Examples of BodyContentHandler

org.apache.tika.sax.BodyContentHandler
Content handler decorator that only passes everything inside the XHTML <body/> tag to the underlying handler. Note that the <body/> tag itself is not passed on.

Examples of org.apache.tika.sax.BodyContentHandler

            + "<body></body></html>";


        Metadata metadata = new Metadata();
        new HtmlParser().parse (
                new ByteArrayInputStream(test.getBytes("UTF-8")),
                new BodyContentHandler(),  metadata, new ParseContext());
        assertEquals("UTF-8", metadata.get(Metadata.CONTENT_ENCODING));


        metadata = new Metadata();
        metadata.set(Metadata.CONTENT_TYPE, "text/html; charset=ISO-8859-1");
        new HtmlParser().parse (
                new ByteArrayInputStream(test.getBytes("UTF-8")),
                new BodyContentHandler(),  metadata, new ParseContext());
        assertEquals("ISO-8859-1", metadata.get(Metadata.CONTENT_ENCODING));
    }

View Full Code Here

Examples of org.apache.tika.sax.BodyContentHandler

        String test = "<html><title>Simple Content</title><body></body></html>";
        Metadata metadata = new Metadata();
        metadata.add(Metadata.CONTENT_LANGUAGE, "en");
        new HtmlParser().parse (
                new ByteArrayInputStream(test.getBytes("UTF-8")),
                new BodyContentHandler(),  metadata, new ParseContext());


        assertEquals("en", metadata.get(Metadata.CONTENT_LANGUAGE));
    }

View Full Code Here

Examples of org.apache.tika.sax.BodyContentHandler

            + "<title>the name is \u00e1ndre</title>"
            + "</head><body></body></html>";
        Metadata metadata = new Metadata();
        new HtmlParser().parse (
                new ByteArrayInputStream(test1.getBytes("UTF-8")),
                new BodyContentHandler(),  metadata, new ParseContext());
        assertEquals("ISO-8859-1", metadata.get(Metadata.CONTENT_ENCODING));


        // Some HTML pages have errors like ';;' versus '; ' as separator
        String test2 =
            "<html><head><meta http-equiv=\"content-type\""
            + " content=\"text/html;;charset=ISO-8859-1\" />"
            + "<title>the name is \u00e1ndre</title>"
            + "</head><body></body></html>";
        metadata = new Metadata();
        new HtmlParser().parse (
                new ByteArrayInputStream(test2.getBytes("UTF-8")),
                new BodyContentHandler(),  metadata, new ParseContext());
        assertEquals("ISO-8859-1", metadata.get(Metadata.CONTENT_ENCODING));
    }

View Full Code Here

Examples of org.apache.tika.sax.BodyContentHandler

            + "<body></body></html>";


        Metadata metadata = new Metadata();
        new HtmlParser().parse (
                new ByteArrayInputStream(test.getBytes("UTF-8")),
                new BodyContentHandler(),  metadata, new ParseContext());
        assertEquals("UTF-8", metadata.get(Metadata.CONTENT_ENCODING));


        metadata = new Metadata();
        metadata.set(Metadata.CONTENT_TYPE, "charset=ISO-8859-1;text/html");
        new HtmlParser().parse (
                new ByteArrayInputStream(test.getBytes("UTF-8")),
                new BodyContentHandler(),  metadata, new ParseContext());
        assertEquals("ISO-8859-1", metadata.get(Metadata.CONTENT_ENCODING));
    }

View Full Code Here

Examples of org.apache.tika.sax.BodyContentHandler

    public void testMetaHttpEquivWithLotsOfPreambleText() throws Exception {
        String path = "/test-documents/big-preamble.html";
        Metadata metadata = new Metadata();
        new HtmlParser().parse(
                HtmlParserTest.class.getResourceAsStream(path),
                new BodyContentHandler(),  metadata, new ParseContext());


        assertEquals("windows-1251", metadata.get(Metadata.CONTENT_ENCODING));
    }

View Full Code Here

Examples of org.apache.tika.sax.BodyContentHandler


        try {
            Metadata metadata = new Metadata();
            metadata.set(Metadata.RESOURCE_NAME_KEY, tp.resourceStatedName);
            metadata.set(Metadata.CONTENT_TYPE, tp.statedType);
            ContentHandler handler = new BodyContentHandler();
            new AutoDetectParser().parse(input, handler, metadata);


            assertEquals("Bad content type: " + tp,
                    tp.realType, metadata.get(Metadata.CONTENT_TYPE));


            assertTrue("Expected content not found: " + tp,
                    handler.toString().contains(tp.expectedContentFragment));
        } finally {
            input.close();
        }
    }

View Full Code Here

Examples of org.apache.tika.sax.BodyContentHandler

    public void testZipBombPrevention() throws Exception {
        InputStream tgz = AutoDetectParserTest.class.getResourceAsStream(
                "/test-documents/TIKA-216.tgz");
        try {
            Metadata metadata = new Metadata();
            ContentHandler handler = new BodyContentHandler(-1);
            new AutoDetectParser().parse(tgz, handler, metadata);
            fail("Zip bomb was not detected");
        } catch (TikaException e) {
            // expected
        } finally {

View Full Code Here

Examples of org.apache.tika.sax.BodyContentHandler

            }
            // Use the delegate parser to parse the compressed document
            parser.parse(
                    new CloseShieldInputStream(stream),
                    new EmbeddedContentHandler(
                            new BodyContentHandler(xhtml)),
                    entrydata, context);
        } finally {
            stream.close();
        }
    }

View Full Code Here

Examples of org.apache.tika.sax.BodyContentHandler

                    try {
                        // Use the delegate parser to parse this entry
                        parser.parse(
                                new CloseShieldInputStream(archive),
                                new EmbeddedContentHandler(
                                        new BodyContentHandler(xhtml)),
                                        entrydata, context);
                    } catch (TikaException e) {
                        // Could not parse the entry, just skip the content
                    }
                    xhtml.endElement("div");

View Full Code Here

Examples of org.apache.tika.sax.BodyContentHandler

            }
        };
    }


    private ContentHandler getTextContentHandler(Writer writer) {
        return new BodyContentHandler(writer);
    }

View Full Code Here

0 1 2 3 4 5

TOP

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.