Search in sources :

Example 1 with EncodingType

use of org.pentaho.di.trans.steps.fileinput.text.EncodingType in project pentaho-kettle by pentaho.

the class TextFileInputDialog method getFirst.

// Get the first x lines
private List<String> getFirst(int nrlines, boolean skipHeaders) throws KettleException {
    TextFileInputMeta meta = new TextFileInputMeta();
    getInfo(meta, true);
    FileInputList textFileList = meta.getFileInputList(transMeta);
    InputStream fi;
    CompressionInputStream f = null;
    StringBuilder lineStringBuilder = new StringBuilder(256);
    int fileFormatType = meta.getFileFormatTypeNr();
    List<String> retval = new ArrayList<>();
    if (textFileList.nrOfFiles() > 0) {
        FileObject file = textFileList.getFile(0);
        try {
            fi = KettleVFS.getInputStream(file);
            CompressionProvider provider = CompressionProviderFactory.getInstance().createCompressionProviderInstance(meta.content.fileCompression);
            f = provider.createInputStream(fi);
            InputStreamReader reader;
            if (meta.getEncoding() != null && meta.getEncoding().length() > 0) {
                reader = new InputStreamReader(f, meta.getEncoding());
            } else {
                reader = new InputStreamReader(f);
            }
            EncodingType encodingType = EncodingType.guessEncodingType(reader.getEncoding());
            int linenr = 0;
            int maxnr = nrlines + (meta.content.header ? meta.content.nrHeaderLines : 0);
            if (skipHeaders) {
                // Skip the header lines first if more then one, it helps us position
                if (meta.content.layoutPaged && meta.content.nrLinesDocHeader > 0) {
                    int skipped = 0;
                    String line = TextFileInputUtils.getLine(log, reader, encodingType, fileFormatType, lineStringBuilder);
                    while (line != null && skipped < meta.content.nrLinesDocHeader - 1) {
                        skipped++;
                        line = TextFileInputUtils.getLine(log, reader, encodingType, fileFormatType, lineStringBuilder);
                    }
                }
                // Skip the header lines first if more then one, it helps us position
                if (meta.content.header && meta.content.nrHeaderLines > 0) {
                    int skipped = 0;
                    String line = TextFileInputUtils.getLine(log, reader, encodingType, fileFormatType, lineStringBuilder);
                    while (line != null && skipped < meta.content.nrHeaderLines - 1) {
                        skipped++;
                        line = TextFileInputUtils.getLine(log, reader, encodingType, fileFormatType, lineStringBuilder);
                    }
                }
            }
            String line = TextFileInputUtils.getLine(log, reader, encodingType, fileFormatType, lineStringBuilder);
            while (line != null && (linenr < maxnr || nrlines == 0)) {
                retval.add(line);
                linenr++;
                line = TextFileInputUtils.getLine(log, reader, encodingType, fileFormatType, lineStringBuilder);
            }
        } catch (Exception e) {
            throw new KettleException(BaseMessages.getString(PKG, "TextFileInputDialog.Exception.ErrorGettingFirstLines", "" + nrlines, file.getName().getURI()), e);
        } finally {
            try {
                if (f != null) {
                    f.close();
                }
            } catch (Exception e) {
            // Ignore errors
            }
        }
    }
    return retval;
}
Also used : KettleException(org.pentaho.di.core.exception.KettleException) InputStreamReader(java.io.InputStreamReader) CompressionInputStream(org.pentaho.di.core.compress.CompressionInputStream) CompressionInputStream(org.pentaho.di.core.compress.CompressionInputStream) InputStream(java.io.InputStream) ArrayList(java.util.ArrayList) EncodingType(org.pentaho.di.trans.steps.fileinput.text.EncodingType) ValueMetaString(org.pentaho.di.core.row.value.ValueMetaString) KettleException(org.pentaho.di.core.exception.KettleException) IOException(java.io.IOException) CompressionProvider(org.pentaho.di.core.compress.CompressionProvider) TextFileInputMeta(org.pentaho.di.trans.steps.fileinput.text.TextFileInputMeta) FileObject(org.apache.commons.vfs2.FileObject) FileInputList(org.pentaho.di.core.fileinput.FileInputList)

Example 2 with EncodingType

use of org.pentaho.di.trans.steps.fileinput.text.EncodingType in project pentaho-kettle by pentaho.

the class TextFileInputDialog method getCSV.

// Get the data layout
private void getCSV() {
    TextFileInputMeta meta = new TextFileInputMeta();
    getInfo(meta, true);
    // CSV without separator defined
    if (meta.content.fileType.equalsIgnoreCase("CSV") && (meta.content.separator == null || meta.content.separator.isEmpty())) {
        MessageBox mb = new MessageBox(shell, SWT.OK | SWT.ICON_ERROR);
        mb.setMessage(BaseMessages.getString(PKG, "TextFileInput.Exception.NoSeparator"));
        mb.setText(BaseMessages.getString(PKG, "TextFileInputDialog.DialogTitle"));
        mb.open();
        return;
    }
    TextFileInputMeta previousMeta = (TextFileInputMeta) meta.clone();
    FileInputList textFileList = meta.getFileInputList(transMeta);
    InputStream fileInputStream;
    CompressionInputStream inputStream = null;
    StringBuilder lineStringBuilder = new StringBuilder(256);
    int fileFormatType = meta.getFileFormatTypeNr();
    String delimiter = transMeta.environmentSubstitute(meta.content.separator);
    String enclosure = transMeta.environmentSubstitute(meta.content.enclosure);
    String escapeCharacter = transMeta.environmentSubstitute(meta.content.escapeCharacter);
    if (textFileList.nrOfFiles() > 0) {
        int clearFields = meta.content.header ? SWT.YES : SWT.NO;
        int nrInputFields = meta.inputFields.length;
        if (nrInputFields > 0) {
            MessageBox mb = new MessageBox(shell, SWT.YES | SWT.NO | SWT.CANCEL | SWT.ICON_QUESTION);
            mb.setMessage(BaseMessages.getString(PKG, "TextFileInputDialog.ClearFieldList.DialogMessage"));
            mb.setText(BaseMessages.getString(PKG, "TextFileInputDialog.ClearFieldList.DialogTitle"));
            clearFields = mb.open();
            if (clearFields == SWT.CANCEL) {
                return;
            }
        }
        try {
            wFields.table.removeAll();
            FileObject fileObject = textFileList.getFile(0);
            fileInputStream = KettleVFS.getInputStream(fileObject);
            Table table = wFields.table;
            CompressionProvider provider = CompressionProviderFactory.getInstance().createCompressionProviderInstance(meta.content.fileCompression);
            inputStream = provider.createInputStream(fileInputStream);
            InputStreamReader reader;
            if (meta.getEncoding() != null && meta.getEncoding().length() > 0) {
                reader = new InputStreamReader(inputStream, meta.getEncoding());
            } else {
                reader = new InputStreamReader(inputStream);
            }
            EncodingType encodingType = EncodingType.guessEncodingType(reader.getEncoding());
            // Scan the header-line, determine fields...
            String line = TextFileInputUtils.getLine(log, reader, encodingType, fileFormatType, lineStringBuilder);
            if (line != null) {
                // Estimate the number of input fields...
                // Chop up the line using the delimiter
                String[] fields = TextFileInputUtils.guessStringsFromLine(transMeta, log, line, meta, delimiter, enclosure, escapeCharacter);
                for (int i = 0; i < fields.length; i++) {
                    String field = fields[i];
                    if (field == null || field.length() == 0 || !meta.content.header) {
                        field = "Field" + (i + 1);
                    } else {
                        // Trim the field
                        field = Const.trim(field);
                        // Replace all spaces & - with underscore _
                        field = Const.replace(field, " ", "_");
                        field = Const.replace(field, "-", "_");
                    }
                    TableItem item = new TableItem(table, SWT.NONE);
                    item.setText(1, field);
                    // The default type is String...
                    item.setText(2, "String");
                }
                wFields.setRowNums();
                wFields.optWidth(true);
                // Copy it...
                getInfo(meta, true);
                // Sample a few lines to determine the correct type of the fields...
                String shellText = BaseMessages.getString(PKG, "TextFileInputDialog.LinesToSample.DialogTitle");
                String lineText = BaseMessages.getString(PKG, "TextFileInputDialog.LinesToSample.DialogMessage");
                EnterNumberDialog end = new EnterNumberDialog(shell, 100, shellText, lineText);
                int samples = end.open();
                if (samples >= 0) {
                    getInfo(meta, true);
                    TextFileCSVImportProgressDialog pd = new TextFileCSVImportProgressDialog(shell, meta, transMeta, reader, samples, clearFields == SWT.YES);
                    String message = pd.open();
                    if (message != null) {
                        wFields.removeAll();
                        // OK, what's the result of our search?
                        getData(meta);
                        // 
                        if (clearFields == SWT.NO) {
                            getFieldsData(previousMeta, true);
                            wFields.table.setSelection(previousMeta.inputFields.length, wFields.table.getItemCount() - 1);
                        }
                        wFields.removeEmptyRows();
                        wFields.setRowNums();
                        wFields.optWidth(true);
                        EnterTextDialog etd = new EnterTextDialog(shell, BaseMessages.getString(PKG, "TextFileInputDialog.ScanResults.DialogTitle"), BaseMessages.getString(PKG, "TextFileInputDialog.ScanResults.DialogMessage"), message, true);
                        etd.setReadOnly();
                        etd.open();
                    }
                }
            } else {
                MessageBox mb = new MessageBox(shell, SWT.OK | SWT.ICON_ERROR);
                mb.setMessage(BaseMessages.getString(PKG, "TextFileInputDialog.UnableToReadHeaderLine.DialogMessage"));
                mb.setText(BaseMessages.getString(PKG, "System.Dialog.Error.Title"));
                mb.open();
            }
        } catch (IOException e) {
            new ErrorDialog(shell, BaseMessages.getString(PKG, "TextFileInputDialog.IOError.DialogTitle"), BaseMessages.getString(PKG, "TextFileInputDialog.IOError.DialogMessage"), e);
        } catch (KettleException e) {
            new ErrorDialog(shell, BaseMessages.getString(PKG, "System.Dialog.Error.Title"), BaseMessages.getString(PKG, "TextFileInputDialog.ErrorGettingFileDesc.DialogMessage"), e);
        } finally {
            try {
                if (inputStream != null) {
                    inputStream.close();
                }
            } catch (Exception e) {
            // Ignore errors
            }
        }
    } else {
        MessageBox mb = new MessageBox(shell, SWT.OK | SWT.ICON_ERROR);
        mb.setMessage(BaseMessages.getString(PKG, "TextFileInputDialog.NoValidFileFound.DialogMessage"));
        mb.setText(BaseMessages.getString(PKG, "System.Dialog.Error.Title"));
        mb.open();
    }
}
Also used : KettleException(org.pentaho.di.core.exception.KettleException) Table(org.eclipse.swt.widgets.Table) InputStreamReader(java.io.InputStreamReader) CompressionInputStream(org.pentaho.di.core.compress.CompressionInputStream) CompressionInputStream(org.pentaho.di.core.compress.CompressionInputStream) InputStream(java.io.InputStream) TableItem(org.eclipse.swt.widgets.TableItem) EncodingType(org.pentaho.di.trans.steps.fileinput.text.EncodingType) ErrorDialog(org.pentaho.di.ui.core.dialog.ErrorDialog) ValueMetaString(org.pentaho.di.core.row.value.ValueMetaString) IOException(java.io.IOException) KettleException(org.pentaho.di.core.exception.KettleException) IOException(java.io.IOException) MessageBox(org.eclipse.swt.widgets.MessageBox) CompressionProvider(org.pentaho.di.core.compress.CompressionProvider) TextFileInputMeta(org.pentaho.di.trans.steps.fileinput.text.TextFileInputMeta) EnterTextDialog(org.pentaho.di.ui.core.dialog.EnterTextDialog) FileObject(org.apache.commons.vfs2.FileObject) EnterNumberDialog(org.pentaho.di.ui.core.dialog.EnterNumberDialog) FileInputList(org.pentaho.di.core.fileinput.FileInputList)

Aggregations

IOException (java.io.IOException)2 InputStream (java.io.InputStream)2 InputStreamReader (java.io.InputStreamReader)2 FileObject (org.apache.commons.vfs2.FileObject)2 CompressionInputStream (org.pentaho.di.core.compress.CompressionInputStream)2 CompressionProvider (org.pentaho.di.core.compress.CompressionProvider)2 KettleException (org.pentaho.di.core.exception.KettleException)2 FileInputList (org.pentaho.di.core.fileinput.FileInputList)2 ValueMetaString (org.pentaho.di.core.row.value.ValueMetaString)2 EncodingType (org.pentaho.di.trans.steps.fileinput.text.EncodingType)2 TextFileInputMeta (org.pentaho.di.trans.steps.fileinput.text.TextFileInputMeta)2 ArrayList (java.util.ArrayList)1 MessageBox (org.eclipse.swt.widgets.MessageBox)1 Table (org.eclipse.swt.widgets.Table)1 TableItem (org.eclipse.swt.widgets.TableItem)1 EnterNumberDialog (org.pentaho.di.ui.core.dialog.EnterNumberDialog)1 EnterTextDialog (org.pentaho.di.ui.core.dialog.EnterTextDialog)1 ErrorDialog (org.pentaho.di.ui.core.dialog.ErrorDialog)1