public String text()

in core/src/main/java/org/apache/stormcrawler/parse/TextExtractor.java [81:112]

• 23 lines of code
• 8 McCabe index (conditional complexity)

    public String text(Element element) {
        // not interested in getting any text?
        if (noText) return "";

        final StringBuilder accum = new StringBuilder();

        // no patterns at all - return the text from the whole document
        if (inclusionPatterns.size() == 0 && excludedTags.size() == 0) {
            _text(element, accum);
        } else {
            Elements matches = new Elements();

            for (String pattern : inclusionPatterns) {
                matches = element.select(pattern);
                if (!matches.isEmpty()) {
                    break;
                }
            }

            // if nothing matches or no patterns were defined use the whole doc
            if (matches.isEmpty()) {
                matches.add(element);
            }

            for (Element node : matches) {
                _text(node, accum);
                accum.append("\n");
            }
        }

        return accum.toString().trim();
    }