C# simple odata client with no $metadata needed for someone only use OdataQueryOptions for Query api like me
Install-Package Hcs.Extensions.OdataClient
var httpClient = new HttpClient { BaseAddress = new Uri("https://localhost:44326/") };
var maxId = 100;
var req = httpClient.GetOdataClient<File>("/api/FileManage")
.Where(x => x.CategoryId.HasValue)
.Where(x => x.CategoryId.HasValue && x.Category.Id + 10 < maxId)
.OrderBy(x => x.Path).ThenBy(x => x.Name)
.Take(100)
.Select(x => new { brand_new_id_property = x.Id, x.Path, xxxx = new { x.Category.Name } });
Console.WriteLine(req.GetQueryString(encode: false));
foreach (var result in await req.SendReqeust())
{
Console.WriteLine($"{result.brand_new_id_property} {result.xxxx.Name}");
}
?$filter=CategoryId ne null and (CategoryId ne null and (Category/Id add 10) lt 100)&$orderby=Path asc,Name asc&$top=100&$count=true&$select=Id,Path&$expand=Category($select=Name)
3 A
2 A
4 A
5 A
for current version of this lib IS NOT IQueryable
implementation
use Take
and Skip
will give you new query instance but just simple replace skip/take value,behavior is not like linq (eq [1,2,3,4,5].Take(5).Take(3) will get [1,2,3]
)
same as Take/Skip
but you still can use ThenBy/ThenByDesc
for $orderby=A asc,B asc
combine with take/skip will work like following
// if all data is [{a:1},{a:2},{a:3},{a:4},{a:5}]
query.Take(4) //this will not affect query result
.OrderBy(x=>x.a) //this will not affect query result
.Take(2) // replace $top value to 2
.Skip(1)
.OrderByDesc(x=>x.a) // replace old order expressions
will give you $orderby=a desc&$top=2&$skip=1
[{a=5},{a=4}]
for reduce complexity of expression parser Select
can only apply once to the query, select expression
is only for generate odata $select/$expand,then lib compile the expression for local projection use(Enumerable.Select
),lib will insert nested member access null check for you , for example
x = >new{ x.Category.Name }
will modify to
x => new { Name = x.Category != null ? x.Category.Name : default(string) }
when use on result projection, so don't worry about null check.
Microsoft.AspNetCore.OData
services.AddOData();
services.AddODataQueryFilter();
services.AddSingleton<ODataUriResolver, StringAsEnumResolver>();
//for .net core 3.1 you need replace built in json serializer for odata $select/$expand working
services.AddControllersWithViews().AddNewtonsoftJson(options =>
{
options.SerializerSettings.NullValueHandling = Newtonsoft.Json.NullValueHandling.Ignore;
options.SerializerSettings.ContractResolver = new Newtonsoft.Json.Serialization.DefaultContractResolver();
options.SerializerSettings.StringEscapeHandling = Newtonsoft.Json.StringEscapeHandling.EscapeHtml;
});
app.UseEndpoints(endpoints =>
{
endpoints.EnableDependencyInjectionForOdata();
endpoints.Count().Filter().OrderBy().Expand().Select().MaxTop(100); //allow complex query function
// ....
});
[Route("api/[controller]")]
[ApiController]
public class FileManageController : ControllerBase
{
DbContext context;
public FileManageController(DbContext context)
{
this.context = context;
}
[EnableQuery]
public IActionResult Get()
{
return Ok(context.Set<File>().AsQueryable());
}
}
Default ResultParser get count from response header
you can replace odata's EnableQueryAttribute to custom ActionFilter
like this
public class HcsEnableQueryAttribute : EnableQueryAttribute
{
public override void OnActionExecuted(ActionExecutedContext actionExecutedContext)
{
base.OnActionExecuted(actionExecutedContext);
var odata = actionExecutedContext.HttpContext.ODataFeature();
if (odata.TotalCount.HasValue)
{
actionExecutedContext.HttpContext.Response.Headers.Add("x-total-count", odata.TotalCount.Value.ToString("#0"));
}
}
}
you can pass other result parser to OdataClient constructor for your response format
new OdataClient<File>(httpClient,"/api/FileManage",resultParser:new PageResultParser<File>())
PageResultParser is built-in parser for starndard odata endpoint
I wrote this lib in rush (about 20ish hours),so use with your own risk, feel free to file a PR if you encounter any bugs