1berry/openjdk
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

8357281: sun.util.Locale.LanguageTag should be immutable

Browse Source 
Reviewed-by: naoto, liach
This commit is contained in:
Justin Lu 2025-05-23 17:44:39 +00:00
parent 917c1546f3
commit 2f530f89e0
3 changed files with 111 additions and 169 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
src/java.base/share/classes/java/util/Locale.java
View File

@ -1691,37 +1691,37 @@ public final class Locale implements Cloneable, Serializable {
LanguageTag tag = LanguageTag.parseLocale(baseLocale, localeExtensions); LanguageTag tag = LanguageTag.parseLocale(baseLocale, localeExtensions);
StringBuilder buf = new StringBuilder(); StringBuilder buf = new StringBuilder();
String subtag = tag.getLanguage(); String subtag = tag.language();
if (!subtag.isEmpty()) { if (!subtag.isEmpty()) {
buf.append(LanguageTag.canonicalizeLanguage(subtag)); buf.append(LanguageTag.canonicalizeLanguage(subtag));
} }
subtag = tag.getScript(); subtag = tag.script();
if (!subtag.isEmpty()) { if (!subtag.isEmpty()) {
buf.append(LanguageTag.SEP); buf.append(LanguageTag.SEP);
buf.append(LanguageTag.canonicalizeScript(subtag)); buf.append(LanguageTag.canonicalizeScript(subtag));
} }
subtag = tag.getRegion(); subtag = tag.region();
if (!subtag.isEmpty()) { if (!subtag.isEmpty()) {
buf.append(LanguageTag.SEP); buf.append(LanguageTag.SEP);
buf.append(LanguageTag.canonicalizeRegion(subtag)); buf.append(LanguageTag.canonicalizeRegion(subtag));
} }
List<String>subtags = tag.getVariants(); List<String>subtags = tag.variants();
for (String s : subtags) { for (String s : subtags) {
buf.append(LanguageTag.SEP); buf.append(LanguageTag.SEP);
// preserve casing // preserve casing
buf.append(s); buf.append(s);
} }
subtags = tag.getExtensions(); subtags = tag.extensions();
for (String s : subtags) { for (String s : subtags) {
buf.append(LanguageTag.SEP); buf.append(LanguageTag.SEP);
buf.append(LanguageTag.canonicalizeExtension(s)); buf.append(LanguageTag.canonicalizeExtension(s));
} }
subtag = tag.getPrivateuse(); subtag = tag.privateuse();
if (!subtag.isEmpty()) { if (!subtag.isEmpty()) {
if (buf.length() > 0) { if (buf.length() > 0) {
buf.append(LanguageTag.SEP); buf.append(LanguageTag.SEP);

16
src/java.base/share/classes/sun/util/locale/InternalLocaleBuilder.java
View File

@ -1,5 +1,5 @@
/* /*
* Copyright (c) 2010, 2022, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2010, 2025, Oracle and/or its affiliates. All rights reserved.
* DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER. * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
* *
* This code is free software; you can redistribute it and/or modify it * This code is free software; you can redistribute it and/or modify it
@ -347,18 +347,18 @@ public final class InternalLocaleBuilder {
*/ */
public InternalLocaleBuilder setLanguageTag(LanguageTag langtag) { public InternalLocaleBuilder setLanguageTag(LanguageTag langtag) {
clear(); clear();
if (!langtag.getExtlangs().isEmpty()) { if (!langtag.extlangs().isEmpty()) {
language = langtag.getExtlangs().get(0); language = langtag.extlangs().get(0);
} else { } else {
String lang = langtag.getLanguage(); String lang = langtag.language();
if (!lang.equals(LanguageTag.UNDETERMINED)) { if (!lang.equals(LanguageTag.UNDETERMINED)) {
language = lang; language = lang;
} }
} }
script = langtag.getScript(); script = langtag.script();
region = langtag.getRegion(); region = langtag.region();
List<String> bcpVariants = langtag.getVariants(); List<String> bcpVariants = langtag.variants();
if (!bcpVariants.isEmpty()) { if (!bcpVariants.isEmpty()) {
StringBuilder var = new StringBuilder(bcpVariants.get(0)); StringBuilder var = new StringBuilder(bcpVariants.get(0));
int size = bcpVariants.size(); int size = bcpVariants.size();
@ -368,7 +368,7 @@ public final class InternalLocaleBuilder {
variant = var.toString(); variant = var.toString();
} }
setExtensions(langtag.getExtensions(), langtag.getPrivateuse()); setExtensions(langtag.extensions(), langtag.privateuse());
return this; return this;
} }

250
src/java.base/share/classes/sun/util/locale/LanguageTag.java
View File

@ -42,30 +42,21 @@ import java.util.Map;
import java.util.Set; import java.util.Set;
import java.util.StringJoiner; import java.util.StringJoiner;
public class LanguageTag { // List fields are unmodifiable
// public record LanguageTag(String language, String script, String region, String privateuse,
// static fields List<String> extlangs, List<String> variants, List<String> extensions) {
//
public static final String SEP = "-"; public static final String SEP = "-";
public static final String PRIVATEUSE = "x"; public static final String PRIVATEUSE = "x";
public static final String UNDETERMINED = "und"; public static final String UNDETERMINED = "und";
public static final String PRIVUSE_VARIANT_PREFIX = "lvariant"; public static final String PRIVUSE_VARIANT_PREFIX = "lvariant";
private static final String EMPTY_SUBTAG = "";
private static final List<String> EMPTY_SUBTAGS = Collections.emptyList();
//
// Language subtag fields
//
private String language = ""; // language subtag
private String script = ""; // script subtag
private String region = ""; // region subtag
private String privateuse = ""; // privateuse
private List<String> extlangs = Collections.emptyList(); // extlang subtags
private List<String> variants = Collections.emptyList(); // variant subtags
private List<String> extensions = Collections.emptyList(); // extensions
// Map contains legacy language tags and its preferred mappings from // Map contains legacy language tags and its preferred mappings from
// http://www.ietf.org/rfc/rfc5646.txt // http://www.ietf.org/rfc/rfc5646.txt
// Keys are lower-case strings. // Keys are lower-case strings.
private static final Map<String, String[]> LEGACY = new HashMap<>(); private static final Map<String, String[]> LEGACY;
static { static {
// grandfathered = irregular ; non-redundant tags registered // grandfathered = irregular ; non-redundant tags registered
@ -128,14 +119,12 @@ public class LanguageTag {
{"zh-min-nan", "nan"}, {"zh-min-nan", "nan"},
{"zh-xiang", "hsn"}, {"zh-xiang", "hsn"},
}; };
LEGACY = HashMap.newHashMap(entries.length);
for (String[] e : entries) { for (String[] e : entries) {
LEGACY.put(LocaleUtils.toLowerString(e[0]), e); LEGACY.put(LocaleUtils.toLowerString(e[0]), e);
} }
} }
private LanguageTag() {
}
/* /*
* BNF in RFC5646 * BNF in RFC5646
* *
@ -195,17 +184,27 @@ public class LanguageTag {
itr = new StringTokenIterator(languageTag, SEP); itr = new StringTokenIterator(languageTag, SEP);
} }
LanguageTag tag = new LanguageTag(); String language = parseLanguage(itr, pp);
List<String> extlangs;
String script;
String region;
List<String> variants;
List<String> extensions;
// langtag must start with either language or privateuse // langtag must start with either language or privateuse
if (tag.parseLanguage(itr, pp)) { if (!language.isEmpty()) {
tag.parseExtlangs(itr, pp); extlangs = parseExtlangs(itr, pp);
tag.parseScript(itr, pp); script = parseScript(itr, pp);
tag.parseRegion(itr, pp); region = parseRegion(itr, pp);
tag.parseVariants(itr, pp); variants = parseVariants(itr, pp);
tag.parseExtensions(itr, pp, errorMsg); extensions = parseExtensions(itr, pp, errorMsg);
} else {
extlangs = EMPTY_SUBTAGS;
script = EMPTY_SUBTAG;
region = EMPTY_SUBTAG;
variants = EMPTY_SUBTAGS;
extensions = EMPTY_SUBTAGS;
} }
tag.parsePrivateuse(itr, pp, errorMsg); String privateuse = parsePrivateuse(itr, pp, errorMsg);
if (!itr.isDone() && pp.getErrorIndex() == -1) { if (!itr.isDone() && pp.getErrorIndex() == -1) {
String s = itr.current(); String s = itr.current();
@ -221,110 +220,94 @@ public class LanguageTag {
throw new IllformedLocaleException(errorMsg.toString(), pp.getErrorIndex()); throw new IllformedLocaleException(errorMsg.toString(), pp.getErrorIndex());
} }
return tag; return new LanguageTag(language, script, region, privateuse, extlangs, variants, extensions);
} }
// //
// Language subtag parsers // Language subtag parsers
// //
private boolean parseLanguage(StringTokenIterator itr, ParsePosition pp) { private static String parseLanguage(StringTokenIterator itr, ParsePosition pp) {
if (itr.isDone() || pp.getErrorIndex() != -1) { if (itr.isDone() || pp.getErrorIndex() != -1) {
return false; return EMPTY_SUBTAG;
} }
boolean found = false;
String s = itr.current(); String s = itr.current();
if (isLanguage(s)) { if (isLanguage(s)) {
found = true;
language = s;
pp.setIndex(itr.currentEnd()); pp.setIndex(itr.currentEnd());
itr.next(); itr.next();
return s;
} }
return found; return EMPTY_SUBTAG;
} }
private boolean parseExtlangs(StringTokenIterator itr, ParsePosition pp) { private static List<String> parseExtlangs(StringTokenIterator itr, ParsePosition pp) {
if (itr.isDone() || pp.getErrorIndex() != -1) { if (itr.isDone() || pp.getErrorIndex() != -1) {
return false; return EMPTY_SUBTAGS;
} }
List<String> extlangs = null;
boolean found = false;
while (!itr.isDone()) { while (!itr.isDone()) {
String s = itr.current(); String s = itr.current();
if (!isExtlang(s)) { if (!isExtlang(s)) {
break; break;
} }
found = true; if (extlangs == null) {
if (extlangs.isEmpty()) {
extlangs = new ArrayList<>(3); extlangs = new ArrayList<>(3);
} }
extlangs.add(s); extlangs.add(s);
pp.setIndex(itr.currentEnd()); pp.setIndex(itr.currentEnd());
itr.next(); itr.next();
if (extlangs.size() == 3) { if (extlangs.size() == 3) {
// Maximum 3 extlangs // Maximum 3 extlangs
break; break;
} }
} }
return extlangs == null ? EMPTY_SUBTAGS :
return found; Collections.unmodifiableList(extlangs);
} }
private boolean parseScript(StringTokenIterator itr, ParsePosition pp) { private static String parseScript(StringTokenIterator itr, ParsePosition pp) {
if (itr.isDone() || pp.getErrorIndex() != -1) { if (itr.isDone() || pp.getErrorIndex() != -1) {
return false; return EMPTY_SUBTAG;
} }
boolean found = false;
String s = itr.current(); String s = itr.current();
if (isScript(s)) { if (isScript(s)) {
found = true;
script = s;
pp.setIndex(itr.currentEnd()); pp.setIndex(itr.currentEnd());
itr.next(); itr.next();
return s;
} }
return found; return EMPTY_SUBTAG;
} }
private boolean parseRegion(StringTokenIterator itr, ParsePosition pp) { private static String parseRegion(StringTokenIterator itr, ParsePosition pp) {
if (itr.isDone() || pp.getErrorIndex() != -1) { if (itr.isDone() || pp.getErrorIndex() != -1) {
return false; return EMPTY_SUBTAG;
} }
boolean found = false;
String s = itr.current(); String s = itr.current();
if (isRegion(s)) { if (isRegion(s)) {
found = true;
region = s;
pp.setIndex(itr.currentEnd()); pp.setIndex(itr.currentEnd());
itr.next(); itr.next();
return s;
} }
return found; return EMPTY_SUBTAG;
} }
private boolean parseVariants(StringTokenIterator itr, ParsePosition pp) { private static List<String> parseVariants(StringTokenIterator itr, ParsePosition pp) {
if (itr.isDone() || pp.getErrorIndex() != -1) { if (itr.isDone() || pp.getErrorIndex() != -1) {
return false; return EMPTY_SUBTAGS;
} }
List<String> variants = null;
boolean found = false;
while (!itr.isDone()) { while (!itr.isDone()) {
String s = itr.current(); String s = itr.current();
if (!isVariant(s)) { if (!isVariant(s)) {
break; break;
} }
found = true; if (variants == null) {
if (variants.isEmpty()) {
variants = new ArrayList<>(3); variants = new ArrayList<>(3);
} }
variants.add(s); variants.add(s);
@ -332,16 +315,16 @@ public class LanguageTag {
itr.next(); itr.next();
} }
return found; return variants == null ? EMPTY_SUBTAGS :
Collections.unmodifiableList(variants);
} }
private boolean parseExtensions(StringTokenIterator itr, ParsePosition pp, private static List<String> parseExtensions(StringTokenIterator itr, ParsePosition pp,
StringBuilder err) { StringBuilder err) {
if (itr.isDone() || pp.getErrorIndex() != -1) { if (itr.isDone() || pp.getErrorIndex() != -1) {
return false; return EMPTY_SUBTAGS;
} }
List<String> extensions = null;
boolean found = false;
while (!itr.isDone()) { while (!itr.isDone()) {
String s = itr.current(); String s = itr.current();
@ -368,26 +351,24 @@ public class LanguageTag {
break; break;
} }
if (extensions.isEmpty()) { if (extensions == null) {
extensions = new ArrayList<>(4); extensions = new ArrayList<>(4);
} }
extensions.add(sb.toString()); extensions.add(sb.toString());
found = true;
} else { } else {
break; break;
} }
} }
return found; return extensions == null ? EMPTY_SUBTAGS :
Collections.unmodifiableList(extensions);
} }
private boolean parsePrivateuse(StringTokenIterator itr, ParsePosition pp, private static String parsePrivateuse(StringTokenIterator itr, ParsePosition pp,
StringBuilder err) { StringBuilder err) {
if (itr.isDone() || pp.getErrorIndex() != -1) { if (itr.isDone() || pp.getErrorIndex() != -1) {
return false; return EMPTY_SUBTAG;
} }
boolean found = false;
String s = itr.current(); String s = itr.current();
if (isPrivateusePrefix(s)) { if (isPrivateusePrefix(s)) {
int start = itr.currentStart(); int start = itr.currentStart();
@ -410,12 +391,11 @@ public class LanguageTag {
pp.setErrorIndex(start); pp.setErrorIndex(start);
err.append("Incomplete privateuse"); err.append("Incomplete privateuse");
} else { } else {
privateuse = sb.toString(); return sb.toString();
found = true;
} }
} }
return found; return EMPTY_SUBTAG;
} }
public static String caseFoldTag(String tag) { public static String caseFoldTag(String tag) {
@ -462,48 +442,50 @@ public class LanguageTag {
} }
public static LanguageTag parseLocale(BaseLocale baseLocale, LocaleExtensions localeExtensions) { public static LanguageTag parseLocale(BaseLocale baseLocale, LocaleExtensions localeExtensions) {
LanguageTag tag = new LanguageTag();
String language = baseLocale.getLanguage(); String language = EMPTY_SUBTAG;
String script = baseLocale.getScript(); String script = EMPTY_SUBTAG;
String region = baseLocale.getRegion(); String region = EMPTY_SUBTAG;
String variant = baseLocale.getVariant();
String baseLanguage = baseLocale.getLanguage();
String baseScript = baseLocale.getScript();
String baseRegion = baseLocale.getRegion();
String baseVariant = baseLocale.getVariant();
boolean hasSubtag = false; boolean hasSubtag = false;
String privuseVar = null; // store ill-formed variant subtags String privuseVar = null; // store ill-formed variant subtags
if (isLanguage(language)) { if (isLanguage(baseLanguage)) {
// Convert a deprecated language code to its new code // Convert a deprecated language code to its new code
if (language.equals("iw")) { baseLanguage = switch (baseLanguage) {
language = "he"; case "iw" -> "he";
} else if (language.equals("ji")) { case "ji" -> "yi";
language = "yi"; case "in" -> "id";
} else if (language.equals("in")) { default -> baseLanguage;
language = "id"; };
} language = baseLanguage;
tag.language = language;
} }
if (isScript(script)) { if (isScript(baseScript)) {
tag.script = canonicalizeScript(script); script = canonicalizeScript(baseScript);
hasSubtag = true; hasSubtag = true;
} }
if (isRegion(region)) { if (isRegion(baseRegion)) {
tag.region = canonicalizeRegion(region); region = canonicalizeRegion(baseRegion);
hasSubtag = true; hasSubtag = true;
} }
// Special handling for no_NO_NY - use nn_NO for language tag // Special handling for no_NO_NY - use nn_NO for language tag
if (tag.language.equals("no") && tag.region.equals("NO") && variant.equals("NY")) { if (language.equals("no") && region.equals("NO") && baseVariant.equals("NY")) {
tag.language = "nn"; language = "nn";
variant = ""; baseVariant = EMPTY_SUBTAG;
} }
if (!variant.isEmpty()) {
List<String> variants = null; List<String> variants = null;
StringTokenIterator varitr = new StringTokenIterator(variant, BaseLocale.SEP); if (!baseVariant.isEmpty()) {
StringTokenIterator varitr = new StringTokenIterator(baseVariant, BaseLocale.SEP);
while (!varitr.isDone()) { while (!varitr.isDone()) {
String var = varitr.current(); String var = varitr.current();
if (!isVariant(var)) { if (!isVariant(var)) {
@ -516,7 +498,6 @@ public class LanguageTag {
varitr.next(); varitr.next();
} }
if (variants != null) { if (variants != null) {
tag.variants = variants;
hasSubtag = true; hasSubtag = true;
} }
if (!varitr.isDone()) { if (!varitr.isDone()) {
@ -556,7 +537,6 @@ public class LanguageTag {
} }
if (extensions != null) { if (extensions != null) {
tag.extensions = extensions;
hasSubtag = true; hasSubtag = true;
} }
@ -570,59 +550,21 @@ public class LanguageTag {
} }
} }
if (privateuse != null) { if (language.isEmpty() && (hasSubtag || privateuse == null)) {
tag.privateuse = privateuse;
}
if (tag.language.isEmpty() && (hasSubtag || privateuse == null)) {
// use lang "und" when 1) no language is available AND // use lang "und" when 1) no language is available AND
// 2) any of other subtags other than private use are available or // 2) any of other subtags other than private use are available or
// no private use tag is available // no private use tag is available
tag.language = UNDETERMINED; language = UNDETERMINED;
} }
return tag; privateuse = privateuse == null ? EMPTY_SUBTAG : privateuse;
} extensions = extensions == null ? EMPTY_SUBTAGS :
Collections.unmodifiableList(extensions);
variants = variants == null ? EMPTY_SUBTAGS :
Collections.unmodifiableList(variants);
// // extlangs always empty for locale parse
// Getter methods for language subtag fields return new LanguageTag(language, script, region, privateuse, EMPTY_SUBTAGS, variants, extensions);
//
public String getLanguage() {
return language;
}
public List<String> getExtlangs() {
if (extlangs.isEmpty()) {
return Collections.emptyList();
}
return Collections.unmodifiableList(extlangs);
}
public String getScript() {
return script;
}
public String getRegion() {
return region;
}
public List<String> getVariants() {
if (variants.isEmpty()) {
return Collections.emptyList();
}
return Collections.unmodifiableList(variants);
}
public List<String> getExtensions() {
if (extensions.isEmpty()) {
return Collections.emptyList();
}
return Collections.unmodifiableList(extensions);
}
public String getPrivateuse() {
return privateuse;
} }
// //