Examples with CharsetToolkit org.gephi.utils.CharsetToolkit used on opensource projects

Example 1 with CharsetToolkit

use of org.gephi.utils.CharsetToolkit in project gephi by gephi.

the class SpreadsheetUtils method configureCSVParser.

public static CSVParser configureCSVParser(File file, Character fieldSeparator, Charset charset, boolean withFirstRecordAsHeader) throws IOException {
    if (fieldSeparator == null) {
        fieldSeparator = ',';
    }
    CSVFormat csvFormat = CSVFormat.DEFAULT.withDelimiter(fieldSeparator).withEscape('\\').withIgnoreEmptyLines(true).withNullString("").withIgnoreSurroundingSpaces(true).withTrim(true);
    if (withFirstRecordAsHeader) {
        csvFormat = csvFormat.withFirstRecordAsHeader().withAllowMissingColumnNames(false).withIgnoreHeaderCase(false);
    } else {
        csvFormat = csvFormat.withHeader((String[]) null).withSkipHeaderRecord(false);
    }
    boolean hasBOM = false;
    try (FileInputStream is = new FileInputStream(file)) {
        CharsetToolkit charsetToolkit = new CharsetToolkit(is);
        hasBOM = charsetToolkit.hasUTF8Bom() || charsetToolkit.hasUTF16BEBom() || charsetToolkit.hasUTF16LEBom();
    } catch (IOException e) {
    // NOOP
    }
    FileInputStream fileInputStream = new FileInputStream(file);
    InputStreamReader is = new InputStreamReader(fileInputStream, charset);
    if (hasBOM) {
        try {
            is.read();
        } catch (IOException e) {
        // should never happen, as a file with no content
        // but with a BOM has at least one char
        }
    }
    return new CSVParser(is, csvFormat);
}

Example 2 with CharsetToolkit

use of org.gephi.utils.CharsetToolkit in project gephi by gephi.

the class ImportUtils method getTextReader.

/**
 * Returns a <code>LineNumberReader</code> for <code>inputStream</code>. The
 * stream must be a character stream. The charset is detected automatically.
 *
 * @param stream the stream that is to be read
 * @return a reader for the character stream
 * @throws IOException if the stream can't be read
 */
public static LineNumberReader getTextReader(InputStream stream) throws IOException {
    LineNumberReader reader;
    CharsetToolkit charsetToolkit = new CharsetToolkit(stream);
    reader = (LineNumberReader) charsetToolkit.getReader();
    return reader;
}

Example 3 with CharsetToolkit

use of org.gephi.utils.CharsetToolkit in project gephi by gephi.

the class ImporterSpreadsheetCSV method autoDetectCharset.

private void autoDetectCharset() {
    // Try to auto-detect the charset:
    try {
        FileInputStream is = new FileInputStream(file);
        CharsetToolkit charsetToolkit = new CharsetToolkit(is);
        charsetToolkit.setDefaultCharset(Charset.forName("UTF-8"));
        charset = charsetToolkit.getCharset();
    } catch (Exception ex) {
    }
}

Example 4 with CharsetToolkit

use of org.gephi.utils.CharsetToolkit in project gephi by gephi.

the class DesktopImportControllerUI method importFiles.

private void importFiles(final Reader[] readers, final FileImporter[] importers, FileObject[] fileObjects) {
    try {
        File[] files = new File[readers.length];
        Map<ImporterUI, List<FileImporter>> importerUIs = new HashMap<>();
        for (int i = 0; i < importers.length; i++) {
            FileImporter importer = importers[i];
            ImporterUI ui = controller.getUI(importer);
            if (ui != null) {
                List<FileImporter> l = importerUIs.get(ui);
                if (l == null) {
                    l = new ArrayList<>();
                    importerUIs.put(ui, l);
                }
                l.add(importer);
            }
            if (importer instanceof FileImporter.FileAware) {
                try (Reader reader = readers[i]) {
                    File file = null;
                    if (fileObjects != null) {
                        file = FileUtil.toFile(fileObjects[i]);
                    }
                    if (file == null) {
                        // There is no source file but the importer needs it, create temporary copy:
                        String fileName = "tmp_file" + 1;
                        String charset = "UTF-8";
                        if (fileObjects != null && fileObjects[i] != null) {
                            // Netbeans FileUtil.toFile bug returning null?? Try to recover:
                            fileName = fileObjects[i].getNameExt();
                            CharsetToolkit charsetToolkit = new CharsetToolkit(fileObjects[i].getInputStream());
                            charset = charsetToolkit.getCharset().name();
                        }
                        file = TempDirUtils.createTempDir().createFile(fileName);
                        try (FileOutputStream fos = new FileOutputStream(file)) {
                            FileUtil.copy(new ReaderInputStream(reader, charset), fos);
                        }
                    }
                    files[i] = file;
                    ((FileImporter.FileAware) importer).setFile(file);
                }
            }
        }
        for (Map.Entry<ImporterUI, List<FileImporter>> entry : importerUIs.entrySet()) {
            ImporterUI ui = entry.getKey();
            String title = NbBundle.getMessage(DesktopImportControllerUI.class, "DesktopImportControllerUI.file.ui.dialog.title", ui.getDisplayName());
            JPanel panel = ui.getPanel();
            FileImporter[] fi = (FileImporter[]) entry.getValue().toArray((FileImporter[]) Array.newInstance(entry.getValue().get(0).getClass(), 0));
            ui.setup(fi);
            if (panel != null) {
                final DialogDescriptor dd = new DialogDescriptor(panel, title);
                if (panel instanceof ValidationPanel) {
                    ValidationPanel vp = (ValidationPanel) panel;
                    vp.addChangeListener(new ChangeListener() {

                        @Override
                        public void stateChanged(ChangeEvent e) {
                            dd.setValid(!((ValidationPanel) e.getSource()).isProblem());
                        }
                    });
                }
                Object result = DialogDisplayer.getDefault().notify(dd);
                if (!result.equals(NotifyDescriptor.OK_OPTION)) {
                    ui.unsetup(false);
                    return;
                }
            }
            if (ui instanceof ImporterUI.WithWizard) {
                boolean finishedOk = showWizard(ui, ((ImporterUI.WithWizard) ui).getWizardDescriptor());
                if (!finishedOk) {
                    ui.unsetup(false);
                    return;
                }
            }
            ui.unsetup(true);
        }
        final List<Container> results = new ArrayList<>();
        for (int i = 0; i < importers.length; i++) {
            doImport(results, readers[i], files[i], importers[i]);
        }
        executor.execute(null, new Runnable() {

            @Override
            public void run() {
                if (!results.isEmpty()) {
                    finishImport(results.toArray(new Container[0]));
                }
            }
        });
    } catch (Exception ex) {
        Exceptions.printStackTrace(ex);
    }
}