Improve regexp performance by avoiding conversion of input to char[].
git-svn-id: svn://10.0.0.236/trunk@262084 18797224-902f-48f8-a5cc-f745e15eee43
This commit is contained in:
hannes%helma.at
parent
99598a50b1
commit
2b8bf99d0e
@ -1330,12 +1330,12 @@ if (regexp.anchorCh >= 0) {
|
||||
*/
|
||||
private static boolean
|
||||
flatNMatcher(REGlobalData gData, int matchChars,
|
||||
int length, char[] chars, int end)
|
||||
int length, String input, int end)
|
||||
{
|
||||
if ((gData.cp + length) > end)
|
||||
return false;
|
||||
for (int i = 0; i < length; i++) {
|
||||
if (gData.regexp.source[matchChars + i] != chars[gData.cp + i]) {
|
||||
if (gData.regexp.source[matchChars + i] != input.charAt(gData.cp + i)) {
|
||||
return false;
|
||||
}
|
||||
}
|
||||
@ -1345,13 +1345,13 @@ if (regexp.anchorCh >= 0) {
|
||||
|
||||
private static boolean
|
||||
flatNIMatcher(REGlobalData gData, int matchChars,
|
||||
int length, char[] chars, int end)
|
||||
int length, String input, int end)
|
||||
{
|
||||
if ((gData.cp + length) > end)
|
||||
return false;
|
||||
for (int i = 0; i < length; i++) {
|
||||
if (upcase(gData.regexp.source[matchChars + i])
|
||||
!= upcase(chars[gData.cp + i]))
|
||||
!= upcase(input.charAt(gData.cp + i)))
|
||||
{
|
||||
return false;
|
||||
}
|
||||
@ -1385,7 +1385,7 @@ if (regexp.anchorCh >= 0) {
|
||||
*/
|
||||
private static boolean
|
||||
backrefMatcher(REGlobalData gData, int parenIndex,
|
||||
char[] chars, int end)
|
||||
String input, int end)
|
||||
{
|
||||
int len;
|
||||
int i;
|
||||
@ -1399,13 +1399,13 @@ if (regexp.anchorCh >= 0) {
|
||||
|
||||
if ((gData.regexp.flags & JSREG_FOLD) != 0) {
|
||||
for (i = 0; i < len; i++) {
|
||||
if (upcase(chars[parenContent + i]) != upcase(chars[gData.cp + i]))
|
||||
if (upcase(input.charAt(parenContent + i)) != upcase(input.charAt(gData.cp + i)))
|
||||
return false;
|
||||
}
|
||||
}
|
||||
else {
|
||||
for (i = 0; i < len; i++) {
|
||||
if (chars[parenContent + i] != chars[gData.cp + i])
|
||||
if (input.charAt(parenContent + i) != input.charAt(gData.cp + i))
|
||||
return false;
|
||||
}
|
||||
}
|
||||
@ -1674,7 +1674,7 @@ if (regexp.anchorCh >= 0) {
|
||||
}
|
||||
|
||||
private static boolean
|
||||
executeREBytecode(REGlobalData gData, char[] chars, int end)
|
||||
executeREBytecode(REGlobalData gData, String input, int end)
|
||||
{
|
||||
int pc = 0;
|
||||
byte program[] = gData.regexp.program;
|
||||
@ -1685,7 +1685,7 @@ if (regexp.anchorCh >= 0) {
|
||||
currentContinuation_pc = 0;
|
||||
currentContinuation_op = REOP_END;
|
||||
if (debug) {
|
||||
System.out.println("Input = \"" + new String(chars) + "\", start at " + gData.cp);
|
||||
System.out.println("Input = \"" + input + "\", start at " + gData.cp);
|
||||
}
|
||||
int op = program[pc++];
|
||||
for (;;) {
|
||||
@ -1700,7 +1700,7 @@ System.out.println("Testing at " + gData.cp + ", op = " + op);
|
||||
if (gData.cp != 0) {
|
||||
if (gData.multiline ||
|
||||
((gData.regexp.flags & JSREG_MULTILINE) != 0)) {
|
||||
if (!isLineTerm(chars[gData.cp - 1])) {
|
||||
if (!isLineTerm(input.charAt(gData.cp - 1))) {
|
||||
result = false;
|
||||
break;
|
||||
}
|
||||
@ -1716,7 +1716,7 @@ System.out.println("Testing at " + gData.cp + ", op = " + op);
|
||||
if (gData.cp != end) {
|
||||
if (gData.multiline ||
|
||||
((gData.regexp.flags & JSREG_MULTILINE) != 0)) {
|
||||
if (!isLineTerm(chars[gData.cp])) {
|
||||
if (!isLineTerm(input.charAt(gData.cp))) {
|
||||
result = false;
|
||||
break;
|
||||
}
|
||||
@ -1729,51 +1729,51 @@ System.out.println("Testing at " + gData.cp + ", op = " + op);
|
||||
result = true;
|
||||
break;
|
||||
case REOP_WBDRY:
|
||||
result = ((gData.cp == 0 || !isWord(chars[gData.cp - 1]))
|
||||
^ !((gData.cp < end) && isWord(chars[gData.cp])));
|
||||
result = ((gData.cp == 0 || !isWord(input.charAt(gData.cp - 1)))
|
||||
^ !((gData.cp < end) && isWord(input.charAt(gData.cp))));
|
||||
break;
|
||||
case REOP_WNONBDRY:
|
||||
result = ((gData.cp == 0 || !isWord(chars[gData.cp - 1]))
|
||||
^ ((gData.cp < end) && isWord(chars[gData.cp])));
|
||||
result = ((gData.cp == 0 || !isWord(input.charAt(gData.cp - 1)))
|
||||
^ ((gData.cp < end) && isWord(input.charAt(gData.cp))));
|
||||
break;
|
||||
case REOP_DOT:
|
||||
result = (gData.cp != end && !isLineTerm(chars[gData.cp]));
|
||||
result = (gData.cp != end && !isLineTerm(input.charAt(gData.cp)));
|
||||
if (result) {
|
||||
gData.cp++;
|
||||
}
|
||||
break;
|
||||
case REOP_DIGIT:
|
||||
result = (gData.cp != end && isDigit(chars[gData.cp]));
|
||||
result = (gData.cp != end && isDigit(input.charAt(gData.cp)));
|
||||
if (result) {
|
||||
gData.cp++;
|
||||
}
|
||||
break;
|
||||
case REOP_NONDIGIT:
|
||||
result = (gData.cp != end && !isDigit(chars[gData.cp]));
|
||||
result = (gData.cp != end && !isDigit(input.charAt(gData.cp)));
|
||||
if (result) {
|
||||
gData.cp++;
|
||||
}
|
||||
break;
|
||||
case REOP_SPACE:
|
||||
result = (gData.cp != end && isREWhiteSpace(chars[gData.cp]));
|
||||
result = (gData.cp != end && isREWhiteSpace(input.charAt(gData.cp)));
|
||||
if (result) {
|
||||
gData.cp++;
|
||||
}
|
||||
break;
|
||||
case REOP_NONSPACE:
|
||||
result = (gData.cp != end && !isREWhiteSpace(chars[gData.cp]));
|
||||
result = (gData.cp != end && !isREWhiteSpace(input.charAt(gData.cp)));
|
||||
if (result) {
|
||||
gData.cp++;
|
||||
}
|
||||
break;
|
||||
case REOP_ALNUM:
|
||||
result = (gData.cp != end && isWord(chars[gData.cp]));
|
||||
result = (gData.cp != end && isWord(input.charAt(gData.cp)));
|
||||
if (result) {
|
||||
gData.cp++;
|
||||
}
|
||||
break;
|
||||
case REOP_NONALNUM:
|
||||
result = (gData.cp != end && !isWord(chars[gData.cp]));
|
||||
result = (gData.cp != end && !isWord(input.charAt(gData.cp)));
|
||||
if (result) {
|
||||
gData.cp++;
|
||||
}
|
||||
@ -1784,7 +1784,7 @@ System.out.println("Testing at " + gData.cp + ", op = " + op);
|
||||
pc += INDEX_LEN;
|
||||
int length = getIndex(program, pc);
|
||||
pc += INDEX_LEN;
|
||||
result = flatNMatcher(gData, offset, length, chars, end);
|
||||
result = flatNMatcher(gData, offset, length, input, end);
|
||||
}
|
||||
break;
|
||||
case REOP_FLATi:
|
||||
@ -1793,13 +1793,13 @@ System.out.println("Testing at " + gData.cp + ", op = " + op);
|
||||
pc += INDEX_LEN;
|
||||
int length = getIndex(program, pc);
|
||||
pc += INDEX_LEN;
|
||||
result = flatNIMatcher(gData, offset, length, chars, end);
|
||||
result = flatNIMatcher(gData, offset, length, input, end);
|
||||
}
|
||||
break;
|
||||
case REOP_FLAT1:
|
||||
{
|
||||
char matchCh = (char)(program[pc++] & 0xFF);
|
||||
result = (gData.cp != end && chars[gData.cp] == matchCh);
|
||||
result = (gData.cp != end && input.charAt(gData.cp) == matchCh);
|
||||
if (result) {
|
||||
gData.cp++;
|
||||
}
|
||||
@ -1809,7 +1809,7 @@ System.out.println("Testing at " + gData.cp + ", op = " + op);
|
||||
{
|
||||
char matchCh = (char)(program[pc++] & 0xFF);
|
||||
result = (gData.cp != end
|
||||
&& upcase(chars[gData.cp]) == upcase(matchCh));
|
||||
&& upcase(input.charAt(gData.cp)) == upcase(matchCh));
|
||||
if (result) {
|
||||
gData.cp++;
|
||||
}
|
||||
@ -1819,7 +1819,7 @@ System.out.println("Testing at " + gData.cp + ", op = " + op);
|
||||
{
|
||||
char matchCh = (char)getIndex(program, pc);
|
||||
pc += INDEX_LEN;
|
||||
result = (gData.cp != end && chars[gData.cp] == matchCh);
|
||||
result = (gData.cp != end && input.charAt(gData.cp) == matchCh);
|
||||
if (result) {
|
||||
gData.cp++;
|
||||
}
|
||||
@ -1830,7 +1830,7 @@ System.out.println("Testing at " + gData.cp + ", op = " + op);
|
||||
char matchCh = (char)getIndex(program, pc);
|
||||
pc += INDEX_LEN;
|
||||
result = (gData.cp != end
|
||||
&& upcase(chars[gData.cp]) == upcase(matchCh));
|
||||
&& upcase(input.charAt(gData.cp)) == upcase(matchCh));
|
||||
if (result) {
|
||||
gData.cp++;
|
||||
}
|
||||
@ -1889,7 +1889,7 @@ System.out.println("Testing at " + gData.cp + ", op = " + op);
|
||||
{
|
||||
int parenIndex = getIndex(program, pc);
|
||||
pc += INDEX_LEN;
|
||||
result = backrefMatcher(gData, parenIndex, chars, end);
|
||||
result = backrefMatcher(gData, parenIndex, input, end);
|
||||
}
|
||||
break;
|
||||
|
||||
@ -1899,7 +1899,7 @@ System.out.println("Testing at " + gData.cp + ", op = " + op);
|
||||
pc += INDEX_LEN;
|
||||
if (gData.cp != end) {
|
||||
if (classMatcher(gData, gData.regexp.classList[index],
|
||||
chars[gData.cp]))
|
||||
input.charAt(gData.cp)))
|
||||
{
|
||||
gData.cp++;
|
||||
result = true;
|
||||
@ -2201,7 +2201,7 @@ System.out.println("Testing at " + gData.cp + ", op = " + op);
|
||||
|
||||
private static boolean
|
||||
matchRegExp(REGlobalData gData, RECompiled re,
|
||||
char[] chars, int start, int end, boolean multiline)
|
||||
String input, int start, int end, boolean multiline)
|
||||
{
|
||||
if (re.parenCount != 0) {
|
||||
gData.parens = new long[re.parenCount];
|
||||
@ -2233,7 +2233,7 @@ System.out.println("Testing at " + gData.cp + ", op = " + op);
|
||||
if (i == end) {
|
||||
return false;
|
||||
}
|
||||
char matchCh = chars[i];
|
||||
char matchCh = input.charAt(i);
|
||||
if (matchCh == anchorCh ||
|
||||
((gData.regexp.flags & JSREG_FOLD) != 0
|
||||
&& upcase(matchCh) == upcase((char)anchorCh)))
|
||||
@ -2247,7 +2247,7 @@ System.out.println("Testing at " + gData.cp + ", op = " + op);
|
||||
for (int j = 0; j < re.parenCount; j++) {
|
||||
gData.set_parens(j, -1, 0);
|
||||
}
|
||||
boolean result = executeREBytecode(gData, chars, end);
|
||||
boolean result = executeREBytecode(gData, input, end);
|
||||
|
||||
gData.backTrackStackTop = null;
|
||||
gData.stateStackTop = null;
|
||||
@ -2268,24 +2268,21 @@ System.out.println("Testing at " + gData.cp + ", op = " + op);
|
||||
REGlobalData gData = new REGlobalData();
|
||||
|
||||
int start = indexp[0];
|
||||
char[] charArray = str.toCharArray();
|
||||
int end = charArray.length;
|
||||
int end = str.length();
|
||||
if (start > end)
|
||||
start = end;
|
||||
//
|
||||
// Call the recursive matcher to do the real work.
|
||||
//
|
||||
boolean matches = matchRegExp(gData, re, charArray, start, end,
|
||||
boolean matches = matchRegExp(gData, re, str, start, end,
|
||||
res.multiline);
|
||||
if (!matches) {
|
||||
if (matchType != PREFIX) return null;
|
||||
return Undefined.instance;
|
||||
}
|
||||
int index = gData.cp;
|
||||
int i = index;
|
||||
indexp[0] = i;
|
||||
int matchlen = i - (start + gData.skipped);
|
||||
int ep = index;
|
||||
int ep = indexp[0] = index;
|
||||
int matchlen = ep - (start + gData.skipped);
|
||||
index -= matchlen;
|
||||
Object result;
|
||||
Scriptable obj;
|
||||
@ -2308,7 +2305,7 @@ System.out.println("Testing at " + gData.cp + ", op = " + op);
|
||||
result = cx.newArray(scope, 0);
|
||||
obj = (Scriptable) result;
|
||||
|
||||
String matchstr = new String(charArray, index, matchlen);
|
||||
String matchstr = str.substring(index, index + matchlen);
|
||||
obj.put(0, obj, matchstr);
|
||||
}
|
||||
|
||||
@ -2324,7 +2321,7 @@ System.out.println("Testing at " + gData.cp + ", op = " + op);
|
||||
String parstr;
|
||||
if (cap_index != -1) {
|
||||
int cap_length = gData.parens_length(num);
|
||||
parsub = new SubString(charArray, cap_index, cap_length);
|
||||
parsub = new SubString(str, cap_index, cap_length);
|
||||
res.parens[num] = parsub;
|
||||
if (matchType == TEST) continue;
|
||||
parstr = parsub.toString();
|
||||
@ -2352,11 +2349,11 @@ System.out.println("Testing at " + gData.cp + ", op = " + op);
|
||||
res.leftContext = new SubString();
|
||||
res.rightContext = new SubString();
|
||||
}
|
||||
res.lastMatch.charArray = charArray;
|
||||
res.lastMatch.str = str;
|
||||
res.lastMatch.index = index;
|
||||
res.lastMatch.length = matchlen;
|
||||
|
||||
res.leftContext.charArray = charArray;
|
||||
res.leftContext.str = str;
|
||||
if (cx.getLanguageVersion() == Context.VERSION_1_2) {
|
||||
/*
|
||||
* JS1.2 emulated Perl4.0.1.8 (patch level 36) for global regexps used
|
||||
@ -2383,7 +2380,7 @@ System.out.println("Testing at " + gData.cp + ", op = " + op);
|
||||
res.leftContext.length = start + gData.skipped;
|
||||
}
|
||||
|
||||
res.rightContext.charArray = charArray;
|
||||
res.rightContext.str = str;
|
||||
res.rightContext.index = ep;
|
||||
res.rightContext.length = end - ep;
|
||||
|
||||
|
||||
@ -100,7 +100,6 @@ public class RegExpImpl implements RegExpProxy {
|
||||
data.leftIndex = 0;
|
||||
Object val = matchOrReplace(cx, scope, thisObj, args,
|
||||
this, data, true);
|
||||
SubString rc = this.rightContext;
|
||||
|
||||
if (data.charBuf == null) {
|
||||
if (data.global || val == null
|
||||
@ -112,7 +111,8 @@ public class RegExpImpl implements RegExpProxy {
|
||||
SubString lc = this.leftContext;
|
||||
replace_glob(data, cx, scope, this, lc.index, lc.length);
|
||||
}
|
||||
data.charBuf.append(rc.charArray, rc.index, rc.length);
|
||||
SubString rc = this.rightContext;
|
||||
data.charBuf.append(rc.str, rc.index, rc.index + rc.length);
|
||||
return data.charBuf.toString();
|
||||
}
|
||||
|
||||
@ -363,15 +363,15 @@ public class RegExpImpl implements RegExpProxy {
|
||||
}
|
||||
|
||||
int growth = leftlen + replen + reImpl.rightContext.length;
|
||||
StringBuffer charBuf = rdata.charBuf;
|
||||
StringBuilder charBuf = rdata.charBuf;
|
||||
if (charBuf == null) {
|
||||
charBuf = new StringBuffer(growth);
|
||||
charBuf = new StringBuilder(growth);
|
||||
rdata.charBuf = charBuf;
|
||||
} else {
|
||||
charBuf.ensureCapacity(rdata.charBuf.length() + growth);
|
||||
}
|
||||
|
||||
charBuf.append(reImpl.leftContext.charArray, leftIndex, leftlen);
|
||||
charBuf.append(reImpl.leftContext.str, leftIndex, leftIndex + leftlen);
|
||||
if (rdata.lambda != null) {
|
||||
charBuf.append(lambdaStr);
|
||||
} else {
|
||||
@ -475,7 +475,7 @@ public class RegExpImpl implements RegExpProxy {
|
||||
private static void do_replace(GlobData rdata, Context cx,
|
||||
RegExpImpl regExpImpl)
|
||||
{
|
||||
StringBuffer charBuf = rdata.charBuf;
|
||||
StringBuilder charBuf = rdata.charBuf;
|
||||
int cp = 0;
|
||||
String da = rdata.repstr;
|
||||
int dp = rdata.dollar;
|
||||
@ -490,7 +490,7 @@ public class RegExpImpl implements RegExpProxy {
|
||||
if (sub != null) {
|
||||
len = sub.length;
|
||||
if (len > 0) {
|
||||
charBuf.append(sub.charArray, sub.index, len);
|
||||
charBuf.append(sub.str, sub.index, sub.index + len);
|
||||
}
|
||||
cp += skip[0];
|
||||
dp += skip[0];
|
||||
@ -752,6 +752,6 @@ final class GlobData
|
||||
Function lambda; /* replacement function object or null */
|
||||
String repstr; /* replacement string */
|
||||
int dollar = -1; /* -1 or index of first $ in repstr */
|
||||
StringBuffer charBuf; /* result characters, null initially */
|
||||
StringBuilder charBuf; /* result characters, null initially */
|
||||
int leftIndex; /* leftContext index, always 0 for JS1.2 */
|
||||
}
|
||||
|
||||
@ -37,6 +37,9 @@
|
||||
|
||||
package org.mozilla.javascript.regexp;
|
||||
|
||||
/**
|
||||
* A utility class for lazily instantiated substrings.
|
||||
*/
|
||||
public class SubString {
|
||||
|
||||
public SubString()
|
||||
@ -45,30 +48,28 @@ public class SubString {
|
||||
|
||||
public SubString(String str)
|
||||
{
|
||||
this.str = str;
|
||||
index = 0;
|
||||
charArray = str.toCharArray();
|
||||
length = str.length();
|
||||
}
|
||||
|
||||
public SubString(char[] source, int start, int len)
|
||||
public SubString(String source, int start, int len)
|
||||
{
|
||||
index = 0;
|
||||
str = source;
|
||||
index = start;
|
||||
length = len;
|
||||
charArray = new char[len];
|
||||
// is this copy needed?
|
||||
System.arraycopy(source, start, charArray, 0, len);
|
||||
}
|
||||
|
||||
@Override
|
||||
public String toString() {
|
||||
return charArray == null
|
||||
return str == null
|
||||
? ""
|
||||
: new String(charArray, index, length);
|
||||
: str.substring(index, index + length);
|
||||
}
|
||||
|
||||
public static final SubString emptySubString = new SubString();
|
||||
|
||||
char[] charArray;
|
||||
String str;
|
||||
int index;
|
||||
int length;
|
||||
}
|
||||
|
||||
Loading…
x
Reference in New Issue
Block a user