Example usage for org.jsoup.nodes Element child

List of usage examples for org.jsoup.nodes Element child

  1. HOME
  2. Java
  3. org.jsoup
  4. org.jsoup.nodes.*
  5. Element
  6. child

Introduction

In this page you can find the example usage for org.jsoup.nodes Element child.

Prototype

public Element child(int index)

Source Link

Document

Get a child element of this element, by its 0-based index number.

Usage

From source file:de.geeksfactory.opacclient.apis.Pica.java

protected SearchRequestResult parse_search(String html, int page) throws OpacErrorException {
    Document doc = Jsoup.parse(html);

    updateSearchSetValue(doc);/* w w w .  jav a  2s. c o  m*/

    if (doc.select(".error").size() > 0) {
        String error = doc.select(".error").first().text().trim();
        if (error.equals("Es wurde nichts gefunden.") || error.equals("Nothing has been found")
                || error.equals("Er is niets gevonden.") || error.equals("Rien n'a t trouv.")) {
            // nothing found
            return new SearchRequestResult(new ArrayList<SearchResult>(), 0, 1, 1);
        } else {
            // error
            throw new OpacErrorException(error);
        }
    }

    reusehtml = html;

    int results_total;

    String resultnumstr = doc.select(".pages").first().text();
    Pattern p = Pattern.compile("[0-9]+$");
    Matcher m = p.matcher(resultnumstr);
    if (m.find()) {
        resultnumstr = m.group();
    }
    if (resultnumstr.contains("(")) {
        results_total = Integer.parseInt(resultnumstr.replaceAll(".*\\(([0-9]+)\\).*", "$1"));
    } else if (resultnumstr.contains(": ")) {
        results_total = Integer.parseInt(resultnumstr.replaceAll(".*: ([0-9]+)$", "$1"));
    } else {
        results_total = Integer.parseInt(resultnumstr);
    }

    List<SearchResult> results = new ArrayList<>();

    if (results_total == 1) {
        // Only one result
        DetailledItem singleResult = parse_result(html);
        SearchResult sr = new SearchResult();
        sr.setType(getMediaTypeInSingleResult(html));
        sr.setInnerhtml(
                "<b>" + singleResult.getTitle() + "</b><br>" + singleResult.getDetails().get(0).getContent());
        results.add(sr);
    }

    Elements table = doc.select("table[summary=hitlist] tbody tr[valign=top]");
    // identifier = null;

    Elements links = doc.select("table[summary=hitlist] a");
    boolean haslink = false;
    for (int i = 0; i < links.size(); i++) {
        Element node = links.get(i);
        if (node.hasAttr("href") & node.attr("href").contains("SHW?") && !haslink) {
            haslink = true;
            try {
                List<NameValuePair> anyurl = URLEncodedUtils.parse(new URI(node.attr("href")),
                        getDefaultEncoding());
                for (NameValuePair nv : anyurl) {
                    if (nv.getName().equals("identifier")) {
                        // identifier = nv.getValue();
                        break;
                    }
                }
            } catch (Exception e) {
                e.printStackTrace();
            }

        }
    }

    for (int i = 0; i < table.size(); i++) {
        Element tr = table.get(i);
        SearchResult sr = new SearchResult();
        if (tr.select("td.hit img").size() > 0) {
            String[] fparts = tr.select("td img").get(0).attr("src").split("/");
            String fname = fparts[fparts.length - 1];
            if (data.has("mediatypes")) {
                try {
                    sr.setType(MediaType.valueOf(data.getJSONObject("mediatypes").getString(fname)));
                } catch (JSONException | IllegalArgumentException e) {
                    sr.setType(defaulttypes.get(fname.toLowerCase(Locale.GERMAN).replace(".jpg", "")
                            .replace(".gif", "").replace(".png", "")));
                }
            } else {
                sr.setType(defaulttypes.get(fname.toLowerCase(Locale.GERMAN).replace(".jpg", "")
                        .replace(".gif", "").replace(".png", "")));
            }
        }
        Element middlething = tr.child(2);

        List<Node> children = middlething.childNodes();
        int childrennum = children.size();

        List<String[]> strings = new ArrayList<>();
        for (int ch = 0; ch < childrennum; ch++) {
            Node node = children.get(ch);
            if (node instanceof TextNode) {
                String text = ((TextNode) node).text().trim();
                if (text.length() > 3) {
                    strings.add(new String[] { "text", "", text });
                }
            } else if (node instanceof Element) {

                List<Node> subchildren = node.childNodes();
                for (int j = 0; j < subchildren.size(); j++) {
                    Node subnode = subchildren.get(j);
                    if (subnode instanceof TextNode) {
                        String text = ((TextNode) subnode).text().trim();
                        if (text.length() > 3) {
                            strings.add(new String[] { ((Element) node).tag().getName(), "text", text,
                                    ((Element) node).className(), node.attr("style") });
                        }
                    } else if (subnode instanceof Element) {
                        String text = ((Element) subnode).text().trim();
                        if (text.length() > 3) {
                            strings.add(new String[] { ((Element) node).tag().getName(),
                                    ((Element) subnode).tag().getName(), text, ((Element) node).className(),
                                    node.attr("style") });
                        }
                    }
                }
            }
        }

        StringBuilder description = new StringBuilder();

        int k = 0;
        for (String[] part : strings) {
            if (part[0].equals("a") && k == 0) {
                description.append("<b>").append(part[2]).append("</b>");
            } else if (k < 3) {
                description.append("<br />").append(part[2]);
            }
            k++;
        }
        sr.setInnerhtml(description.toString());

        sr.setNr(10 * (page - 1) + i);
        sr.setId(null);
        results.add(sr);
    }
    resultcount = results.size();
    return new SearchRequestResult(results, results_total, page);
}

From source file:de.geeksfactory.opacclient.apis.Bibliotheca.java

protected SearchRequestResult parse_search(String html, int page) {
    Document doc = Jsoup.parse(html);
    doc.setBaseUri(opac_url);//from w w  w.  j ava 2s .  c o m
    Elements table = doc.select(".resulttab tr.result_trefferX, .resulttab tr.result_treffer");
    List<SearchResult> results = new ArrayList<>();
    for (int i = 0; i < table.size(); i++) {
        Element tr = table.get(i);
        SearchResult sr = new SearchResult();
        int contentindex = 1;
        if (tr.select("td a img").size() > 0) {
            String[] fparts = tr.select("td a img").get(0).attr("src").split("/");
            String fname = fparts[fparts.length - 1];
            if (data.has("mediatypes")) {
                try {
                    sr.setType(MediaType.valueOf(data.getJSONObject("mediatypes").getString(fname)));
                } catch (JSONException | IllegalArgumentException e) {
                    sr.setType(defaulttypes.get(fname.toLowerCase(Locale.GERMAN).replace(".jpg", "")
                            .replace(".gif", "").replace(".png", "")));
                }
            } else {
                sr.setType(defaulttypes.get(fname.toLowerCase(Locale.GERMAN).replace(".jpg", "")
                        .replace(".gif", "").replace(".png", "")));
            }
        } else {
            if (tr.children().size() == 3) {
                contentindex = 2;
            }
        }
        sr.setInnerhtml(tr.child(contentindex).child(0).html());

        sr.setNr(i);
        Element link = tr.child(contentindex).select("a").first();
        try {
            if (link != null && link.attr("href").contains("detmediennr")) {
                Map<String, String> params = getQueryParamsFirst(link.attr("abs:href"));
                String nr = params.get("detmediennr");
                if (Integer.parseInt(nr) > i + 1) {
                    // Seems to be an ID
                    if (params.get("detDB") != null) {
                        sr.setId("&detmediennr=" + nr + "&detDB=" + params.get("detDB"));
                    } else {
                        sr.setId("&detmediennr=" + nr);
                    }
                }
            }
        } catch (Exception e) {
        }
        try {
            if (tr.child(1).childNode(0) instanceof Comment) {
                Comment c = (Comment) tr.child(1).childNode(0);
                String comment = c.getData().trim();
                String id = comment.split(": ")[1];
                sr.setId(id);
            }
        } catch (Exception e) {
            e.printStackTrace();
        }
        results.add(sr);
    }
    int results_total = -1;
    if (doc.select(".result_gefunden").size() > 0) {
        try {
            results_total = Integer.parseInt(
                    doc.select(".result_gefunden").text().trim().replaceAll(".*[^0-9]+([0-9]+).*", "$1"));
        } catch (NumberFormatException e) {
            e.printStackTrace();
            results_total = -1;
        }
    }
    return new SearchRequestResult(results, results_total, page);
}

From source file:de.geeksfactory.opacclient.apis.Bibliotheca.java

@Override
public ProlongAllResult prolongAll(Account account, int useraction, String selection) throws IOException {

    if (!initialised) {
        start();/*  w w w . ja  v a  2s  . com*/
    }
    if (System.currentTimeMillis() - logged_in > SESSION_LIFETIME || logged_in_as == null) {
        try {
            account(account);
        } catch (JSONException e) {
            e.printStackTrace();
            return new ProlongAllResult(MultiStepResult.Status.ERROR,
                    stringProvider.getString(StringProvider.CONNECTION_ERROR));
        } catch (OpacErrorException e) {
            return new ProlongAllResult(MultiStepResult.Status.ERROR, e.getMessage());
        }
    } else if (logged_in_as.getId() != account.getId()) {
        try {
            account(account);
        } catch (JSONException e) {
            e.printStackTrace();
            return new ProlongAllResult(MultiStepResult.Status.ERROR,
                    stringProvider.getString(StringProvider.CONNECTION_ERROR));
        } catch (OpacErrorException e) {
            return new ProlongAllResult(MultiStepResult.Status.ERROR, e.getMessage());
        }
    }
    String html = httpGet(opac_url + "/index.asp?target=alleverl", getDefaultEncoding());
    Document doc = Jsoup.parse(html);

    if (doc.getElementsByClass("kontomeldung").size() == 1) {
        String err = doc.getElementsByClass("kontomeldung").get(0).text();
        return new ProlongAllResult(MultiStepResult.Status.ERROR, err);
    }

    if (doc.select(".kontozeile table").size() == 1) {
        Map<Integer, String> colmap = new HashMap<>();
        List<Map<String, String>> result = new ArrayList<>();
        for (Element tr : doc.select(".kontozeile table tr")) {
            if (tr.select(".tabHeaderKonto").size() > 0) {
                int i = 0;
                for (Element th : tr.select("th")) {
                    if (th.text().contains("Verfasser")) {
                        colmap.put(i, OpacApi.ProlongAllResult.KEY_LINE_AUTHOR);
                    } else if (th.text().contains("Titel")) {
                        colmap.put(i, OpacApi.ProlongAllResult.KEY_LINE_TITLE);
                    } else if (th.text().contains("Neue")) {
                        colmap.put(i, OpacApi.ProlongAllResult.KEY_LINE_NEW_RETURNDATE);
                    } else if (th.text().contains("Frist")) {
                        colmap.put(i, OpacApi.ProlongAllResult.KEY_LINE_OLD_RETURNDATE);
                    } else if (th.text().contains("Status")) {
                        colmap.put(i, OpacApi.ProlongAllResult.KEY_LINE_MESSAGE);
                    }
                    i++;
                }
            } else {
                Map<String, String> line = new HashMap<>();
                for (Entry<Integer, String> entry : colmap.entrySet()) {
                    line.put(entry.getValue(), tr.child(entry.getKey()).text().trim());
                }
                result.add(line);
            }
        }

        if (doc.select("input#make_allvl").size() > 0) {
            List<NameValuePair> nameValuePairs = new ArrayList<>(2);
            nameValuePairs.add(new BasicNameValuePair("target", "make_allvl_flag"));
            nameValuePairs.add(new BasicNameValuePair("make_allvl", "Bestaetigung"));
            httpPost(opac_url + "/index.asp", new UrlEncodedFormEntity(nameValuePairs), getDefaultEncoding());
        }

        return new ProlongAllResult(MultiStepResult.Status.OK, result);
    }

    return new ProlongAllResult(MultiStepResult.Status.ERROR,
            stringProvider.getString(StringProvider.INTERNAL_ERROR));
}

From source file:de.geeksfactory.opacclient.apis.BiBer1992.java

private SearchRequestResult parse_search(String html, int page) {
    List<SearchResult> results = new ArrayList<>();
    Document doc = Jsoup.parse(html);

    if (doc.select("h3").text().contains("Es wurde nichts gefunden")) {
        return new SearchRequestResult(results, 0, page);
    }/*from ww  w . j av a2  s.  c om*/

    Elements trList = doc.select("form table tr[valign]"); // <tr
    // valign="top">
    Elements elem;
    int rows_per_hit = 2;
    if (trList.size() == 1 || (trList.size() > 1 && trList.get(0).select("input[type=checkbox]").size() > 0
            && trList.get(1).select("input[type=checkbox]").size() > 0)) {
        rows_per_hit = 1;
    }

    try {
        rows_per_hit = data.getInt("rows_per_hit");
    } catch (JSONException e) {
    }

    // Overall search results
    // are very differently layouted, but have always the text:
    // "....Treffer Gesamt (nnn)"
    int results_total;
    Pattern pattern = Pattern.compile("Treffer Gesamt \\(([0-9]+)\\)");
    Matcher matcher = pattern.matcher(html);
    if (matcher.find()) {
        results_total = Integer.parseInt(matcher.group(1));
    } else {
        results_total = -1;
    }

    // limit to 20 entries
    int numOfEntries = trList.size() / rows_per_hit; // two rows per entry
    if (numOfEntries > numOfResultsPerPage) {
        numOfEntries = numOfResultsPerPage;
    }

    for (int i = 0; i < numOfEntries; i++) {
        Element tr = trList.get(i * rows_per_hit);
        SearchResult sr = new SearchResult();

        // ID as href tag
        elem = tr.select("td a");
        if (elem.size() > 0) {
            String hrefID = elem.get(0).attr("href");
            sr.setId(hrefID);
        } else {
            // no ID as href found, look for the ID in the input form
            elem = tr.select("td input");
            if (elem.size() > 0) {
                String nameID = elem.get(0).attr("name").trim();
                String hrefID = "/" + opacDir + "/ftitle" + opacSuffix + "?LANG=de&FUNC=full&" + nameID
                        + "=YES";
                sr.setId(hrefID);
            }
        }

        // media type
        elem = tr.select("td img");
        if (elem.size() > 0) {
            setMediaTypeFromImageFilename(sr, elem.get(0).attr("src"));
        }

        // description
        String desc = "";
        try {
            // array "searchtable" list the column numbers of the
            // description
            JSONArray searchtable = data.getJSONArray("searchtable");
            for (int j = 0; j < searchtable.length(); j++) {
                int colNum = searchtable.getInt(j);
                if (j > 0) {
                    desc = desc + "<br />";
                }
                String c = tr.child(colNum).html();
                if (tr.child(colNum).childNodes().size() == 1
                        && tr.child(colNum).select("a[href*=ftitle.]").size() > 0) {
                    c = tr.select("a[href*=ftitle.]").text();
                }
                desc = desc + c;
            }
        } catch (Exception e) {
            e.printStackTrace();
        }
        // remove links "<a ...>...</a>
        // needed for Friedrichshafen: "Warenkorb", "Vormerkung"
        // Herford: "Medienkorb"
        desc = desc.replaceAll("<a .*?</a>", "");
        sr.setInnerhtml(desc);

        if (tr.select("font.p04x09b").size() > 0 && tr.select("font.p02x09b").size() == 0) {
            sr.setStatus(Status.GREEN);
        } else if (tr.select("font.p04x09b").size() == 0 && tr.select("font.p02x09b").size() > 0) {
            sr.setStatus(Status.RED);
        } else if (tr.select("font.p04x09b").size() > 0 && tr.select("font.p02x09b").size() > 0) {
            sr.setStatus(Status.YELLOW);
        }

        // number
        sr.setNr(i / rows_per_hit);
        results.add(sr);
    }

    // m_resultcount = results.size();
    return new SearchRequestResult(results, results_total, page);
}

From source file:de.geeksfactory.opacclient.apis.SISIS.java

protected SearchRequestResult parse_search(String html, int page) throws OpacErrorException {
    Document doc = Jsoup.parse(html);
    doc.setBaseUri(opac_url + "/searchfoo");

    if (doc.select(".error").size() > 0) {
        throw new OpacErrorException(doc.select(".error").text().trim());
    } else if (doc.select(".nohits").size() > 0) {
        throw new OpacErrorException(doc.select(".nohits").text().trim());
    } else if (doc.select(".box-header h2, #nohits").text().contains("keine Treffer")) {
        return new SearchRequestResult(new ArrayList<SearchResult>(), 0, 1, 1);
    }/*w  w  w. jav  a 2  s  . co  m*/

    int results_total = -1;

    String resultnumstr = doc.select(".box-header h2").first().text();
    if (resultnumstr.contains("(1/1)") || resultnumstr.contains(" 1/1")) {
        reusehtml = html;
        throw new OpacErrorException("is_a_redirect");
    } else if (resultnumstr.contains("(")) {
        results_total = Integer.parseInt(resultnumstr.replaceAll(".*\\(([0-9]+)\\).*", "$1"));
    } else if (resultnumstr.contains(": ")) {
        results_total = Integer.parseInt(resultnumstr.replaceAll(".*: ([0-9]+)$", "$1"));
    }

    Elements table = doc.select("table.data tbody tr");
    identifier = null;

    Elements links = doc.select("table.data a");
    boolean haslink = false;
    for (int i = 0; i < links.size(); i++) {
        Element node = links.get(i);
        if (node.hasAttr("href") & node.attr("href").contains("singleHit.do") && !haslink) {
            haslink = true;
            try {
                List<NameValuePair> anyurl = URLEncodedUtils
                        .parse(new URI(node.attr("href").replace(" ", "%20").replace("&amp;", "&")), ENCODING);
                for (NameValuePair nv : anyurl) {
                    if (nv.getName().equals("identifier")) {
                        identifier = nv.getValue();
                        break;
                    }
                }
            } catch (Exception e) {
                e.printStackTrace();
            }

        }
    }

    List<SearchResult> results = new ArrayList<>();
    for (int i = 0; i < table.size(); i++) {
        Element tr = table.get(i);
        SearchResult sr = new SearchResult();
        if (tr.select("td img[title]").size() > 0) {
            String title = tr.select("td img").get(0).attr("title");
            String[] fparts = tr.select("td img").get(0).attr("src").split("/");
            String fname = fparts[fparts.length - 1];
            MediaType default_by_fname = defaulttypes.get(fname.toLowerCase(Locale.GERMAN).replace(".jpg", "")
                    .replace(".gif", "").replace(".png", ""));
            MediaType default_by_title = defaulttypes.get(title);
            MediaType default_name = default_by_title != null ? default_by_title : default_by_fname;
            if (data.has("mediatypes")) {
                try {
                    sr.setType(MediaType.valueOf(data.getJSONObject("mediatypes").getString(fname)));
                } catch (JSONException | IllegalArgumentException e) {
                    sr.setType(default_name);
                }
            } else {
                sr.setType(default_name);
            }
        }
        String alltext = tr.text();
        if (alltext.contains("eAudio") || alltext.contains("eMusic")) {
            sr.setType(MediaType.MP3);
        } else if (alltext.contains("eVideo")) {
            sr.setType(MediaType.EVIDEO);
        } else if (alltext.contains("eBook")) {
            sr.setType(MediaType.EBOOK);
        } else if (alltext.contains("Munzinger")) {
            sr.setType(MediaType.EDOC);
        }

        if (tr.children().size() > 3 && tr.child(3).select("img[title*=cover]").size() == 1) {
            sr.setCover(tr.child(3).select("img[title*=cover]").attr("abs:src"));
            if (sr.getCover().contains("showCover.do")) {
                downloadCover(sr);
            }
        }

        Element middlething;
        if (tr.children().size() > 2 && tr.child(2).select("a").size() > 0) {
            middlething = tr.child(2);
        } else {
            middlething = tr.child(1);
        }

        List<Node> children = middlething.childNodes();
        if (middlething.select("div").not("#hlrightblock,.bestellfunktionen").size() == 1) {
            Element indiv = middlething.select("div").not("#hlrightblock,.bestellfunktionen").first();
            if (indiv.children().size() > 1) {
                children = indiv.childNodes();
            }
        } else if (middlething.select("span.titleData").size() == 1) {
            children = middlething.select("span.titleData").first().childNodes();
        }
        int childrennum = children.size();

        List<String[]> strings = new ArrayList<>();
        for (int ch = 0; ch < childrennum; ch++) {
            Node node = children.get(ch);
            if (node instanceof TextNode) {
                String text = ((TextNode) node).text().trim();
                if (text.length() > 3) {
                    strings.add(new String[] { "text", "", text });
                }
            } else if (node instanceof Element) {

                List<Node> subchildren = node.childNodes();
                for (int j = 0; j < subchildren.size(); j++) {
                    Node subnode = subchildren.get(j);
                    if (subnode instanceof TextNode) {
                        String text = ((TextNode) subnode).text().trim();
                        if (text.length() > 3) {
                            strings.add(new String[] { ((Element) node).tag().getName(), "text", text,
                                    ((Element) node).className(), node.attr("style") });
                        }
                    } else if (subnode instanceof Element) {
                        String text = ((Element) subnode).text().trim();
                        if (text.length() > 3) {
                            strings.add(new String[] { ((Element) node).tag().getName(),
                                    ((Element) subnode).tag().getName(), text, ((Element) node).className(),
                                    node.attr("style") });
                        }
                    }
                }
            }
        }

        StringBuilder description = null;
        if (tr.select("span.Z3988").size() == 1) {
            // Sometimes there is a <span class="Z3988"> item which provides
            // data in a standardized format.
            List<NameValuePair> z3988data;
            boolean hastitle = false;
            try {
                description = new StringBuilder();
                z3988data = URLEncodedUtils
                        .parse(new URI("http://dummy/?" + tr.select("span.Z3988").attr("title")), "UTF-8");
                for (NameValuePair nv : z3988data) {
                    if (nv.getValue() != null) {
                        if (!nv.getValue().trim().equals("")) {
                            if (nv.getName().equals("rft.btitle") && !hastitle) {
                                description.append("<b>").append(nv.getValue()).append("</b>");
                                hastitle = true;
                            } else if (nv.getName().equals("rft.atitle") && !hastitle) {
                                description.append("<b>").append(nv.getValue()).append("</b>");
                                hastitle = true;
                            } else if (nv.getName().equals("rft.au")) {
                                description.append("<br />").append(nv.getValue());
                            } else if (nv.getName().equals("rft.date")) {
                                description.append("<br />").append(nv.getValue());
                            }
                        }
                    }
                }
            } catch (URISyntaxException e) {
                description = null;
            }
        }
        boolean described = false;
        if (description != null && description.length() > 0) {
            sr.setInnerhtml(description.toString());
            described = true;
        } else {
            description = new StringBuilder();
        }
        int k = 0;
        boolean yearfound = false;
        boolean titlefound = false;
        boolean sigfound = false;
        for (String[] part : strings) {
            if (!described) {
                if (part[0].equals("a") && (k == 0 || !titlefound)) {
                    if (k != 0) {
                        description.append("<br />");
                    }
                    description.append("<b>").append(part[2]).append("</b>");
                    titlefound = true;
                } else if (part[2].matches("\\D*[0-9]{4}\\D*") && part[2].length() <= 10) {
                    yearfound = true;
                    if (k != 0) {
                        description.append("<br />");
                    }
                    description.append(part[2]);
                } else if (k == 1 && !yearfound && part[2].matches("^\\s*\\([0-9]{4}\\)$")) {
                    if (k != 0) {
                        description.append("<br />");
                    }
                    description.append(part[2]);
                } else if (k == 1 && !yearfound && part[2].matches("^\\s*\\([0-9]{4}\\)$")) {
                    if (k != 0) {
                        description.append("<br />");
                    }
                    description.append(part[2]);
                } else if (k > 1 && k < 4 && !sigfound && part[0].equals("text")
                        && part[2].matches("^[A-Za-z0-9,\\- ]+$")) {
                    description.append("<br />");
                    description.append(part[2]);
                }
            }
            if (part.length == 4) {
                if (part[0].equals("span") && part[3].equals("textgruen")) {
                    sr.setStatus(SearchResult.Status.GREEN);
                } else if (part[0].equals("span") && part[3].equals("textrot")) {
                    sr.setStatus(SearchResult.Status.RED);
                }
            } else if (part.length == 5) {
                if (part[4].contains("purple")) {
                    sr.setStatus(SearchResult.Status.YELLOW);
                }
            }
            if (sr.getStatus() == null) {
                if ((part[2].contains("entliehen")
                        && part[2].startsWith("Vormerkung ist leider nicht mglich"))
                        || part[2].contains("nur in anderer Zweigstelle ausleihbar und nicht bestellbar")) {
                    sr.setStatus(SearchResult.Status.RED);
                } else if (part[2].startsWith("entliehen")
                        || part[2].contains("Ein Exemplar finden Sie in einer anderen Zweigstelle")) {
                    sr.setStatus(SearchResult.Status.YELLOW);
                } else if ((part[2].startsWith("bestellbar") && !part[2].contains("nicht bestellbar"))
                        || (part[2].startsWith("vorbestellbar") && !part[2].contains("nicht vorbestellbar"))
                        || (part[2].startsWith("vorbestellbar") && !part[2].contains("nicht vorbestellbar"))
                        || (part[2].startsWith("vormerkbar") && !part[2].contains("nicht vormerkbar"))
                        || (part[2].contains("heute zurckgebucht"))
                        || (part[2].contains("ausleihbar") && !part[2].contains("nicht ausleihbar"))) {
                    sr.setStatus(SearchResult.Status.GREEN);
                }
                if (sr.getType() != null) {
                    if (sr.getType().equals(MediaType.EBOOK) || sr.getType().equals(MediaType.EVIDEO)
                            || sr.getType().equals(MediaType.MP3))
                    // Especially Onleihe.de ebooks are often marked
                    // green though they are not available.
                    {
                        sr.setStatus(SearchResult.Status.UNKNOWN);
                    }
                }
            }
            k++;
        }
        if (!described) {
            sr.setInnerhtml(description.toString());
        }

        sr.setNr(10 * (page - 1) + i);
        sr.setId(null);
        results.add(sr);
    }
    resultcount = results.size();
    return new SearchRequestResult(results, results_total, page);
}

From source file:org.apdplat.superword.extract.ChineseSynonymAntonymExtractor.java

public static SynonymAntonym parseSynonymAntonym(String html, String word) {
    SynonymAntonym synonymAntonym = new SynonymAntonym();
    synonymAntonym.setWord(new Word(word, ""));
    try {//  w  ww . j av  a  2s .  co  m
        for (Element element : Jsoup.parse(html).select(SYNONYM_ANTONYM_CSS_PATH)) {
            int size = element.children().size();
            LOGGER.debug("element size:" + size);
            for (int i = 0; i < size / 2; i++) {
                String type = element.child(i * 2).text();
                LOGGER.debug("type:" + type);
                if ("??".equals(type)) {
                    String synonym = element.child(i * 2 + 1).text();
                    LOGGER.debug("synonym:" + synonym);
                    for (String w : synonym.split("\\s+")) {
                        w = w.replaceAll("\\s+", "");
                        if (w.length() < 2) {
                            continue;
                        }
                        if (isNotChineseChar(w)) {
                            LOGGER.debug("?" + w);
                            continue;
                        }
                        if (w.equals(word)) {
                            continue;
                        }
                        LOGGER.debug("word:" + w);
                        synonymAntonym.addSynonym(new Word(w, ""));
                    }
                }
                if ("???".equals(type)) {
                    String antonym = element.child(i * 2 + 1).text();
                    LOGGER.debug("antonym:" + antonym);
                    for (String w : antonym.split("\\s+")) {
                        w = w.replaceAll("\\s+", "");
                        if (w.length() < 2) {
                            continue;
                        }
                        if (isNotChineseChar(w)) {
                            LOGGER.debug("?" + w);
                            continue;
                        }
                        LOGGER.debug("word:" + w);
                        synonymAntonym.addAntonym(new Word(w, ""));
                    }
                }
            }
        }
        if (!synonymAntonym.getAntonym().isEmpty() || !synonymAntonym.getSynonym().isEmpty()) {
            LOGGER.info("??????" + synonymAntonym);
        }
    } catch (Exception e) {
        LOGGER.error("??????", e);
    }
    return synonymAntonym;
}

From source file:org.apdplat.superword.extract.SentenceExtractor.java

public static Map<String, String> parse2(String html) {
    Map<String, String> sentences = new HashMap<>();
    try {//from   ww w .  j  av  a 2  s  .c  o  m
        Document document = Jsoup.parse(html);
        String title = document.select("html head title").text();
        if (!title.startsWith("??")) {
            LOGGER.error("???" + title);
            return sentences;
        }
        for (Element element : document.select("html body div#main div.main_sl div.info div.info-body")) {
            String en = element.child(3).text().trim();
            LOGGER.info("???:" + en);
            if (en.split("\\s+").length < 2) {
                LOGGER.debug("???");
                continue;
            }
            String cn = element.child(4).text().trim() + element.child(5).text().trim();
            LOGGER.info("???:" + cn);
            if (StringUtils.isNotBlank(en) && StringUtils.isNotBlank(cn)) {
                sentences.put(en, cn);
                //?
                TextAnalyzer.seg(en).forEach(w -> {
                    Word word = new Word(w, "");
                    WORD_FREQUENCE.putIfAbsent(word, new AtomicInteger());
                    WORD_FREQUENCE.get(word).incrementAndGet();
                });
            }
        }
    } catch (Exception e) {
        LOGGER.error("???", e);
    }
    return sentences;
}

From source file:org.arb.extractor.DomTreeWalker.java

private void collectIdsOnElement(Element element, AbstractCodeUnit codeUnit) {
    if (element.hasAttr("id")) {
        elementIdSet.add(element.attr("id"));
    }/*www. j  a v  a2 s.  c  o m*/
    if (element.hasAttr("arb:id")) {
        arbIdSet.add(element.attr("arb:id"));
    }
    for (int i = 0; i < element.children().size(); i++) {
        collectIdsOnElement(element.child(i), codeUnit);
    }
}

From source file:org.arb.extractor.DomTreeWalker.java

/**
 * Extract resource from an element and all its children.
 *  //  w w w .j a  v a2  s.co  m
 * @param element the target element.
 * @param codeUnit used to record all found replacement.
 */
private void extractResourceOnElement(Element element, AbstractCodeUnit codeUnit) {
    String ownText = element.ownText();
    if (!ownText.isEmpty() && !hasResource(element)) {
        DomCodeReplacement replacement = new DomCodeReplacement();
        replacement.setElement(element);
        replacement.setResourceText(ownText);
        replacement.setResourceId(getElementResourceId(element));
        if (replacement.getResourceId() == null) {
            replacement.setNewId(true);
            replacement.setUseArbId(true);
        }
        codeUnit.addReplacement(replacement);
    }
    for (int i = 0; i < element.children().size(); i++) {
        extractResourceOnElement(element.child(i), codeUnit);
    }
}

From source file:org.asqatasun.rules.elementselector.ImageElementSelector.java

/**
 * /*  w  w  w.  j a  v  a2  s . co m*/
 * @param imageParent
 * @param image
 * @return whether the current image is an image link
 */
private boolean isImageLink(Element imageParent, Element image) {
    if (imageParent == null || !StringUtils.equals(imageParent.text(), image.text())) {
        return false;
    }
    if (imageParent.children().size() == 1) {
        return isImageLink(imageParent.child(0), image);
    } else if (imageParent.children().isEmpty() && imageParent.equals(image)) {
        return true;
    }
    return false;
}