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

Example 1 with Element

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

the class AreaPageParser method parseAreaPage.

/**
 * Parse the region page at the given URL.
 * @param identifier Identifier of the area.
 * @return An area or <code>null</code> if an error occurred.
 */
public Area parseAreaPage(String identifier) {
    Area ret = null;
    String url = "http://lorebook.lotro.com/wiki/Area:" + identifier;
    try {
        DownloadService downloader = DownloadService.getInstance();
        String page = downloader.getPage(url);
        Source source = new Source(page);
        // <div class="lorebooktitle">Area: Bindbole Wood</div>
        Element titleTag = JerichoHtmlUtils.findElementByTagNameAndAttributeValue(source, HTMLElementName.DIV, "class", "lorebooktitle");
        String name = "";
        if (titleTag != null) {
            name = CharacterReference.decodeCollapseWhiteSpace(titleTag.getContent());
            if (name.startsWith(TITLE_SEED)) {
                name = name.substring(TITLE_SEED.length()).trim();
            }
        }
        _area = new Area(identifier, name);
        // <div id="regionQuests"
        Element questsTable = JerichoHtmlUtils.findElementByTagNameAndAttributeValue(source, HTMLElementName.TABLE, "id", "region_quests_table");
        if (questsTable != null) {
            parseQuests(questsTable);
        }
        ret = _area;
        _area = null;
    } catch (Exception e) {
        _logger.error("Cannot parse region page [" + url + "]", e);
    }
    return ret;
}

Example 2 with Element

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

the class AreaPageParser method parseQuestRow.

private void parseQuestRow(Element questRow) {
    /*
    <tr>
    <td style="text-align: left;">
    <a href="/wiki/Quest:Bundle_for_Bywater">Bundle for Bywater</a>
    </td>
    <td>9</td>
    </tr>
     */
    // String questName=JerichoHtmlUtils.getTagContents(questRow,HTMLElementName.A);
    Element firstA = questRow.getFirstElement(HTMLElementName.A);
    if (firstA != null) {
        String url = firstA.getAttributeValue("href");
        if ((url != null) && (url.startsWith(QUEST_URL_SEED))) {
            String questIdentifier = url.substring(QUEST_URL_SEED.length()).trim();
            _area.addQuest(questIdentifier);
        }
    // System.out.println("Quest: ["+questName+"], URL=["+url+"]");
    }
}

Example 3 with Element

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

the class RegionPageParser method parseArea.

private void parseArea(Element area) {
    /*
<div>
<img id="1879064022" class="territoryMarker" src="http://content.turbine.com/sites/google_map/search_result_icon.gif" title="Highlight on Map">
<a href="/wiki/Area:Bindbole_Wood">Bindbole Wood</a>
</div>
     */
    String areaName = JerichoHtmlUtils.getTagContents(area, HTMLElementName.A);
    Element firstA = area.getFirstElement(HTMLElementName.A);
    if (firstA != null) {
        String url = firstA.getAttributeValue("href");
        if ((url != null) && (url.startsWith(AREA_URL_SEED))) {
            String areaIdentifier = url.substring(AREA_URL_SEED.length()).trim();
            Area a = new Area(areaIdentifier, areaName);
            _region.addArea(a);
        }
    // System.out.println("Area: ["+areaName+"], URL=["+url+"]");
    }
}

Example 4 with Element

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

the class RegionPageParser method parseRegionPage.

/**
 * Parse a region page.
 * @param identifier Identifier of the region.
 * @return A region or <code>null</code> if an error occurred.
 */
public Region parseRegionPage(String identifier) {
    Region ret = null;
    String url = "http://lorebook.lotro.com/wiki/Region:" + identifier;
    try {
        DownloadService downloader = DownloadService.getInstance();
        String page = downloader.getPage(url);
        Source source = new Source(page);
        // <div class="lorebooktitle">Region: The Shire</div>
        Element titleTag = JerichoHtmlUtils.findElementByTagNameAndAttributeValue(source, HTMLElementName.DIV, "class", "lorebooktitle");
        String name = "";
        if (titleTag != null) {
            name = CharacterReference.decodeCollapseWhiteSpace(titleTag.getContent());
            if (name.startsWith(TITLE_SEED)) {
                name = name.substring(TITLE_SEED.length()).trim();
            }
        }
        _region = new Region(identifier, name);
        // <div class="regionAreas widget ui-corner-all">
        Element regionAreas = JerichoHtmlUtils.findElementByTagNameAndAttributeValue(source, HTMLElementName.DIV, "class", "regionAreas widget ui-corner-all");
        if (regionAreas != null) {
            parseAreas(regionAreas);
        }
        ret = _region;
        _region = null;
    } catch (Exception e) {
        _logger.error("Cannot parse region page [" + url + "]", e);
    }
    return ret;
}

Example 5 with Element

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

the class RelicsIndexPageParser method handleTableRow.

private Relic handleTableRow(Element tr, Integer level, RelicType defaultType) {
    Relic relic = null;
    List<Element> tds = JerichoHtmlUtils.findElementsByTagName(tr, HTMLElementName.TD);
    if (tds.size() == 2) {
        // Icon & name
        Element iconAndNameElement = tds.get(0);
        List<Element> as = JerichoHtmlUtils.findElementsByTagName(iconAndNameElement, HTMLElementName.A);
        Element iconElement = as.get(0);
        String iconPath = extractIcon(iconElement);
        // Name
        Element nameElement = as.get(1);
        String name = JerichoHtmlUtils.getTagContents(nameElement, HTMLElementName.SPAN);
        if (name == null) {
            name = JerichoHtmlUtils.getTagContents(nameElement, HTMLElementName.A);
        }
        // Type
        RelicType type = getTypeFromName(name);
        relic = new Relic(name, type, level);
        relic.setIconFilename(iconPath);
        // Stats
        Element statsElement = tds.get(1);
        String statsStr = JerichoHtmlUtils.getTextFromTag(statsElement);
        statsStr = statsStr.replaceAll(",", "\n");
        BasicStatsSet stats = parseStats(statsStr);
        relic.getStats().setStats(stats);
    } else if (tds.size() == 3) {
        // Icon
        Element iconElement = tds.get(0);
        String iconPath = extractIcon(iconElement);
        // Name
        Element nameElement = tds.get(1);
        String name = JerichoHtmlUtils.getTagContents(nameElement, HTMLElementName.A);
        // Type
        RelicType type = getTypeFromName(name);
        relic = new Relic(name, type, level);
        relic.setIconFilename(iconPath);
        // Stats
        Element statsElement = tds.get(2);
        String statsStr = JerichoHtmlUtils.getTextFromTag(statsElement);
        BasicStatsSet stats = parseStats(statsStr);
        relic.getStats().setStats(stats);
    } else if (// Retired relics
    tds.size() == 4) {
        // Icon
        Element iconElement = tds.get(0);
        String iconPath = extractIcon(iconElement);
        // Name
        Element nameElement = tds.get(1);
        String name = JerichoHtmlUtils.getTagContents(nameElement, HTMLElementName.A);
        if (name == null) {
            name = JerichoHtmlUtils.getTagContents(nameElement, HTMLElementName.TD);
        }
        if (name.startsWith("Relic:")) {
            name = name.substring(6).trim();
        }
        // Tier
        /*
      Element tierElement=tds.get(2);
      String tierStr=JerichoHtmlUtils.getTextFromTag(tierElement);
      Integer tier=null;
      if (tierStr.length()>0)
      {
        tier=NumericTools.parseInteger(tierStr.trim());
      }
      else
      {
        tier=null;
      }
      */
        relic = new Relic(name, defaultType, level);
        relic.setIconFilename(iconPath);
        // Stats
        Element statsElements = tds.get(3);
        String statsStr = JerichoHtmlUtils.getTextFromTag(statsElements);
        statsStr = statsStr.replace(", ", "\n");
        BasicStatsSet stats = parseStats(statsStr);
        relic.getStats().setStats(stats);
    } else if (// Crafted relics
    tds.size() == 5) {
        // Icon
        Element iconElement = tds.get(0);
        String iconPath = extractIcon(iconElement);
        // Name
        Element nameElement = tds.get(1);
        String name = JerichoHtmlUtils.getTagContents(nameElement, HTMLElementName.A);
        // Type
        RelicType type = RelicType.CRAFTED_RELIC;
        // Level
        Element levelElement = tds.get(2);
        String levelStr = JerichoHtmlUtils.getTextFromTag(levelElement);
        if (levelStr.length() > 0) {
            level = NumericTools.parseInteger(levelStr.trim());
        } else {
            level = null;
        }
        relic = new Relic(name, type, level);
        relic.setIconFilename(iconPath);
        // Stats
        Element statsElements = tds.get(3);
        String statsStr = JerichoHtmlUtils.getTextFromTag(statsElements);
        BasicStatsSet stats = parseStats(statsStr);
        relic.getStats().setStats(stats);
    }
    // System.out.println(relic);
    return relic;
}