private static string SearchWord(Word word)
{
int attempts = 3;
while (attempts > 0)
{
try
{
if (word.GrammaticalClass != null)
{
break;
}
string htmlString = new System.Net.WebClient().DownloadString(string.Format("http://www.dicionarioinformal.com.br/{0}/", word.Token));
HtmlDocument doc = new HtmlDocument();
doc.LoadHtml(htmlString);
ClassifyWord(word, doc);
GetMoreWords(doc);
//System.Diagnostics.Debug.WriteLine(string.Format("http://www.dicionarioinformal.com.br/{0}/", word.Token));
break;
}
catch (Exception)
{
attempts--;
}
}
numberOfWordsProccessed++;
if (attempts <= 0)
{
return "Error: " + word.Token;
}
else
{
return string.Format("http://www.dicionarioinformal.com.br/{0}", word.Token);
}
}
private static void ClassifyWord(Word word, HtmlDocument doc)
{
var span = doc.DocumentNode.SelectSingleNode("//span[@class='textoDefinicao']");
if (span != null)
{
List<HtmlNode> meaningfullNodes = span.ChildNodes.Where(node => node.Name == "#text").ToList();
foreach (var node in meaningfullNodes)
{
string innerText = node.InnerText.ToLower();
Grammar grammaticalClass = null;
//verifica as classes gramaticais que a palavra pode assumir
#region Classe Gramatical
if (innerText.ToLower().LastIndexOf(GrammaticalClassEnum.Substantivo.ToString().ToLower()) > 0)
{
grammaticalClass = new Grammar()
{
GrammaticalClass = GrammaticalClassEnum.Substantivo,
};
}
if (innerText.ToLower().LastIndexOf(GrammaticalClassEnum.Artigo.ToString().ToLower()) > 0)
{
grammaticalClass = new Grammar()
{
GrammaticalClass = GrammaticalClassEnum.Artigo,
};
}
if (innerText.ToLower().LastIndexOf(GrammaticalClassEnum.Adjetivo.ToString().ToLower()) > 0)
{
grammaticalClass = new Grammar()
{
GrammaticalClass = GrammaticalClassEnum.Adjetivo,
};
}
if (innerText.ToLower().LastIndexOf(GrammaticalClassEnum.Numeral.ToString().ToLower()) > 0)
{
grammaticalClass = new Grammar()
{
GrammaticalClass = GrammaticalClassEnum.Numeral,
};
}
if (innerText.ToLower().LastIndexOf(GrammaticalClassEnum.Pronome.ToString().ToLower()) > 0)
{
grammaticalClass = new Grammar()
{
GrammaticalClass = GrammaticalClassEnum.Pronome,
};
}
if (innerText.ToLower().LastIndexOf(GrammaticalClassEnum.Verbo.ToString().ToLower()) > 0)
{
grammaticalClass = new Grammar()
{
GrammaticalClass = GrammaticalClassEnum.Verbo,
};
}
if (innerText.ToLower().LastIndexOf(GrammaticalClassEnum.Advérbio.ToString().ToLower()) > 0)
{
grammaticalClass = new Grammar()
{
GrammaticalClass = GrammaticalClassEnum.Advérbio,
};
}
if (innerText.ToLower().LastIndexOf(GrammaticalClassEnum.Preposição.ToString().ToLower()) > 0)
{
grammaticalClass = new Grammar()
{
GrammaticalClass = GrammaticalClassEnum.Preposição,
};
}
if (innerText.ToLower().LastIndexOf(GrammaticalClassEnum.Conjunção.ToString().ToLower()) > 0)
{
grammaticalClass = new Grammar()
{
GrammaticalClass = GrammaticalClassEnum.Conjunção,
};
}
if (innerText.ToLower().LastIndexOf(GrammaticalClassEnum.Interjeição.ToString().ToLower()) > 0)
{
grammaticalClass = new Grammar()
{
GrammaticalClass = GrammaticalClassEnum.Interjeição,
};
}
#endregion
//verifica a variação de número da palavra
#region Variação de Número
if (grammaticalClass != null)
{
if (innerText.ToLower().LastIndexOf(NumberVariationEnum.Plural.ToString().ToLower()) > 0)
{
grammaticalClass.NumberVariation = NumberVariationEnum.Plural;
}
if (innerText.ToLower().LastIndexOf(NumberVariationEnum.Singular.ToString().ToLower()) > 0)
{
grammaticalClass.NumberVariation = NumberVariationEnum.Singular;
}
}
#endregion
//verifica o genero da palavra
#region Variação de Genero
if (grammaticalClass != null)
{
if (innerText.ToLower().LastIndexOf(GenderVariationEnum.Feminino.ToString().ToLower()) > 0)
{
grammaticalClass.GenderVariation = GenderVariationEnum.Feminino;
}
if (innerText.ToLower().LastIndexOf(GenderVariationEnum.Masculino.ToString().ToLower()) > 0)
{
grammaticalClass.GenderVariation = GenderVariationEnum.Masculino;
}
}
#endregion
if (grammaticalClass != null)
{
if (word.GrammaticalClass == null)
{
word.GrammaticalClass = new Dictionary<GrammaticalClassEnum, Grammar>();
word.GrammaticalClass.Add(grammaticalClass.GrammaticalClass, grammaticalClass);
}
else
{
if (!word.GrammaticalClass.ContainsKey(grammaticalClass.GrammaticalClass))
{
word.GrammaticalClass.Add(grammaticalClass.GrammaticalClass, grammaticalClass);
}
}
//System.Diagnostics.Debug.WriteLine(Newtonsoft.Json.JsonConvert.SerializeObject(grammaticalClass));
}
}
}
if (!wordDictionary.ContainsKey(word.Token))
{
wordDictionary.TryAdd(word.Token, word);
}
else
{
if (word.GrammaticalClass != null)
{
foreach (GrammaticalClassEnum key in word.GrammaticalClass.Keys)
{
if (!wordDictionary[word.Token].GrammaticalClass.ContainsKey(key))
{
wordDictionary[word.Token].GrammaticalClass.Add(key, word.GrammaticalClass[key]);
}
}
}
}
}
private static void SaveWords()
{
//System.Xml.Serialization.XmlSerializer serializer = new System.Xml.Serialization.XmlSerializer(WordRepository);
List<Word> listOfWords = wordDictionary.Values.OrderBy(w => w.Token).ToList();
WordRepository repository = new WordRepository()
{
WordList = new List<Word>(),
};
foreach (Word word in listOfWords)
{
Dictionary<string, FriendlyGrammar> friendlyGrammaticalDictionary = null;
if (word.GrammaticalClass != null)
{
friendlyGrammaticalDictionary = new Dictionary<string, FriendlyGrammar>();
foreach (Grammar grammar in word.GrammaticalClass.Values)
{
friendlyGrammaticalDictionary.Add(grammar.GrammaticalClass.ToString(), new FriendlyGrammar()
{
GenderVariation = grammar.GenderVariation.ToString(),
NumberVariation = grammar.NumberVariation.ToString(),
GrammaticalClass = grammar.GrammaticalClass.ToString(),
});
}
}
Word repositoryWord = new Word()
{
Token = word.Token,
GrammaticalClass = word.GrammaticalClass,
FriendlyGrammaticalClass = friendlyGrammaticalDictionary,
GrammaticalClassAsJson = Newtonsoft.Json.JsonConvert.SerializeObject(word.GrammaticalClass),
};
repository.WordList.Add(repositoryWord);
}
string serializedJson = Newtonsoft.Json.JsonConvert.SerializeObject(repository);
File.WriteAllText(jsonRepoPath, serializedJson);
//using (Stream stream = new FileStream(xmlRepoPath, FileMode.Create))
//{
// System.Xml.Serialization.XmlSerializer xmlserializer = new System.Xml.Serialization.XmlSerializer(typeof(WordRepository));
// xmlserializer.Serialize(stream, repository);
//}
//string serializedXml = ServiceStack.Text.XmlSerializer.SerializeToString<WordRepository>(repository);
//File.WriteAllText(repoPath, serializedXml);
}
