/// <summary>
/// Ctor: deserializes binary data.
/// </summary>
public Index(BinReader br)
{
WordHolder = new CedictEngine.WordHolder(br);
SenseIndex = new Dictionary<int, SenseIndexItem>();
int senseIndexKeyCount = br.ReadInt();
for (int i = 0; i != senseIndexKeyCount; ++i)
{
int tokenId = br.ReadInt();
SenseIndexItem sii = new SenseIndexItem(br);
SenseIndex[tokenId] = sii;
}
IdeoIndex = new Dictionary<char, IdeoIndexItem>();
PinyinIndex = new Dictionary<string, PinyinIndexItem>();
int ideoIndexKeyCount = br.ReadInt();
for (int i = 0; i != ideoIndexKeyCount; ++i)
{
char c = br.ReadChar();
IdeoIndexItem iii = new IdeoIndexItem(br);
IdeoIndex[c] = iii;
}
int pinyinIndexKeyCount = br.ReadInt();
for (int i = 0; i != pinyinIndexKeyCount; ++i)
{
string str = br.ReadString();
PinyinIndexItem pyi = new PinyinIndexItem(br);
PinyinIndex[str] = pyi;
}
}
/// <summary>
/// Indexes one parsed Cedict entry (hanzi, pinyin and target-language indexes).
/// </summary>
private void indexEntry(CedictEntry entry, int id)
{
// Index character of simplified headword
foreach (char c in entry.ChSimpl)
{
IdeoIndexItem ii;
if (index.IdeoIndex.ContainsKey(c)) ii = index.IdeoIndex[c];
else
{
ii = new IdeoIndexItem();
index.IdeoIndex[c] = ii;
}
// Avoid indexing same entry twice if a char occurs multiple times
if (ii.EntriesHeadwordSimp.Count == 0 ||
ii.EntriesHeadwordSimp[ii.EntriesHeadwordSimp.Count - 1] != id)
ii.EntriesHeadwordSimp.Add(id);
}
// Index characters of traditional headword
foreach (char c in entry.ChTrad)
{
IdeoIndexItem ii;
if (index.IdeoIndex.ContainsKey(c)) ii = index.IdeoIndex[c];
else
{
ii = new IdeoIndexItem();
index.IdeoIndex[c] = ii;
}
// Avoid indexing same entry twice if a char occurs multiple times
if (ii.EntriesHeadwordTrad.Count == 0 ||
ii.EntriesHeadwordTrad[ii.EntriesHeadwordTrad.Count - 1] != id)
ii.EntriesHeadwordTrad.Add(id);
}
// Index pinyin syllables
foreach (PinyinSyllable pys in entry.Pinyin)
{
PinyinIndexItem pi;
// Index contains lower-case syllables
string textLo = pys.Text.ToLowerInvariant();
if (index.PinyinIndex.ContainsKey(textLo)) pi = index.PinyinIndex[textLo];
else
{
pi = new PinyinIndexItem();
index.PinyinIndex[textLo] = pi;
}
// Figure out which list in index item - by tone
List<int> entryList;
if (pys.Tone == -1) entryList = pi.EntriesNT;
else if (pys.Tone == 0) entryList = pi.Entries0;
else if (pys.Tone == 1) entryList = pi.Entries1;
else if (pys.Tone == 2) entryList = pi.Entries2;
else if (pys.Tone == 3) entryList = pi.Entries3;
else if (pys.Tone == 4) entryList = pi.Entries4;
else throw new Exception("Invalid tone: " + pys.Tone.ToString());
// Avoid indexing same entry twice if a syllable occurs multiple times
if (entryList.Count == 0 || entryList[entryList.Count - 1] != id)
entryList.Add(id);
}
// Index equiv of each sense
int senseIx = -1;
foreach (CedictSense sense in entry.Senses)
{
++senseIx;
// Empty equiv: nothing to index
if (sense.Equiv.IsEmpty) continue;
// Tokenize
ReadOnlyCollection<EquivToken> tokens = tokenizer.Tokenize(sense.Equiv);
// Index sense
indexSense(tokens, id, senseIx);
}
}