public override void AddToIndex(Dictionary <Tuple <string, string>, Term> index)
{
var task = db.Table <CedictEntry>().ToListAsync();
var entries = task.Result;
var indices = new List <Term>();
foreach (var entry in entries)
{
var key = Tuple.Create(entry.Traditional, entry.Pinyin);
Term term;
if (index.TryGetValue(key, out term))
{
term.CedictEntryId = entry.Id;
}
else
{
term = new Term()
{
Traditional = entry.Traditional,
Simplified = entry.Simplified,
Pinyin = Pinyin.ConvertToAccents(entry.Pinyin),
PinyinNumbered = entry.Pinyin,
PinyinNoNumbers = Pinyin.RemoveNumbersAndUnderscore(entry.Pinyin),
Length = entry.Traditional.Length,
CedictEntryId = entry.Id
};
index[key] = term;
}
}
}
public override void AddToIndex(Dictionary <Tuple <string, string>, Term> index)
{
var entries = db.QueryAsync <MoedictHeteronymLookupForm>(
"SELECT h.Id AS Id, Headword AS Traditional, Pinyin, PinyinNumbered "
+ "FROM MoedictHeteronym h LEFT JOIN MoedictEntry e ON h.EntryId = e.Id").Result;
foreach (var entry in entries)
{
var key = Tuple.Create(entry.Traditional, entry.PinyinNumbered);
Term term;
if (index.TryGetValue(key, out term))
{
term.MoedictHeteronymId = entry.Id;
}
else
{
term = new Term()
{
Traditional = entry.Traditional,
Simplified = TradToSimp.Convert(entry.Traditional).FirstOrDefault(),
Pinyin = entry.Pinyin,
PinyinNumbered = entry.PinyinNumbered,
PinyinNoNumbers = Pinyin.RemoveNumbersAndUnderscore(entry.PinyinNumbered),
Length = entry.Traditional.Length,
MoedictHeteronymId = entry.Id,
};
index[key] = term;
}
}
}
public override async Task <IEnumerable <SearchResult> > Search(CancellationToken ct, string query)
{
var results = new List <SearchResult>();
//try
{
//var entries = await db.Table<CedictEntry>().Where(p => p.Traditional.StartsWith(query)).ToListAsync();
var entries = await db.QueryAsync <CedictEntry>(ct, "SELECT * FROM CedictEntry WHERE Traditional LIKE ? ESCAPE '\\'", query + "%");
foreach (var q in Pinyin.ToQueryForms(query))
{
var newEntries = await db.QueryAsync <CedictEntry>(ct,
"SELECT * FROM (SELECT * FROM CedictEntry WHERE PinyinNoNumbers LIKE ? ESCAPE '\\') "
+ "WHERE Pinyin LIKE ? ESCAPE '\\'", Pinyin.RemoveNumbersAndUnderscore(q) + "%", q + "%");
entries.AddRange(newEntries);
}
foreach (var s in entries)
{
var definitions = await db.QueryAsync <CedictDefinition>(ct, "SELECT * FROM CedictDefinition WHERE EntryId = ?", s.Id);
//var definitions = await db.Table<CedictDefinition>().Where(d => d.EntryId == s.Id).ToListAsync();
results.Add(new SearchResult()
{
Traditional = s.Traditional,
Simplified = s.Simplified,
Pinyin = Pinyin.ConvertToAccents(s.Pinyin),
PinyinNumbered = s.Pinyin,
Definitions = new List <List <string> >()
{
new List <string>(from d in definitions select d.Definition)
}
});
}
}
return(results);
}
public async Task <IEnumerable <Term> > Search(CancellationToken ct, string query)
{
var results = new List <TermWithFreq>();
results.AddRange(await db.QueryAsync <TermWithFreq>(ct, "SELECT * FROM Term LEFT JOIN Frequency ON Simplified = Hanzi OR Traditional = Hanzi WHERE Traditional LIKE ? OR Simplified LIKE ? ESCAPE '\\'", query + "%", query + "%"));
var queryForms = Pinyin.ToQueryForms(query);
int numberOfForms = queryForms.Count();
bool allowMultiCharacter = numberOfForms > 1 || (numberOfForms > 0 && queryForms.First().IndexOf(' ') != -1);
string limitLength = allowMultiCharacter ? "" : "Length = 1 AND ";
foreach (var q in queryForms)
{
if (!string.IsNullOrWhiteSpace(q))
{
results.AddRange(await db.QueryAsync <TermWithFreq>(ct,
"SELECT * FROM (SELECT * FROM (SELECT * FROM Term WHERE " + limitLength + "PinyinNoNumbers LIKE ? ESCAPE '\\') "
+ " WHERE PinyinNumbered LIKE ?) LEFT JOIN Frequency ON Simplified = Hanzi OR Traditional = Hanzi", Pinyin.RemoveNumbersAndUnderscore(q) + "%", q + "%"));
}
}
return(results.OrderBy(r => r.Length)
.ThenBy(r => r.PinyinNumbered != null ? r.PinyinNumbered.Length : int.MaxValue)
.ThenByDescending(r => r.Score, freqComparer)
.ThenBy(r => r.PinyinNumbered));
}
public override void Build()
{
char[] space = new char[] { ' ' };
db.DropTableAsync <MoedictEntry>().Wait();
db.DropTableAsync <MoedictHeteronym>().Wait();
db.DropTableAsync <MoedictDefinition>().Wait();
db.DropTableAsync <MoedictQuote>().Wait();
db.DropTableAsync <MoedictExample>().Wait();
db.DropTableAsync <MoedictLink>().Wait();
db.CreateTableAsync <MoedictEntry>().Wait();
db.CreateTableAsync <MoedictHeteronym>().Wait();
db.CreateTableAsync <MoedictDefinition>().Wait();
db.CreateTableAsync <MoedictQuote>().Wait();
db.CreateTableAsync <MoedictExample>().Wait();
db.CreateTableAsync <MoedictLink>().Wait();
string[] lines = System.IO.File.ReadAllLines("a.txt");
var entries = new List <MoedictEntry>();
var heteronyms = new List <MoedictHeteronym>();
var definitions = new List <MoedictDefinition>();
var quotes = new List <MoedictQuote>();
var examples = new List <MoedictExample>();
var links = new List <MoedictLink>();
foreach (string line in lines)
{
string[] tokens = line.Split(space, 3);
MoedictEntry entry = JsonConvert.DeserializeObject <MoedictEntry>(tokens[2]);
entry.Headword = entry.Title.Replace("`", "").Replace("~", "").Trim();
Match match = extractAlternateHeadword.Match(entry.Headword);
if (match.Success)
{
entry.Headword = match.Groups[1].Value;
entry.AlternateHeadword = match.Groups[2].Value;
}
entries.Add(entry);
}
db.InsertAllAsync(entries).Wait();
foreach (var e in entries)
{
foreach (var h in e.Heteronyms)
{
h.EntryId = e.Id;
if (h.Pinyin != null)
{
Match match = extractAlternatePinyin.Match(h.Pinyin);
if (match.Success)
{
h.Pinyin = match.Groups[1].Value.Trim();
h.AlternatePinyinNote = match.Groups[2].Value;
h.AlternatePinyin = match.Groups[3].Value.Trim();
}
h.PinyinNumbered = Pinyin.ConvertToNumbers(h.Pinyin);
h.AlternatePinyinNumbered = Pinyin.ConvertToNumbers(h.AlternatePinyin);
}
heteronyms.Add(h);
}
}
db.InsertAllAsync(heteronyms).Wait();
foreach (var h in heteronyms)
{
foreach (var d in h.Definitions)
{
d.HeteronymId = h.Id;
definitions.Add(d);
}
}
db.InsertAllAsync(definitions).Wait();
foreach (var d in definitions)
{
if (d.Quotes != null)
{
foreach (var quote in d.QuoteStrings)
{
//quote.DefinitionId = d.Id;
//quotes.Add(quote);
quotes.Add(new MoedictQuote()
{
DefinitionId = d.Id, Quote = quote
});
}
}
if (d.Examples != null)
{
foreach (var example in d.ExampleStrings)
{
//example.DefinitionId = d.Id;
//examples.Add(example);
examples.Add(new MoedictExample()
{
DefinitionId = d.Id, Example = example
});
}
}
if (d.Links != null)
{
foreach (var link in d.LinkStrings)
{
//link.DefinitionId = d.Id;
//links.Add(link);
links.Add(new MoedictLink()
{
DefinitionId = d.Id, Link = link
});
}
}
}
db.InsertAllAsync(quotes).Wait();
db.InsertAllAsync(examples).Wait();
db.InsertAllAsync(links).Wait();
}