/// <summary>
/// Parses the GenBank features from the GenBank file.
/// </summary>
/// <param name="line">parse line</param>
/// <param name="sequence">The sequence.</param>
/// <param name="stream">The stream reader.</param>
/// <returns>The parsed line.</returns>
private string ParseFeatures(string line, ref Sequence sequence, StreamReader stream)
{
ILocationBuilder locBuilder = LocationBuilder;
if (locBuilder == null)
{
throw new InvalidOperationException(Properties.Resource.NullLocationBuild);
}
// set data indent for features
string lineData;
// The sub-items of a feature are referred to as qualifiers. These do not have unique
// keys, so they are stored as lists in the SubItems dictionary.
SequenceFeatures features = new SequenceFeatures();
IList <FeatureItem> featureList = features.All;
while (line != null)
{
string lineHeader = GetLineHeader(line, FeatureDataIndent);
if (String.IsNullOrEmpty(line) || lineHeader == "FEATURES")
{
line = GoToNextLine(line, stream);
continue;
}
if (line[0] != ' ')
{
// start of non-feature text
break;
}
if (lineHeader == null)
{
string message = Properties.Resource.GenbankEmptyFeature;
Trace.Report(message);
throw new InvalidDataException(message);
}
// check for multi-line location string
lineData = GetLineData(line, FeatureDataIndent);
string featureKey = lineHeader;
string location = lineData;
line = GoToNextLine(line, stream);
lineData = GetLineData(line, FeatureDataIndent);
lineHeader = GetLineHeader(line, FeatureDataIndent);
while ((line != null) && (lineHeader == string.Empty) &&
(lineData != string.Empty) && !lineData.StartsWith("/", StringComparison.Ordinal))
{
location += lineData;
GetLineData(line, FeatureDataIndent);
line = GoToNextLine(line, stream);
lineData = GetLineData(line, FeatureDataIndent);
lineHeader = GetLineHeader(line, FeatureDataIndent);
}
// create features as MetadataListItems
FeatureItem feature = new FeatureItem(featureKey, locBuilder.GetLocation(location));
// process the list of qualifiers, which are each in the form of
// /key="value"
string qualifierKey = string.Empty;
string qualifierValue = string.Empty;
bool quotationMarkStarted = false;
while (line != null)
{
lineData = GetLineData(line, FeatureDataIndent);
lineHeader = GetLineHeader(line, FeatureDataIndent);
if ((lineHeader == string.Empty) && (lineData != null))
{
// '/' denotes a continuation of the previous line
// Note that, if there are multiple lines of qualifierValue,
// sometimes a line break will happen such that a "/" which is
// part of the qualifierValue will start a continuation line.
// This is identified by verifying open and closing double quotes.
if (lineData.StartsWith("/", StringComparison.Ordinal) && !quotationMarkStarted)
{
// new qualifier; save previous if this isn't the first
if (!String.IsNullOrEmpty(qualifierKey))
{
AddQualifierToFeature(feature, qualifierKey, qualifierValue);
}
// set the key and value of this qualifier
int equalsIndex = lineData.IndexOf('=');
if (equalsIndex < 0)
{
// no value, just key (this is allowed, see NC_005213.gbk)
qualifierKey = lineData.Substring(1);
qualifierValue = string.Empty;
}
else if (equalsIndex > 0)
{
qualifierKey = lineData.Substring(1, equalsIndex - 1);
qualifierValue = lineData.Substring(equalsIndex + 1);
quotationMarkStarted = qualifierValue[0] == '"';
if (qualifierValue[qualifierValue.Length - 1] == '"')
{
quotationMarkStarted = false;
}
}
else
{
string message = String.Format(
CultureInfo.CurrentCulture,
Properties.Resource.GenbankInvalidFeature,
line);
Trace.Report(message);
throw new InvalidDataException(message);
}
}
else
{
// Continuation of previous line; "note" gets a line break, and
// everything else except "translation" and "transl_except" gets a
// space to separate words.
if (qualifierKey == "note")
{
qualifierValue += Environment.NewLine;
}
else if (qualifierKey != "translation" && qualifierKey != "transl_except")
{
qualifierValue += " ";
}
qualifierValue += lineData;
if (qualifierValue[qualifierValue.Length - 1] == '"')
{
quotationMarkStarted = false;
}
}
line = GoToNextLine(line, stream);
}
else if (line.StartsWith("\t", StringComparison.Ordinal))
{
// this seems to be data corruption; but BioPerl test set includes
// (old, 2003) NT_021877.gbk which has this problem, so we
// handle it
ApplicationLog.WriteLine("WARN: nonstandard line format at line {0}: '{1}'", lineNumber, line);
qualifierValue += " " + line.Trim();
if (qualifierValue[qualifierValue.Length - 1] == '"')
{
quotationMarkStarted = false;
}
line = GoToNextLine(line, stream);
}
else
{
break;
}
}
// add last qualifier
if (!String.IsNullOrEmpty(qualifierKey))
{
AddQualifierToFeature(feature, qualifierKey, qualifierValue);
}
// still add feature, even if it has no qualifiers
featureList.Add(StandardFeatureMap.GetStandardFeatureItem(feature));
}
if (featureList.Count > 0)
{
((GenBankMetadata)sequence.Metadata[Helper.GenBankMetadataKey]).Features = features;
}
return(line);
}
/// <summary>
/// Creates a new FeatureItem that is a copy of the current FeatureItem.
/// </summary>
/// <returns>A new FeatureItem that is a copy of this FeatureItem.</returns>
public FeatureItem Clone()
{
return(StandardFeatureMap.GetStandardFeatureItem(new FeatureItem(this)));
}
private void ParseFeatures(BioTextReader bioReader, ref Sequence sequence)
{
ILocationBuilder locBuilder = LocationBuilder;
if (locBuilder == null)
{
throw new InvalidOperationException(Resource.NullLocationBuild);
}
// set data indent for features
bioReader.DataIndent = _featureDataIndent;
// The sub-items of a feature are referred to as qualifiers. These do not have unique
// keys, so they are stored as lists in the SubItems dictionary.
SequenceFeatures features = new SequenceFeatures();
IList <FeatureItem> featureList = features.All;
while (bioReader.HasLines)
{
if (String.IsNullOrEmpty(bioReader.Line) || bioReader.LineHeader == "FEATURES")
{
bioReader.GoToNextLine();
continue;
}
if (bioReader.Line[0] != ' ')
{
// start of non-feature text
break;
}
if (!bioReader.LineHasHeader)
{
string message = Properties.Resource.GenbankEmptyFeature;
Trace.Report(message);
throw new InvalidDataException(message);
}
// check for multi-line location string
string featureKey = bioReader.LineHeader;
string location = bioReader.LineData;
bioReader.GoToNextLine();
while (bioReader.HasLines && !bioReader.LineHasHeader &&
bioReader.LineHasData && !bioReader.LineData.StartsWith("/", StringComparison.Ordinal))
{
location += bioReader.LineData;
bioReader.GoToNextLine();
}
// create features as MetadataListItems
FeatureItem feature = new FeatureItem(featureKey, locBuilder.GetLocation(location));
// process the list of qualifiers, which are each in the form of
// /key="value"
string qualifierKey = string.Empty;
string qualifierValue = string.Empty;
while (bioReader.HasLines)
{
if (!bioReader.LineHasHeader && bioReader.LineHasData)
{
// '/' denotes a continuation of the previous line
if (bioReader.LineData.StartsWith("/", StringComparison.Ordinal))
{
// new qualifier; save previous if this isn't the first
if (!String.IsNullOrEmpty(qualifierKey))
{
AddQualifierToFeature(feature, qualifierKey, qualifierValue);
}
// set the key and value of this qualifier
int equalsIndex = bioReader.LineData.IndexOf('=');
if (equalsIndex < 0)
{
// no value, just key (this is allowed, see NC_005213.gbk)
qualifierKey = bioReader.LineData.Substring(1);
qualifierValue = string.Empty;
}
else if (equalsIndex > 0)
{
qualifierKey = bioReader.LineData.Substring(1, equalsIndex - 1);
qualifierValue = bioReader.LineData.Substring(equalsIndex + 1);
}
else
{
string message = String.Format(
CultureInfo.CurrentCulture,
Properties.Resource.GenbankInvalidFeature,
bioReader.Line);
Trace.Report(message);
throw new InvalidDataException(message);
}
}
else
{
// Continuation of previous line; "note" gets a line break, and
// everything else except "translation" and "transl_except" gets a
// space to separate words.
if (qualifierKey == "note")
{
qualifierValue += Environment.NewLine;
}
else if (qualifierKey != "translation" && qualifierKey != "transl_except")
{
qualifierValue += " ";
}
qualifierValue += bioReader.LineData;
}
bioReader.GoToNextLine();
}
else if (bioReader.Line.StartsWith("\t", StringComparison.Ordinal))
{
// this seems to be data corruption; but BioPerl test set includes
// (old, 2003) NT_021877.gbk which has this problem, so we
// handle it
ApplicationLog.WriteLine("WARN: nonstandard line format at line {0}: '{1}'",
bioReader.LineNumber, bioReader.Line);
qualifierValue += " " + bioReader.Line.Trim();
bioReader.GoToNextLine();
}
else
{
break;
}
}
// add last qualifier
if (!String.IsNullOrEmpty(qualifierKey))
{
AddQualifierToFeature(feature, qualifierKey, qualifierValue);
}
// still add feature, even if it has no qualifiers
featureList.Add(StandardFeatureMap.GetStandardFeatureItem(feature));
}
if (featureList.Count > 0)
{
((GenBankMetadata)sequence.Metadata[Helper.GenBankMetadataKey]).Features = features;
}
}
