List of usage examples for org.jsoup.nodes Element child
public Element child(int index)
From source file:de.geeksfactory.opacclient.apis.Zones.java
private DetailledItem parse_result(String id, String html) { Document doc = Jsoup.parse(html); DetailledItem result = new DetailledItem(); result.setTitle(""); boolean title_is_set = false; result.setId(id);/*from w w w .j av a2s. c om*/ String detailTrsQuery = version18 ? ".inRoundBox1 table table tr" : ".DetailDataCell table table:not(.inRecordHeader) tr"; Elements detailtrs1 = doc.select(detailTrsQuery); for (int i = 0; i < detailtrs1.size(); i++) { Element tr = detailtrs1.get(i); int s = tr.children().size(); if (tr.child(0).text().trim().equals("Titel") && !title_is_set) { result.setTitle(tr.child(s - 1).text().trim()); title_is_set = true; } else if (s > 1) { Element valchild = tr.child(s - 1); if (valchild.select("table").isEmpty()) { String val = valchild.text().trim(); if (val.length() > 0) { result.addDetail(new Detail(tr.child(0).text().trim(), val)); } } } } for (Element a : doc.select("a.SummaryActionLink")) { if (a.text().contains("Vormerken")) { result.setReservable(true); result.setReservation_info(a.attr("href")); } } Elements detaildiv = doc.select("div.record-item-new"); if (!detaildiv.isEmpty()) { for (int i = 0; i < detaildiv.size(); i++) { Element dd = detaildiv.get(i); String text = ""; for (Node node : dd.childNodes()) { if (node instanceof TextNode) { String snip = ((TextNode) node).text(); if (snip.length() > 0) { text += snip; } } else if (node instanceof Element) { if (((Element) node).tagName().equals("br")) { text += "\n"; } else { String snip = ((Element) node).text().trim(); if (snip.length() > 0) { text += snip; } } } } result.addDetail(new Detail("", text)); } } if (doc.select("span.z3988").size() > 0) { // Sometimes there is a <span class="Z3988"> item which provides // data in a standardized format. String z3988data = doc.select("span.z3988").first().attr("title").trim(); for (String pair : z3988data.split("&")) { String[] nv = pair.split("=", 2); if (nv.length == 2) { if (!nv[1].trim().equals("")) { if (nv[0].equals("rft.btitle") && result.getTitle().length() == 0) { result.setTitle(nv[1]); } else if (nv[0].equals("rft.atitle") && result.getTitle().length() == 0) { result.setTitle(nv[1]); } else if (nv[0].equals("rft.au")) { result.addDetail(new Detail("Author", nv[1])); } } } } } // Cover if (doc.select(".BookCover, .LargeBookCover").size() > 0) { result.setCover(doc.select(".BookCover, .LargeBookCover").first().attr("src")); } Elements copydivs = doc.select("div[id^=stock_]"); String pop = ""; for (int i = 0; i < copydivs.size(); i++) { Element div = copydivs.get(i); if (div.attr("id").startsWith("stock_head")) { pop = div.text().trim(); continue; } Copy copy = new Copy(); DateTimeFormatter fmt = DateTimeFormat.forPattern("dd.MM.yyyy").withLocale(Locale.GERMAN); // This is getting very ugly - check if it is valid for libraries which are not Hamburg. // Seems to also work in Kiel (Zones 1.8, checked 10.10.2015) int j = 0; for (Node node : div.childNodes()) { try { if (node instanceof Element) { if (((Element) node).tag().getName().equals("br")) { copy.setBranch(pop); result.addCopy(copy); j = -1; } else if (((Element) node).tag().getName().equals("b") && j == 1) { copy.setLocation(((Element) node).text()); } else if (((Element) node).tag().getName().equals("b") && j > 1) { copy.setStatus(((Element) node).text()); } j++; } else if (node instanceof TextNode) { if (j == 0) { copy.setDepartment(((TextNode) node).text()); } if (j == 2) { copy.setBarcode(((TextNode) node).getWholeText().trim().split("\n")[0].trim()); } if (j == 6) { String text = ((TextNode) node).text().trim(); String date = text.substring(text.length() - 10); try { copy.setReturnDate(fmt.parseLocalDate(date)); } catch (IllegalArgumentException e) { e.printStackTrace(); } } j++; } } catch (Exception e) { e.printStackTrace(); } } } return result; }
From source file:de.geeksfactory.opacclient.apis.Bibliotheca.java
protected DetailledItem parse_result(String html) { Document doc = Jsoup.parse(html); doc.setBaseUri(opac_url);//from w w w. ja v a 2 s .c o m DetailledItem result = new DetailledItem(); if (doc.select(".detail_cover img").size() == 1) { result.setCover(doc.select(".detail_cover img").get(0).attr("src")); } result.setTitle(doc.select(".detail_titel").text()); Elements detailtrs = doc.select(".detailzeile table tr"); for (int i = 0; i < detailtrs.size(); i++) { Element tr = detailtrs.get(i); if (tr.child(0).hasClass("detail_feld")) { String title = tr.child(0).text(); String content = tr.child(1).text(); if (title.equals("Gesamtwerk:") || title.equals("Erschienen in:")) { try { if (tr.child(1).select("a").size() > 0) { Element link = tr.child(1).select("a").first(); List<NameValuePair> query = URLEncodedUtils.parse(new URI(link.absUrl("href")), "UTF-8"); for (NameValuePair q : query) { if (q.getName().equals("MedienNr")) { result.setCollectionId(q.getValue()); } } } } catch (URISyntaxException e) { } } else { if (content.contains("hier klicken") && tr.child(1).select("a").size() > 0) { content += " " + tr.child(1).select("a").first().attr("href"); } result.addDetail(new Detail(title, content)); } } } Elements detailcenterlinks = doc.select(".detailzeile_center a.detail_link"); for (int i = 0; i < detailcenterlinks.size(); i++) { Element a = detailcenterlinks.get(i); result.addDetail(new Detail(a.text().trim(), a.absUrl("href"))); } try { JSONObject copymap = new JSONObject(); if (data.has("copiestable")) { copymap = data.getJSONObject("copiestable"); } else { Elements ths = doc.select(".exemplartab .exemplarmenubar th"); for (int i = 0; i < ths.size(); i++) { Element th = ths.get(i); String head = th.text().trim(); if (head.equals("Zweigstelle")) { copymap.put("branch", i); } else if (head.equals("Abteilung")) { copymap.put("department", i); } else if (head.equals("Bereich") || head.equals("Standort")) { copymap.put("location", i); } else if (head.equals("Signatur")) { copymap.put("signature", i); } else if (head.equals("Barcode") || head.equals("Medien-Nummer")) { copymap.put("barcode", i); } else if (head.equals("Status")) { copymap.put("status", i); } else if (head.equals("Frist") || head.matches("Verf.+gbar")) { copymap.put("returndate", i); } else if (head.equals("Vorbestellungen") || head.equals("Reservierungen")) { copymap.put("reservations", i); } } } Elements exemplartrs = doc.select(".exemplartab .tabExemplar, .exemplartab .tabExemplar_"); DateTimeFormatter fmt = DateTimeFormat.forPattern("dd.MM.yyyy").withLocale(Locale.GERMAN); for (int i = 0; i < exemplartrs.size(); i++) { Element tr = exemplartrs.get(i); Copy copy = new Copy(); Iterator<?> keys = copymap.keys(); while (keys.hasNext()) { String key = (String) keys.next(); int index; try { index = copymap.has(key) ? copymap.getInt(key) : -1; } catch (JSONException e1) { index = -1; } if (index >= 0) { try { copy.set(key, tr.child(index).text(), fmt); } catch (IllegalArgumentException e) { e.printStackTrace(); } } } result.addCopy(copy); } } catch (Exception e) { e.printStackTrace(); } try { Elements bandtrs = doc.select("table .tabBand a"); for (int i = 0; i < bandtrs.size(); i++) { Element tr = bandtrs.get(i); Volume volume = new Volume(); volume.setId(tr.attr("href").split("=")[1]); volume.setTitle(tr.text()); result.addVolume(volume); } } catch (Exception e) { e.printStackTrace(); } if (doc.select(".detail_vorbest a").size() == 1) { result.setReservable(true); result.setReservation_info(doc.select(".detail_vorbest a").attr("href")); } return result; }
From source file:org.abondar.experimental.eventsearch.EventFinder.java
public void getEvent(String eventId, String evType) { try {// w w w. j av a 2 s .co m Document dc = Jsoup.connect("https://afisha.yandex.ru/msk/events/" + eventId + "/").get(); Event eb = new Event(); eb.setEventID(eventId); eb.setCategory(eventTypes.get(evType)); Elements elems = dc.select("meta"); for (Element e : elems) { if (e.attributes().get("property").contains("og:description")) { eb.setDescription(e.attributes().get("content")); } } elems = dc.select("title"); for (Element e : elems) { eb.setName(e.html().substring(0, e.html().indexOf(""))); } elems = dc.select("a[href]"); for (Element e : elems) { for (Attribute attr : e.attributes().asList()) { if (attr.getValue().contains("/msk/places/")) { eb.setPlace(getEventPlaces(attr.getValue())); } } } elems = dc.select("tr[id]"); for (Element e : elems) { for (Attribute attr : e.attributes().asList()) { if (attr.getValue().contains("f")) { eb.setDate(e.children().first().html()); try { Element e1 = e.child(1).children().first(); Element e2 = e1.children().first(); Element e3 = e2.children().first(); Element e4 = e3.children().first(); eb.setTime(e4.html()); } catch (NullPointerException ex) { Element e1 = e.child(2).children().first(); Element e2 = e1.children().first(); Element e3 = e2.children().first(); Element e4 = e3.children().first(); eb.setTime(e4.html()); } } } } geoCode(eb); formJson(eb); } catch (IOException ex) { Logger.getLogger(EventFinder.class.getName()).log(Level.SEVERE, null, ex); } }
From source file:me.vertretungsplan.parser.UntisCommonParser.java
/** * Parses an Untis substitution schedule table * * @param table the <code>table</code> Element from the HTML document * @param data {@link SubstitutionScheduleData#getData()} * @param day the {@link SubstitutionScheduleDay} where the substitutions will be stored * @param defaultClass the class that should be set if there is no class column in the table *//*ww w . j a v a 2 s . c o m*/ private void parseSubstitutionScheduleTable(Element table, JSONObject data, SubstitutionScheduleDay day, String defaultClass) throws JSONException, CredentialInvalidException { if (data.optBoolean(PARAM_CLASS_IN_EXTRA_LINE) || data.optBoolean("class_in_extra_line")) { // backwards compatibility for (Element element : table.select("td.inline_header")) { String className = getClassName(element.text(), data); if (isValidClass(className)) { Element zeile = null; try { zeile = element.parent().nextElementSibling(); if (zeile.select("td") == null) { zeile = zeile.nextElementSibling(); } int skipLines = 0; while (zeile != null && !zeile.select("td").attr("class").equals("list inline_header")) { if (skipLines > 0) { skipLines--; zeile = zeile.nextElementSibling(); continue; } Substitution v = new Substitution(); int i = 0; for (Element spalte : zeile.select("td")) { String text = spalte.text(); if (isEmpty(text)) { i++; continue; } int skipLinesForThisColumn = 0; Element nextLine = zeile.nextElementSibling(); boolean continueSkippingLines = true; while (continueSkippingLines) { if (nextLine != null && nextLine.children().size() == zeile.children().size()) { Element columnInNextLine = nextLine.child(spalte.elementSiblingIndex()); if (columnInNextLine.text().replaceAll("\u00A0", "").trim() .equals(nextLine.text().replaceAll("\u00A0", "").trim())) { // Continued in the next line text += " " + columnInNextLine.text(); skipLinesForThisColumn++; nextLine = nextLine.nextElementSibling(); } else { continueSkippingLines = false; } } else { continueSkippingLines = false; } } if (skipLinesForThisColumn > skipLines) skipLines = skipLinesForThisColumn; String type = data.getJSONArray(PARAM_COLUMNS).getString(i); switch (type) { case "lesson": v.setLesson(text); break; case "subject": handleSubject(v, spalte); break; case "previousSubject": v.setPreviousSubject(text); break; case "type": v.setType(text); v.setColor(colorProvider.getColor(text)); break; case "type-entfall": if (text.equals("x")) { v.setType("Entfall"); v.setColor(colorProvider.getColor("Entfall")); } else { v.setType("Vertretung"); v.setColor(colorProvider.getColor("Vertretung")); } break; case "room": handleRoom(v, spalte); break; case "teacher": handleTeacher(v, spalte, data); break; case "previousTeacher": v.setPreviousTeachers(splitTeachers(text, data)); break; case "desc": v.setDesc(text); break; case "desc-type": v.setDesc(text); String recognizedType = recognizeType(text); v.setType(recognizedType); v.setColor(colorProvider.getColor(recognizedType)); break; case "previousRoom": v.setPreviousRoom(text); break; case "substitutionFrom": v.setSubstitutionFrom(text); break; case "teacherTo": v.setTeacherTo(text); break; case "ignore": break; case "date": // used by UntisSubstitutionParser break; default: throw new IllegalArgumentException("Unknown column type: " + type); } i++; } autoDetectType(data, zeile, v); v.getClasses().add(className); if (v.getLesson() != null && !v.getLesson().equals("")) { day.addSubstitution(v); } zeile = zeile.nextElementSibling(); } } catch (Throwable e) { e.printStackTrace(); } } } } else { boolean hasType = false; for (int i = 0; i < data.getJSONArray(PARAM_COLUMNS).length(); i++) { if (data.getJSONArray(PARAM_COLUMNS).getString(i).equals("type")) { hasType = true; } } int skipLines = 0; for (Element zeile : table.select("tr.list.odd:not(:has(td.inline_header)), " + "tr.list.even:not(:has(td.inline_header)), " + "tr:has(td[align=center]):gt(0)")) { if (skipLines > 0) { skipLines--; continue; } Substitution v = new Substitution(); String klassen = defaultClass != null ? defaultClass : ""; int i = 0; for (Element spalte : zeile.select("td")) { String text = spalte.text(); String type = data.getJSONArray(PARAM_COLUMNS).getString(i); if (isEmpty(text) && !type.equals("type-entfall")) { i++; continue; } int skipLinesForThisColumn = 0; Element nextLine = zeile.nextElementSibling(); boolean continueSkippingLines = true; while (continueSkippingLines) { if (nextLine != null && nextLine.children().size() == zeile.children().size()) { Element columnInNextLine = nextLine.child(spalte.elementSiblingIndex()); if (columnInNextLine.text().replaceAll("\u00A0", "").trim() .equals(nextLine.text().replaceAll("\u00A0", "").trim())) { // Continued in the next line text += " " + columnInNextLine.text(); skipLinesForThisColumn++; nextLine = nextLine.nextElementSibling(); } else { continueSkippingLines = false; } } else { continueSkippingLines = false; } } if (skipLinesForThisColumn > skipLines) skipLines = skipLinesForThisColumn; switch (type) { case "lesson": v.setLesson(text); break; case "subject": handleSubject(v, spalte); break; case "previousSubject": v.setPreviousSubject(text); break; case "type": v.setType(text); v.setColor(colorProvider.getColor(text)); break; case "type-entfall": if (text.equals("x")) { v.setType("Entfall"); v.setColor(colorProvider.getColor("Entfall")); } else if (!hasType) { v.setType("Vertretung"); v.setColor(colorProvider.getColor("Vertretung")); } break; case "room": handleRoom(v, spalte); break; case "previousRoom": v.setPreviousRoom(text); break; case "desc": v.setDesc(text); break; case "desc-type": v.setDesc(text); String recognizedType = recognizeType(text); v.setType(recognizedType); v.setColor(colorProvider.getColor(recognizedType)); break; case "teacher": handleTeacher(v, spalte, data); break; case "previousTeacher": v.setPreviousTeachers(splitTeachers(text, data)); break; case "substitutionFrom": v.setSubstitutionFrom(text); break; case "teacherTo": v.setTeacherTo(text); break; case "class": klassen = getClassName(text, data); break; case "ignore": break; case "date": // used by UntisSubstitutionParser break; default: throw new IllegalArgumentException("Unknown column type: " + type); } i++; } if (v.getLesson() == null || v.getLesson().equals("")) { continue; } autoDetectType(data, zeile, v); List<String> affectedClasses; // Detect things like "7" Pattern singlePattern = Pattern.compile("(\\d+)"); Matcher singleMatcher = singlePattern.matcher(klassen); // Detect things like "5-12" Pattern rangePattern = Pattern.compile("(\\d+) ?- ?(\\d+)"); Matcher rangeMatcher = rangePattern.matcher(klassen); Pattern pattern2 = Pattern.compile("^(\\d+).*"); if (rangeMatcher.matches()) { affectedClasses = new ArrayList<>(); int min = Integer.parseInt(rangeMatcher.group(1)); int max = Integer.parseInt(rangeMatcher.group(2)); try { for (String klasse : getAllClasses()) { Matcher matcher2 = pattern2.matcher(klasse); if (matcher2.matches()) { int num = Integer.parseInt(matcher2.group(1)); if (min <= num && num <= max) affectedClasses.add(klasse); } } } catch (IOException e) { e.printStackTrace(); } } else if (singleMatcher.matches()) { affectedClasses = new ArrayList<>(); int grade = Integer.parseInt(singleMatcher.group(1)); try { for (String klasse : getAllClasses()) { Matcher matcher2 = pattern2.matcher(klasse); if (matcher2.matches() && grade == Integer.parseInt(matcher2.group(1))) { affectedClasses.add(klasse); } } } catch (IOException e) { e.printStackTrace(); } } else { if (data.optBoolean(PARAM_CLASSES_SEPARATED, true) && data.optBoolean("classes_separated", true)) { // backwards compatibility affectedClasses = Arrays.asList(klassen.split(", ")); } else { affectedClasses = new ArrayList<>(); try { for (String klasse : getAllClasses()) { // TODO: is there a better way? StringBuilder regex = new StringBuilder(); for (char character : klasse.toCharArray()) { if (character == '?') { regex.append("\\?"); } else { regex.append(character); } regex.append(".*"); } if (klassen.matches(regex.toString())) { affectedClasses.add(klasse); } } } catch (IOException e) { e.printStackTrace(); } } } for (String klasse : affectedClasses) { if (isValidClass(klasse)) { v.getClasses().add(klasse); } } if (data.optBoolean(PARAM_MERGE_WITH_DIFFERENT_TYPE, false)) { boolean found = false; for (Substitution subst : day.getSubstitutions()) { if (subst.equalsExcludingType(v)) { found = true; if (v.getType().equals("Vertretung")) { subst.setType("Vertretung"); subst.setColor(colorProvider.getColor("Vertretung")); } break; } } if (!found) { day.addSubstitution(v); } } else { day.addSubstitution(v); } } } }
From source file:com.astamuse.asta4d.render.RenderUtil.java
private final static void apply(Element target, List<Renderer> rendererList, RenderAction renderAction, int startIndex, int count) { // The renderer list have to be applied recursively because the // transformer will always return a new Element clone. if (startIndex >= count) { return;/*from ww w . j av a 2 s .c om*/ } final Renderer currentRenderer = rendererList.get(startIndex); RendererType rendererType = currentRenderer.getRendererType(); switch (rendererType) { case GO_THROUGH: apply(target, rendererList, renderAction, startIndex + 1, count); return; /* case DEBUG: currentRenderer.getTransformerList().get(0).invoke(target); apply(target, rendererList, renderAction, startIndex + 1, count); return; */ case RENDER_ACTION: ((RenderActionRenderer) currentRenderer).getStyle().apply(renderAction); apply(target, rendererList, renderAction, startIndex + 1, count); return; default: // do nothing break; } String selector = currentRenderer.getSelector(); List<Transformer<?>> transformerList = currentRenderer.getTransformerList(); List<Element> elemList; if (PSEUDO_ROOT_SELECTOR.equals(selector)) { elemList = new LinkedList<Element>(); elemList.add(target); } else { elemList = new ArrayList<>(target.select(selector)); } if (elemList.isEmpty()) { if (rendererType == RendererType.ELEMENT_NOT_FOUND_HANDLER) { elemList.add(target); transformerList.clear(); transformerList.add( new RendererTransformer(((ElementNotFoundHandler) currentRenderer).alternativeRenderer())); } else if (renderAction.isOutputMissingSelectorWarning()) { String creationInfo = currentRenderer.getCreationSiteInfo(); if (creationInfo == null) { creationInfo = ""; } else { creationInfo = " at [ " + creationInfo + " ]"; } logger.warn( "There is no element found for selector [{}]{}, if it is deserved, try Renderer#disableMissingSelectorWarning() " + "to disable this message and Renderer#enableMissingSelectorWarning could enable this warning again in " + "your renderer chain", selector, creationInfo); apply(target, rendererList, renderAction, startIndex + 1, count); return; } } else { if (rendererType == RendererType.ELEMENT_NOT_FOUND_HANDLER) { apply(target, rendererList, renderAction, startIndex + 1, count); return; } } Element delayedElement = null; Element resultNode; // TODO we suppose that the element is listed as the order from parent // to children, so we reverse it. Perhaps we need a real order process // to ensure the wanted order. Collections.reverse(elemList); boolean renderForRoot; for (Element elem : elemList) { renderForRoot = PSEUDO_ROOT_SELECTOR.equals(selector) || rendererType == RendererType.ELEMENT_NOT_FOUND_HANDLER; if (!renderForRoot) { // faked group node will be not applied by renderers(only when the current selector is not the pseudo :root) if (elem.tagName().equals(ExtNodeConstants.GROUP_NODE_TAG) && ExtNodeConstants.GROUP_NODE_ATTR_TYPE_FAKE .equals(elem.attr(ExtNodeConstants.GROUP_NODE_ATTR_TYPE))) { continue; } } if (elem == target) { delayedElement = elem; continue; } for (Transformer<?> transformer : transformerList) { resultNode = transformer.invoke(elem); elem.before(resultNode); } // for transformer elem.remove(); } // for element // if the root element is one of the process targets, we can not apply // the left renderers to original element because it will be replaced by // a new element even it is not necessary (that is how Transformer // works). if (delayedElement == null) { apply(target, rendererList, renderAction, startIndex + 1, count); } else { if (rendererType == RendererType.ELEMENT_NOT_FOUND_HANDLER && delayedElement instanceof Document) { delayedElement = delayedElement.child(0); } for (Transformer<?> transformer : transformerList) { resultNode = transformer.invoke(delayedElement); delayedElement.before(resultNode); apply(resultNode, rendererList, renderAction, startIndex + 1, count); } // for transformer delayedElement.remove(); } }
From source file:com.jimplush.goose.ContentExtractor.java
/** * adds any siblings that may have a decent score to this node * * @param node/*from w ww.java2 s . c o m*/ * @return */ private Element addSiblings(Element node) { if (logger.isDebugEnabled()) { logger.debug("Starting to add siblings"); } int baselineScoreForSiblingParagraphs = getBaselineScoreForSiblings(node); Element currentSibling = node.previousElementSibling(); while (currentSibling != null) { if (logger.isDebugEnabled()) { logger.debug("SIBLINGCHECK: " + debugNode(currentSibling)); } if (currentSibling.tagName().equals("p")) { node.child(0).before(currentSibling.outerHtml()); currentSibling = currentSibling.previousElementSibling(); continue; } // check for a paraph embedded in a containing element int insertedSiblings = 0; Elements potentialParagraphs = currentSibling.getElementsByTag("p"); if (potentialParagraphs.first() == null) { currentSibling = currentSibling.previousElementSibling(); continue; } for (Element firstParagraph : potentialParagraphs) { WordStats wordStats = StopWords.getStopWordCount(firstParagraph.text()); int paragraphScore = wordStats.getStopWordCount(); if ((float) (baselineScoreForSiblingParagraphs * .30) < paragraphScore) { if (logger.isDebugEnabled()) { logger.debug("This node looks like a good sibling, adding it"); } node.child(insertedSiblings).before("<p>" + firstParagraph.text() + "<p>"); insertedSiblings++; } } currentSibling = currentSibling.previousElementSibling(); } return node; }
From source file:com.liato.bankdroid.banking.banks.AbsIkanoPartner.java
@Override public void update() throws BankException, LoginException, BankChoiceException { super.update(); if (username == null || password == null || username.length() == 0 || password.length() == 0) { throw new LoginException(res.getText(R.string.invalid_username_password).toString()); }//from w ww . j a v a 2s.c om urlopen = login(); Document d = Jsoup.parse(response); Element element = d.select("#primary-nav > li:eq(1) > a").first(); if (element != null && element.attr("href") != null) { String myAccountUrl = element.attr("href"); try { response = urlopen.open("https://partner.ikanobank.se/" + myAccountUrl); d = Jsoup.parse(response); Elements es = d.select("#CustomerAccountInformationSpan > span > span"); int accId = 0; for (Element el : es) { Element name = el.select("> span > span:eq(0)").first(); Element balance = el.select("> span:eq(1)").first(); Element currency = el.select("> span:eq(2)").first(); if (name != null && balance != null && currency != null) { Account account = new Account(name.text().trim(), Helpers.parseBalance(balance.text()), Integer.toString(accId)); account.setCurrency(Helpers.parseCurrency(currency.text(), "SEK")); if (accId > 0) { account.setAliasfor("0"); } accounts.add(account); accId++; } } if (accounts.isEmpty()) { throw new BankException(res.getText(R.string.no_accounts_found).toString()); } // Use the amount from "Kvar att handla fr" which should be the // last account in the list. this.balance = accounts.get(accounts.size() - 1).getBalance(); ArrayList<Transaction> transactions = new ArrayList<Transaction>(); es = d.select("#ShowCustomerTransactionPurchasesInformationDiv table tr:has(td)"); for (Element el : es) { if (el.childNodeSize() == 6) { Transaction transaction = new Transaction(el.child(0).text().trim(), el.child(1).text().trim(), Helpers.parseBalance(el.child(2).text())); transaction.setCurrency(Helpers.parseCurrency(el.child(3).text().trim(), "SEK")); transactions.add(transaction); } } accounts.get(0).setTransactions(transactions); } catch (ClientProtocolException e) { throw new BankException(e.getMessage()); } catch (IOException e) { throw new BankException(e.getMessage()); } } if (accounts.isEmpty()) { throw new BankException(res.getText(R.string.no_accounts_found).toString()); } super.updateComplete(); }
From source file:de.geeksfactory.opacclient.apis.SISIS.java
protected void parse_medialist(List<LentItem> media, Document doc, int offset) { Elements copytrs = doc.select(".data tr"); doc.setBaseUri(opac_url);/*w ww. j a v a 2 s . c o m*/ DateTimeFormatter fmt = DateTimeFormat.forPattern("dd.MM.yyyy").withLocale(Locale.GERMAN); int trs = copytrs.size(); if (trs == 1) { return; } assert (trs > 0); for (int i = 1; i < trs; i++) { Element tr = copytrs.get(i); LentItem item = new LentItem(); if (tr.text().contains("keine Daten")) { return; } item.setTitle(tr.child(1).select("strong").text().trim()); try { item.setAuthor(tr.child(1).html().split("<br[ /]*>")[1].trim()); String[] col2split = tr.child(2).html().split("<br[ /]*>"); String deadline = col2split[0].trim(); if (deadline.contains("-")) { deadline = deadline.split("-")[1].trim(); } try { item.setDeadline(fmt.parseLocalDate(deadline).toString()); } catch (IllegalArgumentException e1) { e1.printStackTrace(); } if (col2split.length > 1) { item.setHomeBranch(col2split[1].trim()); } if (tr.select("a").size() > 0) { for (Element link : tr.select("a")) { String href = link.attr("abs:href"); Map<String, String> hrefq = getQueryParamsFirst(href); if (hrefq.get("methodToCall").equals("renewalPossible")) { item.setProlongData(offset + "$" + href.split("\\?")[1]); item.setRenewable(true); break; } } } else if (tr.select(".textrot, .textgruen, .textdunkelblau").size() > 0) { item.setProlongData("" + tr.select(".textrot, .textgruen, .textdunkelblau").text()); item.setRenewable(false); } } catch (Exception ex) { ex.printStackTrace(); } media.add(item); } assert (media.size() == trs - 1); }
From source file:de.geeksfactory.opacclient.apis.SISIS.java
protected void parse_reslist(String type, List<ReservedItem> reservations, Document doc, int offset) { Elements copytrs = doc.select(".data tr"); doc.setBaseUri(opac_url);//w w w . j a v a 2 s. c o m int trs = copytrs.size(); if (trs == 1) { return; } assert (trs > 0); for (int i = 1; i < trs; i++) { Element tr = copytrs.get(i); ReservedItem item = new ReservedItem(); if (tr.text().contains("keine Daten") || tr.children().size() == 1) { return; } item.setTitle(tr.child(1).select("strong").text().trim()); try { String[] rowsplit1 = tr.child(1).html().split("<br[ /]*>"); String[] rowsplit2 = tr.child(2).html().split("<br[ /]*>"); if (rowsplit1.length > 1) item.setAuthor(rowsplit1[1].trim()); if (rowsplit2.length > 2) item.setBranch(rowsplit2[2].trim()); if (rowsplit2.length > 2) item.setStatus(rowsplit2[0].trim()); if (tr.select("a").size() == 1) { item.setCancelData(type + "$" + offset + "$" + tr.select("a").attr("abs:href").split("\\?")[1]); } } catch (Exception e) { e.printStackTrace(); } reservations.add(item); } assert (reservations.size() == trs - 1); }
From source file:de.geeksfactory.opacclient.apis.SISIS.java
protected DetailledItem parse_result(String html) throws IOException { Document doc = Jsoup.parse(html); doc.setBaseUri(opac_url);/* w ww . j a va 2 s . c o m*/ String html2 = httpGet(opac_url + "/singleHit.do?methodToCall=activateTab&tab=showTitleActive", ENCODING); Document doc2 = Jsoup.parse(html2); doc2.setBaseUri(opac_url); String html3 = httpGet(opac_url + "/singleHit.do?methodToCall=activateTab&tab=showAvailabilityActive", ENCODING); Document doc3 = Jsoup.parse(html3); doc3.setBaseUri(opac_url); DetailledItem result = new DetailledItem(); try { result.setId(doc.select("#bibtip_id").text().trim()); } catch (Exception ex) { ex.printStackTrace(); } List<String> reservationlinks = new ArrayList<>(); for (Element link : doc3.select("#vormerkung a, #tab-content a")) { String href = link.absUrl("href"); Map<String, String> hrefq = getQueryParamsFirst(href); if (result.getId() == null) { // ID retrieval String key = hrefq.get("katkey"); if (key != null) { result.setId(key); break; } } // Vormerken if (hrefq.get("methodToCall") != null) { if (hrefq.get("methodToCall").equals("doVormerkung") || hrefq.get("methodToCall").equals("doBestellung")) { reservationlinks.add(href.split("\\?")[1]); } } } if (reservationlinks.size() == 1) { result.setReservable(true); result.setReservation_info(reservationlinks.get(0)); } else if (reservationlinks.size() == 0) { result.setReservable(false); } else { // TODO: Multiple options - handle this case! } if (doc.select(".data td img").size() == 1) { result.setCover(doc.select(".data td img").first().attr("abs:src")); try { downloadCover(result); } catch (Exception e) { } } if (doc.select(".aw_teaser_title").size() == 1) { result.setTitle(doc.select(".aw_teaser_title").first().text().trim()); } else if (doc.select(".data td strong").size() > 0) { result.setTitle(doc.select(".data td strong").first().text().trim()); } else { result.setTitle(""); } if (doc.select(".aw_teaser_title_zusatz").size() > 0) { result.addDetail(new Detail("Titelzusatz", doc.select(".aw_teaser_title_zusatz").text().trim())); } String title = ""; String text = ""; boolean takeover = false; Element detailtrs = doc2.select(".box-container .data td").first(); for (Node node : detailtrs.childNodes()) { if (node instanceof Element) { if (((Element) node).tagName().equals("strong")) { title = ((Element) node).text().trim(); text = ""; } else { if (((Element) node).tagName().equals("a") && (((Element) node).text().trim().contains("hier klicken") || title.equals("Link:"))) { text = text + node.attr("href"); takeover = true; break; } } } else if (node instanceof TextNode) { text = text + ((TextNode) node).text(); } } if (!takeover) { text = ""; title = ""; } detailtrs = doc2.select("#tab-content .data td").first(); if (detailtrs != null) { for (Node node : detailtrs.childNodes()) { if (node instanceof Element) { if (((Element) node).tagName().equals("strong")) { if (!text.equals("") && !title.equals("")) { result.addDetail(new Detail(title.trim(), text.trim())); if (title.equals("Titel:")) { result.setTitle(text.trim()); } text = ""; } title = ((Element) node).text().trim(); } else { if (((Element) node).tagName().equals("a") && (((Element) node).text().trim().contains("hier klicken") || title.equals("Link:"))) { text = text + node.attr("href"); } else { text = text + ((Element) node).text(); } } } else if (node instanceof TextNode) { text = text + ((TextNode) node).text(); } } } else { if (doc2.select("#tab-content .fulltitle tr").size() > 0) { Elements rows = doc2.select("#tab-content .fulltitle tr"); for (Element tr : rows) { if (tr.children().size() == 2) { Element valcell = tr.child(1); String value = valcell.text().trim(); if (valcell.select("a").size() == 1) { value = valcell.select("a").first().absUrl("href"); } result.addDetail(new Detail(tr.child(0).text().trim(), value)); } } } else { result.addDetail(new Detail(stringProvider.getString(StringProvider.ERROR), stringProvider.getString(StringProvider.COULD_NOT_LOAD_DETAIL))); } } if (!text.equals("") && !title.equals("")) { result.addDetail(new Detail(title.trim(), text.trim())); if (title.equals("Titel:")) { result.setTitle(text.trim()); } } for (Element link : doc3.select("#tab-content a")) { Map<String, String> hrefq = getQueryParamsFirst(link.absUrl("href")); if (result.getId() == null) { // ID retrieval String key = hrefq.get("katkey"); if (key != null) { result.setId(key); break; } } } for (Element link : doc3.select(".box-container a")) { if (link.text().trim().equals("Download")) { result.addDetail( new Detail(stringProvider.getString(StringProvider.DOWNLOAD), link.absUrl("href"))); } } Map<String, Integer> copy_columnmap = new HashMap<>(); // Default values copy_columnmap.put("barcode", 1); copy_columnmap.put("branch", 3); copy_columnmap.put("status", 4); Elements copy_columns = doc.select("#tab-content .data tr#bg2 th"); for (int i = 0; i < copy_columns.size(); i++) { Element th = copy_columns.get(i); String head = th.text().trim(); if (head.contains("Status")) { copy_columnmap.put("status", i); } if (head.contains("Zweigstelle")) { copy_columnmap.put("branch", i); } if (head.contains("Mediennummer")) { copy_columnmap.put("barcode", i); } if (head.contains("Standort")) { copy_columnmap.put("location", i); } if (head.contains("Signatur")) { copy_columnmap.put("signature", i); } } Pattern status_lent = Pattern.compile( "^(entliehen) bis ([0-9]{1,2}.[0-9]{1,2}.[0-9]{2," + "4}) \\(gesamte Vormerkungen: ([0-9]+)\\)$"); Pattern status_and_barcode = Pattern.compile("^(.*) ([0-9A-Za-z]+)$"); Elements exemplartrs = doc.select("#tab-content .data tr").not("#bg2"); DateTimeFormatter fmt = DateTimeFormat.forPattern("dd.MM.yyyy").withLocale(Locale.GERMAN); for (Element tr : exemplartrs) { try { Copy copy = new Copy(); Element status = tr.child(copy_columnmap.get("status")); Element barcode = tr.child(copy_columnmap.get("barcode")); String barcodetext = barcode.text().trim().replace(" Wegweiser", ""); // STATUS String statustext; if (status.getElementsByTag("b").size() > 0) { statustext = status.getElementsByTag("b").text().trim(); } else { statustext = status.text().trim(); } if (copy_columnmap.get("status").equals(copy_columnmap.get("barcode"))) { Matcher matcher1 = status_and_barcode.matcher(statustext); if (matcher1.matches()) { statustext = matcher1.group(1); barcodetext = matcher1.group(2); } } Matcher matcher = status_lent.matcher(statustext); if (matcher.matches()) { copy.setStatus(matcher.group(1)); copy.setReservations(matcher.group(3)); copy.setReturnDate(fmt.parseLocalDate(matcher.group(2))); } else { copy.setStatus(statustext); } copy.setBarcode(barcodetext); if (status.select("a[href*=doVormerkung]").size() == 1) { copy.setResInfo(status.select("a[href*=doVormerkung]").attr("href").split("\\?")[1]); } String branchtext = tr.child(copy_columnmap.get("branch")).text().trim().replace(" Wegweiser", ""); copy.setBranch(branchtext); if (copy_columnmap.containsKey("location")) { copy.setLocation( tr.child(copy_columnmap.get("location")).text().trim().replace(" Wegweiser", "")); } if (copy_columnmap.containsKey("signature")) { copy.setShelfmark( tr.child(copy_columnmap.get("signature")).text().trim().replace(" Wegweiser", "")); } result.addCopy(copy); } catch (Exception ex) { ex.printStackTrace(); } } try { Element isvolume = null; Map<String, String> volume = new HashMap<>(); Elements links = doc.select(".data td a"); int elcount = links.size(); for (int eli = 0; eli < elcount; eli++) { List<NameValuePair> anyurl = URLEncodedUtils.parse(new URI(links.get(eli).attr("href")), "UTF-8"); for (NameValuePair nv : anyurl) { if (nv.getName().equals("methodToCall") && nv.getValue().equals("volumeSearch")) { isvolume = links.get(eli); } else if (nv.getName().equals("catKey")) { volume.put("catKey", nv.getValue()); } else if (nv.getName().equals("dbIdentifier")) { volume.put("dbIdentifier", nv.getValue()); } } if (isvolume != null) { volume.put("volume", "true"); result.setVolumesearch(volume); break; } } } catch (Exception e) { e.printStackTrace(); } return result; }