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

Example 21 with Node

use of org.jsoup.nodes.Node in project structr by structr.

the class MicroformatParser method extractChildContent.

private Object extractChildContent(final Element element) {
    final List<String> parts = new LinkedList<>();
    element.traverse(new NodeVisitor() {

        @Override
        public void head(Node node, int depth) {
            if (node instanceof Element) {
                final Element element = (Element) node;
                final Set<String> classes = element.classNames();
                removeEmpty(classes);
                if (classes.isEmpty()) {
                    parts.add(element.ownText());
                }
            }
        }

        @Override
        public void tail(Node node, int depth) {
        }
    });
    if (parts.isEmpty()) {
        final String ownText = element.ownText();
        if (StringUtils.isNotBlank(ownText)) {
            parts.add(element.ownText());
        }
    }
    if (parts.isEmpty()) {
        return null;
    }
    if (parts.size() == 1) {
        return parts.get(0);
    }
    return parts;
}

Example 22 with Node

use of org.jsoup.nodes.Node in project substitution-schedule-parser by vertretungsplanme.

the class SVPlanParser method parseSVPlanSchedule.

@NotNull
SubstitutionSchedule parseSVPlanSchedule(List<Document> docs) throws IOException, JSONException {
    SubstitutionSchedule v = SubstitutionSchedule.fromData(scheduleData);
    for (Document doc : docs) {
        if (doc.select(".svp").size() > 0) {
            for (Element svp : doc.select(".svp")) {
                parseSvPlanDay(v, svp, doc);
            }
        } else if (doc.select(".Trennlinie").size() > 0) {
            Element div = new Element(Tag.valueOf("div"), "");
            for (Node node : doc.body().childNodesCopy()) {
                if (node instanceof Element && ((Element) node).hasClass("Trennlinie") && div.select("table").size() > 0) {
                    parseSvPlanDay(v, div, doc);
                    div = new Element(Tag.valueOf("div"), "");
                } else {
                    div.appendChild(node);
                }
            }
            parseSvPlanDay(v, div, doc);
        } else {
            parseSvPlanDay(v, doc, doc);
        }
    }
    v.setClasses(getAllClasses());
    v.setTeachers(getAllTeachers());
    return v;
}

Example 23 with Node

use of org.jsoup.nodes.Node 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 24 with Node

use of org.jsoup.nodes.Node 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 25 with Node

use of org.jsoup.nodes.Node 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++;
                }
            }
        }
    }
}