public static int ParseFile(string inputFile, string outFile)
{
if (CheckJson && File.Exists(outFile))
{
Logger.Info("JSON file {0} already exist", outFile);
return(0);
}
if (!File.Exists(inputFile))
{
Logger.Info("ERROR: {0} file NOT exists", inputFile);
return(0);
}
ColumnPredictor.InitializeIfNotAlready();
string logFile = Path.Combine(Path.GetDirectoryName(inputFile),
Path.GetFileName(inputFile) + ".log");
Logger.SetSecondLogFileName(Path.GetFullPath(logFile));
Logger.Info(String.Format("Parsing {0}", inputFile));
IAdapter adapter = GetAdapter(inputFile);
Logger.Info(String.Format("TablesCount = {0}", adapter.GetTablesCount()));
Logger.Info(String.Format("RowsCount = {0}", adapter.GetRowsCount()));
if (adapter.GetTablesCount() == 0 && !inputFile.EndsWith(".toloka_json"))
{
throw new SmartParserException("No tables found in document");
}
if (HtmlFileName != "")
{
adapter.WriteHtmlFile(HtmlFileName);
}
if (adapter.GetWorkSheetCount() > 1)
{
Logger.Info(String.Format("File has multiple ({0}) worksheets", adapter.GetWorkSheetCount()));
Declaration allDeclarations = null;
for (int sheetIndex = 0; sheetIndex < adapter.GetWorkSheetCount(); sheetIndex++)
{
adapter.SetCurrentWorksheet(sheetIndex);
try
{
if (DeclarationSerializer.SmartParserJsonFormat == SmartParserJsonFormatEnum.Disclosures)
{
var sheetDeclarations = BuildDeclarations(adapter, inputFile);
if (allDeclarations == null)
{
allDeclarations = sheetDeclarations;
}
else
{
allDeclarations.AddDeclarations(sheetDeclarations);
}
}
else
{
string curOutFile = outFile.Replace(".json", "_" + sheetIndex.ToString() + ".json");
Logger.Info(String.Format("Parsing worksheet {0} into file {1}", sheetIndex, curOutFile));
WriteOutputJson(inputFile, BuildDeclarations(adapter, inputFile), curOutFile);
}
}
catch (ColumnDetectorException)
{
Logger.Info(String.Format("Skipping empty sheet {0} (No headers found exception thrown)",
sheetIndex));
}
if (allDeclarations != null)
{
WriteOutputJson(inputFile, allDeclarations, outFile);
}
}
}
else
{
WriteOutputJson(inputFile, BuildDeclarations(adapter, inputFile), outFile);
}
return(0);
}
