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

Example 11 with TextNode

use of org.jsoup.nodes.TextNode in project McWiki by skylerdev.

the class CommandWiki method buildPages.

private List<String> buildPages(Document doc, String title, String url) {
    Elements main = doc.select("div[id=mw-content-text] > p, h2, h3");
    ArrayList<String> pages = new ArrayList<String>();
    pages.add(titlePage(title, url));
    pages.add("Table of contents placeholder");
    JSONArray contentsPage = newPage();
    MCJson contentsHead = new MCJson("Contents\n\n", "dark_gray");
    contentsHead.setBold(true);
    contentsPage.add(contentsHead);
    JSONArray currentPage = newPage();
    int currentPageSize = 0;
    int maxChars = 230;
    MCJson space = new MCJson(" ");
    for (Element mainchild : main) {
        if (currentPageSize > maxChars) {
            pages.add(currentPage.toString());
            currentPageSize = 0;
            currentPage = newPage();
        }
        if (mainchild.is("h2")) {
            if (isOmitted(mainchild)) {
                continue;
            }
            pages.add(currentPage.toString());
            currentPage = newPage();
            currentPageSize = 20;
            String h = mainchild.text().replaceAll("\\[edit\\]", "");
            currentPage.add(new MCJson(h, header2));
            currentPage.add(space);
            MCJson backButton = new MCJson("«", link);
            backButton.setHover("show_text", "Back to contents");
            backButton.setClick("change_page", "2");
            currentPage.add(backButton);
            currentPage.add(new MCJson("\n"));
            MCJson contentslink = new MCJson(h, link);
            contentslink.setHover("show_text", "Jump to this section");
            contentslink.setClick("change_page", "" + (pages.size() + 1));
            contentsPage.add(contentslink);
            contentsPage.add(new MCJson("\n"));
        } else if (mainchild.is("h3")) {
            String h = mainchild.text().replaceAll("\\[edit\\]", "");
            currentPage.add(new MCJson(h, header3));
            currentPage.add(space);
            currentPageSize += h.length() + 2;
        } else if (mainchild.is("p")) {
            List<Node> inner = mainchild.childNodes();
            for (Node n : inner) {
                if (currentPageSize > maxChars) {
                    pages.add(currentPage.toString());
                    currentPageSize = 0;
                    currentPage = newPage();
                }
                if (n instanceof Element) {
                    Element e = (Element) n;
                    if (e.is("a")) {
                        String linkto = e.attr("href");
                        MCJson a = new MCJson(e.text(), link);
                        if (linkto.startsWith("/")) {
                            a.setClick("run_command", "/wiki " + linkto.substring(1));
                            a.setHover("show_text", "Click to show this article.");
                        } else {
                            a.setClick("open_url", linkto);
                            a.setHover("show_text", "External Link");
                        }
                        currentPage.add(a);
                    } else if (e.is("b")) {
                        currentPage.add(new MCJson(e.text(), bold));
                    } else if (e.is("i")) {
                        currentPage.add(new MCJson(e.text(), italic));
                    }
                    currentPageSize += e.text().length();
                }
                if (n instanceof TextNode) {
                    TextNode t = (TextNode) n;
                    String s = t.text();
                    if (currentPageSize + s.length() > maxChars - 10) {
                        // karate chop
                        int splitAt = s.lastIndexOf(" ", maxChars - currentPageSize);
                        if (splitAt < 0) {
                            pages.add(currentPage.toString());
                            currentPage = newPage();
                            currentPage.add(new MCJson(s));
                            currentPageSize = s.length();
                        } else {
                            String firstString = s.substring(0, splitAt);
                            String nextString = s.substring(splitAt + 1);
                            currentPage.add(new MCJson(firstString));
                            pages.add(currentPage.toString());
                            currentPageSize = nextString.length();
                            currentPage = newPage();
                            currentPage.add(new MCJson(nextString));
                        }
                    } else {
                        currentPage.add(new MCJson(s));
                        currentPageSize += s.length();
                    }
                }
            }
            currentPage.add("\n");
            currentPageSize += 20;
        }
    }
    pages.set(1, contentsPage.toString());
    pages.add(endPage(title, url));
    return pages;
}

Example 12 with TextNode

use of org.jsoup.nodes.TextNode in project ultimate-java by pantinor.

the class HtmlUtils method buildStringFromNode.

public static StringBuffer buildStringFromNode(Node node) {
    StringBuffer buffer = new StringBuffer();
    if (node instanceof TextNode) {
        TextNode textNode = (TextNode) node;
        buffer.append(textNode.text().trim());
    }
    for (Node childNode : node.childNodes()) {
        buffer.append(buildStringFromNode(childNode));
    }
    if (node instanceof Element) {
        Element element = (Element) node;
        String tagName = element.tagName();
        if ("p".equals(tagName) || "br".equals(tagName)) {
            buffer.append("\n");
        }
        if ("h1".equals(tagName) || "h2".equals(tagName) || "h3".equals(tagName) || "h4".equals(tagName) || "dt".equals(tagName) || "dl".equals(tagName)) {
            buffer.append("\n\n");
        }
    }
    return buffer;
}

Example 13 with TextNode

use of org.jsoup.nodes.TextNode in project NetDiscovery by fengzhizi715.

the class CssSelector method getText.

protected String getText(Element element) {
    StringBuilder accum = new StringBuilder();
    for (Node node : element.childNodes()) {
        if (node instanceof TextNode) {
            TextNode textNode = (TextNode) node;
            accum.append(textNode.text());
        }
    }
    return accum.toString();
}

Example 14 with TextNode

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

the class DaVinciParser method parsePage.

@NotNull
static void parsePage(Element doc, SubstitutionSchedule schedule, ColorProvider colorProvider) throws IOException {
    SubstitutionScheduleDay day = new SubstitutionScheduleDay();
    Element titleElem;
    if (doc.select("h1.list-table-caption").size() > 0) {
        titleElem = doc.select("h1.list-table-caption").first();
    } else {
        // DaVinci 5
        titleElem = doc.select("h2").first();
    }
    String title = titleElem.text();
    String klasse = null;
    // title can either be date or class
    Pattern datePattern = Pattern.compile("\\d+\\.\\d+.\\d{4}");
    Matcher dateMatcher = datePattern.matcher(title);
    if (dateMatcher.find()) {
        day.setDateString(dateMatcher.group());
        day.setDate(ParserUtils.parseDate(dateMatcher.group()));
    } else {
        klasse = title;
        String nextText = titleElem.nextElementSibling().text();
        if (nextText.matches("\\w+ \\d+\\.\\d+.\\d{4}")) {
            day.setDateString(nextText);
            day.setDate(ParserUtils.parseDate(nextText));
        } else {
            // could not find date, must be multiple days
            day = null;
        }
    }
    for (Element p : doc.select(".row:has(h1.list-table-caption) p")) {
        for (TextNode node : p.textNodes()) {
            if (!node.text().trim().isEmpty() && day != null)
                day.addMessage(node.text().trim());
        }
    }
    for (Element message : doc.select(".callout")) {
        for (TextNode node : message.textNodes()) {
            if (!node.text().trim().isEmpty())
                day.addMessage(node.text().trim());
        }
    }
    Element lastChangeElem = doc.select(".row.copyright div").first();
    if (lastChangeElem == null) {
        // DaVinci 5
        lastChangeElem = doc.select("h1").first();
    }
    if (lastChangeElem != null) {
        String lastChange = lastChangeElem.ownText();
        Pattern pattern = Pattern.compile("(\\d{2}-\\d{2}-\\d{4} \\d{2}:\\d{2}) \\|");
        Matcher matcher = pattern.matcher(lastChange);
        if (matcher.find()) {
            LocalDateTime lastChangeTime = DateTimeFormat.forPattern("dd-MM-yyyy HH:mm").parseLocalDateTime(matcher.group(1));
            if (day != null) {
                day.setLastChange(lastChangeTime);
            } else {
                schedule.setLastChange(lastChangeTime);
            }
        } else {
            Pattern pattern2 = Pattern.compile("(\\d{2}.\\d{2}.\\d{4} \\| \\d+:\\d{2})");
            Matcher matcher2 = pattern2.matcher(lastChange);
            if (matcher2.find()) {
                LocalDateTime lastChangeTime = DateTimeFormat.forPattern("dd.MM.yyyy | HH:mm").parseLocalDateTime(matcher2.group(1));
                if (day != null) {
                    day.setLastChange(lastChangeTime);
                } else {
                    schedule.setLastChange(lastChangeTime);
                }
            }
        }
    } else {
        Pattern pattern = Pattern.compile("<!-- Created by daVinci 5 \\| (\\d+\\.\\d+\\.\\d+ \\| \\d+:\\d+) \\| " + "www.stueber.de -->");
        Matcher matcher = pattern.matcher(doc.html());
        if (matcher.find()) {
            String str = matcher.group(1);
            LocalDateTime date = DateTimeFormat.forPattern("dd.MM.yyyy | HH:mm").parseLocalDateTime(str);
            if (day != null) {
                day.setLastChange(date);
            } else {
                schedule.setLastChange(date);
            }
        }
    }
    if (doc.select(".list-table").size() > 0 || !doc.select(".callout").text().contains("Es liegen keine")) {
        Element table = doc.select(".list-table, table").first();
        parseDaVinciTable(table, schedule, klasse, day, colorProvider);
    }
    if (day != null) {
        schedule.addDay(day);
    }
}

Example 15 with TextNode

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

the class Bibliotheca method parseSearchFields.

@Override
public List<SearchField> parseSearchFields() throws IOException, JSONException {
    if (!initialised) {
        start();
    }
    List<SearchField> fields = new ArrayList<>();
    // Read branches and media types
    FormBody.Builder formData = new FormBody.Builder(Charset.forName(getDefaultEncoding()));
    formData.add("link_profis.x", "0");
    formData.add("link_profis.y", "1");
    String html = httpPost(opac_url + "/index.asp", formData.build(), getDefaultEncoding());
    Document doc = Jsoup.parse(html);
    Elements fieldElems = doc.select(".suchfeldinhalt");
    for (Element fieldElem : fieldElems) {
        String name = fieldElem.select(".suchfeld_inhalt_titel label").text();
        String hint = "";
        if (fieldElem.select(".suchfeld_inhalt_input").size() > 0) {
            List<TextNode> textNodes = fieldElem.select(".suchfeld_inhalt_input").first().textNodes();
            if (textNodes.size() > 0) {
                for (TextNode node : textNodes) {
                    String text = node.getWholeText().replace("\n", "");
                    if (!text.equals("")) {
                        hint = node.getWholeText().replace("\n", "");
                        break;
                    }
                }
            }
        }
        Elements inputs = fieldElem.select(".suchfeld_inhalt_input input[type=text], " + ".suchfeld_inhalt_input select");
        if (inputs.size() == 1) {
            SearchField field = createSearchField(name, hint, inputs.get(0));
            Elements radios = fieldElem.select("input[type=radio]");
            if (field instanceof TextSearchField && radios.size() > 0) {
                TextSearchField tf = (TextSearchField) field;
                if (radios.get(0).attr("value").equals("stich")) {
                    tf.setFreeSearch(true);
                    if (fieldElem.select("label[for=stichtit_sich]").size() > 0) {
                        tf.setHint(fieldElem.select("label[for=stichtit_sich]").text().trim());
                    }
                    JSONObject addData = new JSONObject();
                    JSONObject params = new JSONObject();
                    params.put("stichtit", "stich");
                    addData.put("additional_params", params);
                    tf.setData(addData);
                }
                if (radios.size() == 2 && radios.get(1).attr("value").equals("titel")) {
                    TextSearchField tf2 = new TextSearchField();
                    tf2.setId(tf.getId());
                    if (fieldElem.select("label[for=stichtit_titel]").size() > 0) {
                        tf2.setDisplayName(fieldElem.select("label[for=stichtit_titel]").text().trim());
                    }
                    JSONObject addData = new JSONObject();
                    JSONObject params = new JSONObject();
                    params.put("stichtit", "titel");
                    addData.put("additional_params", params);
                    tf2.setData(addData);
                    fields.add(tf2);
                }
            }
            fields.add(field);
        } else if (inputs.size() == 2 && inputs.select("input[type=text]").size() == 2) {
            // Two text fields, e.g. year from/to or two keywords
            fields.add(createSearchField(name, hint, inputs.get(0)));
            TextSearchField secondField = (TextSearchField) createSearchField(name, hint, inputs.get(1));
            secondField.setHalfWidth(true);
            fields.add(secondField);
        } else if (inputs.size() == 2 && inputs.get(0).tagName().equals("select") && inputs.get(1).tagName().equals("input") && inputs.get(0).attr("name").equals("feld1")) {
            // Break it down into single text fields.
            for (Element option : inputs.get(0).select("option")) {
                TextSearchField field = new TextSearchField();
                field.setHint(hint);
                field.setDisplayName(option.text());
                field.setId(inputs.get(1).attr("name") + "$" + option.attr("value"));
                JSONObject data = new JSONObject();
                JSONObject params = new JSONObject();
                params.put(inputs.get(0).attr("name"), option.attr("value"));
                data.put("additional_params", params);
                field.setData(data);
                fields.add(field);
            }
        }
    }
    if (fields.size() > 0) {
        DropdownSearchField orderField = new DropdownSearchField("orderselect", stringProvider.getString(StringProvider.ORDER), false, null);
        orderField.addDropdownValue("1", stringProvider.getString(StringProvider.ORDER_DEFAULT));
        orderField.addDropdownValue("2:desc", stringProvider.getString(StringProvider.ORDER_YEAR_DESC));
        orderField.addDropdownValue("2:asc", stringProvider.getString(StringProvider.ORDER_YEAR_ASC));
        orderField.addDropdownValue("3:desc", stringProvider.getString(StringProvider.ORDER_CATEGORY_DESC));
        orderField.addDropdownValue("3:asc", stringProvider.getString(StringProvider.ORDER_CATEGORY_ASC));
        orderField.setMeaning(Meaning.ORDER);
        fields.add(orderField);
    }
    return fields;
}