Package org.apache.tika.parser

Examples of org.apache.tika.parser.Parser

44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
*/
public class OutlookParserTest {

    @Test
    public void testOutlookParsing() throws Exception {
        Parser parser = new AutoDetectParser(); // Should auto-detect!
        ContentHandler handler = new BodyContentHandler();
        Metadata metadata = new Metadata();

        InputStream stream = OutlookParserTest.class.getResourceAsStream(
                "/test-documents/test-outlook.msg");
        try {
            parser.parse(stream, handler, metadata, new ParseContext());
        } finally {
            stream.close();
        }

        assertEquals(
View Full Code Here
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
public class RFC822ParserTest {

    @Test
    public void testSimple() {
        Parser parser = new RFC822Parser();
        Metadata metadata = new Metadata();
        InputStream stream = getStream("test-documents/testRFC822");
        ContentHandler handler = mock(DefaultHandler.class);

        try {
            parser.parse(stream, handler, metadata, new ParseContext());
            verify(handler).startDocument();
            //just one body
            verify(handler).startElement(eq(XHTMLContentHandler.XHTML), eq("p"), eq("p"), any(Attributes.class));
            verify(handler).endElement(XHTMLContentHandler.XHTML, "p", "p");
            //no multi-part body parts
View Full Code Here
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
        }
    }

    @Test
    public void testMultipart() {
        Parser parser = new RFC822Parser();
        Metadata metadata = new Metadata();
        InputStream stream = getStream("test-documents/testRFC822-multipart");
        ContentHandler handler = mock(XHTMLContentHandler.class);

        try {
            parser.parse(stream, handler, metadata, new ParseContext());
            verify(handler).startDocument();
            //4 body-part divs -- two outer bodies and two inner bodies
            verify(handler, times(4)).startElement(eq(XHTMLContentHandler.XHTML), eq("div"), eq("div"), any(Attributes.class));
            verify(handler, times(4)).endElement(XHTMLContentHandler.XHTML, "div", "div");
            //5 paragraph elements, 4 for body-parts and 1 for encompassing message
            verify(handler, times(5)).startElement(eq(XHTMLContentHandler.XHTML), eq("p"), eq("p"), any(Attributes.class));
            verify(handler, times(5)).endElement(XHTMLContentHandler.XHTML, "p", "p");
            verify(handler).endDocument();
        } catch (Exception e) {
            fail("Exception thrown: " + e.getMessage());
        }
       
        //repeat, this time looking at content
        parser = new RFC822Parser();
        metadata = new Metadata();
        stream = getStream("test-documents/testRFC822-multipart");
        handler = new BodyContentHandler();
        try {
            parser.parse(stream, handler, metadata, new ParseContext());
            //tests correct decoding of quoted printable text, including UTF-8 bytes into Unicode
            String bodyText = handler.toString();
            assertTrue(bodyText.contains("body 1"));
            assertTrue(bodyText.contains("body 2"));
            assertFalse(bodyText.contains("R0lGODlhNgE8AMQAA")); //part of encoded gif
View Full Code Here
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
     *
     * @see <a href="https://issues.apache.org/jira/browse/TIKA-197">TIKA-197</a>
     */
    @Test
    public void testMultipleCopies() throws Exception {
        Parser parser = new AutoDetectParser();
        ContentHandler handler = new BodyContentHandler();
        Metadata metadata = new Metadata();

        InputStream stream = OutlookParserTest.class.getResourceAsStream(
                "/test-documents/testMSG.msg");
        try {
            parser.parse(stream, handler, metadata, new ParseContext());
        } finally {
            stream.close();
        }

        assertEquals(
View Full Code Here
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
     *  properly reported
     */
    @Test
    public void testParsingErrorInForkedParserShouldBeReported() throws Exception {
        BrokenParser brokenParser = new BrokenParser();
        Parser parser = new ForkParser(ForkParser.class.getClassLoader(), brokenParser);
        InputStream stream = getClass().getResourceAsStream("/test-documents/testTXT.txt");
       
        // With a serializable error, we'll get that back
        try {
            ContentHandler output = new BodyContentHandler();
            ParseContext context = new ParseContext();
            parser.parse(stream, output, new Metadata(), context);
            fail("Expected TikaException caused by Error");
        } catch (TikaException e) {
            assertEquals(brokenParser.err, e.getCause());
        }
       
View Full Code Here
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
     *
     * @see <a href="https://issues.apache.org/jira/browse/TIKA-395">TIKA-395</a>
     */
    @Test
    public void testOutlookNew() throws Exception {
        Parser parser = new AutoDetectParser();
        ContentHandler handler = new BodyContentHandler();
        Metadata metadata = new Metadata();

        InputStream stream = OutlookParserTest.class.getResourceAsStream(
                "/test-documents/test-outlook2003.msg");
        try {
            parser.parse(stream, handler, metadata, new ParseContext());
        } finally {
            stream.close();
        }

        assertEquals(
View Full Code Here
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
        }
    }

    @Test
    public void testQuotedPrintable() {
        Parser parser = new RFC822Parser();
        Metadata metadata = new Metadata();
        InputStream stream = getStream("test-documents/testRFC822_quoted");
        ContentHandler handler = new BodyContentHandler();

        try {
            parser.parse(stream, handler, metadata, new ParseContext());
            //tests correct decoding of quoted printable text, including UTF-8 bytes into Unicode
            String bodyText = handler.toString();
            assertTrue(bodyText.contains("D\u00FCsseldorf has non-ascii."));
            assertTrue(bodyText.contains("Lines can be split like this."));
            assertTrue(bodyText.contains("Spaces at the end of a line \r\nmust be encoded.\r\n"));
View Full Code Here
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
        }
    }

    @Test
    public void testBase64() {
        Parser parser = new RFC822Parser();
        Metadata metadata = new Metadata();
        InputStream stream = getStream("test-documents/testRFC822_base64");
        ContentHandler handler = new BodyContentHandler();

        try {
            parser.parse(stream, handler, metadata, new ParseContext());
            //tests correct decoding of base64 text, including ISO-8859-1 bytes into Unicode
            assertTrue(handler.toString().contains("Here is some text, with international characters, voil\u00E0!"));
        } catch (Exception e) {
            fail("Exception thrown: " + e.getMessage());
        }
View Full Code Here
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
        assertTrue(content.contains("Navigation Pane"));
    }
    
    @Test
    public void testOutlookHTMLVersion() throws Exception {
        Parser parser = new AutoDetectParser();
        Metadata metadata = new Metadata();
      
        // Check the HTML version
        StringWriter sw = new StringWriter();
        SAXTransformerFactory factory = (SAXTransformerFactory)
                 SAXTransformerFactory.newInstance();
        TransformerHandler handler = factory.newTransformerHandler();
        handler.getTransformer().setOutputProperty(OutputKeys.METHOD, "xml");
        handler.getTransformer().setOutputProperty(OutputKeys.INDENT, "yes");
        handler.setResult(new StreamResult(sw));

        InputStream stream = OutlookParserTest.class.getResourceAsStream(
               "/test-documents/testMSG_chinese.msg");
        try {
           parser.parse(stream, handler, metadata, new ParseContext());
        } finally {
           stream.close();
        }
        
        // As the HTML version should have been processed, ensure
View Full Code Here
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
        }
    }

    @Test
    public void testI18NHeaders() {
        Parser parser = new RFC822Parser();
        Metadata metadata = new Metadata();
        InputStream stream = getStream("test-documents/testRFC822_i18nheaders");
        ContentHandler handler = mock(DefaultHandler.class);

        try {
            parser.parse(stream, handler, metadata, new ParseContext());
            //tests correct decoding of internationalized headers, both
            //quoted-printable (Q) and Base64 (B).
            assertEquals("Keld J\u00F8rn Simonsen <keld@dkuug.dk>",
                    metadata.get(TikaCoreProperties.CREATOR));
            assertEquals("If you can read this you understand the example.",
View Full Code Here
TOP

Related Classes of org.apache.tika.parser.Parser

  • bixo.parser.TikaCallableTest
  • com.baasbox.controllers.File
  • com.cloudera.cdk.morphline.solrcell.SolrCellBuilder$SolrCell
  • com.digitalpebble.behemoth.tika.TikaProcessor
  • com.findwise.hydra.stage.tika.SimpleFetchingTikaStageTest
  • com.ikanow.infinit.e.harvest.extraction.document.file.FileHarvester
  • com.ikanow.infinit.e.harvest.extraction.text.legacy.TextExtractorTika
  • com.nidhinova.tika.server.TikaService
  • com.tamingtext.tika.TikaTest
  • ipcress.Extraction
    • 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.