Examples with TextNode org.jsoup.nodes.TextNode used on opensource projects

Example 16 with TextNode

use of org.jsoup.nodes.TextNode in project opacclient by opacapp.

the class Heidi method parse_reservations.

protected List<ReservedItem> parse_reservations(String html) {
    Document doc = Jsoup.parse(html);
    List<ReservedItem> reservations = new ArrayList<>();
    DateTimeFormatter fmt = DateTimeFormat.forPattern("dd.MM.yyyy").withLocale(Locale.GERMAN);
    for (Element tr : doc.select("table.kontopos tr")) {
        ReservedItem item = new ReservedItem();
        Element desc = tr.child(1).select("label").first();
        Element pos = tr.child(3);
        if (tr.child(1).select("a").size() > 0) {
            String kk = getQueryParamsFirst(tr.child(1).select("a").first().absUrl("href")).get("katkey");
            item.setId(kk);
        }
        if (tr.child(0).select("input").size() > 0) {
            item.setCancelData(tr.child(0).select("input").first().val());
        }
        int i = 0;
        for (Node node : desc.childNodes()) {
            if (node instanceof TextNode) {
                String text = ((TextNode) node).text().trim();
                if (i == 0) {
                    item.setAuthor(text);
                } else if (i == 1) {
                    item.setTitle(text);
                }
                i++;
            }
        }
        i = 0;
        for (Node node : pos.childNodes()) {
            if (node instanceof TextNode) {
                String text = ((TextNode) node).text().trim();
                if (i == 0 && text.contains("")) {
                    try {
                        item.setReadyDate(fmt.parseLocalDate(text));
                    } catch (IllegalArgumentException e) {
                        item.setStatus(text);
                    }
                } else if (i == 1) {
                    item.setBranch(text);
                }
                i++;
            }
        }
        reservations.add(item);
    }
    return reservations;
}

Example 17 with TextNode

use of org.jsoup.nodes.TextNode in project opacclient by opacapp.

the class Heidi method account.

@Override
public AccountData account(Account account) throws IOException, JSONException, OpacErrorException {
    login(account);
    String html;
    Document doc;
    AccountData adata = new AccountData(account.getId());
    DateTimeFormatter fmt = DateTimeFormat.forPattern("dd.MM.yyyy").withLocale(Locale.GERMAN);
    html = httpGet(opac_url + "/konto.cgi?sess=" + sessid, getDefaultEncoding());
    doc = Jsoup.parse(html);
    doc.setBaseUri(opac_url + "/");
    for (Element td : doc.select("table.konto td")) {
        if (td.text().contains("Offene")) {
            String text = td.text().trim().replaceAll("Offene[^0-9]+Geb.+hren:[^0-9]+([0-9.," + "]+)[^0-9€A-Z]*(€|EUR|CHF|Fr.)", "$1 $2");
            adata.setPendingFees(text);
        }
    }
    List<LentItem> lent = new ArrayList<>();
    for (Element tr : doc.select("table.kontopos tr")) {
        LentItem item = new LentItem();
        Element desc = tr.child(1).select("label").first();
        String dates = tr.child(2).text().trim();
        if (tr.child(1).select("a").size() > 0) {
            String kk = getQueryParamsFirst(tr.child(1).select("a").first().absUrl("href")).get("katkey");
            item.setId(kk);
        }
        int i = 0;
        for (Node node : desc.childNodes()) {
            if (node instanceof TextNode) {
                String text = ((TextNode) node).text().trim();
                if (i == 0) {
                    item.setAuthor(text);
                } else if (i == 1) {
                    item.setTitle(text);
                } else if (text.contains("Mediennummer")) {
                    item.setBarcode(text.replace("Mediennummer: ", ""));
                }
                i++;
            }
        }
        if (tr.child(0).select("input").size() == 1) {
            item.setProlongData(tr.child(0).select("input").first().val());
            item.setRenewable(true);
        } else {
            item.setProlongData("§" + tr.child(0).select("span").first().attr("class"));
            item.setRenewable(false);
        }
        String todate = dates;
        if (todate.contains("-")) {
            String[] datesplit = todate.split("-");
            todate = datesplit[1].trim();
        }
        try {
            item.setDeadline(fmt.parseLocalDate(todate.substring(0, 10)));
        } catch (IllegalArgumentException e) {
            e.printStackTrace();
        }
        lent.add(item);
    }
    adata.setLent(lent);
    List<ReservedItem> reservations = new ArrayList<>();
    html = httpGet(opac_url + "/konto.cgi?konto=v&sess=" + sessid, getDefaultEncoding());
    reservations.addAll(parse_reservations(html));
    html = httpGet(opac_url + "/konto.cgi?konto=b&sess=" + sessid, getDefaultEncoding());
    reservations.addAll(parse_reservations(html));
    adata.setReservations(reservations);
    return adata;
}

Example 18 with TextNode

use of org.jsoup.nodes.TextNode in project opacclient by opacapp.

the class VuFind method parseCopies.

static void parseCopies(DetailedItem res, Document doc, JSONObject data) throws JSONException {
    if ("doublestacked".equals(data.optString("copystyle"))) {
        // e.g. http://vopac.nlg.gr/Record/393668/Holdings#tabnav
        // for Athens_GreekNationalLibrary
        Element container = doc.select(".tab-container").first();
        String branch = "";
        for (Element child : container.children()) {
            if (child.tagName().equals("h5")) {
                branch = child.text();
            } else if (child.tagName().equals("table")) {
                int i = 0;
                String callNumber = "";
                for (Element row : child.select("tr")) {
                    if (i == 0) {
                        callNumber = row.child(1).text();
                    } else {
                        Copy copy = new Copy();
                        copy.setBranch(branch);
                        copy.setShelfmark(callNumber);
                        copy.setBarcode(row.child(0).text());
                        copy.setStatus(row.child(1).text());
                        res.addCopy(copy);
                    }
                    i++;
                }
            }
        }
    } else if ("stackedtable".equals(data.optString("copystyle"))) {
        // e.g. http://search.lib.auth.gr/Record/376356
        // or https://katalog.ub.uni-leipzig.de/Record/0000196115
        // or https://www.stadt-muenster.de/opac2/Record/0367968
        Element container = doc.select(".recordsubcontent, .tab-container").first();
        // .tab-container is used in Muenster.
        String branch = "";
        JSONObject copytable = data.getJSONObject("copytable");
        for (Element child : container.children()) {
            if (child.tagName().equals("div")) {
                child = child.child(0);
            }
            if (child.tagName().equals("h3")) {
                branch = child.text();
            } else if (child.tagName().equals("table")) {
                if (child.select("caption").size() > 0) {
                    // Leipzig_Uni
                    branch = child.select("caption").first().ownText();
                }
                int i = 0;
                String callNumber = null;
                if ("headrow".equals(copytable.optString("signature"))) {
                    callNumber = child.select("tr").get(0).child(1).text();
                }
                for (Element row : child.select("tr")) {
                    if (i < copytable.optInt("_offset", 0)) {
                        i++;
                        continue;
                    }
                    Copy copy = new Copy();
                    if (callNumber != null) {
                        copy.setShelfmark(callNumber);
                    }
                    copy.setBranch(branch);
                    Iterator<?> keys = copytable.keys();
                    while (keys.hasNext()) {
                        String key = (String) keys.next();
                        if (key.startsWith("_"))
                            continue;
                        if (copytable.optString(key, "").contains("/")) {
                            // Leipzig_Uni
                            String[] splitted = copytable.getString(key).split("/");
                            int col = Integer.parseInt(splitted[0]);
                            int line = Integer.parseInt(splitted[1]);
                            int j = 0;
                            for (Node node : row.child(col).childNodes()) {
                                if (node instanceof Element) {
                                    if (((Element) node).tagName().equals("br")) {
                                        j++;
                                    } else if (j == line) {
                                        copy.set(key, ((Element) node).text());
                                    }
                                } else if (node instanceof TextNode && j == line && !((TextNode) node).text().trim().equals("")) {
                                    copy.set(key, ((TextNode) node).text());
                                }
                            }
                        } else {
                            // Thessaloniki_University
                            if (copytable.optInt(key, -1) == -1)
                                continue;
                            String value = row.child(copytable.getInt(key)).text();
                            copy.set(key, value);
                        }
                    }
                    res.addCopy(copy);
                    i++;
                }
            }
        }
    }
}

Example 19 with TextNode

use of org.jsoup.nodes.TextNode in project opacclient by opacapp.

the class PicaOld method parseMediaList.

static void parseMediaList(List<LentItem> media, Document doc, StringProvider stringProvider, List<String> renewalCounts) throws OpacErrorException {
    Elements copytrs = doc.select("table[summary^=list] > tbody > tr[valign=top]");
    DateTimeFormatter fmt = DateTimeFormat.forPattern("dd-MM-yyyy").withLocale(Locale.GERMAN);
    int trs = copytrs.size();
    if (trs < 1) {
        throw new OpacErrorException(stringProvider.getString(StringProvider.COULD_NOT_LOAD_ACCOUNT));
    }
    assert (trs > 0);
    for (int i = 0; i < trs; i++) {
        Element tr = copytrs.get(i);
        if (tr.select("table[summary=title data]").size() > 0) {
            // According to HTML code from Bug reports (Server TU Darmstadt,
            // Berlin Ibero-Amerikanisches Institut)
            LentItem item = new LentItem();
            // Check if there is a checkbox to prolong this item
            if (tr.select("input").size() > 0) {
                item.setProlongData(tr.select("input").attr("value"));
            } else {
                item.setRenewable(false);
            }
            Elements datatrs = tr.select("table[summary=title data] tr");
            item.setTitle(datatrs.get(0).text());
            String reservations = null;
            for (Element td : datatrs.get(1).select("td")) {
                List<TextNode> textNodes = td.textNodes();
                Elements titles = td.select("span.label-small");
                List<String> values = new ArrayList<>();
                if (td.select("span[name=xxxxx]").size() > 0) {
                    for (Element span : td.select("span[name=xxxxx]")) {
                        values.add(span.text());
                    }
                } else {
                    for (TextNode node : textNodes) {
                        if (!node.text().equals(" ")) {
                            values.add(node.text());
                        }
                    }
                }
                assert (values.size() == titles.size());
                for (int j = 0; j < values.size(); j++) {
                    String title = titles.get(j).text();
                    String value = values.get(j).trim().replace(";", "");
                    // noinspection StatementWithEmptyBody
                    if (title.contains("Signatur") || title.contains("shelf mark") || title.contains("signatuur")) {
                    // not supported
                    } else if (title.contains("Status") || title.contains("status") || title.contains("statut")) {
                        item.setStatus(value);
                    } else if (title.contains("Leihfristende") || title.contains("expiry date") || title.contains("vervaldatum") || title.contains("date d'expiration")) {
                        try {
                            item.setDeadline(fmt.parseLocalDate(value));
                        } catch (IllegalArgumentException e1) {
                            e1.printStackTrace();
                        }
                    } else // noinspection StatementWithEmptyBody
                    if (title.contains("Vormerkungen") || title.contains("reservations") || title.contains("reserveringen") || title.contains("réservations")) {
                        reservations = value;
                    }
                }
            }
            media.add(item);
        } else {
            // like in Kiel
            String prolongCount = "";
            if (renewalCounts.size() == trs && renewalCounts.get(i) != null) {
                prolongCount = renewalCounts.get(i);
            }
            String reminderCount = tr.child(13).text().trim();
            if (reminderCount.contains(" Mahn") && reminderCount.contains("(") && reminderCount.indexOf("(") < reminderCount.indexOf(" Mahn")) {
                reminderCount = reminderCount.substring(reminderCount.indexOf("(") + 1, reminderCount.indexOf(" Mahn"));
            } else {
                reminderCount = "";
            }
            LentItem item = new LentItem();
            if (tr.child(4).text().trim().length() < 5 && tr.child(5).text().trim().length() > 4) {
                item.setTitle(tr.child(5).text().trim());
            } else {
                item.setTitle(tr.child(4).text().trim());
            }
            String status = tr.child(13).text().trim();
            if (!reminderCount.equals("0") && !reminderCount.equals("")) {
                if (!status.equals(""))
                    status += ", ";
                status += reminderCount + " " + stringProvider.getString(StringProvider.REMINDERS) + ", ";
            }
            if (!"".equals(prolongCount)) {
                if (!status.equals(""))
                    status += ", ";
                status += prolongCount + "x " + stringProvider.getString(StringProvider.PROLONGED_ABBR);
            }
            if (tr.children().size() >= 26 && !"".equals(tr.child(25).text().trim())) {
                if (!status.equals(""))
                    status += ", ";
                try {
                    status += stringProvider.getQuantityString(StringProvider.RESERVATIONS_NUMBER, Integer.parseInt(tr.child(25).text().trim()), Integer.parseInt(tr.child(25).text().trim()));
                } catch (NumberFormatException e) {
                // pass
                }
            }
            // + tr.child(25).text().trim() + " Vormerkungen");
            item.setStatus(status);
            try {
                item.setDeadline(fmt.parseLocalDate(tr.child(21).text().trim()));
            } catch (IllegalArgumentException e) {
                e.printStackTrace();
            }
            if (tr.child(1).select("input").size() > 0) {
                // If there is no checkbox, the medium is not renewable
                item.setProlongData(tr.child(1).select("input").attr("value"));
            }
            media.add(item);
        }
    }
    assert (media.size() == trs);
}

Example 20 with TextNode

use of org.jsoup.nodes.TextNode in project opacclient by opacapp.

the class SISIS method parseDetail.

static DetailedItem parseDetail(String html, String html2, String html3, String coverJs, JSONObject data, StringProvider stringProvider) throws IOException {
    Document doc = Jsoup.parse(html);
    String opac_url = data.optString("baseurl", "");
    doc.setBaseUri(opac_url);
    Document doc2 = Jsoup.parse(html2);
    doc2.setBaseUri(opac_url);
    Document doc3 = Jsoup.parse(html3);
    doc3.setBaseUri(opac_url);
    DetailedItem result = new DetailedItem();
    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 (result.getId() == null && doc.select("#permalink_link").size() > 0) {
        result.setId(doc.select("#permalink_link").text());
    }
    if (coverJs != null) {
        Pattern srcPattern = Pattern.compile("<img .* src=\"([^\"]+)\">");
        Matcher matcher = srcPattern.matcher(coverJs);
        if (matcher.find()) {
            result.setCover(matcher.group(1));
        }
    } else if (doc.select(".data td img").size() == 1) {
        result.setCover(doc.select(".data td img").first().attr("abs:src"));
    }
    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) {
            Element element = (Element) node;
            if (element.tagName().equals("strong")) {
                if (element.hasClass("c2")) {
                    if (!title.equals("")) {
                        result.addDetail(new Detail(title, text.trim()));
                    }
                    title = element.text().trim();
                    text = "";
                } else {
                    text = text + element.text();
                }
            } else {
                if (element.tagName().equals("a")) {
                    if (element.text().trim().contains("hier klicken") || title.contains("Link")) {
                        text = text + node.attr("href");
                        takeover = true;
                        break;
                    } else {
                        text = text + element.text();
                    }
                }
            }
        } 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")));
        }
    }
    if (doc3.select("#tab-content .textrot").size() > 0) {
        result.addDetail(new Detail(stringProvider.getString(StringProvider.STATUS), doc3.select("#tab-content .textrot").text()));
    }
    Map<String, Integer> copy_columnmap = new HashMap<>();
    // Default values
    copy_columnmap.put("barcode", 1);
    copy_columnmap.put("branch", 3);
    copy_columnmap.put("status", 4);
    Element table = doc.select("#tab-content .data").first();
    Elements copy_columns = table != null ? table.select("tr#bg2 th") : new Elements();
    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 = table != null ? table.select("tr").not("#bg2") : new Elements();
    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.trim().replace(" Wegweiser", ""));
            }
            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;
}