private void PopulateCVEnumData()
{
// WARNING: '-' must be at beginning or end, in middle it must be escaped, or it is interpreted as a range
var invalidSymbols = @" @/[():^?*+<=!~`#$%&{}|;'.,>\";
var invalidSymbolsEscaped = System.Text.RegularExpressions.Regex.Escape(invalidSymbols);
// add all whitespace matching, manually escape the ']', since above call doesn't
var invalidSymbolsRegex = @"[\]\s" + invalidSymbolsEscaped + "\\-\\\"]";
// Add "CVID_Unknown" to the list first
var unknown = new OBO_Term
{
Id = "??:0000000",
Name = "CVID_Unknown",
EnumName = "CVID_Unknown",
Def = "CVID_Unknown [Unknown]",
IsObsolete = false
};
_cvEnumData.Add("CVID_Unknown", unknown);
_cvMapData.Add("??", new Dictionary <string, OBO_Term>()
{
{ "CVID_Unknown", unknown }
});
const string obsolete = "_OBSOLETE";
foreach (var obo in _allObo)
{
var id = obo.Id;
var parent = new Dictionary <string, OBO_Term>();
_cvMapData.Add(id, parent);
foreach (var term in obo.Terms.Values)
{
var name = id + "_";
if (!string.IsNullOrWhiteSpace(term.Id_Namespace))
{
name = term.Id_Namespace + "_";
}
//name += term.Name.Replace(' ', '_');
name += System.Text.RegularExpressions.Regex.Replace(term.Name, invalidSymbolsRegex, "_");
//name += System.Text.RegularExpressions.Regex.Replace(term.Name.Replace(' ', '_'), invalidSymbolsRegex, "_");
if (term.IsObsolete)
{
name += obsolete;
}
var tName = name;
var counter = 0;
while (_cvEnumData.ContainsKey(name))
{
counter++;
name = tName + counter;
}
_cvEnumData.Add(name, term);
parent.Add(name, term);
term.EnumName = name;
}
}
}
public void Read(string url)
{
Console.WriteLine("Downloading data from " + url);
var fileData = (new WebClient()).DownloadString(url);
FileData = new OBO_File(url);
using (var reader = new StringReader(fileData))
{
var line = reader.ReadLine();
while (string.IsNullOrWhiteSpace(line) && line != null)
{
line = reader.ReadLine();
}
var type = "header";
var data = new List <KeyValuePair <string, string> >();
while (reader.Peek() != -1 && !string.IsNullOrWhiteSpace(line))
{
if (line.StartsWith("["))
{
type = line;
line = reader.ReadLine();
while (string.IsNullOrWhiteSpace(line) && line != null)
{
line = reader.ReadLine();
}
}
data.Clear();
while (!string.IsNullOrWhiteSpace(line) && !line.StartsWith("["))
{
var keyStop = line.IndexOf(':');
var valueStart = line[keyStop + 1] == ' ' ? keyStop + 2 : keyStop + 1;
data.Add(new KeyValuePair <string, string>(line.Substring(0, keyStop), line.Substring(valueStart)));
line = reader.ReadLine();
while (string.IsNullOrWhiteSpace(line) && line != null)
{
line = reader.ReadLine();
}
}
switch (type.ToLower())
{
case "header":
FileData.Header = new OBO_Header(data);
break;
case "[term]":
var term = new OBO_Term(data);
if (FileData.Terms.ContainsKey(term.Id))
{
Console.WriteLine("Warning: Duplicate term id found");
Console.WriteLine("\tFirst term: \t\"" + FileData.Terms[term.Id].Id + "\": \"" +
FileData.Terms[term.Id].Def + "\".");
Console.WriteLine("\tConflict term: \t\"" + term.Id + "\": \"" + term.Def + "\".");
Console.WriteLine("\tChanging conflict id to \"" + term.Id + "_\"");
term.Id += "_";
}
FileData.Terms.Add(term.Id, term);
break;
case "[typedef]":
var typeDef = new OBO_Typedef(data);
FileData.Typedefs.Add(typeDef.Id, typeDef);
break;
case "[instance]":
var instance = new OBO_Instance(data);
FileData.Instances.Add(instance.Id, instance);
break;
}
}
}
if (FileData.Terms.Count > 0)
{
var namespaces = FileData.Terms.Values.Select(x => x.Id_Namespace).Distinct();
foreach (var ns in namespaces)
{
if (FileData.IsGeneratedId)
{
FileData.Id = ns;
}
else if (!FileData.Id.Equals(ns))
{
FileData.AdditionalNamespaces.Add(ns);
}
}
}
foreach (var import in FileData.Header.Import)
{
var reader = new OBO_Reader();
reader.Read(import);
ImportedFileData.Add(reader.FileData);
ImportedFileData.AddRange(reader.ImportedFileData);
}
}
