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

Example 1 with Source

use of net.htmlparser.jericho.Source in project vue-gwt by Axellience.

the class TemplateParser method parseHtmlTemplate.

/**
 * Parse a given HTML template and return the a result object containing the expressions
 * and a transformed HTML.
 * @param htmlTemplate The HTML template to process, as a String
 * @param context Context of the Component we are currently processing
 * @param messager Used to report errors in template during Annotation Processing
 * @return A {@link TemplateParserResult} containing the processed template and expressions
 */
public TemplateParserResult parseHtmlTemplate(String htmlTemplate, TemplateParserContext context, Messager messager) {
    this.context = context;
    this.logger = new TemplateParserLogger(context, messager);
    initJerichoConfig(this.logger);
    Source source = new Source(htmlTemplate);
    outputDocument = new OutputDocument(source);
    result = new TemplateParserResult(context);
    processImports(source);
    source.getChildElements().forEach(this::processElement);
    result.setProcessedTemplate(outputDocument.toString());
    return result;
}

Example 2 with Source

use of net.htmlparser.jericho.Source 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 3 with Source

use of net.htmlparser.jericho.Source 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 4 with Source

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

the class ItemPageParser method findIdentifiers.

private void findIdentifiers(List<Item> items) {
    String url = "http://lorebook.lotro.com/index.php?title=" + _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");
        if (pageSource != null) {
            String text = JerichoHtmlUtils.getTextFromTag(pageSource);
            parsePageSource(text, items);
        } else {
            _logger.warn("Cannot find identifiers!");
        }
    } catch (Exception e) {
        _logger.error("Parsing error", e);
    }
}

Example 5 with Source

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

the class QuestPageParser method parseQuestPage.

/**
 * Parse the quest page at the given URL.
 * @param url URL of quest page.
 * @return A list of quests or <code>null</code> if an error occurred.
 */
public List<QuestDescription> parseQuestPage(String url) {
    List<QuestDescription> quests = 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) {
            // identifier
            // <a id="ca-nstab-quest" class="lorebook_action_link" href="/wiki/Quest:A_Feminine_Curve_to_the_Steel">Article</a>
            _key = null;
            Element articleLink = JerichoHtmlUtils.findElementByTagNameAndAttributeValue(source, HTMLElementName.A, "id", "ca-nstab-quest");
            if (articleLink != null) {
                String thisURL = articleLink.getAttributeValue("href");
                if ((thisURL != null) && (thisURL.startsWith(QUEST_URL_SEED))) {
                    _key = thisURL.substring(QUEST_URL_SEED.length()).trim();
                }
            }
            quests = new ArrayList<QuestDescription>();
            List<Element> questSections = JerichoHtmlUtils.findElementsByTagNameAndAttributeValue(lorebook, HTMLElementName.DIV, "class", "lorebookquest");
            if ((questSections != null) && (questSections.size() > 0)) {
                for (Element questSection : questSections) {
                    QuestDescription quest = parseQuestSection(questSection);
                    if (quest != null) {
                        // System.out.println(quest.dump());
                        quests.add(quest);
                        quest.setKey(_key);
                    }
                }
            }
            findIdentifiers(quests);
        }
    } catch (Exception e) {
        quests = null;
        _logger.error("Cannot parse quest page [" + url + "]", e);
    }
    return quests;
}