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

Example 61 with Element

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

the class RewardsHTMLParser method parseTraitReward.

private void parseTraitReward(Element rewardDiv, Rewards rewards) {
    List<Element> as = rewardDiv.getAllElements(HTMLElementName.A);
    int size = (as != null) ? as.size() : 0;
    if (size == 2) {
        Element secondA = as.get(1);
        String name = CharacterReference.decodeCollapseWhiteSpace(secondA.getContent());
        // String iconURL=null;
        Element firstA = as.get(0);
        String url = firstA.getAttributeValue("href");
        if ((url != null) && (url.startsWith(URL_SEED))) {
            String qualifiedIdentifier = url.substring(URL_SEED.length()).trim();
            if (qualifiedIdentifier.startsWith(TRAIT_URL_SEED)) {
                String identifier = qualifiedIdentifier.substring(TRAIT_URL_SEED.length()).trim();
                if ((identifier.startsWith(EMOTE_SEED)) && (name.startsWith(EMOTE_SEED))) {
                    // identifier=identifier.substring(EMOTE_SEED.length()).trim();
                    name = name.substring(EMOTE_SEED.length()).trim();
                    Emote emote = new Emote(/*identifier,*/
                    name);
                    rewards.addEmote(emote);
                // System.out.println(emote.dump());
                } else {
                    Trait trait = new Trait(/*identifier,*/
                    name);
                    rewards.addTrait(trait);
                // System.out.println(trait.dump());
                }
            } else if (qualifiedIdentifier.startsWith(PASSIVE_SKILL_URL_SEED)) {
                // String identifier=qualifiedIdentifier.substring(PASSIVE_SKILL_URL_SEED.length()).trim();
                Skill skill = new Skill(/*SkillType.PASSIVE,identifier,*/
                name);
                rewards.addSkill(skill);
            // System.out.println(skill.dump());
            } else {
                _logger.warn(_objectId + ": unmanaged trait/skill identifier [" + qualifiedIdentifier + "]");
            }
        /*
        List<Element> imgs=firstA.getAllElements(HTMLElementName.IMG);
        if ((imgs!=null) && (imgs.size()>=1))
        {
          Element img=imgs.get(0);
          iconURL=img.getAttributeValue("src");
        }
        */
        } else {
            _logger.warn(_objectId + ": malformed URL [" + url + "]");
        }
    } else {
        _logger.warn(_objectId + ": trait reward with " + size + " anchor tags!");
    }
/*
    <div class="questReward">
    <div>
    <strong>Traits:</strong>
    </div>
    <div>
    <a href="/wiki/Trait:Expert_Woodworker_Proficiency">
    <img class="icon" src="http://content.turbine.com/sites/lorebook.lotro.com/images/icons/trait/trait_c_craft_woodworker_complete_proficiency_3.png">
    </a>
    <a href="/wiki/Trait:Expert_Woodworker_Proficiency">Expert Woodworker Proficiency</a>
    </div>
    </div>
     */
}

Example 62 with Element

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

the class RewardsHTMLParser method parseVirtuesReward.

private void parseVirtuesReward(Element rewardDiv, Rewards rewards) {
    List<Element> as = rewardDiv.getAllElements(HTMLElementName.A);
    int size = (as != null) ? as.size() : 0;
    int nbItems = size / 2;
    if (nbItems * 2 == size) {
        for (int i = 0; i < nbItems; i++) {
            String name = null;
            String virtueIdentifier = null;
            // String iconURL=null;
            Element firstA = as.get(2 * i + 1);
            String url = firstA.getAttributeValue("href");
            if ((url != null) && (url.startsWith(VIRTUE_URL_SEED))) {
                virtueIdentifier = url.substring(VIRTUE_URL_SEED.length()).trim();
            }
            name = CharacterReference.decodeCollapseWhiteSpace(firstA.getContent());
            if ((name != null) && (virtueIdentifier != null)) {
                VirtueId virtueId = VirtueId.valueOf(virtueIdentifier.toUpperCase());
                Virtue virtue = new Virtue(virtueId, 1);
                rewards.addVirtue(virtue);
            }
        }
    } else {
        _logger.warn(_objectId + ": virtue with " + size + " anchor tags!");
    }
/*
<div class="questReward">
<div>
<strong>Virtues:</strong>
</div>
<div>
<a href="/wiki/Trait:Patience">
<img class="icon" src="http://content.turbine.com/sites/lorebook.lotro.com/images/icons/trait/trait_virtue_patience.png">
</a>
<a href="/wiki/Trait:Patience">Patience</a>
</div>
</div>
 */
}

Example 63 with Element

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

the class LotroWikiDeedPageParser method parseDeeds.

/**
 * Parse the lotro wiki deed page for the given deed ID.
 * @param from Source page.
 * @param deedId Deed ID.
 * @return A deed or <code>null</code> if an error occurred.
 */
public List<DeedDescription> parseDeeds(File from, String deedId) {
    _currentFile = from;
    List<DeedDescription> deeds = null;
    try {
        FileInputStream inputStream = new FileInputStream(from);
        Source source = new Source(inputStream);
        String name = null;
        Element backElement = JerichoHtmlUtils.findElementByTagNameAndAttributeValue(source, HTMLElementName.DIV, "id", "contentSub");
        if (backElement != null) {
            Element a = JerichoHtmlUtils.findElementByTagName(backElement, HTMLElementName.A);
            if (a != null) {
                name = a.getAttributeValue("title");
            }
        }
        Element deedSource = JerichoHtmlUtils.findElementByTagNameAndAttributeValue(source, HTMLElementName.TEXTAREA, "id", "wpTextbox1");
        if (deedSource != null) {
            Segment content = deedSource.getContent();
            String text = content.toString();
            deeds = buildDeeds(text, name);
        }
        if (deeds != null) {
            // Fixes
            for (DeedDescription deed : deeds) {
                handleFixes(deed);
            }
        }
    } catch (Exception e) {
        _logger.error("Cannot parse deed page [" + from + "]", e);
    }
    return deeds;
}

Example 64 with Element

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

the class ItemPageParser method parseItemsSet.

private ItemsSet parseItemsSet(Element itemTooltip) {
    ItemsSet ret = null;
    Element itemsSet = JerichoHtmlUtils.findElementByTagNameAndAttributeValue(itemTooltip, HTMLElementName.DIV, "class", "itemset");
    if (itemsSet != null) {
        ret = new ItemsSet();
        String setKey = null;
        String setName = "";
        String setURL = null;
        Element itemsSetName = JerichoHtmlUtils.findElementByTagNameAndAttributeValue(itemsSet, HTMLElementName.DIV, "class", "itemsn");
        if (itemsSetName != null) {
            Element aElement = itemsSetName.getFirstElement(HTMLElementName.A);
            if (aElement != null) {
                setURL = aElement.getAttributeValue("href");
                if ((setURL != null) && (setURL.startsWith(WIKI_SEED))) {
                    setKey = setURL.substring(WIKI_SEED.length());
                    int index = setKey.indexOf(":");
                    if (index != -1) {
                        setKey = setKey.substring(index + 1);
                    }
                }
                setName = JerichoHtmlUtils.getTagContents(itemsSetName, HTMLElementName.A);
            }
        }
        // System.out.println("Set name: "+setName);
        // System.out.println("Set URL: "+setURL);
        ret.setName(setName);
        ret.setKey(setKey);
        // Set items
        Element setItemsElement = JerichoHtmlUtils.findElementByTagNameAndAttributeValue(itemsSet, HTMLElementName.DIV, "class", "itemsps");
        if (setItemsElement != null) {
            List<Element> setItemElements = JerichoHtmlUtils.findElementsByTagNameAndAttributeValue(itemsSet, HTMLElementName.DIV, "class", "itemsp");
            for (Element setItemElement : setItemElements) {
                String itemKey = null;
                String itemURL = null;
                // String itemName=null;
                Element aElement = setItemElement.getFirstElement(HTMLElementName.A);
                if (aElement != null) {
                    itemURL = aElement.getAttributeValue("href");
                    if ((itemURL != null) && (itemURL.startsWith(WIKI_SEED))) {
                        itemKey = itemURL.substring(WIKI_SEED.length());
                    }
                // itemName=JerichoHtmlUtils.getTagContents(setItemElement,HTMLElementName.A);
                }
                if (itemKey != null) {
                    ret.addItem(0, itemKey);
                }
            // System.out.println("Item ["+itemName+"]. URL=["+itemURL+"]");
            }
        }
        // Set effects
        Element setEffectsElement = JerichoHtmlUtils.findElementByTagNameAndAttributeValue(itemsSet, HTMLElementName.DIV, "class", "itemseteff");
        if (setEffectsElement != null) {
            List<Element> divs = setEffectsElement.getAllElements(HTMLElementName.DIV);
            // Remove self
            divs.remove(0);
            int nbOfItems = 0;
            for (Element div : divs) {
                String effectLine = JerichoHtmlUtils.getTagContents(div, HTMLElementName.DIV);
                // System.out.println("Effect line ["+effectLine+"]");
                effectLine = effectLine.trim();
                if (effectLine.length() > 0) {
                    if ((effectLine.startsWith(SET_NUMBER_SEED)) && (effectLine.endsWith(SET_NUMBER_END))) {
                        String nbStr = TextTools.findBetween(effectLine, SET_NUMBER_SEED, SET_NUMBER_END);
                        nbOfItems = NumericTools.parseInt(nbStr, 0);
                    } else {
                        if (nbOfItems != 0) {
                            ret.addBonus(nbOfItems, effectLine);
                        }
                    }
                }
            }
        }
    }
    /*
<div class="itemset">
  <div class="itemsn"><a href="/wiki/Item:Armaments_of_the_Impossible_Shot">Armaments of the Impossible Shot</a></div>
  <div class="itemsps">
    <div class="itemsp"><a href="/wiki/Armour:Helm_of_the_Impossible_Shot">Helm of the Impossible Shot</a></div>
    <div class="itemsp"><a href="/wiki/Armour:Shoulder_guards_of_Impossible_Shot">Shoulder guards of Impossible Shot</a></div>
    <div class="itemsp"><a href="/wiki/Armour:Jacket_of_the_Impossible_Shot">Jacket of the Impossible Shot</a></div>
    <div class="itemsp"><a href="/wiki/Armour:Leggings_of_the_Impossible_Shot">Leggings of the Impossible Shot</a></div>
    <div class="itemsp"><a href="/wiki/Armour:Gloves_of_the_Impossible_Shot">Gloves of the Impossible Shot</a></div>
    <div class="itemsp"><a href="/wiki/Armour:Boots_of_the_Impossible_Shot">Boots of the Impossible Shot</a></div>
  </div>
  <div class="itemseteff">
    <div>Set (2):</div>
    <div>+52.8 in-Combat Power Regen</div>
    <div>Set (3):</div>
    <div>+10% Heart Seeker Damage</div>
    <div>Set (4):</div>
    <div>+600 Physical Mastery Rating</div>
    <div> </div>
    <div>Set (5):</div>
    <div>Intent Concentration recovers all Archer's Art skills</div>
  </div>
  <div class="itemsd"></div>
</div>
     */
    return ret;
}

Example 65 with Element

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

the class ItemPageParser method internalParseItemPage.

/**
 * Parse the item page at the given URL.
 * @param url URL of item page.
 * @return An item or <code>null</code> if an error occurred.
 */
private List<Item> internalParseItemPage(String url) {
    List<Item> items = null;
    try {
        DownloadService downloader = DownloadService.getInstance();
        String page = downloader.getPage(url);
        Source source = new Source(page);
        // <div id="lorebookNoedit">
        // Element lorebook=JerichoHtmlUtils.findElementByTagNameAndAttributeValue(source,HTMLElementName.DIV,"id","lorebookNoedit");
        // if (lorebook!=null)
        {
            items = new ArrayList<Item>();
            // Fetch identifier
            List<Element> itemSections = JerichoHtmlUtils.findElementsByTagNameAndAttributeValue(source, HTMLElementName.TABLE, "class", "tooltip");
            if ((itemSections != null) && (itemSections.size() > 0)) {
                String key = fetchItemKey(source);
                _key = key;
                for (Element itemSection : itemSections) {
                    Item item = parseItemSection(itemSection);
                    if (item != null) {
                        items.add(item);
                        item.setKey(key);
                    }
                }
                findIdentifiers(items);
            }
            List<Element> resourceSections = JerichoHtmlUtils.findElementsByTagNameAndAttributeValue(source, HTMLElementName.DIV, "class", "lorebookresource");
            if ((resourceSections != null) && (resourceSections.size() > 0)) {
                String key = fetchItemKey(source);
                _key = key;
                for (Element resourceSection : resourceSections) {
                    Item item = parseResourceSection(resourceSection);
                    if (item != null) {
                        items.add(item);
                        item.setKey(key);
                    }
                }
                findIdentifiers(items);
            }
        }
    } catch (Exception e) {
        items = null;
        _logger.error("Cannot parse item page [" + url + "]", e);
    }
    return items;
}