Examples with Element net.htmlparser.jericho.Element used on opensource projects

Example 46 with Element

use of net.htmlparser.jericho.Element in project lotro-tools by dmorcellet.

the class JerichoHtmlUtils method findElementByTagName.

/**
 * Find a tag.
 * @param root Parent tag.
 * @param tagName Tag to search.
 * @return A tag element or <code>null</code> if not found.
 */
public static Element findElementByTagName(Segment root, String tagName) {
    Element ret = null;
    List<Element> elements = root.getAllElements(tagName);
    if (elements.size() > 0) {
        ret = elements.get(0);
    }
    return ret;
}

Example 47 with Element

use of net.htmlparser.jericho.Element in project lotro-tools by dmorcellet.

the class JerichoHtmlUtils method findElementsByTagNameAndAttributeValue.

/**
 * Find tags.
 * @param root Parent tag.
 * @param tagName Tag to search.
 * @param attrName Attribute to look.
 * @param expectedValue Expected value for this attribute.
 * @return A possibly empty list of tag elements.
 */
public static List<Element> findElementsByTagNameAndAttributeValue(Segment root, String tagName, String attrName, String expectedValue) {
    List<Element> ret = new ArrayList<Element>();
    List<Element> elements = root.getAllElements(tagName);
    for (Element element : elements) {
        StartTag tag = element.getStartTag();
        String value = tag.getAttributeValue(attrName);
        if (expectedValue.equals(value)) {
            ret.add(element);
        }
    }
    return ret;
}

Example 48 with Element

use of net.htmlparser.jericho.Element in project lotro-tools by dmorcellet.

the class JerichoHtmlUtils method findElementByTagNameAndAttributeValue.

/**
 * Find a tag.
 * @param root Parent tag.
 * @param tagName Tag to search.
 * @param attrName Attribute to look.
 * @param expectedValue Expected value for this attribute.
 * @return A tag element or <code>null</code> if not found.
 */
public static Element findElementByTagNameAndAttributeValue(Segment root, String tagName, String attrName, String expectedValue) {
    Element ret = null;
    List<Element> elements = root.getAllElements(tagName);
    for (Element element : elements) {
        StartTag tag = element.getStartTag();
        String value = tag.getAttributeValue(attrName);
        if (expectedValue.equals(value)) {
            ret = element;
            break;
        }
    }
    return ret;
}

Example 49 with Element

use of net.htmlparser.jericho.Element in project lotro-tools by dmorcellet.

the class QuestPageParser method parseQuestSection.

private QuestDescription parseQuestSection(Element questSection) {
    QuestDescription ret = null;
    try {
        _quest = new QuestDescription();
        Element officialSection = JerichoHtmlUtils.findElementByTagNameAndAttributeValue(questSection, HTMLElementName.DIV, "class", "officialsection");
        if (officialSection != null) {
            parseQuestDescription(officialSection);
        }
        // Texts
        StringBuilder description = new StringBuilder();
        StringBuilder bestower = new StringBuilder();
        StringBuilder bestowerText = new StringBuilder();
        StringBuilder objectives = new StringBuilder();
        List<Element> textSections = JerichoHtmlUtils.findElementsByTagNameAndAttributeValue(questSection, HTMLElementName.DIV, "class", "iteminfosection widget ui-corner-all");
        for (Element textSection : textSections) {
            Element titleSection = JerichoHtmlUtils.findElementByTagNameAndAttributeValue(textSection, HTMLElementName.DIV, "class", "widget-head ui-widget-header ui-corner-top");
            if (titleSection != null) {
                String textSectionTitle = CharacterReference.decodeCollapseWhiteSpace(titleSection.getContent());
                Element contentsSection = JerichoHtmlUtils.findElementByTagNameAndAttributeValue(textSection, HTMLElementName.DIV, "class", "widget-body ui-widget-content ui-corner-bottom");
                if (contentsSection != null) {
                    StringBuilder link = null;
                    StringBuilder text = null;
                    if ("Description".equals(textSectionTitle)) {
                        text = description;
                    } else if ("Bestower".equals(textSectionTitle)) {
                        link = bestower;
                        text = bestowerText;
                    } else if ("Objectives".equals(textSectionTitle)) {
                        text = objectives;
                    }
                    if (text != null) {
                        Element bestowerTag = JerichoHtmlUtils.findElementByTagNameAndAttributeValue(contentsSection, HTMLElementName.DIV, "class", "bestowertext");
                        if (bestowerTag != null) {
                            String contents = JerichoHtmlUtils.getTextFromTag(bestowerTag);
                            text.append(contents);
                        }
                    }
                    if (link != null) {
                        Element bestowerLink = JerichoHtmlUtils.findElementByTagNameAndAttributeValue(contentsSection, HTMLElementName.DIV, "class", "bestowerlink");
                        if (bestowerLink != null) {
                            String contents = JerichoHtmlUtils.getTextFromTag(bestowerLink);
                            link.append(contents);
                        }
                    }
                }
            }
        }
        _quest.setDescription(description.toString().trim());
        _quest.setBestower(bestower.toString().trim());
        _quest.setBestowerText(bestowerText.toString().trim());
        _quest.setObjectives(objectives.toString().trim());
        ret = _quest;
        _quest = null;
    } catch (Exception e) {
        ret = null;
        _logger.error("Quest [" + _key + "]. Cannot parse quest section!", e);
    }
    return ret;
}

Example 50 with Element

use of net.htmlparser.jericho.Element in project lotro-tools by dmorcellet.

the class QuestPageParser method findIdentifiers.

private void findIdentifiers(List<QuestDescription> quests) {
    String url = "http://lorebook.lotro.com/index.php?title=Quest:" + _key + "&action=edit";
    DownloadService downloader = DownloadService.getInstance();
    try {
        String page = downloader.getPage(url);
        Source s = new Source(page);
        // <textarea id="wpTextbox1"
        Element pageSource = JerichoHtmlUtils.findElementByTagNameAndAttributeValue(s, HTMLElementName.TEXTAREA, "id", "wpTextbox1");
        String text = JerichoHtmlUtils.getTextFromTag(pageSource);
        parsePageSource(text, quests);
    } catch (Exception e) {
        _logger.error("Parsing error", e);
    }
}