public void BuildSercherIndexToSQLDB(Action <double, string> IndexesProgress = null)
{
//hashLoadBalance.RemoveAllDBData();
//hashLoadBalance = new ConsistentHashLoadBalance();
SetServerDBCount();
RedBlackTree <string, string> documentIndices_cachList = new RedBlackTree <string, string>();
var DocumentToatalList = documentDB.GetNotIndexDocument();
int remainder = DocumentToatalList.Count;
var remotewords = SercherIndexesDB.GetWords(hashLoadBalance.GetServerNodes());
var localwords = new HashSet <string>();
Dictionary <string, TextComponent> textComponent = new Dictionary <string, TextComponent>();//使用到的时候进行缓存
int curWordCachNum = 0;
for (int i = 0, j = 0; i < DocumentToatalList.Count; i++)
{
var doc = DocumentToatalList[i];
documentDB.UpdateDocumentStateIndexStatus(doc._id, "pro_" + Config.CurrentConfig.IndexesServiceName);
IEnumerable <SegmenterToken> textSplit = Pretreatment(doc);
Dictionary <string, DocumentIndex> documentIndices = new Dictionary <string, DocumentIndex>();
int wordTotal = textSplit.Count();
foreach (var token in textSplit)
{
string word = token.Word.Trim().ToLower();
if (!remotewords.Contains(word))
{
if (!localwords.Contains(word))
{
localwords.Add(word);
remotewords.Add(word);
}
}
//记录一个文档的所有相同词汇
if (documentIndices.TryGetValue(word, out DocumentIndex documentIndex))
{
documentIndex.WordFrequency++;
if (documentIndex.WordFrequency <= Config.CurrentConfig.MaxIndexWordStartLocation)
{
documentIndex.BeginIndex += ',' + token.StartIndex.ToString();
}
documentIndex.DocumentWordTotal = wordTotal;
}
else
{
documentIndices[word] = new DocumentIndex
{
IndexTime = DateTime.Now.Ticks,
DocId = doc._id,
WordFrequency = 1,
BeginIndex = token.StartIndex.ToString(),
DocumentWordTotal = wordTotal,
Permission = doc.Permission == 0 ? Config.CurrentConfig.DefaultPermission : doc.Permission
}
};
}
//转换为脚本并加入全局缓存等待上传
documentIndices.AsParallel().ForAll(kvp =>
{
//UpdateIndex(kvp.Key, kvp.Value);
if (documentIndices_cachList.ContainsKey(kvp.Key.ToString()))
{
string sql = InsetValueIntoMemory(kvp.Key, new DocumentIndex[1] {
kvp.Value
}, false);
lock (lockobj1)//因为此循环内Key唯一，所以只锁了添加代码
{
documentIndices_cachList[kvp.Key] += "," + sql;
}
}
else
{
string sql = InsetValueIntoMemory(kvp.Key, new DocumentIndex[1] {
kvp.Value
}, true);
lock (lockobj1)
{
documentIndices_cachList.Add(kvp.Key, sql);
}
}
});
remainder--;
IndexesProgress?.Invoke(i / (double)DocumentToatalList.Count, "文档：" + doc.Name + " 缓存完成");
curWordCachNum += documentIndices.Count;
documentIndices.Clear();
if (Config.CurrentConfig.MaxIndexCachWordNum < curWordCachNum || i == DocumentToatalList.Count - 1)
{
IndexesProgress?.Invoke(i / (double)DocumentToatalList.Count, "以达缓存上限，开始创建表");
//对每一个同数据库的词汇的脚本进行组合,创建表
var group1 = localwords.GroupBy(w => hashLoadBalance.FindCloseServerDBsByTableName(w).DbName).ToArray();
System.Diagnostics.Stopwatch watch = new Stopwatch();
watch.Start();
Parallel.ForEach(group1, g =>
{
var wordgroup = g.ToArray();
hashLoadBalance.GetServerNodes().First(n => n.DbName == g.Key) //!##GroupKey欠妥，不过数据库比较少的时候影响不大
.CreateIndexTable(wordgroup);
IndexesProgress?.Invoke(i / (double)DocumentToatalList.Count, g.Key + ":一组表创建完成");
});
watch.Stop();
IndexesProgress?.Invoke(i / (double)DocumentToatalList.Count, "表创建完成，用时(s)：" + watch.ElapsedMilliseconds / 1000);
localwords.Clear();
IndexesProgress?.Invoke(i / (double)DocumentToatalList.Count, "开始上传索引");
//对每一个同数据库的词汇的脚本进行组合，上传
var group2 = documentIndices_cachList.AsQueryable().GroupBy(kv => hashLoadBalance.FindCloseServerDBsByTableName(kv.Key).DbName).ToArray();
watch.Restart();
Parallel.ForEach(group2, new ParallelOptions()
{
MaxDegreeOfParallelism = Config.CurrentConfig.UploadThreadNum
}, g =>
{
//上传此db的inser脚本
hashLoadBalance.FindCloseServerDBsByTableName(g.First().Key)
.UploadDocumentIndex(g.Select(s => s.Value + ";").ToArray());
IndexesProgress?.Invoke(i / (double)DocumentToatalList.Count, g.Key + ":一组索引创建完成");
});
watch.Stop();
IndexesProgress?.Invoke(i / (double)DocumentToatalList.Count, "上传索引完成，用时(s)：" + watch.ElapsedMilliseconds / 1000);
documentIndices_cachList.Clear();
while (j <= i)
{
documentDB.UpdateDocumentStateIndexStatus(DocumentToatalList[j]._id, "yes");
j++;
}
curWordCachNum = 0;
IndexesProgress?.Invoke(i / (double)DocumentToatalList.Count, "一批上传完成，刷新缓存");
}
}
}