public static async Task <Word> GetWord(
IIndexHolder _indexHolder,
AdminContext _dbContext,
ZhConverter zhConverter,
DocRef docRef,
Mode mode = Mode.Full)
{
var maybeYngdiengDocument = _indexHolder.GetIndex().YngdiengDocuments
.SingleOrDefault(yDoc => yDoc.DocId == DocRefs.Encode(docRef));
var maybeWordId = string.IsNullOrEmpty(docRef.ZingzeuId)
? (int?)null : int.Parse(docRef.ZingzeuId, NumberStyles.HexNumber);
var maybeWord = maybeWordId == null
? null
: await _dbContext.Words
.Where(w => w.WordId == maybeWordId)
.Include(w => w.PreferredSandhiAudio)
.SingleOrDefaultAsync();
if (maybeYngdiengDocument == null && maybeWord == null)
{
throw new RpcException(new Status(StatusCode.NotFound, "Not found."));
}
var extensions = await _dbContext.Extensions.Where(e => e.WordId == maybeWordId).ToListAsync();
var prons = await _dbContext.Prons.Where(p => p.WordId == maybeWordId).ToListAsync();
var hanzi = zhConverter.tH(await GetHanzi(_dbContext, maybeYngdiengDocument, maybeWordId));
var recommendedProns = mode == Mode.Snippet
? PronChooser.GetSnippetPronunciations(maybeYngdiengDocument, prons, maybeWord?.PreferredSandhiAudio)
: PronChooser.GetRecommendedPronunciations(maybeYngdiengDocument, prons, maybeWord?.PreferredSandhiAudio);
var explanations = mode == Mode.Snippet
? Enumerable.Empty <RichTextNode>()
: GetExplanations(zhConverter, hanzi, maybeYngdiengDocument, extensions);
var audioCards = mode == Mode.Snippet ? Enumerable.Empty <Word.Types.AudioCard>()
: await GetAudioCards(_dbContext, recommendedProns, maybeWordId);
var wordLists = mode == Mode.Full && maybeWordId.HasValue
? await Queries.QueryWordListsByWordId(_dbContext, maybeWordId.Value).Select(wl => Renderers.ToWordList(wl)).ToListAsync()
: Enumerable.Empty <WordList>();
return(new Word
{
Name = ResourceNames.ToWordName(docRef),
Hanzi = hanzi,
Pronunciations = { recommendedProns },
Explanation = { explanations },
Snippet = zhConverter.tH(GetSnippet(maybeYngdiengDocument, extensions)),
AudioCards = { audioCards },
WordLists =
{
wordLists
}
});
}
public static string ToWordName(DocRef docRef)
{
return(string.Format("words/{0}", DocRefs.Encode(docRef)));
}
public static IEnumerable <YngdiengDocument> Combine(
IEnumerable <zingzeudata.ZingzeuData.Models.ZingzeuEntry> zingzeuEntries,
IEnumerable <HistoricalDocument> historicalDocs,
IEnumerable <FengDocument> fengDocs,
IEnumerable <ContribDocument> contribDocs)
{
var results = new List <YngdiengDocument>();
var pendingFeng = new List <FengDocument>(fengDocs);
var pendingHistorical = new List <HistoricalDocument>(historicalDocs);
var pendingContrib = new List <ContribDocument>(contribDocs);
// 第一阶段: 有 zingzeu_id 的，按相同 zingzeu_id 合并成 YngdiengDocument
foreach (var zingzeuWordsEntry in zingzeuEntries)
{
var zingzeuId = zingzeuWordsEntry.Id;
var tmp = new YngdiengDocument
{
DocRef = new Yngdieng.Protos.DocRef {
ZingzeuId = zingzeuId
},
HanziCanonical = new Hanzi {
Regular = zingzeuWordsEntry.Hanzi
},
YngpingSandhi = FindZingzeuWordsPron(zingzeuWordsEntry) ?? string.Empty,
YngpingUnderlying =
FindZingzeuWordsPron(zingzeuWordsEntry,
ZingzeuEntry.Types.ZingzeuPron.Types.SandhiCategory.Bengzi) ?? string.Empty,
HanziAlternatives = { zingzeuWordsEntry.HanziAlt.Select(a => new Hanzi {
Regular = a
}) },
IndexingExtension = new YngdiengDocument.Types.IndexingExtension
{
MandarinWords = { zingzeuWordsEntry.MandarinWords }
}
};
var historicalMatches = pendingHistorical.Where(h => h.ZingzeuId == zingzeuId).ToArray();
foreach (var historicalMatch in historicalMatches)
{
pendingHistorical.Remove(historicalMatch);
tmp.Sources.Add(new YngdiengDocument.Types.Source {
CiklinDfd = historicalMatch
});
}
var fengMatches = pendingFeng.Where(f => f.ZingzeuId == zingzeuId).ToArray();
foreach (var fengMatch in fengMatches)
{
pendingFeng.Remove(fengMatch);
tmp.Sources.Add(new YngdiengDocument.Types.Source {
Feng = fengMatch
});
}
var contribMatches =
pendingContrib.Where(c => c.ZingzeuId == zingzeuId).ToList();
foreach (var contribMatch in contribMatches)
{
pendingContrib.Remove(contribMatch);
tmp.Sources.Add(new YngdiengDocument.Types.Source {
Contrib = contribMatch
});
}
results.Add(tmp);
}
Console.WriteLine($"{results.Count} matched.");
// 第二阶段: 无 zingzeu_id 的，各自单独生成 YngdiengDocument
Console.WriteLine($"{pendingHistorical.Count} unmatched historicalDocs.");
Console.WriteLine($"{pendingFeng.Count} unmatched fengDocs.");
Console.WriteLine($"{pendingContrib.Count} unmatched contribDocs.");
foreach (var h in pendingHistorical)
{
results.Add(new YngdiengDocument
{
DocRef = new Yngdieng.Protos.DocRef {
HistoricalDocId = h.Id
},
Sources = { new YngdiengDocument.Types.Source {
CiklinDfd = h
} }
});
}
foreach (var f in pendingFeng)
{
results.Add(
new YngdiengDocument
{
DocRef = new Yngdieng.Protos.DocRef {
FengId = f.Id
},
Sources = { new YngdiengDocument.Types.Source {
Feng = f
} }
});
}
foreach (var c in pendingContrib)
{
results.Add(new YngdiengDocument
{
DocRef = new Yngdieng.Protos.DocRef {
ContribRowNumber = c.RowNumber
},
Sources = { new YngdiengDocument.Types.Source {
Contrib = c
} }
});
}
foreach (var doc in results)
{
doc.DocId = DocRefs.Encode(doc.DocRef);
doc.HanziCanonical = FindHanziCanonical(doc.Sources, doc.HanziCanonical);
doc.YngpingUnderlying = FindYngpingUnderlying(doc.Sources, doc.YngpingUnderlying);
doc.YngpingSandhi = FindYngpingSandhi(doc.Sources, doc.YngpingSandhi);
doc.IndexingExtension ??= new YngdiengDocument.Types.IndexingExtension {
};
doc.IndexingExtension.YngpingPermutations.AddRange(CollectYngpingPermutations(doc.Sources));
doc.IndexingExtension.HanziMatchable.AddRange(CollectHanziMatchable(doc.Sources));
doc.IndexingExtension.ExplanationText.AddRange(CollectExplanationTexts(doc.Sources));
}
return(results);
}