public static ISearchResponse <Person> Search_Scroll(SearchParam p)
{
/**
* if p.from >= 10000, we can only use this methods to get data
* p.size must be less then 10000, and the better choices are 10, 20, 50, 100, 200, 500, 1000, 2000, 5000
* */
if (p.from % p.size != 0)
{
throw new Exception("p.from must be integrally divided by p.size");
}
var discard_count = p.from / p.size;
int count = 0;
ISearchResponse <Person> resp = null;
resp = ESClientInst.Client.Search <Person>(s => s.Scroll("1m").Index(p.index).Type(p.type).Take(p.size)
.Query(q => q.Match(m => m.Field("company.max").MinimumShouldMatch(MinimumShouldMatch.Percentage(80)).Query(p.keyword)))
.Sort(sd => p.isAsc ? sd.Ascending(p.sort.ToString()) : sd.Descending(p.sort.ToString())));
string scroll_id = resp.ScrollId;
while (count < discard_count && resp.Documents.Any() && scroll_id != null)
{
count++;
resp = ESClientInst.Client.Scroll <Person>("1m", scroll_id);
ESClientInst.Client.ClearScroll(cs => cs.ScrollId(scroll_id));
scroll_id = resp.ScrollId;
}
return(resp);
}
public static ISearchResponse <Person> SearchMany(SearchParam p)
{
var resp = ESClientInst.Client.Search <Person>(s => s.Index(p.index).Type(p.type).From(p.from).Take(p.size)
.Sort(sd => GetHourseSortDescriptor(sd, p))
.Source(sr => GetSourceFilter(sr, p))
.Aggregations(agg => GetAggContainer(agg, p))
.Query(q => q.Terms(t => t.Field(p.batch_cond.Key.ToString()).Terms(p.batch_cond.Value)))
);
return(resp);
}
public static ISearchResponse <Person> SearchByCom(SearchParam p)
{
var resp = ESClientInst.Client.Search <Person>(s => s.Index(p.index).Type(p.type).From(p.from).Take(p.size)
.Sort(sd => GetHourseSortDescriptor(sd, p))
.Source(sr => GetSourceFilter(sr, p))
.Aggregations(agg => GetAggContainer(agg, p))
.Highlight(hl => GetHLDescriptor(hl, p))
.Query(q => q.DisMax(d => d.Queries(dq => dq.Term(t => t.Field(f => f.company).Value(p.keyword).Boost(100)),
dq => dq.MatchPhrase(m => m.Field("company.max").Query(p.keyword).Boost(50)),
dq => dq.MatchPhrase(m => m.Field("company.std").Query(p.keyword).Boost(1)))))
);
return(resp);
}
public static ISearchResponse <Person> SearchHouse(SearchParam p)
{
var resp = ESClientInst.Client.Search <Person>(s => s.Index(p.index).Type(p.type).From(p.from).Take(p.size)
.Sort(sd => GetHourseSortDescriptor(sd, p))
.Source(sr => GetSourceFilter(sr, p))
.Aggregations(agg => GetAggContainer(agg, p))
.Query(q => q.GeoDistance(g => g.Distance(p.distance, DistanceUnit.Kilometers)
.DistanceType(GeoDistanceType.Arc)
.Field(f => f.house)
.Location(p.lat, p.lon)))
);
return(resp);
}
/// <summary>
/// 按城市统计平均薪水
/// </summary>
/// <param name="p"></param>
/// <returns></returns>
public ISearchResponse <Person> AverageSalaryByCity(SearchParam p)
{
var resp = ESClientInst.Client.Search <Person>(s => s.Index(p.index).Type(p.type).Take(0)
.Aggregations(agg => agg
.Terms("city", t => t
.Field(f => f.city)
.Size(100)
.MinimumDocumentCount(50)
.Aggregations(sa => sa
.Average("average_salary", ave => ave
.Field(f => f.salary))
)
)
.AverageBucket("ave_sal_city_buck", ab => ab.BucketsPath("city>average_salary"))
)
);
return(resp);
}
/**
* statistics
*
* */
/// <summary>
/// 给定公司关键词，按毕业年份统计平均薪水
/// </summary>
/// <param name="p"></param>
/// <returns></returns>
public ISearchResponse <Person> AverageSalaryByYear(SearchParam p)
{
var resp = ESClientInst.Client.Search <Person>(s => s.Index(p.index).Type(p.type).Take(0)
.Query(q => q.MatchPhrase(m => m.Field("company.max").Query(p.keyword)))
.Aggregations(agg => agg
.DateHistogram($"average_salary_year", dh => dh
.Field(f => f.graduate)
.Interval(DateInterval.Year)
.Aggregations(sa => sa
.Average("average_salary", ave => ave
.Field(f => f.salary)
//.Script(scr => scr.Source("doc['salary']").Lang("expression"))
)
)
)
)
);
return(resp);
}
private static ISourceFilter GetSourceFilter(SourceFilterDescriptor <Person> sr, SearchParam p)
{
if (p.isInclude)
{
if (p.sources == null || p.sources.Count == 0)
{
return(sr.ExcludeAll());
}
else
{
return(sr.Includes(inc => inc.Fields(p.sources.Select(s => s.ToString()).ToArray())));
}
}
else
{
if (p.sources == null || p.sources.Count == 0)
{
return(sr.IncludeAll());
}
else
{
return(sr.Excludes(exc => exc.Fields(p.sources.Select(s => s.ToString()).ToArray())));
}
}
}
private static SortDescriptor <Person> GetHourseSortDescriptor(SortDescriptor <Person> sd, SearchParam p)
{
return(sd.GeoDistance(g => g.DistanceType(GeoDistanceType.Arc)
.Points(new GeoLocation(p.lat, p.lon))
.Field(f => f.house)
.Order(SortOrder.Ascending)));
}
private static IAggregationContainer GetAggContainer(AggregationContainerDescriptor <Person> agg, SearchParam p)
{
if (p.aggs != null)
{
foreach (var a in p.aggs)
{
switch (a)
{
case PersonEnum.age:
agg.Range("age", r => r.Field(f => f.age).Ranges(rs => rs.To(20),
rs => rs.From(20).To(30),
rs => rs.From(30).To(40),
rs => rs.From(40).To(50),
rs => rs.From(50).To(60),
rs => rs.From(60)));
break;
case PersonEnum.graduate:
agg.DateHistogram("graduate", d => d.Field(f => f.graduate)
.Interval(DateInterval.Year)
.MinimumDocumentCount(1)
.ExtendedBounds("1980", "2050"));
break;
case PersonEnum.isMan:
agg.Terms("isMan", t => t.Field(f => f.isMan));
break;
case PersonEnum.city:
agg.Terms("city", t => t.Field(f => f.city));
break;
}
}
}
return(agg);
}
private static HighlightDescriptor <Person> GetHLDescriptor(HighlightDescriptor <Person> hl, SearchParam p)
{
return(hl.PreTags("<b>")
.PostTags("</b>")
.Fields(f => f.Field("company.std"),
f => f.Field("company.max"),
f => f.Field(fld => fld.company)));
}
public static ESWrapper <Person> SearchAndHandle(
SearchParam p,
Func <SearchParam, ISearchResponse <Person> > search,
Func <ISearchResponse <Person>, ESWrapper <Person> > handle) =>
handle(search(p));