private void ParseGooglePhrase(string word)
{
FireLogMessage("GooglePhrase Parsing " + word);
try
{
string html = (_entities.GoogleDictionaryHtmls.Where(w => w.Word == word).Select(w => w.Html)).FirstOrDefault();
Regex ul = new Regex("<ul class=\"rlt-snt\">(.*?)</ul>", RegexOptions.Singleline);
string list = ul.Match(html).Groups[1].Value;
Regex li = new Regex("<li>.+?<a.+?>(.+?)</a></b>.+?</div>(.+?)</li>", RegexOptions.Singleline);
MatchCollection mc = li.Matches(list);
GreWord gw = GetGreWord(word);
foreach (Match m in mc)
{
string en = m.Groups[1].Value;
string bn = m.Groups[2].Value;
bn = bn.Replace("\n", "").Replace("<b>", "").Replace("</b>", "").Replace(" ", " ");
UpdateGooglePhrase(gw, en, bn);
}
}
catch
{
}
}
String getMnemonics(GreWord gw)
{
var entities = new gredbEntities();
var parser = new ParserHelper(entities);
String mnemonics = "";
List <FeaturedMnemonic> mSelected = (from w in entities.FeaturedMnemonics
where w.GreWord.Word == gw.Word
select w).ToList();
foreach (FeaturedMnemonic ms in mSelected)
{
mnemonics += "##:" + ms.Mnemonic + ", ";
}
List <BasicMnemonic> mAll = (from w in entities.BasicMnemonics
where w.GreWord.Word == gw.Word
select w).ToList();
foreach (BasicMnemonic ms in mAll)
{
mnemonics += "#" + ms.Helpful + "," + ms.NotHelpful + ":" + ms.Mnemonic + ", ";
}
mnemonics = Clean(mnemonics, @"\(Tag.+?\)");
mnemonics = Clean(mnemonics);
return(mnemonics);
}
void UpdateGoogleBengaliWord(GreWord word, string definition)
{
BengaliDefinition gs = (_entities.BengaliDefinitions.Where(w => w.GreWord.Word == word.Word && w.Bengali == definition)).FirstOrDefault();
if (gs == null)
{
gs = new BengaliDefinition()
{
GreWord = word,
Bengali = definition
};
_entities.AddToBengaliDefinitions(gs);
_entities.SaveChanges();
}
}
void UpdateGoogleSynonym(GreWord word, string synonym)
{
GoogleSynonym gs = (_entities.GoogleSynonyms.Where(w => w.GreWord.Word == word.Word && w.Synonym == synonym)).FirstOrDefault();
if (gs == null)
{
gs = new GoogleSynonym()
{
GreWord = word,
Synonym = synonym
};
_entities.AddToGoogleSynonyms(gs);
_entities.SaveChanges();
}
}
private void InsertAffineWord(GreWord word, GreWord affine, int affinity)
{
int count = (from o in _entities.GreWordAffinities
where o.GreWord.Word == word.Word && o.RelevantWord.Word == affine.Word
select o).Count();
if (count == 0)
{
GreWordAffinity gwa = new GreWordAffinity()
{
GreWord = word,
RelevantWord = affine,
Affinity = affinity
};
_entities.AddToGreWordAffinities(gwa);
}
}
void UpdateGooglePhrase(GreWord word, string en, string bn)
{
var gs = (_entities.GooglePhrases.Where(w => w.GreWord.Word == word.Word && w.EnglishPhrase == en)).FirstOrDefault();
if (gs == null)
{
gs = new GooglePhrase()
{
GreWord = word,
EnglishPhrase = en,
BengaliPhrase = bn,
};
_entities.AddToGooglePhrases(gs);
_entities.SaveChanges();
}
}
String getGoogleSynonym(GreWord gw)
{
var entities = new gredbEntities();
var parser = new ParserHelper(entities);
String synnonym = "";
List <GoogleSynonym> gSyn = (from w in entities.GoogleSynonyms
where w.GreWord.Word == gw.Word
select w).ToList();
foreach (GoogleSynonym gs in gSyn)
{
synnonym += gs.Synonym + ", ";
}
return(synnonym);
}
private void ParseGoogleBengali(string word)
{
FireLogMessage("GoogleBengali for " + word);
GreWord greWord = GetGreWord(word);
GoogleDictionaryHtml gd = (_entities.GoogleDictionaryHtmls.Where(w => w.Word == word)).FirstOrDefault();
if (gd == null)
{
return;
}
string text = gd.Html;
Regex regexWord = new Regex(@"<div class=""dct-em"">\s+<span class=""dct-tt"">(.+?)</span>", RegexOptions.Singleline | RegexOptions.IgnoreCase);
MatchCollection meanings = regexWord.Matches(text);
foreach (string def in from Match m in meanings select CleanText(m.Groups[1].Value))
{
UpdateGoogleBengaliWord(greWord, def);
}
}
private void ParseGoogleSynonym(string word)
{
GreWord greWord = GetGreWord(word);
GoogleDictionaryHtml gd = (_entities.GoogleDictionaryHtmls.Where(w => w.Word == word)).FirstOrDefault();
FireLogMessage("GoogleSynonym Parsing " + word);
if (gd == null)
{
return;
}
string text = gd.Html;
Regex regexWord = new Regex(@"<a href=""/dictionary\?hl=en&q=.+?&sl=en&tl=bn&oi=dict_lk"">(.+?)</a>");
MatchCollection meanings = regexWord.Matches(text);
foreach (Match m in meanings)
{
string def = CleanText(m.Groups[1].Value);
UpdateGoogleSynonym(greWord, def);
}
}
private void ParseMnemonicDictionary(string word)
{
FireLogMessage("Parsing Mnemonics for: " + word);
GreWord greWord = GetGreWord(word);
MnemonicsDictionaryHtml gd = (_entities.MnemonicsDictionaryHtmls.Where(w => w.Word == word)).FirstOrDefault();
if (gd == null)
{
return;
}
string text = gd.Html;
Regex regexWordUl = new Regex("<ul class='wordnet'>(.+?)</ul>", RegexOptions.Singleline);
Regex regexWordLi = new Regex("<li>(.+?)</li>", RegexOptions.Singleline);
Regex regexImage = new Regex("<div class=\"floatright\">.+?<img src=\"(.+?)\".+?</div>", RegexOptions.Singleline);
Regex regexTag = new Regex("<div class=\"floatleft\">(.+?)</div>", RegexOptions.Singleline);
Match meaningMatch = regexWordUl.Match(text);
if (meaningMatch.Success)
{
MatchCollection meanings = regexWordLi.Matches(meaningMatch.Value);
foreach (Match m in meanings)
{
string tag = "";
string img = "";
string def = CleanText(m.Groups[1].Value);
MatchCollection mci = regexImage.Matches(def);
foreach (Match mi in mci)
{
Match mt = regexTag.Match(def);
img = mi.Groups[1].Value;
tag = mt.Groups[1].Value.Replace("\r\n", "");
def = regexImage.Replace(def, "");
def = regexTag.Replace(def, "");
def = def.Replace("<div class='clear'></div>", "");
}
WordDefinition wordDefinition = new WordDefinition {
Definition = def, Image = img, Tag = tag, GreWord = greWord,
};
_entities.AddToWordDefinitions(wordDefinition);
_entities.SaveChanges();
}
}
Regex regexSelectedMnemonicDiv = new Regex("<div class='mnemonics'>(.+?)</div>", RegexOptions.Singleline | RegexOptions.IgnoreCase);
Regex regexSelectedMnemonicLi = new Regex(@"<li><p>(.+?)</p><p>\s+added by.+?</li>", RegexOptions.Singleline | RegexOptions.IgnoreCase);
Match selectedMnemonic = regexSelectedMnemonicDiv.Match(text);
if (selectedMnemonic.Success)
{
MatchCollection selectedMnemonics = regexSelectedMnemonicLi.Matches(selectedMnemonic.Groups[1].Value);
foreach (Match m in selectedMnemonics)
{
//listSelectedMnemonics.Items.Add(clarify(m.Groups[1].Value));
string def = CleanText(m.Groups[1].Value);
FeaturedMnemonic mnemonics = new FeaturedMnemonic
{
Mnemonic = def,
GreWord = greWord
};
_entities.AddToFeaturedMnemonics(mnemonics);
_entities.SaveChanges();
}
}
Regex regexAllMnemonicDiv = new Regex("<div class='mnemonic'>(.+?)</div>", RegexOptions.Singleline | RegexOptions.IgnoreCase);
Regex regexAllMnemonicLi = new Regex(@"<li><p>(.+?)</p><p>\s+added by.+?<p>Was this mnemonic useful \? \s+<strong id='hallo\d+'>(.+?)</strong>.+?<strong id='hallo\d+'> (.+?)</strong>", RegexOptions.Singleline | RegexOptions.IgnoreCase);
Match allMnemonic = regexAllMnemonicDiv.Match(text);
if (allMnemonic.Success)
{
MatchCollection meanings = regexAllMnemonicLi.Matches(allMnemonic.Groups[1].Value);
foreach (Match m in meanings)
{
string def = CleanText(m.Groups[1].Value);
BasicMnemonic mnemonics = new BasicMnemonic()
{
Mnemonic = def,
Helpful = m.Groups[2].Value,
NotHelpful = m.Groups[3].Value
};
mnemonics.GreWord = greWord;
_entities.AddToBasicMnemonics(mnemonics);
_entities.SaveChanges();
}
}
}
private void button1_Click(object sender, EventArgs e)
{
long listNameId = -1;
var entities = new gredbEntities();
var list = entities.ListNames.Where(o => o.Name.ToLower().Equals(comboBoxList.Text.Trim().ToLower())).Select(o => o).FirstOrDefault();
if (list == null)
{
ListName listName = new ListName {
Name = comboBoxList.Text.Trim()
};
entities.AddToListNames(listName);
entities.SaveChanges();
listNameId = listName.Id;
}
else
{
listNameId = list.Id;
}
String str = Regex.Replace(textBox1.Text, "([0-9\\.]+|<.+?>)", "");
Regex regex = new Regex("([^\\s]+)");
ListName ln = (from n in entities.ListNames
where n.Id == listNameId
select n).FirstOrDefault();
try
{
MatchCollection mc = regex.Matches(str);
foreach (Match m in mc)
{
try
{
GreWord word = (from w in entities.GreWords
where w.Word.ToLower() == m.Value.ToLower()
select w).FirstOrDefault();
if (word == null)
{
word = new GreWord()
{
Word = m.Value.ToLower(),
};
entities.AddToGreWords(word);
entities.SaveChanges();
}
ListedWord lw = (from w in entities.ListedWords
where w.ListName.Id == ln.Id && w.GreWord.Word == word.Word
select w).FirstOrDefault();
if (lw == null)
{
lw = new ListedWord()
{
GreWord = word,
ListName = ln
};
entities.AddToListedWords(lw);
entities.SaveChanges();
}
}
catch (Exception exp)
{
MessageBox.Show(exp.Message);
}
}
entities.Connection.Close();
}
catch (Exception exp)
{
}
}
private void buttonCollect_Click(object sender, EventArgs e)
{
var entities = new gredbEntities();
long listNameId = -1;
if (comboBoxList.Text.Trim().Length > 0)
{
var list = entities.ListNames.Where(o => o.Name.ToLower().Equals(comboBoxList.Text.Trim().ToLower())).Select(o => o).FirstOrDefault();
if (list == null)
{
ListName listName = new ListName();
listName.Name = comboBoxList.Text.Trim().ToLower();
entities.AddToListNames(listName);
entities.SaveChanges();
listNameId = listName.Id;
}
else
{
listNameId = list.Id;
}
}
else
{
listNameId = Convert.ToInt64(comboBoxList.SelectedValue);
}
String strWord = textBoxWord.Text.Trim().ToLower();
ListName ln = (from n in entities.ListNames
where n.Id == listNameId
select n).FirstOrDefault();
try
{
GreWord word;
word = (from w in entities.GreWords
where w.Word.ToLower() == strWord
select w).FirstOrDefault();
if (word == null)
{
word = new GreWord()
{
Word = strWord,
};
entities.AddToGreWords(word);
entities.SaveChanges();
}
ListedWord lw = (from w in entities.ListedWords
where w.ListName.Id == ln.Id && w.GreWord.Word == word.Word
select w).FirstOrDefault();
if (lw == null)
{
lw = new ListedWord()
{
GreWord = word,
ListName = ln
};
entities.AddToListedWords(lw);
entities.SaveChanges();
}
}
catch (Exception exp)
{
MessageBox.Show(exp.Message);
}
ParserHelper parserHelper = new ParserHelper(entities);
parserHelper.OnLogMessage += (sender2, person) => this.Invoke((MethodInvoker)(() => textBoxInfo.AppendText(person + "\r\n")));
parserHelper.WordToFetch = strWord;
var workerThread = new Thread(parserHelper.FetchAndParseSingleWord)
{
IsBackground = true
};
workerThread.Start();
}
