List of usage examples for org.jsoup.nodes Element getElementsContainingText
public Elements getElementsContainingText(String searchText)
From source file:org.brnvrn.Main.java
/** * Parse a tr HTML element describing the tool * @param tool is to be updated/* w w w. j ava 2s .c om*/ * @param tr brings the data * @return true if successful */ private static boolean parseTrTool(Tool tool, Element tr) { boolean success = true; Element nameLink = tr.select("td:eq(0)").first(); if (nameLink == null) return false; tool.setName(nameLink.text()); tool.setUrl(nameLink.getElementsByTag("a").attr("href")); tool.setLicense(tr.select("td:eq(2)").first().text()); tool.setCompatibility(tr.select("td:eq(3)").first().text()); // More complicated: We will extract and remove known nodes, the rest will be description Element tdDescription = tr.select("td:eq(1)").first(); Elements smalls = tdDescription.getElementsByTag("small"); for (Element small : smalls) { Element author = small.getElementsContainingText("Author").first(); if (author != null) { String authorsString = author.text(); authorsString = authorsString.substring(authorsString.indexOf(":") + 1); tool.addAuthor(authorsString.split(",")); small.remove(); } Element sourceCode = small.getElementsContainingText("ource").last(); if (sourceCode != null) { tool.setUrl_src(sourceCode.attr("href")); small.remove(); } } tdDescription.getElementsByTag("br").remove(); tool.setDescription(Jsoup.clean(tdDescription.html(), Whitelist.relaxed())); // ownText will miss the contained links in the description tool.setDescriptionText(tdDescription.text()); bestEffortThemeLanguage(tool); return success; }
From source file:eu.masconsult.bgbanking.banks.dskbank.DskClient.java
private boolean checkLoggedIn(Document doc) { Elements sup_links = doc.getElementsByClass("supplemental_links"); if (sup_links == null || sup_links.size() == 0) { throw new ParseException("getBankAccounts: can't find .supplemental_links"); }/*from www .ja v a 2 s. com*/ for (Element sup_link : sup_links) { Elements exits = sup_link.getElementsContainingText("Log Out"); if (exits != null && exits.size() > 0) { return true; } } return false; }