Skip to content

Commit 1848bec

Browse files
rbrirbri
committed
do not create new strings if lower/upper conversation for tags/attributes is configured
1 parent 537d3c9 commit 1848bec

File tree

1 file changed

+31
-27
lines changed

1 file changed

+31
-27
lines changed

‎src/main/java/org/htmlunit/cyberneko/HTMLScanner.java‎

Lines changed: 31 additions & 27 deletions
Original file line numberDiff line numberDiff line change
@@ -1129,17 +1129,6 @@ else if (ch1 == '/' && str.charAt(0) == '/') {
11291129
return str;
11301130
}
11311131

1132-
// Modifies the given name based on the specified mode.
1133-
protected static String modifyName(final String name, final short mode) {
1134-
if (NAMES_UPPERCASE == mode) {
1135-
return name.toUpperCase(Locale.ROOT);
1136-
}
1137-
if (NAMES_LOWERCASE == mode) {
1138-
return name.toLowerCase(Locale.ROOT);
1139-
}
1140-
return name;
1141-
}
1142-
11431132
// Converts HTML names string value to constant value.
11441133
//
11451134
// @see #NAMES_NO_CHANGE
@@ -1202,15 +1191,12 @@ protected void scanDoctype() throws IOException {
12021191
String sysid = null;
12031192

12041193
if (fCurrentEntity.skipSpaces()) {
1205-
root = scanName(true);
1194+
root = scanName(true, fNamesElems);
12061195
if (root == null) {
12071196
if (fReportErrors_) {
12081197
fErrorReporter.reportError("HTML1014", null);
12091198
}
12101199
}
1211-
else {
1212-
root = modifyName(root, fNamesElems);
1213-
}
12141200
if (fCurrentEntity.skipSpaces()) {
12151201
if (fCurrentEntity.skip("PUBLIC")) {
12161202
fCurrentEntity.skipSpaces();
@@ -1290,7 +1276,7 @@ else if (c == '<') {
12901276
}
12911277

12921278
// Scan a name.
1293-
protected String scanName(final boolean strict) throws IOException {
1279+
protected String scanName(final boolean strict, final short mode) throws IOException {
12941280
if (DEBUG_BUFFER) {
12951281
fCurrentEntity.debugBufferIfNeeded("(scanName: ");
12961282
}
@@ -1306,6 +1292,7 @@ protected String scanName(final boolean strict) throws IOException {
13061292
while (true) {
13071293
while (fCurrentEntity.hasNext()) {
13081294
final char c = fCurrentEntity.getNextChar();
1295+
13091296
// this has been split up to cater to the needs of branch prediction
13101297
if (strict && (!Character.isLetterOrDigit(c) && c != '-' && c != '.' && c != ':' && c != '_')) {
13111298
fCurrentEntity.rewind();
@@ -1323,6 +1310,16 @@ else if (!strict
13231310
fCurrentEntity.rewind();
13241311
break;
13251312
}
1313+
1314+
if (NAMES_NO_CHANGE == mode) {
1315+
// nothing to do
1316+
}
1317+
else if (NAMES_UPPERCASE == mode && !Character.isUpperCase(c)) {
1318+
fCurrentEntity.buffer_[fCurrentEntity.offset_ - 1] = Character.toUpperCase(c);
1319+
}
1320+
else if (NAMES_LOWERCASE == mode && !Character.isLowerCase(c)) {
1321+
fCurrentEntity.buffer_[fCurrentEntity.offset_ - 1] = Character.toLowerCase(c);
1322+
}
13261323
}
13271324
if (fCurrentEntity.offset_ == fCurrentEntity.length_) {
13281325
final int length = fCurrentEntity.length_ - offset;
@@ -1381,7 +1378,18 @@ protected String scanTagName() throws IOException {
13811378
break;
13821379
}
13831380
}
1381+
1382+
if (NAMES_NO_CHANGE == fNamesElems) {
1383+
// nothing to do
1384+
}
1385+
else if (NAMES_UPPERCASE == fNamesElems && !Character.isUpperCase(c)) {
1386+
fCurrentEntity.buffer_[fCurrentEntity.offset_ - 1] = Character.toUpperCase(c);
1387+
}
1388+
else if (NAMES_LOWERCASE == fNamesElems && !Character.isLowerCase(c)) {
1389+
fCurrentEntity.buffer_[fCurrentEntity.offset_ - 1] = Character.toLowerCase(c);
1390+
}
13841391
}
1392+
13851393
if (fCurrentEntity.offset_ == fCurrentEntity.length_) {
13861394
final int length = fCurrentEntity.length_ - offset;
13871395
System.arraycopy(fCurrentEntity.buffer_, offset, fCurrentEntity.buffer_, 0, length);
@@ -2311,7 +2319,7 @@ else if (ename != null) {
23112319
}
23122320
if (fInsertDoctype_) {
23132321
fDocumentHandler.doctypeDecl(
2314-
modifyName("HTML", fNamesElems),
2322+
NAMES_LOWERCASE == fNamesElems ? "html" : "HTML",
23152323
fDoctypePubid,
23162324
fDoctypeSysid,
23172325
synthesizedAugs());
@@ -2690,7 +2698,7 @@ protected void scanPI() throws IOException {
26902698
}
26912699

26922700
// scan processing instruction
2693-
final String target = scanName(true);
2701+
final String target = scanName(true, fNamesElems);
26942702
if (target != null && !"xml".equalsIgnoreCase(target)) {
26952703
while (true) {
26962704
int c = fCurrentEntity.read();
@@ -2814,7 +2822,7 @@ else if (c == '>') {
28142822
* @throws IOException in case of io problems
28152823
*/
28162824
protected String scanStartElement(final boolean[] empty) throws IOException {
2817-
String ename = scanTagName();
2825+
final String ename = scanTagName();
28182826
final int length = ename != null ? ename.length() : 0;
28192827
if (length == 0) {
28202828
if (fReportErrors_) {
@@ -2826,7 +2834,6 @@ protected String scanStartElement(final boolean[] empty) throws IOException {
28262834
}
28272835
return null;
28282836
}
2829-
ename = modifyName(ename, fNamesElems);
28302837
if (attributes_.getLength() != 0) {
28312838
attributes_.removeAllAttributes();
28322839
}
@@ -3042,7 +3049,7 @@ protected boolean scanAttribute(final XMLAttributesImpl attributes, final boolea
30423049
}
30433050

30443051
fCurrentEntity.rewind();
3045-
String aname = scanName(false);
3052+
String aname = scanName(false, fNamesAttrs);
30463053
if (aname == null) {
30473054
if (fReportErrors_) {
30483055
fErrorReporter.reportError("HTML1011", null);
@@ -3055,7 +3062,7 @@ protected boolean scanAttribute(final XMLAttributesImpl attributes, final boolea
30553062
empty[0] = fCurrentEntity.skipMarkup(false);
30563063
return false;
30573064
}
3058-
aname = '=' + scanName(false);
3065+
aname = '=' + scanName(false, fNamesElems);
30593066
}
30603067
if (fReportErrors_ && !skippedSpaces) {
30613068
fErrorReporter.reportError("HTML1013", new Object[] {aname});
@@ -3068,7 +3075,6 @@ protected boolean scanAttribute(final XMLAttributesImpl attributes, final boolea
30683075
}
30693076
throw new EOFException();
30703077
}
3071-
aname = modifyName(aname, fNamesAttrs);
30723078
if (c == '/') {
30733079
qName_.setValues(null, aname, aname, null);
30743080
attributes.addAttribute(qName_, "CDATA", "", true);
@@ -3293,14 +3299,13 @@ else if (c != currentQuote) {
32933299

32943300
// Scans an end element.
32953301
protected void scanEndElement() throws IOException {
3296-
String ename = scanTagName();
3302+
final String ename = scanTagName();
32973303
if (fReportErrors_ && ename == null) {
32983304
fErrorReporter.reportError("HTML1012", null);
32993305
}
33003306
fCurrentEntity.skipMarkup(false);
33013307
if (ename != null) {
33023308
if (fElementCount >= fElementDepth) {
3303-
ename = modifyName(ename, fNamesElems);
33043309
qName_.setValues(null, ename, ename, null);
33053310
if (DEBUG_CALLBACKS) {
33063311
System.out.println("endElement(" + qName_ + ")");
@@ -3385,14 +3390,13 @@ public boolean scan(final boolean complete) throws IOException {
33853390
case STATE_MARKUP_BRACKET: {
33863391
final int c = fCurrentEntity.read();
33873392
if (c == '/') {
3388-
String ename = scanName(true);
3393+
final String ename = scanName(true, fNamesElems);
33893394
if (ename != null) {
33903395
fCurrentEntity.skipSpaces();
33913396

33923397
if (ename.equalsIgnoreCase(fElementName)) {
33933398
if (fCurrentEntity.read() == '>') {
33943399
if (fElementCount >= fElementDepth) {
3395-
ename = modifyName(ename, fNamesElems);
33963400
fQName_.setValues(null, ename, ename, null);
33973401
if (DEBUG_CALLBACKS) {
33983402
System.out.println("endElement(" + fQName_ + ")");

0 commit comments

Comments
 (0)