Examples with Pattern org.apache.oro.text.regex.Pattern used on opensource projects

Example 26 with Pattern

use of org.apache.oro.text.regex.Pattern in project jmeter by apache.

the class RegexpHTMLParser method getEmbeddedResourceURLs.

/**
 * {@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);
            }
            // Check for a BASE HREF:
            for (int g = 1; g <= NUM_BASE_GROUPS && g <= match.groups(); g++) {
                s = match.group(g);
                if (s != null) {
                    log.debug("new baseUrl: {} - {}", s, baseUrl);
                    try {
                        baseUrl = ConversionUtils.makeRelativeURL(baseUrl, s);
                    } catch (MalformedURLException e) {
                        // Doesn't even look like a URL?
                        // Maybe it isn't: Ignore the exception.
                        log.debug("Can't build base URL from URL {} in page {}", s, 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 | MalformedCachePatternException e) {
        throw new HTMLParseException(e.getMessage(), e);
    } finally {
        JMeterUtils.clearMatcherMemory(matcher, pattern);
    }
}

Example 27 with Pattern

use of org.apache.oro.text.regex.Pattern in project jmeter by apache.

the class RegexExtractor method initTemplate.

private void initTemplate() {
    if (template != null) {
        return;
    }
    // Contains Strings and Integers
    List<Object> combined = new ArrayList<>();
    String rawTemplate = getTemplate();
    PatternMatcher matcher = JMeterUtils.getMatcher();
    Pattern templatePattern = // $NON-NLS-1$
    JMeterUtils.getPatternCache().getPattern(// $NON-NLS-1$
    "\\$(\\d+)\\$", Perl5Compiler.READ_ONLY_MASK & Perl5Compiler.SINGLELINE_MASK);
    if (log.isDebugEnabled()) {
        log.debug("Pattern = '{}', template = '{}'", templatePattern.getPattern(), 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));
        }
        // add match as Integer
        combined.add(Integer.valueOf(currentResult.group(1)));
        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());
        int i = 0;
        for (Object o : combined) {
            log.debug("Template item-{}: {} '{}'", i++, o.getClass(), o);
        }
    }
    template = combined;
}

Example 28 with Pattern

use of org.apache.oro.text.regex.Pattern in project jmeter by apache.

the class HTTPSamplerBase method downloadPageResources.

/**
 * Download the resources of an HTML page.
 *
 * @param pRes
 *            result of the initial request - must contain an HTML response
 * @param container
 *            for storing the results, if any
 * @param frameDepth
 *            Depth of this target in the frame structure. Used only to
 *            prevent infinite recursion.
 * @return res if no resources exist, otherwise the "Container" result with one subsample per request issued
 */
protected HTTPSampleResult downloadPageResources(final HTTPSampleResult pRes, final HTTPSampleResult container, final int frameDepth) {
    HTTPSampleResult res = pRes;
    Iterator<URL> urls = null;
    try {
        final byte[] responseData = res.getResponseData();
        if (responseData.length > 0) {
            // Bug 39205
            final LinkExtractorParser parser = getParser(res);
            if (parser != null) {
                String userAgent = getUserAgent(res);
                urls = parser.getEmbeddedResourceURLs(userAgent, responseData, res.getURL(), res.getDataEncodingWithDefault());
            }
        }
    } catch (LinkExtractorParseException e) {
        // Don't break the world just because this failed:
        res.addSubResult(errorResult(e, new HTTPSampleResult(res)));
        setParentSampleSuccess(res, false);
    }
    HTTPSampleResult lContainer = container;
    // Iterate through the URLs and download each image:
    if (urls != null && urls.hasNext()) {
        if (lContainer == null) {
            lContainer = new HTTPSampleResult(res);
            lContainer.addRawSubResult(res);
        }
        res = lContainer;
        // Get the URL matcher
        String allowRegex = getEmbeddedUrlRE();
        Perl5Matcher localMatcher = null;
        Pattern allowPattern = null;
        if (allowRegex.length() > 0) {
            try {
                allowPattern = JMeterUtils.getPattern(allowRegex);
                // don't fetch unless pattern compiles
                localMatcher = JMeterUtils.getMatcher();
            } catch (MalformedCachePatternException e) {
                // NOSONAR
                log.warn("Ignoring embedded URL match string: {}", e.getMessage());
            }
        }
        Pattern excludePattern = null;
        String excludeRegex = getEmbededUrlExcludeRE();
        if (excludeRegex.length() > 0) {
            try {
                excludePattern = JMeterUtils.getPattern(excludeRegex);
                if (localMatcher == null) {
                    // don't fetch unless pattern compiles
                    localMatcher = JMeterUtils.getMatcher();
                }
            } catch (MalformedCachePatternException e) {
                // NOSONAR
                log.warn("Ignoring embedded URL exclude string: {}", e.getMessage());
            }
        }
        // For concurrent get resources
        final List<Callable<AsynSamplerResultHolder>> list = new ArrayList<>();
        // init with default value
        int maxConcurrentDownloads = CONCURRENT_POOL_SIZE;
        boolean isConcurrentDwn = isConcurrentDwn();
        if (isConcurrentDwn) {
            try {
                maxConcurrentDownloads = Integer.parseInt(getConcurrentPool());
            } catch (NumberFormatException nfe) {
                log.warn(// $NON-NLS-1$
                "Concurrent download resources selected, " + // $NON-NLS-1$
                "but pool size value is bad. Use default value");
            }
            // no need to use another thread, do the sample on the current thread
            if (maxConcurrentDownloads == 1) {
                log.warn("Number of parallel downloads set to 1, (sampler name={})", getName());
                isConcurrentDwn = false;
            }
        }
        while (urls.hasNext()) {
            // See catch clause below
            Object binURL = urls.next();
            try {
                URL url = (URL) binURL;
                if (url == null) {
                    log.warn("Null URL detected (should not happen)");
                } else {
                    try {
                        url = escapeIllegalURLCharacters(url);
                    } catch (Exception e) {
                        // NOSONAR
                        res.addSubResult(errorResult(new Exception(url.toString() + " is not a correct URI", e), new HTTPSampleResult(res)));
                        setParentSampleSuccess(res, false);
                        continue;
                    }
                    log.debug("allowPattern: {}, excludePattern: {}, localMatcher: {}, url: {}", allowPattern, excludePattern, localMatcher, url);
                    // I don't think localMatcher can be null here, but check just in case
                    if (allowPattern != null && localMatcher != null && !localMatcher.matches(url.toString(), allowPattern)) {
                        // we have a pattern and the URL does not match, so skip it
                        continue;
                    }
                    if (excludePattern != null && localMatcher != null && localMatcher.matches(url.toString(), excludePattern)) {
                        // we have a pattern and the URL does not match, so skip it
                        continue;
                    }
                    try {
                        url = url.toURI().normalize().toURL();
                    } catch (MalformedURLException | URISyntaxException e) {
                        res.addSubResult(errorResult(new Exception(url.toString() + " URI can not be normalized", e), new HTTPSampleResult(res)));
                        setParentSampleSuccess(res, false);
                        continue;
                    }
                    if (isConcurrentDwn) {
                        // if concurrent download emb. resources, add to a list for async gets later
                        list.add(new ASyncSample(url, HTTPConstants.GET, false, frameDepth + 1, getCookieManager(), this));
                    } else {
                        // default: serial download embedded resources
                        HTTPSampleResult binRes = sample(url, HTTPConstants.GET, false, frameDepth + 1);
                        res.addSubResult(binRes);
                        setParentSampleSuccess(res, res.isSuccessful() && (binRes == null || binRes.isSuccessful()));
                    }
                }
            } catch (ClassCastException e) {
                // NOSONAR
                res.addSubResult(errorResult(new Exception(binURL + " is not a correct URI", e), new HTTPSampleResult(res)));
                setParentSampleSuccess(res, false);
            }
        }
        // IF for download concurrent embedded resources
        if (isConcurrentDwn && !list.isEmpty()) {
            ResourcesDownloader resourcesDownloader = ResourcesDownloader.getInstance();
            try {
                // sample all resources
                final List<Future<AsynSamplerResultHolder>> retExec = resourcesDownloader.invokeAllAndAwaitTermination(maxConcurrentDownloads, list);
                CookieManager cookieManager = getCookieManager();
                // add result to main sampleResult
                for (Future<AsynSamplerResultHolder> future : retExec) {
                    // this call will not block as the futures return by invokeAllAndAwaitTermination
                    // are either done or cancelled
                    AsynSamplerResultHolder binRes = future.get();
                    if (cookieManager != null) {
                        CollectionProperty cookies = binRes.getCookies();
                        for (JMeterProperty jMeterProperty : cookies) {
                            Cookie cookie = (Cookie) jMeterProperty.getObjectValue();
                            cookieManager.add(cookie);
                        }
                    }
                    res.addSubResult(binRes.getResult());
                    setParentSampleSuccess(res, res.isSuccessful() && (binRes.getResult() != null ? binRes.getResult().isSuccessful() : true));
                }
            } catch (InterruptedException ie) {
                // $NON-NLS-1$
                log.warn("Interrupted fetching embedded resources", ie);
                Thread.currentThread().interrupt();
            } catch (ExecutionException ee) {
                // $NON-NLS-1$
                log.warn("Execution issue when fetching embedded resources", ee);
            }
        }
    }
    return res;
}

Example 29 with Pattern

use of org.apache.oro.text.regex.Pattern in project jmeter by apache.

the class LogFilter method excPattern.

/**
 * The method assumes by default the text is not excluded. If the text
 * matches the pattern, it will then return true.
 *
 * @param text text to be checked
 * @return <code>true</code> if text is excluded
 */
protected boolean excPattern(String text) {
    this.USEFILE = true;
    boolean exc = false;
    for (Pattern excludePattern : this.EXCPATTERNS) {
        if (JMeterUtils.getMatcher().contains(text, excludePattern)) {
            exc = true;
            this.USEFILE = false;
            break;
        }
    }
    return exc;
}

Example 30 with Pattern

use of org.apache.oro.text.regex.Pattern in project jmeter by apache.

the class ReplaceFunctionsWithStrings method transformValue.

@Override
public JMeterProperty transformValue(JMeterProperty prop) throws InvalidVariableException {
    PatternMatcher pm = JMeterUtils.getMatcher();
    PatternCompiler compiler = new Perl5Compiler();
    String input = prop.getStringValue();
    if (input == null) {
        return prop;
    }
    for (Map.Entry<String, String> entry : getVariables().entrySet()) {
        String key = entry.getKey();
        String value = entry.getValue();
        if (regexMatch) {
            try {
                Pattern pattern = compiler.compile(constructPattern(value));
                input = Util.substitute(pm, pattern, new StringSubstitution(FUNCTION_REF_PREFIX + key + FUNCTION_REF_SUFFIX), input, Util.SUBSTITUTE_ALL);
            } catch (MalformedPatternException e) {
                log.warn("Malformed pattern: {}", value);
            }
        } else {
            input = StringUtilities.substitute(input, value, FUNCTION_REF_PREFIX + key + FUNCTION_REF_SUFFIX);
        }
    }
    return new StringProperty(prop.getName(), input);
}