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

Example 41 with Element

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

the class RecipePageParser method parseRecipeSection.

private Recipe parseRecipeSection(Element recipeSection) {
    Recipe ret = null;
    try {
        _recipe = new Recipe();
        // Element mainIconElement=JerichoHtmlUtils.findElementByTagNameAndAttributeValue(recipeSection,HTMLElementName.DIV,"class","mainicon");
        List<Element> officialSections = JerichoHtmlUtils.findElementsByTagNameAndAttributeValue(recipeSection, HTMLElementName.DIV, "class", "officialsection");
        // 1 - summary
        // 2 - ingredients
        // 3 - results
        int nbSections = officialSections.size();
        if (nbSections != 3) {
            System.out.println("Warning: found " + nbSections + " sections!");
        }
        Element summarySection = officialSections.get(0);
        parseRecipeSummary(summarySection);
        // System.out.println("Ingredients:");
        Element ingredientsSection = officialSections.get(1);
        List<Ingredient> ingredients = parseIngredients(ingredientsSection);
        _recipe.setIngredients(ingredients);
        // System.out.println("Results:");
        Element resultsSection = officialSections.get(2);
        List<RecipeVersion> results = parseResults(resultsSection);
        _recipe.setVersions(results);
        ret = _recipe;
        _recipe = null;
    } catch (Exception e) {
        ret = null;
        _logger.error("Recipe [" + _key + "]. Cannot parse recipe section!", e);
    }
    return ret;
}

Example 42 with Element

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

the class RecipePageParser method parseResultItem.

private CraftingResult parseResultItem(Element resultRow, boolean critical) {
    CraftingResult result = new CraftingResult();
    result.setCriticalResult(critical);
    ItemReference itemRef = new ItemReference();
    result.setItem(itemRef);
    Element ingredientIcon = JerichoHtmlUtils.findElementByTagNameAndAttributeValue(resultRow, HTMLElementName.TD, "class", "resicon");
    if (ingredientIcon != null) {
        Element img = JerichoHtmlUtils.findElementByTagName(ingredientIcon, HTMLElementName.IMG);
        if (img != null) {
            String itemIcon = img.getAttributeValue("src");
            // System.out.println("Icon: "+itemIcon);
            itemRef.setIcon(itemIcon);
        }
    }
    Element ingredientQuantity = JerichoHtmlUtils.findElementByTagNameAndAttributeValue(resultRow, HTMLElementName.TD, "class", "resquan");
    if (ingredientQuantity != null) {
        String quantityStr = CharacterReference.decodeCollapseWhiteSpace(ingredientQuantity.getContent());
        int quantity = parseQuantityString(quantityStr);
        // System.out.println("Quantity: "+quantity);
        result.setQuantity(quantity);
    }
    Element ingredientName = JerichoHtmlUtils.findElementByTagNameAndAttributeValue(resultRow, HTMLElementName.TD, "class", "resname");
    if (ingredientName != null) {
        Element a = JerichoHtmlUtils.findElementByTagName(ingredientName, HTMLElementName.A);
        String itemURL = a.getAttributeValue("href");
        String itemId = extractItemIdentifier(itemURL);
        itemRef.setItemKey(itemId);
        String itemName = CharacterReference.decodeCollapseWhiteSpace(a.getContent());
        itemRef.setName(itemName);
    }
    // System.out.println(critical?"Critical result":"Regular result");
    return result;
}

Example 43 with Element

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

the class RecipePageParser method parseResults.

private List<RecipeVersion> parseResults(Element resultsSection) {
    /*
<tr class="resultrow"> // or "resultrow critical"
<td class="resicon"><a href="/wiki/Tool:Ancient_Steel_Scholar's_Glass"><img class="icon" rel="" src="http://content.turbine.com/sites/lorebook.lotro.com/images/icons/item/tool/eq_craft_tool_rare_scholars_glass_tier5.png"></img></a></td>
<td class="restype">regular success</td>  // or critical success
<td class="resquan">x1</td>
<td class="resname"><a href="/wiki/Tool:Ancient_Steel_Scholar's_Glass"></a></td></tr>
     */
    List<RecipeVersion> versions = new ArrayList<RecipeVersion>();
    RecipeVersion version = new RecipeVersion();
    versions.add(version);
    List<Element> rows = resultsSection.getAllElements(HTMLElementName.TR);
    for (Element row : rows) {
        StartTag tag = row.getStartTag();
        String value = tag.getAttributeValue("class");
        if ("resultrow".equals(value)) {
            Element header = JerichoHtmlUtils.findElementByTagName(row, HTMLElementName.TH);
            if (header != null) {
                version = new RecipeVersion();
                versions.add(version);
            // System.out.println("OR");
            } else {
                CraftingResult regular = parseResultItem(row, false);
                version.setRegular(regular);
            }
        } else if ("resultrow critical".equals(value)) {
            CraftingResult critical = parseResultItem(row, true);
            version.setCritical(critical);
        }
    }
    return versions;
}

Example 44 with Element

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

the class RecipePageParser method findIdentifiers.

private void findIdentifiers(List<Recipe> recipes) {
    String url = "http://lorebook.lotro.com/index.php?title=Recipe:" + _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, recipes);
        } else {
            _logger.warn("Cannot find identifiers!");
        }
    } catch (Exception e) {
        _logger.error("Parsing error", e);
    }
}

Example 45 with Element

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

the class RecipePageParser method findTier.

private int findTier(Segment root) {
    int tier = 0;
    Element links = JerichoHtmlUtils.findElementByTagNameAndAttributeValue(root, HTMLElementName.DIV, "id", "mw-normal-catlinks");
    if (links != null) {
        List<Element> as = JerichoHtmlUtils.findElementsByTagName(links, HTMLElementName.A);
        for (Element a : as) {
            String title = a.getAttributeValue("title");
            if (title != null) {
                int index = title.indexOf(CATEGORY_TIER);
                if (index != -1) {
                    int startIndex = index + CATEGORY_TIER.length();
                    String tierStr = title.substring(startIndex, startIndex + 2).trim();
                    tier = NumericTools.parseInt(tierStr, 0);
                    break;
                }
            }
        }
    }
    return tier;
}