String input = content.replaceAll("[\\r\\n]", " ");
        if (ignoreCommentTag) {
            input = input.replaceAll("<!--[^>]+-->", "");
        } else {
            input = input.replace("<!--", "").replace("-->", "");
        }
        final Matcher matcher = getTagPattern().matcher(input);
        final StringBuffer sb = new StringBuffer();
        final Pattern attrPattern = Pattern.compile("\\s[^ ]+=\"([^\"]*)\"");

        final File targetFile = ResourceUtil.getResourceAsFile("test");
        String path = targetFile.getAbsolutePath();
        if (!path.startsWith("/")) {
            path = "/" + path.replace('\\', '/');
        }
        final String url = "file:" + path;

        final int maxCount = 3;
        final int numOfThread = 2;

        final File file = File.createTempFile("crawler-", "");

     * @return the escaped query string
     */
    public static String escapeWildcardQuery(final String query) {
        return query.replace("*", "\\*").replace("?", "\\?");
    }

            final String text = source.get(FieldNames.TEXT).toString();
            if (skipDuplicateWords) {
                final String duplicateCheckStr = text.replace(" ", "");
                if (words.stream().map(word -> word.replace(" ", "")).anyMatch(word -> word.equals(duplicateCheckStr))) {
                    // skip duplicate word.
                    continue;
                }
            }

        final File targetFile = ResourceUtil.getResourceAsFile("test");
        String path = targetFile.getAbsolutePath();
        if (!path.startsWith("/")) {
            path = "/" + path.replace('\\', '/');
        }
        final String url = "file:" + path;

        final int maxCount = 3;
        final int numOfThread = 2;

        final File file = File.createTempFile("crawler-", "");

        String newPath = path.replace(".", "\\.").replace("?", "\\?").replace("*", ".*");
        if (newPath.charAt(0) != '/') {
            newPath = ".*" + newPath;
        }
        if (!newPath.endsWith("$") && !newPath.endsWith(".*")) {
            newPath = newPath + ".*";
        }
        return newPath.replace(".*.*", ".*");
    }

    /*
     * (non-Javadoc)
     *

                ScoreFunctionBuilders.fieldValueFactorFunction(FieldNames.QUERY_FREQ).missing(0));
        functionScoreQueryBuilder.boostMode(CombineFunction.REPLACE);
        return functionScoreQueryBuilder;
    }

    /**
     * Builds the OpenSearch rescorer for popular words.
     * @return The QueryRescorerBuilder instance.
     */

            if (StringUtil.isBlank(s)) {
                throw new CrawlingAccessException("uri is blank.");
            }

            try {
                uri = new URI(normalize(s).replace("%", "%25").replace(" ", "%20"));
            } catch (final URISyntaxException e) {
                throw new CrawlingAccessException("Invalid URL: " + s, e);
            }

            if (!"ftp".equals(uri.getScheme())) {

            url = url.substring(0, idx);
        }

        url = url.replace("/./", "/");

        idx = url.indexOf(";jsessionid");
        if (idx >= 0) {
            url = url.replaceFirst(";jsessionid=[a-zA-Z0-9\\.]*", "");
        }

        if (url.indexOf(' ') >= 0) {
            url = url.replace(" ", "%20");
        }

        String oldUrl = null;

        ruleManager.addRule(oldRule);

        ResponseData responseData = new ResponseData();
        assertNull(ruleManager.getRule(responseData)); // Old rule doesn't match

        // Replace old rule with new rule
        ruleManager.removeRule(oldRule);
        ruleManager.addRule(newRule);

        Rule matched = ruleManager.getRule(responseData);
        assertNotNull(matched);