public static bool SerializeToXml(TrainedDataSet dataset, string filename)
{
if (dataset == null || dataset._wordDictionary == null || dataset._wordDictionary.UniqueWordCount < 1)
{
return(false);
}
// Sort every Word's internal dictionary
foreach (Word word in dataset._wordDictionary.Words)
{
word.OrderInternalDictionary();
}
// Sort the NextWordDictionary
dataset._wordDictionary =
new WordDictionary(
dataset._wordDictionary._internalDictionary.OrderByDescending(kvp => kvp.Value.TotalWordsSeen).ToDictionary(kvp => kvp.Key, kvp => kvp.Value)
);
XDocument doc = new XDocument(
new XElement(XmlElementNames.RootNode,
new XElement(XmlElementNames.TotalWordsProcessedNode, dataset._wordDictionary.TotalWordsProcessed),
dataset._wordDictionary.Words.Select(word =>
new XElement(XmlElementNames.WordNode,
new XElement(XmlElementNames.ValueNode, word.Value),
new XElement(XmlElementNames.DictionarySizeNode, word.TotalWordsSeen),
new XElement(XmlElementNames.DictionaryNode,
word._nextWordDictionary._internalDictionary.Select(kvp =>
new XElement(XmlElementNames.KeyValuePairNode,
new XElement(XmlElementNames.KeyNode, kvp.Key.Value),
new XElement(XmlElementNames.ValueNode, kvp.Value)
)
)
)
)
)
)
);
if (doc != null)
{
doc.Save(filename, SaveOptions.None);
return(File.Exists(filename));
}
return(false);
}
public static TrainedDataSet DeserializeFromXml(string filename)
{
if (!File.Exists(filename))
{
return(new TrainedDataSet());
}
XDocument doc = XDocument.Parse(File.ReadAllText(filename), LoadOptions.None);
if (doc == null)
{
return(new TrainedDataSet());
}
XElement rootNode = doc.XPathSelectElement(XmlElementNames.RootNode);
if (rootNode == null)
{
return(new TrainedDataSet());
}
XElement totalWordsNode = rootNode.XPathSelectElement(XmlElementNames.TotalWordsProcessedNode);
if (totalWordsNode == null)
{
return(new TrainedDataSet());
}
int totalWordsProcessed = 0; int.TryParse(totalWordsNode.Value, out totalWordsProcessed);
List <XElement> wordNodes = rootNode.XPathSelectElements(XmlElementNames.WordNode).ToList();
if (wordNodes == null || wordNodes.Count < 1)
{
return(new TrainedDataSet());
}
// Dictionary
Dictionary <string, Word> dictionary = new Dictionary <string, Word>();
// Create a Word object for each Word before populating NextWordFrequencyDictionary
foreach (XElement wordNode in wordNodes)
{
XElement textNode = wordNode.XPathSelectElement(XmlElementNames.ValueNode);
XElement countNode = wordNode.XPathSelectElement(XmlElementNames.DictionarySizeNode);
if (textNode == null || countNode == null)
{
continue;
}
string text = textNode.Value;
int ttlWordCount = 0; int.TryParse(countNode.Value, out ttlWordCount);
Word newWord = new Word(text);
//newWord.TotalWordsSeen = ttlWordCount;
dictionary.Add(text, newWord);
}
// Now populate NextWordFrequencyDictionary
foreach (XElement wordNode in wordNodes)
{
XElement textNode = wordNode.XPathSelectElement(XmlElementNames.ValueNode);
XElement dictNode = wordNode.XPathSelectElement(XmlElementNames.DictionaryNode);
if (textNode == null || dictNode == null)
{
continue;
}
string text = textNode.Value;
if (!dictionary.Keys.Contains(text))
{
continue;
}
Word word = dictionary[text];
List <XElement> kvpNodes = dictNode.XPathSelectElements(XmlElementNames.KeyValuePairNode).ToList();
foreach (XElement kvpNode in kvpNodes)
{
XElement keyNode = kvpNode.XPathSelectElement(XmlElementNames.KeyNode);
XElement valueNode = kvpNode.XPathSelectElement(XmlElementNames.ValueNode);
string keyText = keyNode.Value;
int valueInt = 0; int.TryParse(valueNode.Value, out valueInt);
if (!dictionary.Keys.Contains(keyText))
{
continue;
}
Word keyWord = dictionary[keyText];
word._nextWordDictionary._internalDictionary.Add(keyWord, valueInt);
}
}
if (dictionary != null)
{
TrainedDataSet result = new TrainedDataSet(new WordDictionary(dictionary));
return(result);
}
else
{
return(new TrainedDataSet());
}
}
public static TrainedDataSet DeserializeFromXml(string filename)
{
if (!File.Exists(filename)) { return new TrainedDataSet(); }
XDocument doc = XDocument.Parse(File.ReadAllText(filename), LoadOptions.None);
if (doc == null) { return new TrainedDataSet(); }
XElement rootNode = doc.XPathSelectElement(XmlElementNames.RootNode);
if (rootNode == null) { return new TrainedDataSet(); }
XElement totalWordsNode = rootNode.XPathSelectElement(XmlElementNames.TotalWordsProcessedNode);
if (totalWordsNode == null) { return new TrainedDataSet(); }
int totalWordsProcessed = 0; int.TryParse(totalWordsNode.Value, out totalWordsProcessed);
List<XElement> wordNodes = rootNode.XPathSelectElements(XmlElementNames.WordNode).ToList();
if (wordNodes == null || wordNodes.Count < 1) { return new TrainedDataSet(); }
// Dictionary
Dictionary<string, Word> dictionary = new Dictionary<string, Word>();
// Create a Word object for each Word before populating NextWordFrequencyDictionary
foreach (XElement wordNode in wordNodes)
{
XElement textNode = wordNode.XPathSelectElement(XmlElementNames.ValueNode);
XElement countNode = wordNode.XPathSelectElement(XmlElementNames.DictionarySizeNode);
if (textNode == null || countNode == null)
{
continue;
}
string text = textNode.Value;
int ttlWordCount = 0; int.TryParse(countNode.Value, out ttlWordCount);
Word newWord = new Word(text);
//newWord.TotalWordsSeen = ttlWordCount;
dictionary.Add(text, newWord);
}
// Now populate NextWordFrequencyDictionary
foreach (XElement wordNode in wordNodes)
{
XElement textNode = wordNode.XPathSelectElement(XmlElementNames.ValueNode);
XElement dictNode = wordNode.XPathSelectElement(XmlElementNames.DictionaryNode);
if (textNode == null || dictNode == null)
{
continue;
}
string text = textNode.Value;
if (!dictionary.Keys.Contains(text))
{
continue;
}
Word word = dictionary[text];
List<XElement> kvpNodes = dictNode.XPathSelectElements(XmlElementNames.KeyValuePairNode).ToList();
foreach (XElement kvpNode in kvpNodes)
{
XElement keyNode = kvpNode.XPathSelectElement(XmlElementNames.KeyNode);
XElement valueNode = kvpNode.XPathSelectElement(XmlElementNames.ValueNode);
string keyText = keyNode.Value;
int valueInt = 0; int.TryParse(valueNode.Value, out valueInt);
if (!dictionary.Keys.Contains(keyText))
{
continue;
}
Word keyWord = dictionary[keyText];
word._nextWordDictionary._internalDictionary.Add(keyWord, valueInt);
}
}
if (dictionary != null)
{
TrainedDataSet result = new TrainedDataSet(new WordDictionary(dictionary));
return result;
}
else
{
return new TrainedDataSet();
}
}
public static bool SerializeToXml(TrainedDataSet dataset, string filename)
{
if (dataset == null || dataset._wordDictionary == null || dataset._wordDictionary.UniqueWordCount < 1)
{
return false;
}
// Sort every Word's internal dictionary
foreach (Word word in dataset._wordDictionary.Words)
{
word.OrderInternalDictionary();
}
// Sort the NextWordDictionary
dataset._wordDictionary =
new WordDictionary(
dataset._wordDictionary._internalDictionary.OrderByDescending(kvp => kvp.Value.TotalWordsSeen).ToDictionary(kvp => kvp.Key, kvp => kvp.Value)
);
XDocument doc = new XDocument(
new XElement(XmlElementNames.RootNode,
new XElement(XmlElementNames.TotalWordsProcessedNode, dataset._wordDictionary.TotalWordsProcessed),
dataset._wordDictionary.Words.Select(word =>
new XElement(XmlElementNames.WordNode,
new XElement(XmlElementNames.ValueNode, word.Value),
new XElement(XmlElementNames.DictionarySizeNode, word.TotalWordsSeen),
new XElement(XmlElementNames.DictionaryNode,
word._nextWordDictionary._internalDictionary.Select(kvp =>
new XElement(XmlElementNames.KeyValuePairNode,
new XElement(XmlElementNames.KeyNode, kvp.Key.Value),
new XElement(XmlElementNames.ValueNode, kvp.Value)
)
)
)
)
)
)
);
if (doc != null)
{
doc.Save(filename, SaveOptions.None);
return File.Exists(filename);
}
return false;
}
private void NewDataSet()
{
if (AskIfSaveFirst())
{
dataSet = new TrainedDataSet();
OnDataSetLoaded();
}
}
public MainForm()
{
InitializeComponent();
dataSet = new TrainedDataSet();
IsDatasetDirty = false;
}
