Package org.htmlparser

Examples of org.htmlparser.Node


        assertType("html tag", Html.class, node[0]);
        Html html = (Html) node[0];
        NodeList nodeList = new NodeList();
        html.collectInto(nodeList, TitleTag.class);
        assertEquals("nodelist size", 1, nodeList.size());
        Node node = nodeList.elementAt(0);
        assertType("expected title tag", TitleTag.class, node);
        TitleTag titleTag = (TitleTag) node;
        assertStringEquals("title", "Some Title", titleTag.getTitle());
    }
View Full Code Here


    protected URL[] extractLinks(String url) throws ParserException
    {
        Parser parser;
        Vector vector;
        Node node;
        LinkTag link;
        URL[] ret;

        parser = new Parser(url);
        parser.registerScanners();
View Full Code Here

                + "<td nowrap> &nbsp;\n"
                + "<a href=s/7509><b>Yahoo! Movies</b></a>"
                + "</td>",
            "http://www.yahoo.com");
        parser.registerScanners();
        Node linkNodes[] = parser.extractAllNodesThatAre(LinkTag.class);

        assertEquals("number of links", 2, linkNodes.length);
        LinkTag linkTag = (LinkTag) linkNodes[0];
        assertStringEquals(
            "Link",
View Full Code Here

        parseAndAssertNodeCount(1);
        assertTrue("Node should be a HTMLLinkTag", node[0] instanceof LinkTag);
        LinkTag linkTag = (LinkTag) node[0];
        // Get the image tag from the link

        Node insideNodes[] = new Node[10];
        int j = 0;
        for (SimpleNodeIterator e = linkTag.children(); e.hasMoreNodes();)
        {
            insideNodes[j++] = (Node) e.nextNode();
        }
View Full Code Here

    public void testParseParameter3() throws ParserException
    {
        Tag tag;
        EndTag etag;
        StringNode snode;
        Node node = null;
        String lin1 =
            "<DIV class=\"userData\" id=\"oLayout\" name=\"oLayout\"></DIV>";
        createParser(lin1);
        NodeIterator en = parser.elements();
        Hashtable h;
        boolean testEnd = true; // test end of first part
        String a, href, myPara, myValue, nice;

        try
        {

            if (en.hasMoreNodes())
            {
                node = en.nextNode();

                tag = (Tag) node;
                h = tag.getAttributes();
                String classValue = (String) h.get("CLASS");
                assertEquals(
                    "The class value should be ",
                    "userData",
                    classValue);
            }

        }
        catch (ClassCastException ce)
        {
            fail("Bad class element = " + node.getClass().getName());
        }
    }
View Full Code Here

    public void testParseParameterA() throws ParserException
    {
        Tag tag;
        EndTag etag;
        StringNode snode;
        Node node = null;
        String lin1 =
            "<A href=\"http://www.iki.fi/kaila\" myParameter yourParameter=\"Kaarle Kaaila\">Kaarle's homepage</A><p>Paragraph</p>";
        createParser(lin1);
        NodeIterator en = parser.elements();
        Hashtable h;
        boolean testEnd = true; // test end of first part
        String a, href, myPara, myValue, nice;

        try
        {

            if (en.hasMoreNodes())
            {
                node = en.nextNode();

                tag = (Tag) node;
                h = tag.getAttributes();
                a = (String) h.get(Tag.TAGNAME);
                href = (String) h.get("HREF");
                myValue = (String) h.get("MYPARAMETER");
                nice = (String) h.get("YOURPARAMETER");
                assertEquals("Link tag (A)", "A", a);
                assertEquals("href value", "http://www.iki.fi/kaila", href);
                assertEquals("myparameter value", "", myValue);
                assertEquals("yourparameter value", "Kaarle Kaaila", nice);
            }
            if (!(node instanceof LinkTag))
            {
                // linkscanner has eaten up this piece
                if (en.hasMoreNodes())
                {
                    node = en.nextNode();
                    snode = (StringNode) node;
                    assertEquals(
                        "Value of element",
                        snode.getText(),
                        "Kaarle's homepage");
                }

                if (en.hasMoreNodes())
                {
                    node = en.nextNode();
                    etag = (EndTag) node;
                    assertEquals("endtag of link", etag.getText(), "A");
                }
            }
            // testing rest
            if (en.hasMoreNodes())
            {
                node = en.nextNode();

                tag = (Tag) node;
                assertEquals("following paragraph begins", tag.getText(), "p");
            }
            if (en.hasMoreNodes())
            {
                node = en.nextNode();
                snode = (StringNode) node;
                assertEquals(
                    "paragraph contents",
                    snode.getText(),
                    "Paragraph");
            }
            if (en.hasMoreNodes())
            {
                node = en.nextNode();
                etag = (EndTag) node;
                assertEquals("paragrapg endtag", etag.getText(), "p");
            }

        }
        catch (ClassCastException ce)
        {
            fail("Bad class element = " + node.getClass().getName());
        }
    }
View Full Code Here

    public void testParseParameterG() throws ParserException
    {
        Tag tag;
        EndTag etag;
        StringNode snode;
        Node node = null;
        String lin1 =
            "<G href=\"http://www.iki.fi/kaila\" myParameter yourParameter=\"Kaila\">Kaarle's homepage</G><p>Paragraph</p>";
        createParser(lin1);
        NodeIterator en = parser.elements();
        Hashtable h;
        boolean testEnd = true; // test end of first part
        String a, href, myPara, myValue, nice;

        try
        {

            if (en.hasMoreNodes())
            {
                node = en.nextNode();

                tag = (Tag) node;
                h = tag.getAttributes();
                a = (String) h.get(Tag.TAGNAME);
                href = (String) h.get("HREF");
                myValue = (String) h.get("MYPARAMETER");
                nice = (String) h.get("YOURPARAMETER");
                assertEquals("The tagname should be G", a, "G");
                assertEquals(
                    "Check the http address",
                    href,
                    "http://www.iki.fi/kaila");
                assertEquals("myValue is empty", myValue, "");
                assertEquals("The second parameter value", nice, "Kaila");
            }
            if (en.hasMoreNodes())
            {
                node = en.nextNode();
                snode = (StringNode) node;
                assertEquals(
                    "The text of the element",
                    snode.getText(),
                    "Kaarle's homepage");
            }

            if (en.hasMoreNodes())
            {
                node = en.nextNode();
                etag = (EndTag) node;
                assertEquals("Endtag is G", etag.getText(), "G");
            }
            // testing rest
            if (en.hasMoreNodes())
            {
                node = en.nextNode();

                tag = (Tag) node;
                assertEquals("Follow up by p-tag", tag.getText(), "p");
            }
            if (en.hasMoreNodes())
            {
                node = en.nextNode();
                snode = (StringNode) node;
                assertEquals(
                    "Verify the paragraph text",
                    snode.getText(),
                    "Paragraph");
            }
            if (en.hasMoreNodes())
            {
                node = en.nextNode();
                etag = (EndTag) node;
                assertEquals("Still patragraph endtag", etag.getText(), "p");
            }

        }
        catch (ClassCastException ce)
        {
            fail("Bad class element = " + node.getClass().getName());
        }
    }
View Full Code Here

    public void testParseParameterSpace() throws ParserException
    {
        Tag tag;
        EndTag etag;
        StringNode snode;
        Node node = null;
        String lin1 = "<A yourParameter = \"Kaarle\">Kaarle's homepage</A>";
        createParser(lin1);
        NodeIterator en = parser.elements();
        Hashtable h;
        boolean testEnd = true; // test end of first part
        String a, href, myPara, myValue, nice;

        try
        {

            if (en.hasMoreNodes())
            {
                node = en.nextNode();

                tag = (Tag) node;
                h = tag.getAttributes();
                a = (String) h.get(Tag.TAGNAME);
                nice = (String) h.get("YOURPARAMETER");
                assertEquals("Link tag (A)", a, "A");
                assertEquals("yourParameter value", "Kaarle", nice);
            }
            if (!(node instanceof LinkTag))
            {
                // linkscanner has eaten up this piece
                if (en.hasMoreNodes())
                {
                    node = en.nextNode();
                    snode = (StringNode) node;
                    assertEquals(
                        "Value of element",
                        snode.getText(),
                        "Kaarle's homepage");
                }

                if (en.hasMoreNodes())
                {
                    node = en.nextNode();
                    etag = (EndTag) node;
                    assertEquals(
                        "Still patragraph endtag",
                        etag.getText(),
                        "A");
                }
            }
            // testing rest

        }
        catch (ClassCastException ce)
        {
            fail("Bad class element = " + node.getClass().getName());
        }
    }
View Full Code Here

    public void testWithoutParseParameter() throws ParserException
    {
        Tag tag;
        EndTag etag;
        StringNode snode;
        Node node = null;
        String testHTML =
            "<A href=\"http://www.iki.fi/kaila\" myParameter yourParameter=\"Kaarle\">Kaarle's homepage</A><p>Paragraph</p>";
        createParser(testHTML);
        NodeIterator en = parser.elements();
        String result = "";
        try
        {
            while (en.hasMoreNodes())
            {
                node = en.nextNode();
                result += node.toHtml();
            }
            String expected =
                "<A YOURPARAMETER=\"Kaarle\" MYPARAMETER=\"\" HREF=\"http://www.iki.fi/kaila\">Kaarle's homepage</A><P>Paragraph</P>";
            assertStringEquals(
                "Check collected contents to original",
                expected,
                result);
        }
        catch (ClassCastException ce)
        {
            fail("Bad class element = " + node.getClass().getName());
        }
    }
View Full Code Here

    public void testEmptyTagParseParameter() throws ParserException
    {
        Tag tag;
        EndTag etag;
        StringNode snode;
        Node node = null;
        String testHTML =
            "<INPUT name=\"foo\" value=\"foobar\" type=\"text\" />";

        createParser(testHTML);
        NodeIterator en = parser.elements();
        String result = "";
        try
        {
            while (en.hasMoreNodes())
            {
                node = en.nextNode();
                result = node.toHtml();
            }
            String expected =
                "<INPUT VALUE=\"foobar\" NAME=\"foo\" TYPE=\"text\"/>";
            assertStringEquals(
                "Check collected contents to original",
                expected,
                result);
        }
        catch (ClassCastException ce)
        {

            fail("Bad class element = " + node.getClass().getName());
        }
    }
View Full Code Here

TOP

Related Classes of org.htmlparser.Node

Copyright © 2018 www.massapicom. All rights reserved.
All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.