Examples with PropertyValue org.apache.poi.hsmf.datatypes.PropertyValue used on opensource projects

Example 1 with PropertyValue

use of org.apache.poi.hsmf.datatypes.PropertyValue in project poi by apache.

the class MAPIMessage method guess7BitEncoding.

/**
    * Tries to identify the correct encoding for 7-bit (non-unicode)
    *  strings in the file.
    * <p>Many messages store their strings as unicode, which is
    *  nice and easy. Some use one-byte encodings for their
    *  strings, but don't always store the encoding anywhere
    *  helpful in the file.</p>
    * <p>This method checks for codepage properties, and failing that
    *  looks at the headers for the message, and uses these to 
    *  guess the correct encoding for your file.</p>
    * <p>Bug #49441 has more on why this is needed</p>
    */
public void guess7BitEncoding() {
    // First choice is a codepage property
    for (MAPIProperty prop : new MAPIProperty[] { MAPIProperty.MESSAGE_CODEPAGE, MAPIProperty.INTERNET_CPID }) {
        List<PropertyValue> val = mainChunks.getProperties().get(prop);
        if (val != null && val.size() > 0) {
            int codepage = ((LongPropertyValue) val.get(0)).getValue();
            try {
                String encoding = CodePageUtil.codepageToEncoding(codepage, true);
                set7BitEncoding(encoding);
                return;
            } catch (UnsupportedEncodingException e) {
                logger.log(POILogger.WARN, "Invalid codepage ID ", codepage, " set for the message via ", prop, ", ignoring");
            }
        }
    }
    // Second choice is a charset on a content type header
    try {
        String[] headers = getHeaders();
        if (headers != null && headers.length > 0) {
            // Look for a content type with a charset
            Pattern p = Pattern.compile("Content-Type:.*?charset=[\"']?([^;'\"]+)[\"']?", Pattern.CASE_INSENSITIVE);
            for (String header : headers) {
                if (header.startsWith("Content-Type")) {
                    Matcher m = p.matcher(header);
                    if (m.matches()) {
                        // Found it! Tell all the string chunks
                        String charset = m.group(1);
                        if (!charset.equalsIgnoreCase("utf-8")) {
                            set7BitEncoding(charset);
                        }
                        return;
                    }
                }
            }
        }
    } catch (ChunkNotFoundException e) {
    }
    // Nothing suitable in the headers, try HTML
    try {
        String html = getHtmlBody();
        if (html != null && html.length() > 0) {
            // Look for a content type in the meta headers
            Pattern p = Pattern.compile("<META\\s+HTTP-EQUIV=\"Content-Type\"\\s+CONTENT=\"text/html;\\s+charset=(.*?)\"");
            Matcher m = p.matcher(html);
            if (m.find()) {
                // Found it! Tell all the string chunks
                String charset = m.group(1);
                set7BitEncoding(charset);
                return;
            }
        }
    } catch (ChunkNotFoundException e) {
    }
}

Example 2 with PropertyValue

use of org.apache.poi.hsmf.datatypes.PropertyValue in project poi by apache.

the class HSMFDump method dump.

public void dump(PrintStream out) throws IOException {
    ChunkGroup[] chunkGroups = POIFSChunkParser.parse(fs);
    for (ChunkGroup chunks : chunkGroups) {
        out.println(chunks.getClass().getSimpleName());
        for (Chunk chunk : chunks.getChunks()) {
            MAPIProperty attr = MAPIProperty.get(chunk.getChunkId());
            if (chunk instanceof PropertiesChunk) {
                PropertiesChunk props = (PropertiesChunk) chunk;
                out.println("   Properties - " + props.getProperties().size() + ":");
                for (MAPIProperty prop : props.getProperties().keySet()) {
                    out.println("       * " + prop);
                    for (PropertyValue v : props.getValues(prop)) {
                        out.println("        = " + v);
                    }
                }
            } else {
                String idName = attr.id + " - " + attr.name;
                if (attr == MAPIProperty.UNKNOWN) {
                    idName = chunk.getChunkId() + " - (unknown)";
                }
                out.println("   " + idName + " - " + chunk.getType().getName());
                out.println("       " + chunk);
            }
        }
        out.println();
    }
}

Example 3 with PropertyValue

use of org.apache.poi.hsmf.datatypes.PropertyValue in project poi by apache.

the class TestFixedSizedProperties method testPropertyValueTypes.

/**
    * Check we find properties of a variety of different types
    */
@Test
public void testPropertyValueTypes() throws Exception {
    Chunks mainChunks = mapiMessageSucceeds.getMainChunks();
    // Ask to have the values looked up
    Map<MAPIProperty, List<PropertyValue>> props = mainChunks.getProperties();
    HashSet<Class<? extends PropertyValue>> seenTypes = new HashSet<Class<? extends PropertyValue>>();
    for (List<PropertyValue> pvs : props.values()) {
        for (PropertyValue pv : pvs) {
            seenTypes.add(pv.getClass());
        }
    }
    assertTrue(seenTypes.toString(), seenTypes.size() > 3);
    assertTrue(seenTypes.toString(), seenTypes.contains(LongPropertyValue.class));
    assertTrue(seenTypes.toString(), seenTypes.contains(TimePropertyValue.class));
    assertFalse(seenTypes.toString(), seenTypes.contains(ChunkBasedPropertyValue.class));
    // Ask for the raw values
    seenTypes.clear();
    for (PropertyValue pv : mainChunks.getRawProperties().values()) {
        seenTypes.add(pv.getClass());
    }
    assertTrue(seenTypes.toString(), seenTypes.size() > 3);
    assertTrue(seenTypes.toString(), seenTypes.contains(LongPropertyValue.class));
    assertTrue(seenTypes.toString(), seenTypes.contains(TimePropertyValue.class));
    assertTrue(seenTypes.toString(), seenTypes.contains(ChunkBasedPropertyValue.class));
}

Example 4 with PropertyValue

use of org.apache.poi.hsmf.datatypes.PropertyValue in project poi by apache.

the class TestFixedSizedProperties method testClientSubmitTime.

/**
    * Will be based on the ClientSubmit time
    */
@Test
public void testClientSubmitTime() throws Exception {
    // Check via the message date
    Calendar clientSubmitTime = mapiMessageSucceeds.getMessageDate();
    assertEquals("Fri, 22 Jun 2012 18:32:54", messageDateFormat.format(clientSubmitTime.getTime()));
    // Fetch the property value directly
    Map<MAPIProperty, List<PropertyValue>> props = mapiMessageSucceeds.getMainChunks().getProperties();
    List<PropertyValue> pv = props.get(MAPIProperty.CLIENT_SUBMIT_TIME);
    assertNotNull(pv);
    assertEquals(1, pv.size());
    clientSubmitTime = (Calendar) pv.get(0).getValue();
    assertEquals("Fri, 22 Jun 2012 18:32:54", messageDateFormat.format(clientSubmitTime.getTime()));
}

Example 5 with PropertyValue

use of org.apache.poi.hsmf.datatypes.PropertyValue in project tika by apache.

the class OutlookExtractor method guess7BitEncoding.

/**
     * Tries to identify the correct encoding for 7-bit (non-unicode)
     *  strings in the file.
     * <p>Many messages store their strings as unicode, which is
     *  nice and easy. Some use one-byte encodings for their
     *  strings, but don't always store the encoding anywhere
     *  helpful in the file.</p>
     * <p>This method checks for codepage properties, and failing that
     *  looks at the headers for the message, and uses these to
     *  guess the correct encoding for your file.</p>
     * <p>Bug #49441 has more on why this is needed</p>
     * <p>This is taken verbatim from POI (TIKA-1238)
     * as a temporary workaround to prevent unsupported encoding exceptions</p>
     */
private void guess7BitEncoding(MAPIMessage msg) {
    Chunks mainChunks = msg.getMainChunks();
    //sanity check
    if (mainChunks == null) {
        return;
    }
    Map<MAPIProperty, List<PropertyValue>> props = mainChunks.getProperties();
    if (props != null) {
        // First choice is a codepage property
        for (MAPIProperty prop : new MAPIProperty[] { MAPIProperty.MESSAGE_CODEPAGE, MAPIProperty.INTERNET_CPID }) {
            List<PropertyValue> val = props.get(prop);
            if (val != null && val.size() > 0) {
                int codepage = ((PropertyValue.LongPropertyValue) val.get(0)).getValue();
                String encoding = null;
                try {
                    encoding = CodePageUtil.codepageToEncoding(codepage, true);
                } catch (UnsupportedEncodingException e) {
                //swallow
                }
                if (tryToSet7BitEncoding(msg, encoding)) {
                    return;
                }
            }
        }
    }
    // Second choice is a charset on a content type header
    try {
        String[] headers = msg.getHeaders();
        if (headers != null && headers.length > 0) {
            // Look for a content type with a charset
            Pattern p = Pattern.compile("Content-Type:.*?charset=[\"']?([^;'\"]+)[\"']?", Pattern.CASE_INSENSITIVE);
            for (String header : headers) {
                if (header.startsWith("Content-Type")) {
                    Matcher m = p.matcher(header);
                    if (m.matches()) {
                        // Found it! Tell all the string chunks
                        String charset = m.group(1);
                        if (tryToSet7BitEncoding(msg, charset)) {
                            return;
                        }
                    }
                }
            }
        }
    } catch (ChunkNotFoundException e) {
    }
    // meta header if there is no other information?
    try {
        String html = msg.getHtmlBody();
        if (html != null && html.length() > 0) {
            Charset charset = null;
            try {
                charset = detector.detect(new ByteArrayInputStream(html.getBytes(UTF_8)), EMPTY_METADATA);
            } catch (IOException e) {
            //swallow
            }
            if (charset != null && tryToSet7BitEncoding(msg, charset.name())) {
                return;
            }
        }
    } catch (ChunkNotFoundException e) {
    }
    //absolute last resort, try charset detector
    StringChunk text = mainChunks.getTextBodyChunk();
    if (text != null) {
        CharsetDetector detector = new CharsetDetector();
        detector.setText(text.getRawValue());
        CharsetMatch match = detector.detect();
        if (match != null && match.getConfidence() > 35 && tryToSet7BitEncoding(msg, match.getName())) {
            return;
        }
    }
}