public static Triple Split(String line, Action<string> errorAction)
{
Triple triple = new Triple();
TRIPLE_TAGS CurrentTag = TRIPLE_TAGS.SUBJECT;
// <http://en.wikipedia.org/wiki/Tony_Benn> <http://purl.org/dc/elements/1.1/title> "Tony Benn" .
// <http://en.wikipedia.org/wiki/Tony_Benn> <http://purl.org/dc/elements/1.1/publisher> "Wikipedia" .
// Hallo <http://test.de/blubb> "Kawumm"@German^^<http://test.de/string> .
#region init local vars
int iCurrent = 0;
StringBuilder strCurrentTag = new StringBuilder();
bool bFirst = true;
bool bMaskChar = false;
char cMaskChar = '0';
#endregion
foreach (char currentChar in line.ToCharArray())
{
#region simply read char, in case it is running in masked mode
if (bMaskChar)
{
strCurrentTag.Append(currentChar);
if (currentChar == cMaskChar)
bMaskChar = false;
}
#endregion
if (!bMaskChar) // no else, because criteria might have changes within the if
{
switch (CurrentTag)
{
#region SUBJECT handling
case TRIPLE_TAGS.SUBJECT:
{
switch (currentChar)
{
case ' ':
{
if (!bFirst)
{
triple.Subject = strCurrentTag.ToString();
strCurrentTag = new StringBuilder();
CurrentTag = TRIPLE_TAGS.PREDICATE;
bFirst = true;
}
break;
}
case '<':
{
if (!bFirst)
errorAction("invalid format: Subject may not contain '<' unless its a URI.");
else
{
// strCurrentTag.Append(currentChar);
bFirst = false;
bMaskChar = true;
cMaskChar = '>';
}
break;
}
case '>':
{
triple.Subject = strCurrentTag.ToString()
.Substring(0, strCurrentTag.Length - 1); // cut >
strCurrentTag = new StringBuilder();
CurrentTag = TRIPLE_TAGS.PREDICATE;
bFirst = true;
break;
}
default:
{
strCurrentTag.Append(currentChar);
bFirst = false;
bMaskChar = true;
cMaskChar = ' ';
break;
}
}
break;
}
#endregion
#region PREDICATE handling
case TRIPLE_TAGS.PREDICATE:
{
switch (currentChar)
{
case ' ': // ignore leading blanks
{
if (!bFirst)
{
errorAction("no blanks allowed in PREDICATE URI!");
throw new Exception();
}
break;
}
case '<':
{
if (!bFirst)
errorAction("invalid format: Subject may not contain '<' unless its a URI.");
else
{
// strCurrentTag.Append(currentChar);
bFirst = false;
bMaskChar = true;
cMaskChar = '>';
}
break;
}
case '>':
{
if (bFirst)
{
errorAction("PREDICATE has top begin with '<'!");
throw new Exception();
}
else
{
triple.Predicate = strCurrentTag.ToString()
.Substring(0, strCurrentTag.Length - 1); // cut >
strCurrentTag = new StringBuilder();
CurrentTag = TRIPLE_TAGS.OBJECT;
bFirst = true;
}
break;
}
default:
{
if (bFirst)
{
errorAction("PREDICATE has top begin with '<'!");
throw new Exception();
}
else strCurrentTag.Append(currentChar);
break;
}
}
break;
}
#endregion
#region OBJECT handling
case TRIPLE_TAGS.OBJECT:
{
switch (currentChar)
{
case ' ':
{
if (!bFirst)
{
errorAction("no blanks allowed in OBJECT (except within '\"')!");
throw new Exception();
}
break;
}
case '"':
{
if (bFirst)
{
// strCurrentTag.Append(currentChar);
bFirst = false;
bMaskChar = true;
cMaskChar = '"';
}
else
{
triple.TripleObject = strCurrentTag.ToString()
.Substring(0, strCurrentTag.Length - 1); ; // cut trailing "
strCurrentTag = new StringBuilder();
CurrentTag = TRIPLE_TAGS.POST_OBJECT;
bFirst = true;
}
break;
}
case '<':
{
if (bFirst)
{
// strCurrentTag.Append(currentChar);
bFirst = false;
bMaskChar = true;
cMaskChar = '>';
}
break;
}
case '>':
{
triple.TripleObject = strCurrentTag.ToString()
.Substring(0, strCurrentTag.Length - 1); ; // ignore >
strCurrentTag = new StringBuilder();
CurrentTag = TRIPLE_TAGS.POST_OBJECT;
bFirst = true;
break;
}
default:
{
if (bFirst)
{
errorAction("OBJECT must either begin with '\"' or '<'!");
throw new Exception();
}
break;
}
}
break;
}
#endregion
#region optional OBJECT data (language and datatype)
case TRIPLE_TAGS.POST_OBJECT:
{
switch (currentChar)
{/*
case ' ':
{
if (!bFirst)
{
errorAction("blank's are not allowed in optional Object data.");
throw new Exception();
} // otherwise ignore
break;
}*/
case '@': // Language
{
CurrentTag = TRIPLE_TAGS.OBJECT_LANG;
break;
}
case '^':
{
CurrentTag = TRIPLE_TAGS.OBJECT_DATATYPE;
break;
}
}
break;
}
#endregion
#region OBJECT_LANG
case TRIPLE_TAGS.OBJECT_LANG:
{
switch (currentChar)
{
case ' ':
case '^':
case '.':
{
triple.Language = strCurrentTag.ToString();
strCurrentTag = new StringBuilder();
CurrentTag = TRIPLE_TAGS.POST_OBJECT;
break;
}
default:
{
strCurrentTag.Append(currentChar);
break;
}
}
break;
}
#endregion
#region OBJECT_DATATYPE
case TRIPLE_TAGS.OBJECT_DATATYPE:
{
switch (currentChar)
{
case '^': // ignore this char
{
break;
}
case '<':
{
// strCurrentTag.Append(currentChar);
bMaskChar = true;
cMaskChar = '>';
bFirst = false;
break;
}
case '>':
{
triple.Datatype = strCurrentTag.ToString()
.Substring(0, strCurrentTag.Length - 1); // remove trailing >
CurrentTag = TRIPLE_TAGS.POST_OBJECT;
break;
}
default:
{
strCurrentTag.Append(currentChar);
break;
}
}
break;
}
#endregion
}
}
iCurrent++;
}
return triple;
}
private static bool SaveTriple(Triple selectedTriple, Dictionary<string, List<string>> dictExistingNodes, Dictionary<string, uint> dictDirectoriesEntries,
Dictionary<string, long> dictVertexIDs, Action<String> LogError)
{
StreamWriter sw = null;
try
{
if (!dictExistingNodes.ContainsKey(selectedTriple.Subject))
//if (!File.Exists(Properties.Settings.Default.WorkDir + Path.DirectorySeparatorChar + Properties.Settings.Default.DataDir + Path.DirectorySeparatorChar + RemoveEvilCharactersForFile(selectedTriple.Subject)))
{
ushort sDirs = 0;
String strPathname = null;
DirectoryInfo dirInfo = null;
/*
foreach (String hashDirName in Directory.EnumerateDirectories(Properties.Settings.Default.WorkDir + Path.DirectorySeparatorChar + Properties.Settings.Default.DataDir))
{
dirInfo = new DirectoryInfo(hashDirName);
if (dirInfo.GetFiles().Length < Properties.Settings.Default.DataDirMaxEntries)
{
strPathname = Properties.Settings.Default.WorkDir + Path.DirectorySeparatorChar
+ Properties.Settings.Default.DataDir + Path.DirectorySeparatorChar
+ dirInfo.Name + Path.DirectorySeparatorChar;
break;
}
else strPathname = null;
sDirs++;
}
*/
foreach (String dirname in dictDirectoriesEntries.Keys)
{
if (dictDirectoriesEntries[dirname] < Properties.Settings.Default.DataDirMaxEntries)
{
strPathname = dirname;
break;
}
sDirs++;
}
if (strPathname == null)
{
if (Directory.Exists(Properties.Settings.Default.WorkDir + Path.DirectorySeparatorChar
+ Properties.Settings.Default.DataDir + Path.DirectorySeparatorChar
+ sDirs))
{
LogError("Directory already exists! " + Properties.Settings.Default.WorkDir + Path.DirectorySeparatorChar
+ Properties.Settings.Default.DataDir + Path.DirectorySeparatorChar
+ sDirs + Path.DirectorySeparatorChar);
return false;
}
else
{
strPathname = Properties.Settings.Default.WorkDir + Path.DirectorySeparatorChar
+ Properties.Settings.Default.DataDir + Path.DirectorySeparatorChar
+ sDirs + Path.DirectorySeparatorChar;
Directory.CreateDirectory(strPathname);
dictDirectoriesEntries.Add(strPathname, 0);
}
}
sw = File.CreateText(strPathname + Path.DirectorySeparatorChar + RemoveEvilCharactersForFile(selectedTriple.Subject));
sw.WriteLine("VertexID=" + dictVertexIDs[selectedTriple.TripleObject].ToString());
List<string> lData = new List<string>();
lData.Add(strPathname + Path.DirectorySeparatorChar + RemoveEvilCharactersForFile(selectedTriple.Subject));
lData.Add(dictVertexIDs[selectedTriple.TripleObject].ToString());
lData.Add(selectedTriple.TripleObject);
dictExistingNodes.Add(selectedTriple.Subject, lData);
dictDirectoriesEntries[strPathname]++;
dictVertexIDs[selectedTriple.TripleObject]++;
}
else
{
sw = File.AppendText(dictExistingNodes[selectedTriple.Subject][0]);
// Properties.Settings.Default.WorkDir + Path.DirectorySeparatorChar + Properties.Settings.Default.DataDir + Path.DirectorySeparatorChar + RemoveEvilCharactersForFile(selectedTriple.Subject));
}
sw.WriteLine(selectedTriple.Subject + "=" + selectedTriple.TripleObject);
sw.Flush();
sw.Close();
}
catch (Exception a)
{
LogError("Error writing Node-File to FS: " + selectedTriple.Subject);
LogError(a.Message);
LogError(a.StackTrace);
return false;
}
return true;
}
