public IEnumerable <IRow> Read()
{
ElasticsearchResponse <DynamicResponse> response;
ElasticsearchDynamicValue hits;
var from = 1;
var size = 10;
string body;
if (_context.Entity.IsPageRequest())
{
from = (_context.Entity.Page * _context.Entity.PageSize) - _context.Entity.PageSize;
body = WriteQuery(_fields, _readFrom, _context, from, _context.Entity.PageSize);
}
else
{
body = WriteQuery(_fields, _readFrom, _context, 0, 0);
response = _client.Search <DynamicResponse>(_context.Connection.Index, _typeName, body);
if (response.Success)
{
hits = response.Body["hits"] as ElasticsearchDynamicValue;
if (hits != null && hits.HasValue)
{
var properties = hits.Value as IDictionary <string, object>;
if (properties != null && properties.ContainsKey("total"))
{
size = Convert.ToInt32(properties["total"]);
body = WriteQuery(_fields, _readFrom, _context, 0, size > 10000 ? 10000 : size);
}
}
}
}
_context.Debug(() => body);
_context.Entity.Query = body;
// move 10000 to configurable limit
response = from + size > 10000
? _client.Search <DynamicResponse>(_context.Connection.Index, _typeName, body, p => p.Scroll(TimeSpan.FromMinutes(1.0)))
: _client.Search <DynamicResponse>(_context.Connection.Index, _typeName, body);
if (!response.Success)
{
LogError(response);
yield break;
}
_context.Entity.Hits = Convert.ToInt32((response.Body["hits"]["total"] as ElasticsearchDynamicValue).Value);
hits = response.Body["hits"]["hits"] as ElasticsearchDynamicValue;
if (hits == null || !hits.HasValue)
{
_context.Warn("No hits from elasticsearch");
yield break;
}
var docs = hits.Value as IList <object>;
if (docs == null)
{
_context.Error("No documents returned from elasticsearch!");
yield break;
}
// if any of the fields do not exist, yield break
if (docs.Count > 0)
{
var doc = docs.First() as IDictionary <string, object>;
var source = doc?["_source"] as IDictionary <string, object>;
if (source == null)
{
_context.Error("Missing _source from elasticsearch response!");
yield break;
}
for (var i = 0; i < _fields.Length; i++)
{
if (source.ContainsKey(_fieldNames[i]))
{
continue;
}
_context.Error($"Field {_fieldNames[i]} does not exist!");
yield break;
}
}
var count = 0;
foreach (var d in docs)
{
var doc = (IDictionary <string, object>)d;
var row = _rowFactory.Create();
var source = (IDictionary <string, object>)doc["_source"];
for (var i = 0; i < _fields.Length; i++)
{
row[_fields[i]] = _fields[i].Convert(source[_fieldNames[i]]);
}
_context.Increment();
yield return(row);
}
count += docs.Count;
// get this from first search response (maybe), unless you have to aggregate it from all...
foreach (var filter in _context.Entity.Filter.Where(f => f.Type == "facet" && !string.IsNullOrEmpty(f.Map)))
{
var map = _context.Process.Maps.First(m => m.Name == filter.Map);
var buckets = response.Body["aggregations"][filter.Key]["buckets"] as ElasticsearchDynamicValue;
if (buckets == null || !buckets.HasValue)
{
continue;
}
var items = buckets.Value as IEnumerable <object>;
if (items == null)
{
continue;
}
foreach (var item in items.OfType <IDictionary <string, object> >())
{
map.Items.Add(new MapItem {
From = $"{item["key"]} ({item["doc_count"]})", To = item["key"]
});
}
}
if (!response.Body.ContainsKey("_scroll_id"))
{
yield break;
}
if (size == count)
{
_client.ClearScroll <DynamicResponse>(new PostData <object>(new { scroll_id = response.Body["_scroll_id"].Value }));
yield break;
}
var scrolls = new HashSet <string>();
do
{
var scrollId = response.Body["_scroll_id"].Value;
scrolls.Add(scrollId);
response = _client.Scroll <DynamicResponse>(new PostData <object>(new { scroll = "1m", scroll_id = scrollId }));
if (response.Success)
{
docs = (IList <object>)response.Body["hits"]["hits"].Value;
foreach (var d in docs)
{
var doc = (IDictionary <string, object>)d;
var row = _rowFactory.Create();
var source = (IDictionary <string, object>)doc["_source"];
for (var i = 0; i < _fields.Length; i++)
{
row[_fields[i]] = _fields[i].Convert(source[_fieldNames[i]]);
}
_context.Increment();
yield return(row);
}
count += docs.Count;
}
else
{
LogError(response);
}
} while (response.Success && count < size);
_client.ClearScroll <DynamicResponse>(new PostData <object>(new { scroll_id = scrolls.ToArray() }));
}
public IEnumerable <IRow> Read()
{
if (_parent.Entities.Sum(e => e.Inserts + e.Updates + e.Deletes) == 0)
{
yield break;
}
;
var batches = _parent.Entities.Select(e => e.BatchId).ToArray();
var minBatchId = batches.Min();
var maxBatchId = batches.Max();
_output.Info("Batch Range: {0} to {1}.", minBatchId, maxBatchId);
var threshold = minBatchId - 1;
var sql = string.Empty;
if (_cf.AdoProvider == AdoProvider.SqlCe)
{
// because SqlCe doesn't support views, re-construct the parent view's definition
var ctx = new PipelineContext(_output.Logger, _parent);
var master = _parent.Entities.First(e => e.IsMaster);
var builder = new StringBuilder();
builder.AppendLine($"SELECT {string.Join(",", _output.Entity.Fields.Where(f => f.Output).Select(f => _cf.Enclose(f.Source.Split('.')[0]) + "." + _cf.Enclose(f.Source.Split('.')[1])))}");
foreach (var from in ctx.SqlStarFroms(_cf))
{
builder.AppendLine(@from);
}
builder.AppendLine($"WHERE {_cf.Enclose(Utility.GetExcelName(master.Index))}.{_cf.Enclose(master.TflBatchId().FieldName())} > @Threshold;");
sql = builder.ToString();
}
else
{
sql = $@"
SELECT {string.Join(",", _output.Entity.Fields.Where(f => f.Output).Select(f => _cf.Enclose(f.Alias)))}
FROM {_cf.Enclose(_output.Process.Star)} {(_cf.AdoProvider == AdoProvider.SqlServer ? "WITH (NOLOCK)" : string.Empty)}
WHERE {_cf.Enclose(Constants.TflBatchId)} > @Threshold;";
}
_output.Debug(() => sql);
using (var cn = _cf.GetConnection()) {
cn.Open();
var cmd = cn.CreateCommand();
cmd.CommandTimeout = 0;
cmd.CommandType = CommandType.Text;
cmd.CommandText = sql;
var min = cmd.CreateParameter();
min.ParameterName = "@Threshold";
min.Value = threshold;
min.Direction = ParameterDirection.Input;
min.DbType = DbType.Int32;
cmd.Parameters.Add(min);
var reader = cmd.ExecuteReader(CommandBehavior.SequentialAccess);
var rowCount = 0;
var fieldArray = _output.Entity.Fields.ToArray();
while (reader.Read())
{
rowCount++;
_output.Increment();
yield return(_rowCreator.Create(reader, fieldArray));
}
_output.Info("{0} from {1}", rowCount, _output.Connection.Name);
}
}
public IEnumerable <IRow> Read()
{
ElasticsearchResponse <DynamicResponse> response;
string body;
if (_context.Entity.IsPageRequest())
{
var from = (_context.Entity.Page * _context.Entity.PageSize) - _context.Entity.PageSize;
body = WriteQuery(_fields, _readFrom, _context, from, _context.Entity.PageSize);
}
else
{
body = WriteQuery(_fields, _readFrom, _context, 0, 0);
response = _client.Search <DynamicResponse>(_context.Connection.Index, _typeName, body);
if (response.Success)
{
var hits = response.Body["hits"] as ElasticsearchDynamicValue;
if (hits != null && hits.HasValue)
{
var properties = hits.Value as IDictionary <string, object>;
if (properties != null && properties.ContainsKey("total"))
{
var size = Convert.ToInt32(properties["total"]) + 1;
body = WriteQuery(_fields, _readFrom, _context, 0, size);
}
}
}
}
_context.Debug(() => body);
_context.Entity.Query = body;
response = _client.Search <DynamicResponse>(_context.Connection.Index, _typeName, body);
if (response.Success)
{
_context.Entity.Hits = Convert.ToInt32((response.Body["hits"]["total"] as ElasticsearchDynamicValue).Value);
var hits = response.Body["hits"]["hits"] as ElasticsearchDynamicValue;
if (hits != null && hits.HasValue)
{
var docs = hits.Value as IEnumerable <object>;
if (docs != null)
{
foreach (var doc in docs.OfType <IDictionary <string, object> >())
{
var row = _rowFactory.Create();
if (doc != null && doc.ContainsKey("_source"))
{
var source = doc["_source"] as IDictionary <string, object>;
if (source != null)
{
for (var i = 0; i < _fields.Length; i++)
{
var field = _fields[i];
row[field] = field.Convert(source[_fieldNames[i]]);
}
}
}
_context.Increment();
yield return(row);
}
}
}
foreach (var filter in _context.Entity.Filter.Where(f => f.Type == "facet" && !string.IsNullOrEmpty(f.Map)))
{
var map = _context.Process.Maps.First(m => m.Name == filter.Map);
var buckets = response.Body["aggregations"][filter.Key]["buckets"] as ElasticsearchDynamicValue;
if (buckets == null || !buckets.HasValue)
{
continue;
}
var items = buckets.Value as IEnumerable <object>;
if (items == null)
{
continue;
}
foreach (var item in items.OfType <IDictionary <string, object> >())
{
map.Items.Add(new MapItem {
From = $"{item["key"]} ({item["doc_count"]})", To = item["key"]
}.WithDefaults());
}
}
}
else
{
_context.Error(response.DebugInformation);
}
}
