasti < i) {
318+ // normal component
319+ if (pass != 0)
320+ res[fillp] = toSourceName(s.substring(lasti, i));
321+ fillp++;
322+ lasti = i+1;
323+ }
324+ if (whichDC >= DANGEROUS_CHAR_FIRST_INDEX) {
325+ if (pass != 0)
326+ res[fillp] = DANGEROUS_CHARS_CA[whichDC];
327+ fillp++;
328+ lasti = i+1;
329+ }
330+ }
331+ if (pass != 0) break;
332+ // between passes, build the result array
333+ res = new Object[fillp];
334+ if (fillp <= 1 && lasti == 0) {
335+ if (fillp != 0) res[0] = toSourceName(s);
336+ break;
337+ }
338+ }
339+ return res;
340+ }
341+
342+ /**
343+ * Given a series of components, create a bytecode name for a classfile.
344+ * This is the inverse of {@link #parseBytecodeName(java.lang.String)}.
345+ * Each component must either be an interned one-character string of
346+ * a dangerous character, or else a safe bytecode name.
347+ * @param components a series of name components
348+ * @return the concatenation of all components
349+ * @throws IllegalArgumentException if any component contains an unsafe
350+ * character, and is not an interned one-character string
351+ * @throws NullPointerException if any component is null
352+ */
353+ public static String unparseBytecodeName(Object[] components) {
354+ Object[] components0 = components;
355+ for (int i = 0; i < components.length; i++) {
356+ Object c = components[i];
357+ if (c instanceof String) {
358+ String mc = toBytecodeName((String) c);
359+ if (i == 0 && components.length == 1)
360+ return mc; // usual case
361+ if ((Object)mc != c) {
362+ if (components == components0)
363+ components = components.clone();
364+ components[i] = c = mc;
365+ }
366+ }
367+ }
368+ return appendAll(components);
369+ }
370+ private static String appendAll(Object[] components) {
371+ if (components.length <= 1) {
372+ if (components.length == 1) {
373+ return String.valueOf(components[0]);
374+ }
375+ return "";
376+ }
377+ int slen = 0;
378+ for (Object c : components) {
379+ if (c instanceof String)
380+ slen += String.valueOf(c).length();
381+ else
382+ slen += 1;
383+ }
384+ StringBuilder sb = new StringBuilder(slen);
385+ for (Object c : components) {
386+ sb.append(c);
387+ }
388+ return sb.toString();
389+ }
390+
391+ /**
392+ * Given a bytecode name, produce the corresponding display name.
393+ * This is the source name, plus quotes if needed.
394+ * If the bytecode name contains dangerous characters,
395+ * assume that they are being used as punctuation,
396+ * and pass them through unchanged.
397+ * Non-empty runs of non-dangerous characters are demangled
398+ * if necessary, and the resulting names are quoted if
399+ * they are not already valid Java identifiers, or if
400+ * they contain a dangerous character (i.e., dollar sign "$").
401+ * Single quotes are used when quoting.
402+ * Within quoted names, embedded single quotes and backslashes
403+ * are further escaped by prepended backslashes.
404+ *
405+ * @param s the original bytecode name (which may be qualified)
406+ * @return a human-readable presentation
407+ */
408+ public static String toDisplayName(String s) {
409+ Object[] components = parseBytecodeName(s);
410+ for (int i = 0; i < components.length; i++) {
411+ if (!(components[i] instanceof String))
412+ continue;
413+ String sn = (String) components[i];
414+ // note that the name is already demangled!
415+ //sn = toSourceName(sn);
416+ if (!isJavaIdent(sn) || sn.indexOf('$') >=0 ) {
417+ components[i] = quoteDisplay(sn);
418+ }
419+ }
420+ return appendAll(components);
421+ }
422+ private static boolean isJavaIdent(String s) {
423+ int slen = s.length();
424+ if (slen == 0) return false;
425+ if (!Character.isJavaIdentifierStart(s.charAt(0)))
426+ return false;
427+ for (int i = 1; i < slen; i++) {
428+ if (!Character.isJavaIdentifierPart(s.charAt(i)))
429+ return false;
430+ }
431+ return true;
432+ }
433+ private static String quoteDisplay(String s) {
434+ // TO DO: Replace wierd characters in s by C-style escapes.
435+ return "'"+s.replaceAll("['\\\\]", "\\\\$0")+"'";
436+ }
437+
438+ private static void checkSafeBytecodeName(String s)
439+ throws IllegalArgumentException {
440+ if (!isSafeBytecodeName(s)) {
441+ throw new IllegalArgumentException(s);
442+ }
443+ }
444+
445+ /**
446+ * Report whether a simple name is safe as a bytecode name.
447+ * Such names are acceptable in class files as class, method, and field names.
448+ * Additionally, they are free of "dangerous" characters, even if those
449+ * characters are legal in some (or all) names in class files.
450+ * @param s the proposed bytecode name
451+ * @return true if the name is non-empty and all of its characters are safe
452+ */
453+ public static boolean isSafeBytecodeName(String s) {
454+ if (s.length() == 0) return false;
455+ // check occurrences of each DANGEROUS char
456+ for (char xc : DANGEROUS_CHARS_A) {
457+ if (xc == ESCAPE_C) continue; // not really that dangerous
458+ if (s.indexOf(xc) >= 0) return false;
459+ }
460+ return true;
461+ }
462+
463+ /**
464+ * Report whether a character is safe in a bytecode name.
465+ * This is true of any unicode character except the following
466+ * <em>dangerous characters</em>: {@code ".;:$[]<>/"}.
467+ * @param s the proposed character
468+ * @return true if the character is safe to use in classfiles
469+ */
470+ public static boolean isSafeBytecodeChar(char c) {
471+ return DANGEROUS_CHARS.indexOf(c) < DANGEROUS_CHAR_FIRST_INDEX;
472+ }
473+
474+ private static boolean looksMangled(String s) {
475+ return s.charAt(0) == ESCAPE_C;
476+ }
477+
478+ private static String mangle(String s) {
479+ if (s.length() == 0)
480+ return NULL_ESCAPE;
481+
482+ // build this lazily, when we first need an escape:
483+ StringBuilder sb = null;
484+
485+ for (int i = 0, slen = s.length(); i < slen; i++) {
486+ char c = s.charAt(i);
487+
488+ boolean needEscape = false;
489+ if (c == ESCAPE_C) {
490+ if (i+1 < slen) {
491+ char c1 = s.charAt(i+1);
492+ if ((i == 0 && c1 == NULL_ESCAPE_C)
493+ || c1 != originalOfReplacement(c1)) {
494+ // an accidental escape
495+ needEscape = true;
496+ }
497+ }
498+ } else {
499+ needEscape = isDangerous(c);
500+ }
501+
502+ if (!needEscape) {
503+ if (sb != null) sb.append(c);
504+ continue;
505+ }
506+
507+ // build sb if this is the first escape
508+ if (sb == null) {
509+ sb = new StringBuilder(s.length()+10);
510+ // mangled names must begin with a backslash:
511+ if (s.charAt(0) != ESCAPE_C && i > 0)
512+ sb.append(NULL_ESCAPE);
513+ // append the string so far, which is unremarkable:
514+ sb.append(s.substring(0, i));
515+ }
516+
517+ // rewrite \ to \-, / to \|, etc.
518+ sb.append(ESCAPE_C);
519+ sb.append(replacementOf(c));
520+ }
521+
522+ if (sb != null) return sb.toString();
523+
524+ return s;
525+ }
526+
527+ private static String demangle(String s) {
528+ // build this lazily, when we first meet an escape:
529+ StringBuilder sb = null;
530+
531+ int stringStart = 0;
532+ if (s.startsWith(NULL_ESCAPE))
533+ stringStart = 2;
534+
535+ for (int i = stringStart, slen = s.length(); i < slen; i++) {
536+ char c = s.charAt(i);
537+
538+ if (c == ESCAPE_C && i+1 < slen) {
539+ // might be an escape sequence
540+ char rc = s.charAt(i+1);
541+ char oc = originalOfReplacement(rc);
542+ if (oc != rc) {
543+ // build sb if this is the first escape
544+ if (sb == null) {
545+ sb = new StringBuilder(s.length());
546+ // append the string so far, which is unremarkable:
547+ sb.append(s.substring(stringStart, i));
548+ }
549+ ++i; // skip both characters
550+ c = oc;
551+ }
552+ }
553+
554+ if (sb != null)
555+ sb.append(c);
556+ }
557+
558+ if (sb != null) return sb.toString();
559+
560+ return s.substring(stringStart);
561+ }
562+
563+ static char ESCAPE_C = '\\';
564+ // empty escape sequence to avoid a null name or illegal prefix
565+ static char NULL_ESCAPE_C = '=';
566+ static String NULL_ESCAPE = ESCAPE_C+""+NULL_ESCAPE_C;
567+
568+ static final String DANGEROUS_CHARS = "\\/.;:$[]<>"; // \\ must be first
569+ static final String REPLACEMENT_CHARS = "-|,?!%{}^_";
570+ static final int DANGEROUS_CHAR_FIRST_INDEX = 1; // index after \\
571+ static char[] DANGEROUS_CHARS_A = DANGEROUS_CHARS.toCharArray();
572+ static char[] REPLACEMENT_CHARS_A = REPLACEMENT_CHARS.toCharArray();
573+ static final Character[] DANGEROUS_CHARS_CA;
574+ static {
575+ Character[] dcca = new Character[DANGEROUS_CHARS.length()];
576+ for (int i = 0; i < dcca.length; i++)
577+ dcca[i] = Character.valueOf(DANGEROUS_CHARS.charAt(i));
578+ DANGEROUS_CHARS_CA = dcca;
579+ }
580+
581+ static final long[] SPECIAL_BITMAP = new long[2]; // 128 bits
582+ static {
583+ String SPECIAL = DANGEROUS_CHARS + REPLACEMENT_CHARS;
584+ //System.out.println("SPECIAL = "+SPECIAL);
585+ for (char c : SPECIAL.toCharArray()) {
586+ SPECIAL_BITMAP[c >>> 6] |= 1L << c;
587+ }
588+ }
589+ static boolean isSpecial(char c) {
590+ if ((c >>> 6) < SPECIAL_BITMAP.length)
591+ return ((SPECIAL_BITMAP[c >>> 6] >> c) & 1) != 0;
592+ else
593+ return false;
594+ }
595+ static char replacementOf(char c) {
596+ if (!isSpecial(c)) return c;
597+ int i = DANGEROUS_CHARS.indexOf(c);
598+ if (i < 0) return c;
599+ return REPLACEMENT_CHARS.charAt(i);
600+ }
601+ static char originalOfReplacement(char c) {
602+ if (!isSpecial(c)) return c;
603+ int i = REPLACEMENT_CHARS.indexOf(c);
604+ if (i < 0) return c;
605+ return DANGEROUS_CHARS.charAt(i);
606+ }
607+ static boolean isDangerous(char c) {
608+ if (!isSpecial(c)) return false;
609+ return (DANGEROUS_CHARS.indexOf(c) >= DANGEROUS_CHAR_FIRST_INDEX);
610+ }
611+ static int indexOfDangerousChar(String s, int from) {
612+ for (int i = from, slen = s.length(); i < slen; i++) {
613+ if (isDangerous(s.charAt(i)))
614+ return i;
615+ }
616+ return -1;
617+ }
618+ static int lastIndexOfDangerousChar(String s, int from) {
619+ for (int i = Math.min(from, s.length()-1); i >= 0; i--) {
620+ if (isDangerous(s.charAt(i)))
621+ return i;
622+ }
623+ return -1;
624+ }
625+
626+
627+}