protected TableDefinition getFicstarTableDefinition(DataFile dataFile) throws IOException {
RetailerSite retailerSite = dataFile.getRetailerSite();
File schemaFile = this.getFicstarSchemaFile();
DataImportConfig dataImportConfig =
applicationContext.getBean("dataImportConfig", DataImportConfig.class);
dataImportConfig.setConfigFilePath(schemaFile.getAbsolutePath());
dataImportConfig.afterPropertiesSet();
// now the config is ready.
Schema schema = dataImportConfig.getSchema();
TableDefinition tableDefinition = schema.getDefinitionByDestination(dataFile.getImportType());
Assert.notNull(
tableDefinition,
"cannot find tableDefinition for :="
+ dataFile.getImportType()
+ " in schema :="
+ schemaFile.getAbsolutePath());
String csvFilePath =
FilenameUtils.normalize(
configurationService.getFileSystem().getFileSystemAsString()
+ "/"
+ dataFile.getFilePath());
tableDefinition.setProperty("path", csvFilePath);
tableDefinition.setProperty("relativePath", dataFile.getFilePath());
tableDefinition.setProperty("isPathAbsolute", "true");
// some feeds like google dont have siteName == RetailerSiteName
tableDefinition.setProperty("columns.siteName.defaultValue", retailerSite.getSiteName());
return tableDefinition;
}
protected int countRows(RetailerSite retailerSite, DataFile dataFile) {
File filePath = null;
try {
filePath = configurationService.getFileSystem().getDirectory(dataFile.getFilePath(), true);
TableDefinition tableDefinition = this.getTableDefinition(retailerSite, dataFile);
String encoding = getDataImportEncoding(tableDefinition, filePath);
LineNumberReader lnr =
new LineNumberReader(new InputStreamReader(new FileInputStream(filePath), encoding));
lnr.readLine(); // Skip header
int numRows = 0;
while (lnr.readLine() != null) {
numRows++;
}
lnr.close();
return numRows;
} catch (Exception e) {
logger.error(String.format("Failed to reading %s data file", filePath), e);
dataFile.setStatus(FileStatus.ERROR_READ);
dataFileRepository.update(dataFile);
return -1;
}
}
public PreprocessorHandlerContext(
RetailerSite retailerSite,
DataFile dataFile,
File fullDataFilePath,
TableDefinition tableDefinition,
String encoding,
String separator,
String quoteChar)
throws IOException, CSVReaderException {
this.dataFile = dataFile;
this.srcFile = dataFile.getSrcFile();
this.dataFilePath = dataFile.getFilePath();
this.fullDataFilePath = fullDataFilePath;
this.base = FilenameUtils.getBaseName(this.fullDataFilePath.getName());
int pos = this.base.indexOf('-');
if (pos == -1) {
throw new RuntimeException("Protocol error, invalid feed file name: " + this.base);
}
this.type = this.base.substring(0, pos);
this.base = this.base.substring(pos + 1);
this.ext = FilenameUtils.getExtension(this.fullDataFilePath.getName());
this.path = this.fullDataFilePath.getParentFile().getPath();
this.encoding = encoding;
this.separator = separator;
this.quoteChar = quoteChar;
this.reader =
new CVSTranslatorReader(this.fullDataFilePath, this.encoding, this.separator, quoteChar);
}
/*
* (non-Javadoc)
*
* @see
* com.dell.acs.dataimport.preprocessor.PreprocessorHandler#preprocess(com
* .dell.acs.persistence.domain.DataFile)
*/
@Override
public void preprocess(DataFile dataFile) {
RetailerSite retailerSite =
retailerSiteRepository.getByName(dataFile.getRetailerSite().getSiteName(), true);
if (configurationService.getBooleanProperty(
DataFilesDownloadManager.class, retailerSite.getSiteName() + ".blocking", false)) {
int blockSize =
configurationService.getIntegerProperty(
DataFilesDownloadManager.class, retailerSite.getSiteName() + ".blockSize", -1);
if (blockSize == -1) {
throw new RuntimeException(
"Protocol Error, Invalid configuration for block size discovered!");
}
this.blockDataFile(dataFile, retailerSite, blockSize);
} else if (configurationService.getBooleanProperty(
DataFilesDownloadManager.class, retailerSite.getSiteName() + ".splitting", false)) {
int splitSize =
configurationService.getIntegerProperty(
DataFilesDownloadManager.class, retailerSite.getSiteName() + ".splitSize", -1);
if (splitSize == -1) {
throw new RuntimeException(
"Protocol Error, Invalid configuration for split size discovered!");
}
this.splitDataFile(dataFile, retailerSite, splitSize);
} else {
this.noSplit(dataFile, retailerSite);
}
}
protected void noSplit(DataFile dataFile, RetailerSite retailerSite) {
if (dataFile.getImportType().compareTo(ImportType.images_zip.getTableName()) != 0) {
int numRows = this.countRows(retailerSite, dataFile);
dataFile.setHost(null);
dataFile.setNumRows(numRows);
this.getDataFileStatisticService().initializeStatistics(dataFile);
this.getDataFileStatisticService().startImport(dataFile);
} else {
dataFile.setNumRows(1);
dataFile.setHost(null);
}
dataFile.setStatus(FileStatus.IN_QUEUE);
dataFileRepository.update(dataFile);
}
private void blockDataFile(DataFile dataFile, RetailerSite retailerSite, int blockSize) {
File filePath = null;
PreprocessorHandlerContext context = null;
try {
dataFile.setStartTime(new Date());
// TODO srfisk 2012/07/31 Only split CSV, might split images zip
// when they are very large.
if (dataFile.getImportType().compareTo(ImportType.images_zip.getTableName()) != 0) {
filePath = configurationService.getFileSystem().getDirectory(dataFile.getFilePath(), true);
TableDefinition tableDefinition = this.getTableDefinition(retailerSite, dataFile);
ImportType importType = ImportType.getTypeFromTableName(dataFile.getImportType());
String encoding = this.getDataImportEncoding(tableDefinition, filePath);
String separator =
Character.toString(
(char) Integer.valueOf(tableDefinition.getProperty("separator", "9")).intValue());
String quotechar = this.getQuoteCharacter(tableDefinition);
int blockCount = 1;
int totalRowCount = 0;
boolean noSplit = false;
boolean done = false;
context =
this.createContext(
retailerSite, dataFile, filePath, tableDefinition, encoding, separator, quotechar);
while (!done) {
context.startWriters(blockCount);
context.writeHeader();
Row row = null;
while ((row = context.reader.readNext()) != null && context.rowCount < blockSize) {
context.writeRow(row);
}
context.closeWriters();
blockCount++;
if (context.isEmptyBlock()) {
done = true;
continue;
}
context.createDataFiles(
retailerSite, importType.getTableName(), importType.getPriority());
context.inQueueDataFiles();
}
dataFile.setHost(null);
dataFile.setNumRows(totalRowCount);
if (!noSplit) {
dataFile.setEndDate(new Date());
dataFile.setEndTime(new Date());
dataFile.setStatus(this.getEndStatus());
} else {
dataFile.setStatus(FileStatus.IN_QUEUE);
}
} else {
dataFile.setHost(null);
dataFile.setStatus(FileStatus.IN_QUEUE);
}
dataFileRepository.update(dataFile);
} catch (IOException e) {
logger.error(String.format("Failed to reading %s data file", filePath), e);
dataFile.setStatus(FileStatus.ERROR_READ);
dataFileRepository.update(dataFile);
} catch (CSVReaderException e) {
logger.error(String.format("Failed to reading %s data file", filePath), e);
dataFile.setStatus(FileStatus.ERROR_WRITE);
dataFileRepository.update(dataFile);
} catch (CSVWriterException e) {
logger.error(String.format("Failed to reading %s data file", filePath), e);
dataFile.setStatus(FileStatus.ERROR_READ);
dataFileRepository.update(dataFile);
} catch (Throwable t) {
logger.error(String.format("Failed to unknown %s data file", filePath), t);
dataFile.setStatus(FileStatus.ERROR_UNKNOWN);
dataFileRepository.update(dataFile);
} finally {
if (context != null) {
try {
context.cleanup();
} catch (IOException e) {
// Ignore, did everything we could to close it.
}
}
}
}
