Examples with Attributes org.xml.sax.Attributes used on opensource projects

Example 51 with Attributes

use of org.xml.sax.Attributes in project tika by apache.

the class HtmlParserTest method testParseAscii.

@Test
public void testParseAscii() throws Exception {
    String path = "/test-documents/testHTML.html";
    final StringWriter href = new StringWriter();
    final StringWriter name = new StringWriter();
    ContentHandler body = new BodyContentHandler();
    Metadata metadata = new Metadata();
    try (InputStream stream = HtmlParserTest.class.getResourceAsStream(path)) {
        ContentHandler link = new DefaultHandler() {

            @Override
            public void startElement(String u, String l, String n, Attributes a) throws SAXException {
                if ("a".equals(l)) {
                    if (a.getValue("href") != null) {
                        href.append(a.getValue("href"));
                    } else if (a.getValue("name") != null) {
                        name.append(a.getValue("name"));
                    }
                }
            }
        };
        new HtmlParser().parse(stream, new TeeContentHandler(body, link), metadata, new ParseContext());
    }
    assertEquals("Title : Test Indexation Html", metadata.get(TikaCoreProperties.TITLE));
    assertEquals("Tika Developers", metadata.get("Author"));
    assertEquals("5", metadata.get("refresh"));
    assertEquals("51.2312", metadata.get(Geographic.LATITUDE));
    assertEquals("-5.1987", metadata.get(Geographic.LONGITUDE));
    assertEquals("http://www.apache.org/", href.toString());
    assertEquals("test-anchor", name.toString());
    String content = body.toString();
    assertTrue("Did not contain expected text:" + "Test Indexation Html", content.contains("Test Indexation Html"));
    assertTrue("Did not contain expected text:" + "Indexation du fichier", content.contains("Indexation du fichier"));
}

Example 52 with Attributes

use of org.xml.sax.Attributes in project tika by apache.

the class HtmlParserTest method assertScriptLink.

private void assertScriptLink(String html, String url) throws Exception {
    // IdentityHtmlMapper is needed to extract <script> tags
    ParseContext context = new ParseContext();
    context.set(HtmlMapper.class, IdentityHtmlMapper.INSTANCE);
    Metadata metadata = new Metadata();
    metadata.set(Metadata.CONTENT_TYPE, "text/html");
    final List<String> links = new ArrayList<String>();
    new HtmlParser().parse(new ByteArrayInputStream(html.getBytes(UTF_8)), new DefaultHandler() {

        @Override
        public void startElement(String u, String l, String name, Attributes atts) {
            if (name.equals("script") && atts.getValue("", "src") != null) {
                links.add(atts.getValue("", "src"));
            }
        }
    }, metadata, context);
    assertEquals(1, links.size());
    assertEquals(url, links.get(0));
}

Example 53 with Attributes

use of org.xml.sax.Attributes in project Lucee by lucee.

the class aprint method _eo.

private static void _eo(PrintStream ps, Object o) {
    if (o instanceof Enumeration)
        _eo(ps, (Enumeration) o);
    else if (o instanceof Object[])
        _eo(ps, (Object[]) o);
    else if (o instanceof boolean[])
        _eo(ps, (boolean[]) o);
    else if (o instanceof byte[])
        _eo(ps, (byte[]) o);
    else if (o instanceof int[])
        _eo(ps, (int[]) o);
    else if (o instanceof float[])
        _eo(ps, (float[]) o);
    else if (o instanceof long[])
        _eo(ps, (long[]) o);
    else if (o instanceof double[])
        _eo(ps, (double[]) o);
    else if (o instanceof char[])
        _eo(ps, (char[]) o);
    else if (o instanceof short[])
        _eo(ps, (short[]) o);
    else if (o instanceof Set)
        _eo(ps, (Set) o);
    else if (o instanceof List)
        _eo(ps, (List) o);
    else if (o instanceof Map)
        _eo(ps, (Map) o);
    else if (o instanceof Collection)
        _eo(ps, (Collection) o);
    else if (o instanceof Iterator)
        _eo(ps, (Iterator) o);
    else if (o instanceof NamedNodeMap)
        _eo(ps, (NamedNodeMap) o);
    else if (o instanceof ResultSet)
        _eo(ps, (ResultSet) o);
    else if (o instanceof Node)
        _eo(ps, (Node) o);
    else if (o instanceof Throwable)
        _eo(ps, (Throwable) o);
    else if (o instanceof Attributes)
        _eo(ps, (Attributes) o);
    else if (o instanceof Cookie) {
        Cookie c = (Cookie) o;
        ps.println("Cookie(name:" + c.getName() + ";domain:" + c.getDomain() + ";maxage:" + c.getMaxAge() + ";path:" + c.getPath() + ";value:" + c.getValue() + ";version:" + c.getVersion() + ";secure:" + c.getSecure() + ")");
    } else if (o instanceof InputSource) {
        InputSource is = (InputSource) o;
        Reader r = is.getCharacterStream();
        try {
            ps.println(IOUtil.toString(is.getCharacterStream()));
        } catch (IOException e) {
        } finally {
            IOUtil.closeEL(r);
        }
    } else
        ps.println(o);
}

Example 54 with Attributes

use of org.xml.sax.Attributes in project spring-security-oauth by spring-projects.

the class SparklrServiceImpl method getSparklrPhotoIds.

public List<String> getSparklrPhotoIds() throws SparklrException {
    try {
        InputStream photosXML = new ByteArrayInputStream(sparklrRestTemplate.getForObject(URI.create(sparklrPhotoListURL), byte[].class));
        final List<String> photoIds = new ArrayList<String>();
        SAXParserFactory parserFactory = SAXParserFactory.newInstance();
        parserFactory.setValidating(false);
        parserFactory.setXIncludeAware(false);
        parserFactory.setNamespaceAware(false);
        SAXParser parser = parserFactory.newSAXParser();
        parser.parse(photosXML, new DefaultHandler() {

            @Override
            public void startElement(String uri, String localName, String qName, Attributes attributes) throws SAXException {
                if ("photo".equals(qName)) {
                    photoIds.add(attributes.getValue("id"));
                }
            }
        });
        return photoIds;
    } catch (IOException e) {
        throw new IllegalStateException(e);
    } catch (SAXException e) {
        throw new IllegalStateException(e);
    } catch (ParserConfigurationException e) {
        throw new IllegalStateException(e);
    }
}

Example 55 with Attributes

use of org.xml.sax.Attributes in project spring-security-oauth by spring-projects.

the class GoogleServiceImpl method getLastTenPicasaPictureURLs.

public List<String> getLastTenPicasaPictureURLs() {
    // byte[] bytes = getGoogleRestTemplate().getForObject(URI.create("https://picasaweb.google.com/data/feed/api/user/default"), byte[].class);
    byte[] bytes = getGoogleRestTemplate().getForObject(URI.create("https://picasaweb.google.com/data/feed/api/user/default?kind=photo&max-results=10"), byte[].class);
    InputStream photosXML = new ByteArrayInputStream(bytes);
    final List<String> photoUrls = new ArrayList<String>();
    SAXParserFactory parserFactory = SAXParserFactory.newInstance();
    parserFactory.setValidating(false);
    parserFactory.setXIncludeAware(false);
    parserFactory.setNamespaceAware(true);
    try {
        SAXParser parser = parserFactory.newSAXParser();
        parser.parse(photosXML, new DefaultHandler() {

            @Override
            public void startElement(String uri, String localName, String qName, Attributes attributes) throws SAXException {
                if ("http://search.yahoo.com/mrss/".equals(uri) && "thumbnail".equalsIgnoreCase(localName)) {
                    int width = 0;
                    try {
                        width = Integer.parseInt(attributes.getValue("width"));
                        if (width > 100 && width < 200) {
                            // just do the thumbnails that are between 100 and 200 px...
                            photoUrls.add(attributes.getValue("url"));
                        }
                    } catch (NumberFormatException e) {
                    // fall through...
                    }
                }
            }
        });
        return photoUrls;
    } catch (ParserConfigurationException e) {
        throw new IllegalStateException(e);
    } catch (SAXException e) {
        throw new IllegalStateException(e);
    } catch (IOException e) {
        throw new IllegalStateException(e);
    }
}