private String getDomain(String url) {
PatternMatcher matcher = new Perl5Matcher();
PatternCompiler compiler = new Perl5Compiler();
PatternMatcherInput input = new PatternMatcherInput(url);
Pattern pattern = null;
String patternString = HTTP_PATTERN;
try {
pattern = compiler.compile(patternString);
} catch (MalformedPatternException e) {
logger.error("Incorrect pattern: " + patternString, e);
}
while (matcher.contains(input, pattern)) {
MatchResult result = matcher.getMatch();
String sResult = result.group(1);
logger.info("Match: " + sResult);
return sResult;
}
patternString = WWW_PATTERN;
try {
pattern = compiler.compile(patternString);
} catch (MalformedPatternException e) {
logger.error("Incorrect pattern: " + patternString, e);
}
while (matcher.contains(input, pattern)) {
MatchResult result = matcher.getMatch();
String sResult = result.group(1);
logger.info("Match: " + sResult);
return sResult;
}
return null;
}
/**
* Creates the variables:<br>
* basename_gn, where n=0...# of groups<br>
* basename_g = number of groups (apart from g0)
*/
private void saveGroups(JMeterVariables vars, String basename, MatchResult match) {
StringBuilder buf = new StringBuilder();
buf.append(basename);
buf.append("_g"); // $NON-NLS-1$
int pfxlen = buf.length();
String prevString = vars.get(buf.toString());
int previous = 0;
if (prevString != null) {
try {
previous = Integer.parseInt(prevString);
} catch (NumberFormatException e) {
log.warn("Could not parse " + prevString + " " + e);
}
}
// Note: match.groups() includes group 0
final int groups = match.groups();
for (int x = 0; x < groups; x++) {
buf.append(x);
vars.put(buf.toString(), match.group(x));
buf.setLength(pfxlen);
}
vars.put(buf.toString(), Integer.toString(groups - 1));
for (int i = groups; i <= previous; i++) {
buf.append(i);
vars.remove(buf.toString()); // remove the remaining _gn vars
buf.setLength(pfxlen);
}
}
/**
* This function returns the matched string in input when the pattern is found.
*
* @param pattern pattern string to match
* @param input input string
* @return returns the portion of the matched string in the input
* @exception throws a FrameworkException when either pattern/input is null.
*/
public static String getMatch(String pattern, String input) throws FrameworkException {
if (input == null || pattern == null) {
throw new FrameworkException(
"RegexUtils: getMatch(): input or pattern is null."
+ "input ="
+ input
+ ", "
+ "pattern = "
+ pattern);
}
Perl5Util util = new Perl5Util();
boolean isMatched = false;
MatchResult matchResult = null;
// default return value is the original input
String result = input;
pattern = makePerl5MatchPattern(pattern);
if (util.match(pattern, input)) {
matchResult = util.getMatch();
result = matchResult.toString();
}
return result;
}
public Value match(Value str) {
PatternMatcher matcher = getMatcher();
// Do the matching
PatternMatcherInput jStr = new PatternMatcherInput(string(str));
Pair result = null;
Pair prev = null;
boolean found = false;
while (matcher.contains(jStr, pattern)) {
found = true;
MatchResult matchResult = matcher.getMatch();
for (int i = 0, length = matchResult.groups(); i < length; i++) {
Pair m = new Pair(new SchemeString(matchResult.group(i)), EMPTYLIST);
if (result == null) result = prev = m;
else {
prev.setCdr(m);
prev = m;
}
}
}
if (!found) return FALSE;
else return result;
}
/** Convert specific starting hand to HoldemAtomicGroup object.
@param groupSpec starting hand (e.g., AhKd, 8h3s)
*/
public HoldemAtomicGroup(String groupSpec) {
myspec = groupSpec;
myhands = new HashSet();
Perl5Compiler compiler = new Perl5Compiler();
Perl5Matcher matcher = new Perl5Matcher();
Pattern atomicPattern;
try {
atomicPattern = compiler.compile
("^([AKQJT98765432])([shdc])([AKQJT98765432])([shdc])$");
} catch (MalformedPatternException e) {
throw new RuntimeException("BUG: " + e.toString());
}
MatchResult result;
if (matcher.matches(groupSpec, atomicPattern)) {
result = matcher.getMatch();
int rank1 = Deck.parseRank(result.group(1));
int suit1 = Deck.parseSuit(result.group(2));
int rank2 = Deck.parseRank(result.group(3));
int suit2 = Deck.parseSuit(result.group(4));
addAtomic(rank1, suit1, rank2, suit2);
} else {
throw new IllegalArgumentException("unable to parse groupSpec: " +
groupSpec);
}
}
public Value matchPositions(Value str) {
PatternMatcher matcher = getMatcher();
// Do the matching
PatternMatcherInput jStr = new PatternMatcherInput(string(str));
Pair result = null;
Pair prev = null;
boolean found = false;
while (matcher.contains(jStr, pattern)) {
found = true;
MatchResult matchResult = matcher.getMatch();
for (int i = 0, length = matchResult.groups(); i < length; i++) {
Pair m =
new Pair(
Quantity.valueOf(matchResult.beginOffset(i)),
Quantity.valueOf(matchResult.endOffset(i)));
Pair elem = new Pair(m, EMPTYLIST);
if (result == null) result = prev = elem;
else {
prev.setCdr(elem);
prev = elem;
}
}
}
if (!found) return FALSE;
else return result;
}
private void saveGroups(MatchResult result) {
if (result != null) {
JMeterVariables vars = getVariables();
for (int x = 0; x < result.groups(); x++) {
vars.put(name + "_g" + x, result.group(x));
}
}
}
public RubyValue post_match() {
String res = "";
int nMatchEnd = result_.endOffset(result_.groups() - 1);
if (nMatchEnd >= 0 && nMatchEnd + 1 < str_.length()) res = str_.substring(nMatchEnd);
if (res == null) return RubyConstant.QNIL;
return ObjectFactory.createString(res);
}
/** 查找 */
public static void simpleContains() throws Exception {
Pattern pattern = new Perl5Compiler().compile("\\d+");
Perl5Matcher matcher = new Perl5Matcher();
PatternMatcherInput matcherInput = new PatternMatcherInput("北京2008年8月08日20时");
System.out.println("simpleContains:");
while (matcher.contains(matcherInput, pattern)) {
MatchResult result = matcher.getMatch();
System.out.println(result.toString());
}
}
public RubyValue to_a() {
RubyArray ar = new RubyArray();
for (int i = 0; i < result_.groups(); i++) {
String str = result_.group(i);
if (str != null) ar.add(ObjectFactory.createString(str));
else ar.add(RubyConstant.QNIL);
}
return ar;
}
public RubyValue captures() {
RubyArray ar = new RubyArray();
for (int i = 1; i < result_.groups(); i++) {
String res = result_.group(i);
if (res == null) ar.add(RubyConstant.QNIL);
else ar.add(ObjectFactory.createString(res));
}
return ar;
}
/**
* This function replace only the first pattern matched with the replacement.
*
* @param pattern pattern string to match
* @param input input string
* @param replacement replacement string
* @return returns the processed string
* @exception throws a FrameworkException when the pattern/replacement is null.
*/
public static String replaceLast(String pattern, String input, String replacement)
throws FrameworkException {
if ((pattern == null) || (input == null) || (replacement == null)) {
throw new FrameworkException(
"RegexUtil: replaceLast(): pattern or input cannot be null. "
+ "pattern = "
+ pattern
+ "input = "
+ input
+ "replacement = "
+ replacement);
}
Perl5Util util = new Perl5Util();
MatchResult matchResult = null;
String result = null;
String pre = null;
String post = null;
StringBuffer resultBuffer = new StringBuffer();
int length = input.length();
String regex = makePerl5SubstitutionPattern(pattern, replacement);
pattern = makePerl5MatchPattern(pattern);
// counts the number of match and grab the last one
while (util.match(pattern, input)) {
// getting the string before match
pre = util.preMatch();
resultBuffer.append(pre);
// get the matched string
matchResult = util.getMatch();
resultBuffer.append(matchResult.toString());
// get the post string after the match
post = util.postMatch();
// the post becomes the new input
input = post;
}
// get the last match found
matchResult = util.getMatch();
// do the string replacement on the pattern found
result = util.substitute(regex, matchResult.toString());
resultBuffer.append(result);
resultBuffer.append(post);
return resultBuffer.toString();
}
/**
* insert正则表达式解析insert语句 如果insert语句与正则表达式匹配，本方法根据表达式对语句的分组
* 将insert语句拆分为6部分,并且将这几部分存放到字符串数组返回 分别为： 1.insert关键字 2.into关键字 3.表名称 4.表字段组
* 5.values关键字 6.插入字段值组
*
* 比如insert语句： Insert into oa_meetingpromptsound ( soundCode , soundName ,
* soundFileName ) values ( '。.尹标平','bb','d()d' ) 将被分解为以下部分： 1.Insert 2.into
* 3.oa_meetingpromptsound 4.( soundCode , soundName , soundFileName )
* 5.values 6.( '。.尹标平','bb','d()d' )
*/
public static String[] parserInsert(String insert)
{
/**
* 定义insert语句的正则表达式 该表达式将insert语句拆分为6部分,分别为： 1.insert关键字 2.into关键字 3.表名称
* 4.表字段组 5.values关键字 6.插入字段值组 比如insert语句： Insert into
* oa_meetingpromptsound ( soundCode , soundName , soundFileName )
* values ( '。.尹标平','bb','d()d' ) 将被分解为以下部分： 1.Insert 2.into
* 3.oa_meetingpromptsound 4.( soundCode , soundName , soundFileName )
* 5.values 6.( '。.尹标平','bb','d()d' )
*/
String patternStr = "\\s*(insert)\\s+" + // 解析insert关键词
"(into)\\s+" + // 解析into关键词
"([^\\(^\\s]+)\\s*" + // 解析表名称
"(\\([^\\)]+\\))\\s*" + // 解析表字段
"(values)\\s*" + // 解析value关键词
"(\\(.*(.*\n*)*.*)"; // 解析字段值
/**
* 编译正则表达式patternStr，并用该表达式与传入的sql语句进行模式匹配,
* 如果匹配正确，则从匹配对象中提取出以上定义好的6部分，存放到数组中并返回 该数组
*/
PatternCompiler compiler = new Perl5Compiler();
Pattern pattern = null;
try
{
pattern = compiler.compile(patternStr,
Perl5Compiler.CASE_INSENSITIVE_MASK);
}
catch (MalformedPatternException e)
{
e.printStackTrace();
return null;
}
PatternMatcher matcher = new Perl5Matcher();
MatchResult result = null;
String[] tokens = null;
boolean match = matcher.matches(insert, pattern);
if (match)
{
result = matcher.getMatch();
tokens = new String[6];
for (int i = 0; i < 6; i++)
{
tokens[i] = result.group(i + 1).trim();
}
}
return tokens;
}
/** 分组 */
public static void simpleResults() throws Exception {
Pattern pattern =
new Perl5Compiler().compile("(\\d+\\.\\d+\\.\\d+\\.\\d+)@(\\d{2}/\\d{2}/\\d{4})");
Perl5Matcher matcher = new Perl5Matcher();
PatternMatcherInput matcherInput = new PatternMatcherInput("202.108.9.38@08/10/2008");
System.out.println("simpleResults:");
while (matcher.contains(matcherInput, pattern)) {
MatchResult result = matcher.getMatch();
for (int i = 0; i < result.groups(); i++) {
System.out.printf("%s : %s\n", i, result.group(i));
}
}
}
public void process() {
Sampler sampler = JMeterContextService.getContext().getCurrentSampler();
SampleResult responseText = JMeterContextService.getContext().getPreviousResult();
if (responseText == null) {
return;
}
initRegex(getArgumentName());
String text = new String(responseText.getResponseData());
Perl5Matcher matcher = JMeterUtils.getMatcher();
String value = "";
if (matcher.contains(text, case1)) {
MatchResult result = matcher.getMatch();
value = result.group(1);
} else if (matcher.contains(text, case2)) {
MatchResult result = matcher.getMatch();
value = result.group(1);
} else if (matcher.contains(text, case3)) {
MatchResult result = matcher.getMatch();
value = result.group(1);
} else if (matcher.contains(text, case4)) {
MatchResult result = matcher.getMatch();
value = result.group(1);
}
modify((HTTPSampler) sampler, value);
}
// @RubyLevelMethod(name="[]")
public RubyValue aref(RubyValue arg) {
if (arg instanceof RubyFixnum) {
int index = arg.toInt();
if (index < 0) index += result_.groups();
String res = result_.group(index);
if (res == null) return RubyConstant.QNIL;
return ObjectFactory.createString(res);
} else if (arg instanceof RubyString) {
// TODO: implement m = /(?<foo>a+)b/.match("ccaaab"); m["foo"] #=> "aaa"
throw new RuntimeException("Not implemented");
}
RubyArray ar = (RubyArray) to_a();
return ar.aref(arg);
}
public RubyValue pre_match() {
String res = "";
int nMatchEnd = result_.beginOffset(0);
if (nMatchEnd >= 0 && nMatchEnd < str_.length()) res = str_.substring(0, nMatchEnd);
if (res == null) return RubyConstant.QNIL;
return ObjectFactory.createString(res);
}
public static String[] parserValues(String values)
{
String patternStr = "([^\\,]*)[\\,]?"; // 解析字段值
String patternStr1 = "('?[^\\,]*'?)[\\,]?"; // 解析字段值
/**
* 编译正则表达式patternStr，并用该表达式与传入的sql语句进行模式匹配,
* 如果匹配正确，则从匹配对象中提取出以上定义好的6部分，存放到数组中并返回 该数组
*/
String[] ret = RegexUtil.containWithPatternMatcherInput(values,patternStr1);
PatternCompiler compiler = new Perl5Compiler();
Pattern pattern = null;
try
{
pattern = compiler.compile(patternStr,
Perl5Compiler.CASE_INSENSITIVE_MASK);
}
catch (MalformedPatternException e)
{
e.printStackTrace();
return null;
}
PatternMatcher matcher = new Perl5Matcher();
MatchResult result = null;
String[] tokens = null;
boolean match = matcher.matches(values.trim(), pattern);
System.out.println(match);
if (match)
{
result = matcher.getMatch();
tokens = new String[6];
for (int i = 0; i < 6; i++)
{
tokens[i] = result.group(i + 1).trim();
System.out.println(tokens[i]);
}
}
return tokens;
}
public TableEntryLocation[] parseTableEntryLocations(String expression) {
resultList.clear();
if (expression != null) {
matcher.setMultiline(true);
if (patternMatcherInput == null) {
patternMatcherInput = new PatternMatcherInput(expression);
} else {
patternMatcherInput.setInput(expression);
}
recompilePatternIfNecessary(locationPattern);
while (matcher.contains(patternMatcherInput, pattern)) {
MatchResult matchResult = matcher.getMatch();
resultList.add(new TableEntryLocation(matchResult.group(1), matchResult.group(2)));
}
}
return resultList.toArray(new TableEntryLocation[0]);
}
private void initTemplate() {
if (template != null) {
return;
}
// Contains Strings and Integers
List<Object> combined = new ArrayList<Object>();
String rawTemplate = getTemplate();
PatternMatcher matcher = JMeterUtils.getMatcher();
Pattern templatePattern =
JMeterUtils.getPatternCache()
.getPattern(
"\\$(\\d+)\\$" // $NON-NLS-1$
,
Perl5Compiler.READ_ONLY_MASK & Perl5Compiler.SINGLELINE_MASK);
if (log.isDebugEnabled()) {
log.debug("Pattern = " + templatePattern.getPattern());
log.debug("template = " + rawTemplate);
}
int beginOffset = 0;
MatchResult currentResult;
PatternMatcherInput pinput = new PatternMatcherInput(rawTemplate);
while (matcher.contains(pinput, templatePattern)) {
currentResult = matcher.getMatch();
final int beginMatch = currentResult.beginOffset(0);
if (beginMatch > beginOffset) { // string is not empty
combined.add(rawTemplate.substring(beginOffset, beginMatch));
}
combined.add(Integer.valueOf(currentResult.group(1))); // add match as Integer
beginOffset = currentResult.endOffset(0);
}
if (beginOffset < rawTemplate.length()) { // trailing string is not empty
combined.add(rawTemplate.substring(beginOffset, rawTemplate.length()));
}
if (log.isDebugEnabled()) {
log.debug("Template item count: " + combined.size());
for (Object o : combined) {
log.debug(o.getClass().getSimpleName() + " '" + o.toString() + "'");
}
}
template = combined;
}
public void appendSubstitution(
StringBuffer stringBuffer,
MatchResult matchResult,
int i,
PatternMatcherInput patternMatcherInput,
PatternMatcher patternMatcher,
Pattern pattern) {
String hex = matchResult.group(1);
char c = (char) Integer.parseInt(hex, 16);
stringBuffer.append(c);
}
/**
* The function checks and handles the situation where another begin token is found in between the
* first begin token and the first end token. Such first beginToken is ignored. The current input
* to be passed in to the replacement function will be the string between the LAST begin token and
* the FIRST end token found.
*
* @param beginToken the pattern to look for
* @param current input string
* @param resultBuffer the buffer to hold the result
* @return returns a Vector containing the substrings of the input that occur
*/
private static final String skipOrphanedBeginToken(
String beginToken, String current, StringBuffer resultBuffer) {
Perl5Util util = new Perl5Util();
boolean nextBeginTokenMatch = true;
MatchResult matchResult = null;
String remainderStr = null;
String subCurrent = null;
StringBuffer currentBuffer = new StringBuffer();
String pre = null;
// subCurrent is copy of the current to check if there is any next beginTokens
subCurrent = current;
while (nextBeginTokenMatch) {
nextBeginTokenMatch = util.match(beginToken, subCurrent);
if (nextBeginTokenMatch == true) {
// pre is the string before the beginToken
pre = util.preMatch();
currentBuffer.append(pre);
// get the matched beginToken
matchResult = util.getMatch();
currentBuffer.append(matchResult.toString());
// get the remaining string after the beginToken
remainderStr = util.postMatch();
subCurrent = remainderStr;
} else // there is no match
{
// appending the string before the last begin token to the result
resultBuffer.append(currentBuffer.toString());
// finally get the string to perform the substitution on
current = subCurrent;
break;
}
} // while
return current;
}
/** Parse a string of the form [x1,y1},[x2,y2},...,[xN,yN} into a list of Points ([int,int]) */
public static List<Point> parseString(String str) throws NumberFormatException {
if (StringUtil.isNullString(str)) {
throw new IllegalArgumentException("Must supply non-empty string");
}
ArrayList<Point> res = new ArrayList<Point>();
Perl5Matcher matcher = RegexpUtil.getMatcher();
while (matcher.contains(str, ONE_POINT_PAT)) {
MatchResult matchResult = matcher.getMatch();
String xstr = matchResult.group(1);
String ystr = matchResult.group(2);
str = matchResult.group(3);
try {
int x = Integer.parseInt(xstr);
int y = Integer.parseInt(ystr);
res.add(new Point(x, y));
} catch (NumberFormatException e) {
throw new IllegalArgumentException("bad point [" + xstr + "," + ystr + "] in " + str);
}
}
res.trimToSize();
return res;
}
private String generateResult(MatchResult match) {
saveGroups(match);
StringBuffer result = new StringBuffer();
for (int a = 0; a < template.length; a++) {
if (template[a] instanceof String) {
result.append(template[a]);
} else {
result.append(match.group(((Integer) template[a]).intValue()));
}
}
JMeterVariables vars = getVariables();
vars.put(name, result.toString());
return result.toString();
}
private String generateResult(MatchResult match) {
StringBuilder result = new StringBuilder();
for (Object obj : template) {
if (log.isDebugEnabled()) {
log.debug(
"RegexExtractor: Template piece " + obj + " (" + obj.getClass().getSimpleName() + ")");
}
if (obj instanceof Integer) {
result.append(match.group(((Integer) obj).intValue()));
} else {
result.append(obj);
}
}
if (log.isDebugEnabled()) {
log.debug("Regex Extractor result = " + result.toString());
}
return result.toString();
}
// Overridden Methods
@Override
protected void inspectContents(File file, String contents) {
StringBuffer buf = new StringBuffer();
// Split it into lines
StringSplitter splitter = new StringSplitter(contents, getLineEnding());
// Scan each line
int totalMatchesForThisFile = 0;
int lineCount = 1;
while (splitter.hasMoreElements()) {
String line = (String) splitter.nextElement();
input = new PatternMatcherInput(line);
try {
while (util.match(regex, input)) {
result = util.getMatch();
totalMatchesForThisFile++;
totalNumberOfMatches++;
buf.append(
" line: " + lineCount + " position: " + result.beginOffset(0) + getLineEnding());
}
} catch (MalformedPerl5PatternException e) {
System.out.println("MalformedPerl5PatternException: " + e.getMessage());
System.out.println("Valid expression: [m]/pattern/[i][m][s][x]");
return;
}
lineCount++;
}
// Output the results
if (totalMatchesForThisFile > 0) {
if (totalMatchesForThisFile == 1) {
System.out.println(
"Found " + totalMatchesForThisFile + " match in file: " + file.getPath());
} else {
System.out.println(
"Found " + totalMatchesForThisFile + " matches in file: " + file.getPath());
}
System.out.print(buf.toString());
System.out.println("");
}
}
public String toString() {
return result_.group(0);
}
private void processComponentStart(
String tagName,
String jwcId,
boolean emptyTag,
int startLine,
int cursorStart,
ILocation startLocation)
throws TemplateParseException {
if (jwcId.equalsIgnoreCase(CONTENT_ID)) {
processContentTag(tagName, startLine, cursorStart, emptyTag);
return;
}
boolean isRemoveId = jwcId.equalsIgnoreCase(REMOVE_ID);
if (_ignoring && !isRemoveId)
templateParseProblem(
Tapestry.format(
"TemplateParser.component-may-not-be-ignored", tagName, Integer.toString(startLine)),
startLocation,
startLine,
cursorStart);
String type = null;
boolean allowBody = false;
if (_patternMatcher.matches(jwcId, _implicitIdPattern)) {
MatchResult match = _patternMatcher.getMatch();
jwcId = match.group(IMPLICIT_ID_PATTERN_ID_GROUP);
type = match.group(IMPLICIT_ID_PATTERN_TYPE_GROUP);
String libraryId = match.group(IMPLICIT_ID_PATTERN_LIBRARY_ID_GROUP);
String simpleType = match.group(IMPLICIT_ID_PATTERN_SIMPLE_TYPE_GROUP);
// If (and this is typical) no actual component id was specified,
// then generate one on the fly.
// The allocated id for anonymous components is
// based on the simple (unprefixed) type, but starts
// with a leading dollar sign to ensure no conflicts
// with user defined component ids (which don't allow dollar signs
// in the id).
if (jwcId == null) jwcId = _idAllocator.allocateId("$" + simpleType);
try {
allowBody = _delegate.getAllowBody(libraryId, simpleType, startLocation);
} catch (ApplicationRuntimeException e) {
// give subclasses a chance to handle and rethrow
templateParseProblem(e, startLine, cursorStart);
}
} else {
if (!isRemoveId) {
if (!_patternMatcher.matches(jwcId, _simpleIdPattern))
templateParseProblem(
Tapestry.format(
"TemplateParser.component-id-invalid",
tagName,
Integer.toString(startLine),
jwcId),
startLocation,
startLine,
cursorStart);
if (!_delegate.getKnownComponent(jwcId))
templateParseProblem(
Tapestry.format(
"TemplateParser.unknown-component-id",
tagName,
Integer.toString(startLine),
jwcId),
startLocation,
startLine,
cursorStart);
try {
allowBody = _delegate.getAllowBody(jwcId, startLocation);
} catch (ApplicationRuntimeException e) {
// give subclasses a chance to handle and rethrow
templateParseProblem(e, startLine, cursorStart);
}
}
}
// Ignore the body if we're removing the entire tag,
// of if the corresponding component doesn't allow
// a body.
boolean ignoreBody = !emptyTag && (isRemoveId || !allowBody);
if (_ignoring && ignoreBody)
templateParseProblem(
Tapestry.format("TemplateParser.nested-ignore", tagName, Integer.toString(startLine)),
new Location(_resourceLocation, startLine),
startLine,
cursorStart);
if (!emptyTag) pushNewTag(tagName, startLine, isRemoveId, ignoreBody);
// End any open block.
addTextToken(cursorStart - 1);
if (!isRemoveId) {
addOpenToken(tagName, jwcId, type, startLocation);
if (emptyTag) _tokens.add(_factory.createCloseToken(tagName, getCurrentLocation()));
}
advance();
}
// @RubyLevelMethod(name="to_s")
public RubyString to_s() {
return ObjectFactory.createString(result_.toString());
}
/** {@inheritDoc} */
@Override
public Iterator<URL> getEmbeddedResourceURLs(
String userAgent, byte[] html, URL baseUrl, URLCollection urls, String encoding)
throws HTMLParseException {
Pattern pattern = null;
Perl5Matcher matcher = null;
try {
matcher = JMeterUtils.getMatcher();
PatternMatcherInput input = localInput.get();
// TODO: find a way to avoid the cost of creating a String here --
// probably a new PatternMatcherInput working on a byte[] would do
// better.
input.setInput(new String(html, encoding));
pattern =
JMeterUtils.getPatternCache()
.getPattern(
REGEXP,
Perl5Compiler.CASE_INSENSITIVE_MASK
| Perl5Compiler.SINGLELINE_MASK
| Perl5Compiler.READ_ONLY_MASK);
while (matcher.contains(input, pattern)) {
MatchResult match = matcher.getMatch();
String s;
if (log.isDebugEnabled()) {
log.debug("match groups " + match.groups() + " " + match.toString());
}
// Check for a BASE HREF:
for (int g = 1; g <= NUM_BASE_GROUPS && g <= match.groups(); g++) {
s = match.group(g);
if (s != null) {
if (log.isDebugEnabled()) {
log.debug("new baseUrl: " + s + " - " + baseUrl.toString());
}
try {
baseUrl = ConversionUtils.makeRelativeURL(baseUrl, s);
} catch (MalformedURLException e) {
// Doesn't even look like a URL?
// Maybe it isn't: Ignore the exception.
if (log.isDebugEnabled()) {
log.debug("Can't build base URL from RL " + s + " in page " + baseUrl, e);
}
}
}
}
for (int g = NUM_BASE_GROUPS + 1; g <= match.groups(); g++) {
s = match.group(g);
if (s != null) {
if (log.isDebugEnabled()) {
log.debug("group " + g + " - " + match.group(g));
}
urls.addURL(s, baseUrl);
}
}
}
return urls.iterator();
} catch (UnsupportedEncodingException e) {
throw new HTMLParseException(e.getMessage(), e);
} catch (MalformedCachePatternException e) {
throw new HTMLParseException(e.getMessage(), e);
} finally {
JMeterUtils.clearMatcherMemory(matcher, pattern);
}
}