List of usage examples for org.jsoup.nodes Element textNodes
public List<TextNode> textNodes()
From source file:me.vertretungsplan.parser.ESchoolParser.java
private String getNewValue(Element cell) { List<TextNode> textNodes = cell.textNodes(); if (textNodes.size() == 1) { return textNodes.get(0).text().trim(); } else if (textNodes.size() == 2) { return textNodes.get(1).text().trim(); } else {// ww w.jav a2 s .c o m return null; } }
From source file:me.vertretungsplan.parser.ESchoolParser.java
private String getPreviousValue(Element cell) { List<TextNode> textNodes = cell.textNodes(); if (textNodes.size() == 1) { return null; } else if (textNodes.size() == 2) { return textNodes.get(0).text().trim(); } else {/*from w w w . j a v a 2s . c o m*/ return null; } }
From source file:me.vertretungsplan.parser.DaVinciParser.java
@NotNull void parsePage(Element doc, SubstitutionSchedule schedule) throws IOException { SubstitutionScheduleDay day = new SubstitutionScheduleDay(); Element titleElem;/*www . ja v a2 s . com*/ if (doc.select("h1.list-table-caption").size() > 0) { titleElem = doc.select("h1.list-table-caption").first(); } else { // DaVinci 5 titleElem = doc.select("h2").first(); } String title = titleElem.text(); String klasse = null; // title can either be date or class Pattern datePattern = Pattern.compile("\\d+\\.\\d+.\\d{4}"); Matcher dateMatcher = datePattern.matcher(title); if (dateMatcher.find()) { day.setDateString(dateMatcher.group()); day.setDate(ParserUtils.parseDate(dateMatcher.group())); } else { klasse = title; String nextText = titleElem.nextElementSibling().text(); if (nextText.matches("\\w+ \\d+\\.\\d+.\\d{4}")) { day.setDateString(nextText); day.setDate(ParserUtils.parseDate(nextText)); } else { // could not find date, must be multiple days day = null; } } for (Element p : doc.select(".row:has(h1.list-table-caption) p")) { for (TextNode node : p.textNodes()) { if (!node.text().trim().isEmpty() && day != null) day.addMessage(node.text().trim()); } } for (Element message : doc.select(".callout")) { for (TextNode node : message.textNodes()) { if (!node.text().trim().isEmpty()) day.addMessage(node.text().trim()); } } Element lastChangeElem = doc.select(".row.copyright div").first(); if (lastChangeElem == null) { // DaVinci 5 lastChangeElem = doc.select("h1").first(); } String lastChange = lastChangeElem.ownText(); Pattern pattern = Pattern.compile("(\\d{2}-\\d{2}-\\d{4} \\d{2}:\\d{2}) \\|"); Matcher matcher = pattern.matcher(lastChange); if (matcher.find()) { LocalDateTime lastChangeTime = DateTimeFormat.forPattern("dd-MM-yyyy HH:mm") .parseLocalDateTime(matcher.group(1)); if (day != null) { day.setLastChange(lastChangeTime); } else { schedule.setLastChange(lastChangeTime); } } else { Pattern pattern2 = Pattern.compile("(\\d{2}.\\d{2}.\\d{4} \\| \\d+:\\d{2})"); Matcher matcher2 = pattern2.matcher(lastChange); if (matcher2.find()) { LocalDateTime lastChangeTime = DateTimeFormat.forPattern("dd.MM.yyyy | HH:mm") .parseLocalDateTime(matcher2.group(1)); if (day != null) { day.setLastChange(lastChangeTime); } else { schedule.setLastChange(lastChangeTime); } } } if (doc.select(".list-table").size() > 0 || !doc.select(".callout").text().contains("Es liegen keine")) { Element table = doc.select(".list-table, table").first(); parseDaVinciTable(table, schedule, klasse, day, colorProvider); } if (day != null) { schedule.addDay(day); } }
From source file:utils.AutoLinkRenderer.java
private AutoLinkRenderer parse(Pattern pattern, ToLink toLink) { Document doc = Jsoup.parse(body); Document.OutputSettings settings = doc.outputSettings(); settings.prettyPrint(false);/* w w w. ja v a2 s . c om*/ Elements elements = doc.getElementsMatchingOwnText(pattern); for (Element el : elements) { if (isIgnoreElement(el)) { continue; } List<TextNode> textNodeList = el.textNodes(); for (TextNode node : textNodeList) { String result = convertLink(node.toString(), pattern, toLink); node.text(StringUtils.EMPTY); node.after(result); } } this.body = doc.body().html(); return this; }
From source file:wo.trade.SearchPageScraper.java
public List<TradeItem> parse() { List<TradeItem> tradeItems = new LinkedList<>(); Document doc = Jsoup.parse(page, "UTF-8"); Element content = doc.getElementById("content"); Elements items = null;//from w w w . j ava2s. c o m if (content == null) { items = doc.getElementsByClass("item"); } else { items = content.getElementsByClass("item"); } for (Element element : items) { TradeItem item = new TradeItem(); item.id = element.attr("id"); item.id = StringUtils.remove(item.id, "item-container-"); item.seller = element.attr("data-seller"); item.thread = element.attr("data-thread"); item.sellerid = element.attr("data-sellerid"); item.buyout = element.attr("data-buyout"); item.ign = element.attr("data-ign"); item.league = element.attr("data-league"); item.name = element.attr("data-name"); item.corrupted = element.getElementsByClass("corrupted").size() > 0; item.identified = element.getElementsByClass("item-unid").size() == 0; // System.out.println(String.format("Now parsing item id %s name %s", item.id, item.name)); Element sockElem = element.getElementsByClass("sockets-raw").get(0); item.socketsRaw = sockElem.text(); Elements accntAgeElement = element.getElementsByAttributeValue("title", "account age and highest level"); if (accntAgeElement != null && !accntAgeElement.isEmpty()) { item.ageAndHighLvl = accntAgeElement.get(0).text(); } // ----- Requirements ----- // Element reqElem = element.getElementsByClass("requirements").get(0); List<TextNode> reqNodes = reqElem.textNodes(); for (TextNode reqNode : reqNodes) { // sample [ Level: 37 , Strength: 42 , Intelligence: 42 ] String req = StringUtils.trimToEmpty(reqNode.getWholeText()); req = req.replaceAll(regex_horizontal_whitespace, ""); req = Util.removeThoseDamnWhiteSpace(req); String separator = ":"; String reqType = trim(substringBefore(req, separator)); switch (reqType) { case "Level": item.reqLvl = trim(substringAfter(req, separator)); break; case "Strength": item.reqStr = trim(substringAfter(req, separator)); break; case "Intelligence": item.reqInt = trim(substringAfter(req, separator)); break; case "Dexterity": item.reqDex = trim(substringAfter(req, separator)); break; } } item.mapQuantity = element.getElementsByAttributeValue("data-name", "mapq").stream().findFirst() .map(n -> n.text()).map(s -> substringAfter(s, "Item quantity:")) .map(s -> StringUtils.removePattern(s, "[^\\d]")).orElse("") .replaceAll(regex_horizontal_whitespace, "").trim(); // ----- Rarity by checking the item name link class ----- // // itemframe0 - normal // itemframe1 - magic // itemframe2 - rare // itemframe3 - unique // itemframe4 - gems // itemframe5 - currency // itemframe6 - divination card String itemframeStr = element.getElementsByClass("title").stream().findFirst().map(n -> n.attr("class")) .orElse(null); itemframeStr = Util.regexMatch("itemframe(\\d)", itemframeStr, 1); if (itemframeStr != null) { int frame = Integer.parseInt(itemframeStr); item.rarity = Rarity.valueOf(frame); } else { item.rarity = Rarity.unknown; } // ----- Verify ----- // item.dataHash = element.getElementsByAttributeValue("onclick", "verify_modern(this)").stream() .findFirst().map(n -> n.attr("data-hash")).orElse("").trim(); // ----- Mods ----- // Elements itemModsElements = element.getElementsByClass("item-mods"); if (itemModsElements != null && itemModsElements.size() > 0) { Element itemMods = itemModsElements.get(0); if (itemMods.getElementsByClass("bullet-item").size() != 0) { Element bulletItem = itemMods.getElementsByClass("bullet-item").get(0); Elements ulMods = bulletItem.getElementsByTag("ul"); if (ulMods.size() == 2) { // implicit mod Elements implicitLIs = ulMods.get(0).getElementsByTag("li"); Element implicitLi = implicitLIs.last(); Mod impMod = new Mod(implicitLi.attr("data-name"), implicitLi.attr("data-value")); item.implicitMod = impMod; } int indexOfExplicitMods = ulMods.size() - 1; Elements modsLi = ulMods.get(indexOfExplicitMods).getElementsByTag("li"); for (Element modLi : modsLi) { // explicit mods Mod mod = new Mod(modLi.attr("data-name"), modLi.attr("data-value")); item.explicitMods.add(mod); } } } // ----- Properties ----- // // this is the third column data (the first col is the image, second is the mods, reqs) item.quality = element.getElementsByAttributeValue("data-name", "q").get(0).text() .replaceAll(regex_horizontal_whitespace, "").trim(); item.physDmgRangeAtMaxQuality = element.getElementsByAttributeValue("data-name", "quality_pd").get(0) .text().replaceAll(regex_horizontal_whitespace, "").trim(); item.eleDmgRange = element.getElementsByAttributeValue("data-name", "ed").get(0).text() .replaceAll(regex_horizontal_whitespace, "").trim(); item.attackSpeed = element.getElementsByAttributeValue("data-name", "aps").get(0).text() .replaceAll(regex_horizontal_whitespace, "").trim(); item.dmgAtMaxQuality = element.getElementsByAttributeValue("data-name", "quality_dps").get(0).text() .replaceAll(regex_horizontal_whitespace, "").trim(); item.physDmgAtMaxQuality = element.getElementsByAttributeValue("data-name", "quality_pdps").get(0) .text().replaceAll(regex_horizontal_whitespace, "").trim(); item.eleDmg = element.getElementsByAttributeValue("data-name", "edps").get(0).text() .replaceAll(regex_horizontal_whitespace, "").trim(); item.armourAtMaxQuality = element.getElementsByAttributeValue("data-name", "quality_armour").get(0) .text().replaceAll(regex_horizontal_whitespace, "").trim(); item.evasionAtMaxQuality = element.getElementsByAttributeValue("data-name", "quality_evasion").get(0) .text().replaceAll(regex_horizontal_whitespace, "").trim(); item.energyShieldAtMaxQuality = element.getElementsByAttributeValue("data-name", "quality_shield") .get(0).text().replaceAll(regex_horizontal_whitespace, "").trim(); item.block = element.getElementsByAttributeValue("data-name", "block").get(0).text() .replaceAll(regex_horizontal_whitespace, "").trim(); item.crit = element.getElementsByAttributeValue("data-name", "crit").get(0).text() .replaceAll(regex_horizontal_whitespace, "").trim(); item.level = element.getElementsByAttributeValue("data-name", "level").get(0).text() .replaceAll(regex_horizontal_whitespace, "").trim(); item.imageUrl = element.getElementsByAttributeValue("alt", "Item icon").get(0).attr("src"); item.stackSize = asList(split(trimToEmpty(item.imageUrl), '&')).stream() .filter(t -> t.startsWith("stackSize=")).findFirst().map(s -> substringAfter(s, "=")) .orElse(null); Elements onlineSpans = element.getElementsMatchingText("online"); if (!onlineSpans.isEmpty()) { item.online = "Online"; } else { item.online = ""; } tradeItems.add(item); } // System.out.println("DONE --- Items"); return tradeItems; }