/// <inheritdoc />
public IEnumerable <IRelatedEntity> GetRelatedEntities(string value)
{
if (string.IsNullOrWhiteSpace(value))
{
return(Enumerable.Empty <IRelatedEntity>());
}
var relatedEntities = new List <IRelatedEntity>();
foreach (var documentUdi in ParserUtilities.GetDocumentUdiFromText(value).ToList())
{
relatedEntities.Add(new RelatedDocumentEntity
{
RelatedEntityUdi = documentUdi
});
}
foreach (var mediaUdi in ParserUtilities.GetMediaUdiFromText(value).ToList())
{
relatedEntities.Add(new RelatedMediaEntity()
{
RelatedEntityUdi = mediaUdi
});
}
return(relatedEntities);
}
public void ParseSingleLine_ParsesTestData()
{
// ASSERT:
// Processor can parse individual line.
Assert.IsNotNull(sourceBytes);
Assert.IsTrue(sourceBytes.Any());
processor = new EbcdicParser();
int count = 0;
int position = 0;
while (position < sourceBytes.Length)
{
byte[] lineBytes = ParserUtilities.ReadBytesRange(
sourceBytes,
length: lineTemplate.LineSize,
startPosition: position);
position += lineTemplate.LineSize;
var parsedLine = processor.ParseSingleLine(lineTemplate, lineBytes);
if (count == 0)
{
AssertDataForTheFirstRecord(parsedLine);
}
Debug.WriteLine(parsedLine.ToXmlString());
count++;
}
Assert.IsTrue(count > 1);
}
internal virtual IReadOnlyList <Document> GetDocuments()
{
var list = new List <Document>();
if (SearchStrings.All(IsInvalidSearchString))
{
return(list);
}
using var luceneDirectory = FSDirectory.Open(LuceneIndexPath);
using var luceneDirectoryReader = DirectoryReader.Open(luceneDirectory);
var searcher = new IndexSearcher(luceneDirectoryReader);
foreach (var searchString in SearchStrings)
{
if (IsInvalidSearchString(searchString))
{
continue;
}
var preparedSearchTerm = PrepareSearchTerm(searchString);
var query = ParserUtilities.ParseQuery(preparedSearchTerm, queryParser);
var hits = searcher.Search(query, Filter, ResultsLimit).ScoreDocs;
list.AddRange(hits.Select(hit => searcher.Doc(hit.Doc)));
}
return(list);
}
public void GivenType_WhenGetPropertyNames_ThenReturnAllNames()
{
var parserUtilities = new ParserUtilities();
var properties = parserUtilities.GetPublicGetSetPropertyNames(typeof(SampleData));
ArePropertiesEqual(properties[0], "Id", "Given Id", 1);
ArePropertiesEqual(properties[1], "Name", "Name Property", 2);
ArePropertiesEqual(properties[2], "Description", "Description Property", 12);
}
public void ParseSingleLine_ParsesTestData()
{
// ASSERT:
// Processor can parse individual line.
var sourceBytes = TestData.GetSampleEbcidicDataWithFixedRecordLength();
var lineTemplate = TestData.GetLineTemplateToParseEbcidicDataWithFixedRecordLength();
Assert.IsNotNull(sourceBytes);
Assert.IsTrue(sourceBytes.Any());
var processor = new EbcdicParser();
int count = 0;
int position = 0;
while (position < sourceBytes.Length)
{
byte[] lineBytes = ParserUtilities.ReadBytesRange(
sourceBytes,
length: lineTemplate.LineSize,
startPosition: position);
position += lineTemplate.LineSize;
var parsedLine = processor.ParseSingleLine(lineTemplate, lineBytes);
if (count == 0)
{
Assert.IsTrue(parsedLine[TestData.FieldReservationNumber].Equals("04416365US2"));
Assert.IsTrue(parsedLine[TestData.FieldCheckInDate].Equals("2015-01-23"));
Assert.IsTrue(parsedLine[TestData.FieldCalcNetAmount].Equals("437.39"));
Assert.IsTrue(parsedLine[TestData.FieldCustomerName].Equals("KAY KENG,LOW"));
Assert.IsTrue(parsedLine[TestData.FieldRunDate].Equals("2015-01-28"));
Assert.IsTrue(parsedLine[TestData.FieldCurrencyConvRate].Equals("0.762728"));
Assert.IsTrue(parsedLine[TestData.FieldUsDollarAmountDue].Equals("220.26"));
Assert.IsTrue(parsedLine[TestData.FieldDateOfBirth].Equals("1972-08-09"));
}
Debug.WriteLine(parsedLine.ToXmlString());
count++;
}
Assert.IsTrue(count > 1);
}
public void ParserUtilities_ReadBytesRange_ReturnsSlicedArray()
{
// ASSERT:
// Method returns a slice of the input array
var input = sourceBytes;
byte[] result = ParserUtilities.ReadBytesRange(input, startPosition: 2, length: 5);
byte[] expected = { 2, 3, 4, 5, 6 };
Assert.IsNotNull(result);
Assert.IsTrue(result.Any());
Assert.IsTrue(result.Length == expected.Length);
for (int i = 0; i < result.Length; i++)
{
Assert.IsTrue(result[i] == expected[i]);
}
}
public void Used_For_Manual_Testing()
{
var lineBytes = ParserUtilities.ConvertHexStringToByteArray("F1-F5-F0-F1-F2-F3");
var lineTemplate = new LineTemplate(lineBytes.Length);
var fieldName = "TEST_ITEM";
lineTemplate.AddFieldTemplate(
new FieldTemplate(
fieldName: fieldName,
fieldType: FieldType.DateString,
startPosition: 0,
fieldSize: lineBytes.Length,
decimalPlaces: 0));
var processor = new EbcdicParser();
var parsedLine = processor.ParseSingleLine(lineTemplate, lineBytes);
var testOutput = parsedLine[fieldName];
Debug.WriteLine(parsedLine.ToXmlString());
}
/// <summary>
/// Sample EBCDIC binary data.
/// </summary>
/// <returns></returns>
public static byte[] GetSampleEbcidicDataWithFixedRecordLength()
{
var sb = new StringBuilder();
//First record
sb.Append("F0-F4-F4-F1-F6-F3-F6-F5-E4-E2-F2"); //RESERVATION-NUMBER (Type="String"): 04416365US2
sb.Append("F1-F5-F0-F1-F2-F3"); //CHECKIN-DATE (Type="DateString"): 2015-01-23
sb.Append("00-00-AA-DB"); //CALC-NET-AMOUNT (Type="BinaryNum"): 437.39
sb.Append("D2-C1-E8-40-D2-C5-D5-C7-6B-D3-D6-E6-40-40-40-40-40-40-40-40-40-40-40-40-40-40-40-40-40-40"); //CUSTOMER-NAME (Type="String"): KAY KENG,LOW
sb.Append("F0-F1-F2-F8-F1-F5"); //RUNDATE (Type="DateStringMMDDYY"): 2015-01-28
sb.Append("00-00-07-62-72-8F"); //CURRENCY-CONV-RATE (Type="Packed"): 0.762728
sb.Append("00-00-00-22-02-6C"); //US-DOLLAR-AMOUNT-DUE (Type="Packed"): 220.26
sb.Append("07-20-80-9C"); //DATE-OF-BIRTH (Type="PackedDate"): 1972-08-09
//Second record
sb.Append("F0-F4-F4-F8-F2-F9-F3-F3-E2-C7-F0"); //04482933SG0
sb.Append("F1-F5-F0-F1-F2-F4"); //2015-01-24
sb.Append("00-00-42-35"); //169.49
sb.Append("C8-D6-6B-D9-D6-E8-E2-E3-D6-D5-40-40-40-40-40-40-40-40-40-40-40-40-40-40-40-40-40-40-40-40"); //HO,ROYSTON
sb.Append("F0-F1-F2-F8-F1-F5"); //2015-01-28
sb.Append("00-00-07-62-72-8F"); //0.762728
sb.Append("00-00-00-12-08-1C"); //120.81
sb.Append("07-50-62-3C"); //1975-06-23
//Third record
sb.Append("F0-F4-F4-F2-F8-F3-F2-F6-E2-C7-F0"); //04428326SG0
sb.Append("F1-F5-F0-F1-F2-F6"); //2015-01-26
sb.Append("00-00-D2-A8"); //539.28
sb.Append("E5-C1-E9-C8-C1-D7-D7-C1-D3-D3-E8-6B-D4-C1-D5-E4-C5-D3-40-C6-D9-C1-D5-C3-C9-E2-40-40-40-40"); //VAZHAPPALLY,MANUEL FRANCIS
sb.Append("F0-F1-F2-F8-F1-F5"); //2015-01-28
sb.Append("00-00-07-62-72-8F"); //0.762728
sb.Append("00-00-00-38-48-9C"); //384.89
sb.Append("06-70-50-2C"); //1967-05-02
byte[] ebcdicBytes = ParserUtilities.ConvertHexStringToByteArray(sb.ToString().Replace("-", ""));
return(ebcdicBytes);
}
/// <inheritdoc />
public IEnumerable <IRelatedEntity> GetRelatedEntities(string value)
{
if (!string.IsNullOrWhiteSpace(value))
{
var entities = new List <IRelatedEntity>();
var model = XDocument.Parse(value).Root;
if (model != null)
{
try
{
var socialImage = model.Descendants("socialImage").FirstOrDefault();
if (socialImage == null)
{
return(Enumerable.Empty <IRelatedEntity>());
}
if (string.IsNullOrEmpty(socialImage.Value))
{
return(Enumerable.Empty <IRelatedEntity>());
}
foreach (var documentUdi in ParserUtilities.GetMediaUdiFromText(socialImage.Value).ToList())
{
entities.Add(new RelatedMediaEntity
{
RelatedEntityUdi = documentUdi,
});
}
return(entities);
}
catch { }
}
}
return(Enumerable.Empty <IRelatedEntity>());
}
public void ParserUtilities_ConvertHexStringToByteArray_ReturnsByteArray()
{
// ASSERT:
// Method converts HEX string into byte array
string[] inputs =
{
"01-42-CC-6A",
"01 42 CC 6A",
"0142CC6A",
"0142cc6a"
};
byte[] expected = { 1, 66, 204, 106 };
foreach (var input in inputs)
{
byte[] result = ParserUtilities.ConvertHexStringToByteArray(input);
Assert.IsNotNull(result);
Assert.IsTrue(result.Any());
Assert.IsTrue(result.Length == expected.Length);
Assert.IsTrue(expected.SequenceEqual(result));
}
}
public void ParseTags_LessThanRequiredColumns_ThrowException()
{
Assert.Throws <UserErrorException>(() => ParserUtilities.ParseTags("#CHROM\tPOS\tREF", "#CHROM", 4));
}
public void ParseGenomeAssembly_IncorrectFormat_ThrowException()
{
Assert.Throws <UserErrorException>(() => ParserUtilities.ParseGenomeAssembly("#assembly-GRCh38", _allowedGenomeAssemblies));
}
public void ParseGenomeAssembly_UnsupportedAssembly_ThrowException()
{
Assert.Throws <UserErrorException>(() => ParserUtilities.ParseGenomeAssembly("#assembly=hg19", _allowedGenomeAssemblies));
}
public void CheckPrefix_InvalidPrefix_ThrowException()
{
Assert.Throws <UserErrorException>(() => ParserUtilities.CheckPrefix("invalidPrefix=someValue", "expectedPrefix"));
}
internal override string PrepareSearchTerm(string inputString)
{
return(ParserUtilities.PrepareSearchTerm(inputString));
}
public void ParseCategories_InvalidValue_ThrowException()
{
Assert.Throws <UserErrorException>(() => ParserUtilities.ParseCategories("#categories\tWOW", 1, 1, null));
}
public void ParseTypes_InvalidType_ThrowException(string type)
{
string typeLine = $"#type\t.\t.\t.\t{type}";
Assert.Throws <UserErrorException>(() => ParserUtilities.ParseTypes(typeLine, 4, 1));
}
public void ParseTitle_IncorrectFormat()
{
Assert.Throws <UserErrorException>(() => ParserUtilities.ParseTitle("#title:customSA"));
}
public void ParseTypes_ValidType_Pass(string type)
{
string typeLine = $"#type\t.\t.\t.\t{type}";
ParserUtilities.ParseTypes(typeLine, 4, 1);
}
public void ParseTitle_Conflict_JsonTag()
{
Assert.Throws <UserErrorException>(() => ParserUtilities.ParseTitle("#title=topmed"));
}