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

Example 56 with Element

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

the class CharacterPageParser method parseMainPage.

/**
 * Parse the character page at the given URL.
 * @param name Toon name.
 * @param url URL of character page.
 * @param retryNumber Number of previous tries.
 * @return A character or <code>null</code> if an error occurred..
 */
private CharacterData parseMainPage(String name, String url, int retryNumber) {
    if (_logger.isInfoEnabled()) {
        _logger.info("Character description page parsing for toon [" + name + "] " + ((retryNumber > 0) ? " try #" + retryNumber : ""));
    }
    CharacterData ret = null;
    try {
        DownloadService downloader = DownloadService.getInstance();
        String page = downloader.getPage(url);
        Source source = new Source(page);
        _character = null;
        // <table class="char_panel freep">
        Element charPanel = JerichoHtmlUtils.findElementByTagNameAndAttributeValue(source, HTMLElementName.TABLE, "class", "char_panel freep");
        if (charPanel != null) {
            _character = new CharacterData();
            parseCharacterDescription(charPanel);
            parseStats(charPanel);
        }
        ret = _character;
        _character = null;
    } catch (Exception e) {
        if (_logger.isInfoEnabled()) {
            _logger.info("Cannot parse character page [" + url + "]", e);
        }
    }
    return ret;
}

Example 57 with Element

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

the class RewardsHTMLParser method parseReputationReward.

private Reputation parseReputationReward(Element rewardDiv) {
    // System.out.println("Reputation reward!");
    Reputation r = new Reputation();
    List<Element> elements = rewardDiv.getChildElements();
    if ((elements != null) && (elements.size() == 2)) {
        Element reputationNode = elements.get(1);
        List<Segment> nodes = JerichoHtmlUtils.getChildNodes(reputationNode);
        int nbNodes = nodes.size();
        int nbItems = nbNodes / 4;
        for (int i = 0; i < nbItems; i++) {
            Segment valueNode = nodes.get(i * 4 + 1);
            Segment factionNode = nodes.get(i * 4 + 3);
            if ((valueNode.getClass() == Segment.class) && (factionNode.getClass() == Segment.class)) {
                String valueStr = valueNode.toString();
                valueStr = valueStr.replace("with", "").trim();
                valueStr = valueStr.replace("+", "").trim();
                int reputation = NumericTools.parseInt(valueStr, 0);
                String factionName = factionNode.toString().trim();
                Faction faction = FactionsRegistry.getInstance().getByName(factionName);
                if (faction != null) {
                    ReputationItem item = new ReputationItem(faction);
                    item.setAmount(reputation);
                    r.add(item);
                } else {
                    _logger.error("Cannot get faction [" + factionName + "]!");
                }
            }
        }
    }
    return r;
/*
<div class="questReward">
<div>
<strong>Reputation:</strong>
</div>
<div>
<img class="icon" src="http://content.turbine.com/sites/lorebook.lotro.com/images/icons/reputation_increase.gif">
+700 with
<a href="/wiki/Faction:Malledhrim">Malledhrim</a>
</div>
</div>
     */
}

Example 58 with Element

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

the class RewardsHTMLParser method parseMoneyReward.

private Money parseMoneyReward(Element rewardDiv) {
    // System.out.println("Money reward!");
    Money m = new Money();
    List<Element> elements = rewardDiv.getChildElements();
    if ((elements != null) && (elements.size() == 2)) {
        Element moneyNode = elements.get(1);
        List<Segment> nodes = JerichoHtmlUtils.getChildNodes(moneyNode);
        int nb = nodes.size();
        for (int i = 0; i < nb; i++) {
            Segment s = nodes.get(i);
            if (s.getClass() == Segment.class) {
                int nbCoins = NumericTools.parseInt(s.toString(), 0);
                if ((nbCoins > 0) && (i < nb - 1)) {
                    Segment tmp = nodes.get(i + 1);
                    if (tmp instanceof StartTag) {
                        String className = ((StartTag) tmp).getAttributeValue("class");
                        if ("coin".equals(className)) {
                            String type = ((StartTag) tmp).getAttributeValue("src");
                            if (type != null) {
                                if (type.contains("silver")) {
                                    m.setSilverCoins(nbCoins);
                                } else if (type.contains("copper")) {
                                    m.setCopperCoins(nbCoins);
                                } else if (type.contains("gold")) {
                                    m.setGoldCoins(nbCoins);
                                } else {
                                    _logger.warn(_objectId + ": unknown coin type [" + type + "]");
                                }
                            }
                        }
                    }
                }
            }
        }
    }
    return m;
/*
    <div class="questReward">
    <div>
    <strong>Money:</strong>
    </div>
    <div>
    29
    <img class="coin" src="http://content.turbine.com/sites/lorebook.lotro.com/images/icons/currency/silver.gif">
    &nbsp;5
    <img class="coin" src="http://content.turbine.com/sites/lorebook.lotro.com/images/icons/currency/copper.gif">
    </div>
    </div>
    */
}

Example 59 with Element

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

the class RewardsHTMLParser method parseItemReward.

private void parseItemReward(Element rewardDiv, Rewards rewards) {
    ObjectsSet objects = null;
    // System.out.println("Item reward!");
    String key = "";
    List<Element> divs = rewardDiv.getAllElements(HTMLElementName.DIV);
    // remove reward div
    divs.remove(0);
    for (Element div : divs) {
        List<Element> strongs = div.getAllElements(HTMLElementName.STRONG);
        if (strongs.size() > 0) {
            Element strong = strongs.get(0);
            key = CharacterReference.decodeCollapseWhiteSpace(strong.getContent());
            key = cleanupFieldName(key);
            if (RECEIVE_KEY.equals(key)) {
                objects = rewards.getObjects();
            } else if (SELECT_ONE_OF_KEY.equals(key)) {
                objects = rewards.getSelectObjects();
            } else {
                _logger.warn(_objectId + ": unmanaged object selection key [" + key + "]");
            }
        } else {
            List<Element> as = div.getAllElements(HTMLElementName.A);
            if ((as != null) && (as.size() == 2)) {
                Element iconItem = as.get(0);
                String iconURL = null;
                List<Element> imgs = iconItem.getAllElements(HTMLElementName.IMG);
                if ((imgs != null) && (imgs.size() == 1)) {
                    Element img = imgs.get(0);
                    iconURL = img.getAttributeValue("src");
                // System.out.println("Icon: "+iconURL);
                }
                Element textItem = as.get(1);
                String itemName = CharacterReference.decodeCollapseWhiteSpace(textItem.getContent());
                String url = textItem.getAttributeValue("href");
                ObjectItem item = new ObjectItem(itemName);
                item.setObjectURL(url);
                item.setIconURL(iconURL);
                int quantity = 1;
                TextExtractor extractor = div.getTextExtractor();
                extractor.excludeElement(iconItem.getStartTag());
                extractor.excludeElement(textItem.getStartTag());
                String text = extractor.toString();
                int factorIndex = text.indexOf("(x");
                if (factorIndex != -1) {
                    int parenthesisIndex = text.indexOf(')', factorIndex + 2);
                    if (parenthesisIndex != -1) {
                        String factorStr = text.substring(factorIndex + 2, parenthesisIndex);
                        quantity = NumericTools.parseInt(factorStr, 1);
                    }
                }
                if (objects != null) {
                    objects.addObject(item, quantity);
                } else {
                    _logger.warn(_objectId + ": ignored object [" + item + "], quantity=" + quantity);
                }
            // System.out.println("Item: "+itemName+", URL: "+url);
            }
        }
    }
/*

<div class="questReward">
<div>
<strong>Receive:</strong>
</div>
<div>
<a href="/wiki/Item:Drownholt_Compass">
<img class="icon" rel="" src="http://content.turbine.com/sites/lorebook.lotro.com/images/icons/item/device/it_tracking_mom_book9_chapter5.png">
</a>
<a href="/wiki/Item:Drownholt_Compass">Drownholt Compass</a>
&nbsp;(x5)
</div>
</div>
     */
}

Example 60 with Element

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

the class RewardsHTMLParser method parseDestinyPoints.

private int parseDestinyPoints(Element rewardDiv) {
    int nbDestinyPoints = 0;
    List<Element> elements = rewardDiv.getChildElements();
    if ((elements != null) && (elements.size() == 2)) {
        Element moneyNode = elements.get(1);
        List<Segment> nodes = JerichoHtmlUtils.getChildNodes(moneyNode);
        int nb = nodes.size();
        for (int i = 0; i < nb; i++) {
            Segment s = nodes.get(i);
            if (s.getClass() == Segment.class) {
                nbDestinyPoints = NumericTools.parseInt(s.toString(), 0);
            }
        }
    }
    return nbDestinyPoints;
/*
<div class="questReward">
<div>
<strong>Destiny Points:</strong>
</div>
<div>
250
<img class="icon" src="http://content.turbine.com/sites/lorebook.lotro.com/images/icons/icon_destiny_points_15.png">
</div>
</div>
 */
}