private static List <DictionaryValue> translations(string verb)
{
string store = "translations/"; // Path to store translations.
if (!Directory.Exists(store))
{
Directory.CreateDirectory(store);
}
foreach (string file in Directory.GetFiles(store, "*.dict"))
{
List <DictionaryValue> entry = ObjectSerializer.FromFile
<List <DictionaryValue> >(file);
if (entry.Count > 0 && entry[0].EnglishWord.Equals(verb,
StringComparison.InvariantCultureIgnoreCase))
{
return(entry);
}
}
List <DictionaryValue> translations =
new WordReference(client).GetENFR_Translations(verb);
ObjectSerializer.ToFile(translations, store + verb + ".dict");
return(translations);
}
public Analysis_VerseExtract(WordReference wordReference)
{
this.WordReference = wordReference;
var verseReference =
SharedData.Document.CorpusDocument[WordReference.ChapterIndex, WordReference.VerseIndex];
int wordIndex = wordReference.WordIndex;
this.PrecedingWords =
verseReference
.Words
.Where(x =>
x.Index >= wordIndex - NumberOfSurroundingWordsToDisplay &&
x.Index < wordIndex)
.ToArray();
this.FollowingWords =
verseReference
.Words
.Where(x =>
x.Index > wordIndex &&
x.Index <= wordIndex + NumberOfSurroundingWordsToDisplay)
.ToArray();
BuckwalterWord = wordReference.BuckwalterText;
}
private async Task AddWordToReferenceTableAsync(Guid instanceId, string wordName, IEnumerable <WordInformation> wordInfo)
{
var wordRef = new WordReference(wordName, wordInfo);
wordRef.Serialize();
var tableEntity = new TableEntityAdapter <WordReference>()
{
PartitionKey = instanceId.ToString(),
RowKey = wordName,
OriginalEntity = wordRef,
};
await this.wordReferencesTable.ExecuteAsync(TableOperation.Insert(tableEntity));
}
public IEnumerable <SearchResult> Search(SearchQuery query)
{
if (query == null)
{
throw new ArgumentNullException("query");
}
IDictionary <string, SearchResult> results = new Dictionary <string, SearchResult>(StringComparer.OrdinalIgnoreCase);
if (!string.IsNullOrWhiteSpace(query.SearchTerms))
{
Regex wordPattern = new Regex(@"[a-z1-9]{3,}", RegexOptions.Compiled | RegexOptions.IgnoreCase);
MatchCollection wordMatches = wordPattern.Matches(query.SearchTerms);
foreach (Match match in wordMatches)
{
if (s_Words.ContainsKey(match.Value))
{
WordReference wordRef = s_Words[match.Value];
foreach (KeyValuePair <string, int> pair in wordRef.Files)
{
FileReference fileRef = s_Files[pair.Key];
SearchResult result;
if (results.ContainsKey(fileRef.Path))
{
result = results[fileRef.Path];
}
else
{
result = new SearchResult();
result.Url = "/Files/" + Path.GetFileName(fileRef.Path);
result.Title = fileRef.Title;
result.Preview = fileRef.Preview;
results.Add(fileRef.Path, result);
}
result.Rank += pair.Value;
}
}
}
}
return(results.Values.OrderByDescending(result => result.Rank));
}
private static void AddTheWordsToTheIndex(IEnumerable <Word> words, T key, TextIndex <T> index)
{
foreach (var word in words)
{
var wordReference = new WordReference <T> {
Word = word.WordText, Key = key, PhraseIndex = word.Index
};
if (ThisWordIsNotInTheIndex(word, index))
{
AddTheNewWordToTheIndex(word, index, wordReference);
}
else
{
AddTheNewUseOfThisWordToTheIndex(word, index, wordReference);
}
}
}
private static void LoadIndex(IFilePathResolver resolver)
{
using (Stream indexStream = new FileStream(resolver.MapPath("~/App_Data/Index.xml"), FileMode.Open, FileAccess.Read, FileShare.None))
{
const string ns = "http://schemas.magurany.com/search";
XmlSchemaSet schemas = new XmlSchemaSet();
schemas.Add(ns, resolver.MapPath("~/App_Data/Index.xsd"));
XmlReaderSettings settings = new XmlReaderSettings();
settings.IgnoreComments = true;
settings.IgnoreWhitespace = true;
settings.Schemas = schemas;
XmlReader reader = XmlReader.Create(indexStream, settings);
reader.ReadStartElement("Index", ns);
reader.ReadStartElement("Words", ns);
while (reader.IsStartElement("Word", ns))
{
WordReference wordRef = new WordReference();
reader.MoveToAttribute("Word");
wordRef.Word = reader.ReadContentAsString();
reader.MoveToAttribute("Count");
wordRef.Count = reader.ReadContentAsInt();
reader.ReadStartElement("Word", ns);
while (reader.IsStartElement("File", ns))
{
reader.MoveToAttribute("Path");
string path = reader.ReadContentAsString();
reader.MoveToAttribute("Count");
int count = reader.ReadContentAsInt();
wordRef.Files.Add(path, count);
reader.ReadStartElement("File", ns);
}
s_Words.Add(wordRef.Word, wordRef);
reader.ReadEndElement();
}
reader.ReadEndElement();
reader.ReadStartElement("Files", ns);
while (reader.IsStartElement("File", ns))
{
FileReference fileRef = new FileReference();
reader.MoveToAttribute("Path");
fileRef.Path = reader.ReadContentAsString();
reader.MoveToAttribute("Title");
fileRef.Title = reader.ReadContentAsString();
reader.MoveToAttribute("Preview");
fileRef.Preview = reader.ReadContentAsString();
s_Files.Add(fileRef.Path, fileRef);
reader.ReadStartElement("File", ns);
}
reader.ReadEndElement();
reader.ReadEndElement();
}
}
private static void AddTheNewWordToTheIndex(Word word, TextIndex <T> index, WordReference <T> wordReference)
{
var wordReferences = new List <WordReference <T> >(new[] { wordReference });
index.Add(word.WordText, wordReferences);
}
private static void AddTheNewUseOfThisWordToTheIndex(Word word, TextIndex <T> index, WordReference <T> wordReference)
{
index[word.WordText].Add(wordReference);
}
public Analysis_WordUsageWithVerseExtract(WordReference wordReference)
{
this.Extract = new Analysis_VerseExtract(wordReference);
this.Reference = wordReference;
}
static void Main(string[] args)
{
if (args.Length != 2)
{
Console.WriteLine("usage:");
Console.WriteLine("mIndexer DirectoryToSearch Index.xml");
return;
}
DirectoryInfo sourceDir = new DirectoryInfo(args[0]);
FileInfo indexFile = new FileInfo(args[1]);
if (!sourceDir.Exists)
{
Console.WriteLine("The directory, '{0},' was not found.", sourceDir);
return;
}
FileInfo[] sourceFiles = sourceDir.GetFiles("*.txt");
if (sourceFiles.Length == 0)
{
Console.WriteLine("The directory, '{0},' does not contain any text files.", sourceDir);
return;
}
bool proceed = false;
while (!proceed && indexFile.Exists)
{
Console.Write("The file, '{0},' exists. Overwrite? [y/n] ", indexFile);
ConsoleKeyInfo yesNo = Console.ReadKey(false);
if (ConsoleKey.N == yesNo.Key)
{
return;
}
proceed = ConsoleKey.Y == yesNo.Key;
Console.WriteLine();
}
if (!indexFile.Directory.Exists)
{
indexFile.Directory.Create();
}
IDictionary <string, WordReference> words = new Dictionary <string, WordReference>(StringComparer.OrdinalIgnoreCase);
IList <FileReference> files = new List <FileReference>(sourceFiles.Length);
Regex wordPattern = new Regex(@"[a-z1-9]{3,}", RegexOptions.Compiled | RegexOptions.IgnoreCase);
Regex previewPattern = new Regex(@"[^ ]+", RegexOptions.Compiled | RegexOptions.IgnoreCase);
foreach (FileInfo sourceFile in sourceFiles)
{
FileReference fileRef = new FileReference();
fileRef.Path = sourceFile.FullName;
fileRef.Preview = string.Empty;
fileRef.Title = sourceFile.Name;
using (Stream sourceStream = sourceFile.OpenRead())
{
TextReader reader = new StreamReader(sourceStream);
string line;
while ((line = reader.ReadLine()) != null)
{
if (!string.IsNullOrWhiteSpace(line))
{
if (fileRef.Preview.Length < 250)
{
MatchCollection previewMatches = previewPattern.Matches(line);
foreach (Match previewMatch in previewMatches)
{
fileRef.Preview += previewMatch.Value + " ";
}
}
MatchCollection wordMatches = wordPattern.Matches(line);
foreach (Match match in wordMatches)
{
string word = match.Value;
WordReference wordRef;
if (words.ContainsKey(word))
{
wordRef = words[word];
}
else
{
wordRef = new WordReference();
wordRef.Word = word;
words.Add(word, wordRef);
}
wordRef.Count++;
if (wordRef.Files.ContainsKey(sourceFile.FullName))
{
wordRef.Files[sourceFile.FullName]++;
}
else
{
wordRef.Files.Add(sourceFile.FullName, 1);
}
}
}
}
files.Add(fileRef);
}
}
using (Stream indexStream = indexFile.Open(FileMode.Create, FileAccess.Write, FileShare.None))
{
const string ns = "http://schemas.magurany.com/search";
XmlWriterSettings settings = new XmlWriterSettings();
settings.Indent = true;
XmlWriter writer = XmlWriter.Create(indexStream, settings);
writer.WriteStartDocument();
writer.WriteStartElement("Index", ns);
writer.WriteStartElement("Words", ns);
foreach (WordReference wordRef in words.Values)
{
writer.WriteStartElement("Word", ns);
writer.WriteAttributeString("Word", wordRef.Word);
writer.WriteAttributeString("Count", wordRef.Count.ToString());
foreach (KeyValuePair <string, int> pair in wordRef.Files)
{
writer.WriteStartElement("File", ns);
writer.WriteAttributeString("Path", pair.Key);
writer.WriteAttributeString("Count", pair.Value.ToString());
writer.WriteEndElement();
}
writer.WriteEndElement();
}
writer.WriteEndElement();
writer.WriteStartElement("Files", ns);
foreach (FileReference wordRef in files)
{
writer.WriteStartElement("File", ns);
writer.WriteAttributeString("Path", wordRef.Path);
writer.WriteAttributeString("Title", wordRef.Title);
writer.WriteAttributeString("Preview", wordRef.Preview);
writer.WriteEndElement();
}
writer.WriteEndElement();
writer.WriteEndElement();
writer.Flush();
}
}