yv01p/Umbraco-CMS
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Updates the ContentRepository to not fetch the published version info in the same query when returning a collection of documents, instead we fetch the published version information in a separate query

Browse Source
This commit is contained in:
Shannon
2017-02-02 17:05:33 +11:00
parent 86b2dac231
commit 32d757b13c
6 changed files with 135 additions and 25 deletions
Download Patch File Download Diff File Expand all files Collapse all files

22
src/Umbraco.Core/Persistence/Factories/ContentFactory.cs
View File

@@ -27,7 +27,20 @@ namespace Umbraco.Core.Persistence.Factories
#region Implementation of IEntityFactory<IContent,DocumentDto>
public static IContent BuildEntity(DocumentDto dto, IContentType contentType)
/// <summary>
/// Builds a IContent item from the dto(s) and content type
/// </summary>
/// <param name="dto">
/// This DTO can contain all of the information to build an IContent item, however in cases where multiple entities are being built,
/// a separate <see cref="DocumentPublishedReadOnlyDto"/> publishedDto entity will be supplied in place of the <see cref="DocumentDto"/>'s own
/// ResultColumn DocumentPublishedReadOnlyDto
/// </param>
/// <param name="contentType"></param>
/// <param name="publishedDto">
/// When querying for multiple content items the main DTO will not contain the ResultColumn DocumentPublishedReadOnlyDto and a separate publishedDto instance will be supplied
/// </param>
/// <returns></returns>
public static IContent BuildEntity(DocumentDto dto, IContentType contentType, DocumentPublishedReadOnlyDto publishedDto = null)
{
var content = new Content(dto.Text, dto.ContentVersionDto.ContentDto.NodeDto.ParentId, contentType);
@@ -52,8 +65,13 @@ namespace Umbraco.Core.Persistence.Factories
content.ExpireDate = dto.ExpiresDate.HasValue ? dto.ExpiresDate.Value : (DateTime?)null;
content.ReleaseDate = dto.ReleaseDate.HasValue ? dto.ReleaseDate.Value : (DateTime?)null;
content.Version = dto.ContentVersionDto.VersionId;
content.PublishedState = dto.Published ? PublishedState.Published : PublishedState.Unpublished;
content.PublishedVersionGuid = dto.DocumentPublishedReadOnlyDto == null ? default(Guid) : dto.DocumentPublishedReadOnlyDto.VersionId;
//Check if the publishedDto has been supplied, if not the use the dto's own DocumentPublishedReadOnlyDto value
content.PublishedVersionGuid = publishedDto == null
? (dto.DocumentPublishedReadOnlyDto == null ? default(Guid) : dto.DocumentPublishedReadOnlyDto.VersionId)
: publishedDto.VersionId;
//on initial construction we don't want to have dirty properties tracked
// http://issues.umbraco.org/issue/U4-1946

24
src/Umbraco.Core/Persistence/Repositories/BaseQueryType.cs
View File

@@ -2,8 +2,30 @@ namespace Umbraco.Core.Persistence.Repositories
{
internal enum BaseQueryType
{
Full,
/// <summary>
/// A query to return all information for a single item
/// </summary>
/// <remarks>
/// In some cases this will be the same as <see cref="FullMultiple"/>
/// </remarks>
FullSingle,
/// <summary>
/// A query to return all information for multiple items
/// </summary>
/// <remarks>
/// In some cases this will be the same as <see cref="FullSingle"/>
/// </remarks>
FullMultiple,
/// <summary>
/// A query to return the ids for items
/// </summary>
Ids,
/// <summary>
/// A query to return the count for items
/// </summary>
Count
}
}

107
src/Umbraco.Core/Persistence/Repositories/ContentRepository.cs
View File

@@ -1,5 +1,6 @@
using System;
using System.Collections.Generic;
using System.Collections.ObjectModel;
using System.Globalization;
using System.Linq;
using System.Xml;
@@ -53,7 +54,7 @@ namespace Umbraco.Core.Persistence.Repositories
protected override IContent PerformGet(int id)
{
var sql = GetBaseQuery(false)
var sql = GetBaseQuery(BaseQueryType.FullSingle)
.Where(GetBaseWhereClause(), new { Id = id })
.Where<DocumentDto>(x => x.Newest, SqlSyntax)
.OrderByDescending<ContentVersionDto>(x => x.VersionDate, SqlSyntax);
@@ -81,7 +82,7 @@ namespace Umbraco.Core.Persistence.Repositories
return s;
};
var sqlBaseFull = GetBaseQuery(BaseQueryType.Full);
var sqlBaseFull = GetBaseQuery(BaseQueryType.FullMultiple);
var sqlBaseIds = GetBaseQuery(BaseQueryType.Ids);
return ProcessQuery(translate(sqlBaseFull), new PagingSqlQuery(translate(sqlBaseIds)));
@@ -89,7 +90,7 @@ namespace Umbraco.Core.Persistence.Repositories
protected override IEnumerable<IContent> PerformGetByQuery(IQuery<IContent> query)
{
var sqlBaseFull = GetBaseQuery(BaseQueryType.Full);
var sqlBaseFull = GetBaseQuery(BaseQueryType.FullMultiple);
var sqlBaseIds = GetBaseQuery(BaseQueryType.Ids);
Func<SqlTranslator<IContent>, Sql> translate = (translator) =>
@@ -110,6 +111,18 @@ namespace Umbraco.Core.Persistence.Repositories
#region Overrides of PetaPocoRepositoryBase<IContent>
/// <summary>
/// Returns the base query to return Content
/// </summary>
/// <param name="queryType"></param>
/// <returns></returns>
/// <remarks>
/// Content queries will differ depending on what needs to be returned:
/// * FullSingle: When querying for a single document, this will include the Outer join to fetch the content item's published version info
/// * FullMultiple: When querying for multiple documents, this will exclude the Outer join to fetch the content item's published version info - this info would need to be fetched separately
/// * Ids: This would essentially be the same as FullMultiple however the columns specified will only return the Ids for the documents
/// * Count: A query to return the count for documents
/// </remarks>
protected override Sql GetBaseQuery(BaseQueryType queryType)
{
var sql = new Sql();
@@ -122,14 +135,14 @@ namespace Umbraco.Core.Persistence.Repositories
.InnerJoin<NodeDto>(SqlSyntax)
.On<ContentDto, NodeDto>(SqlSyntax, left => left.NodeId, right => right.NodeId);
if (queryType == BaseQueryType.Full)
if (queryType == BaseQueryType.FullSingle)
{
//The only reason we apply this left outer join is to be able to pull back the DocumentPublishedReadOnlyDto
//information with the entire data set, so basically this will get both the latest document and also it's published
//version if it has one. When performing a count or when just retrieving Ids like in paging, this is unecessary
//version if it has one. When performing a count or when retrieving Ids like in paging, this is unecessary
//and causes huge performance overhead for the SQL server, especially when sorting the result.
//To fix this perf overhead we'd need another index on :
// CREATE NON CLUSTERED INDEX ON cmsDocument.node + cmsDocument.published
//We also don't include this outer join when querying for multiple entities since it is much faster to fetch this information
//in a separate query. For a single entity this is ok.
var sqlx = string.Format("LEFT OUTER JOIN {0} {1} ON ({1}.{2}={0}.{2} AND {1}.{3}=1)",
SqlSyntax.GetQuotedTableName("cmsDocument"),
@@ -151,7 +164,7 @@ namespace Umbraco.Core.Persistence.Repositories
protected override Sql GetBaseQuery(bool isCount)
{
return GetBaseQuery(isCount ? BaseQueryType.Count : BaseQueryType.Full);
return GetBaseQuery(isCount ? BaseQueryType.Count : BaseQueryType.FullSingle);
}
protected override string GetBaseWhereClause()
@@ -222,7 +235,7 @@ namespace Umbraco.Core.Persistence.Repositories
while (true)
{
// get the next group of nodes
var sqlFull = translate(baseId, GetBaseQuery(BaseQueryType.Full));
var sqlFull = translate(baseId, GetBaseQuery(BaseQueryType.FullMultiple));
var sqlIds = translate(baseId, GetBaseQuery(BaseQueryType.Ids));
var xmlItems = ProcessQuery(SqlSyntax.SelectTop(sqlFull, groupSize), new PagingSqlQuery(SqlSyntax.SelectTop(sqlIds, groupSize)))
@@ -245,7 +258,7 @@ namespace Umbraco.Core.Persistence.Repositories
Logger.Error<MediaRepository>("Could not rebuild XML for nodeId=" + xmlItem.NodeId, e);
}
}
baseId = xmlItems.Last().NodeId;
baseId = xmlItems[xmlItems.Count - 1].NodeId;
}
}
@@ -257,7 +270,7 @@ namespace Umbraco.Core.Persistence.Repositories
.OrderByDescending<ContentVersionDto>(x => x.VersionDate, SqlSyntax);
};
var sqlFull = translate(GetBaseQuery(BaseQueryType.Full));
var sqlFull = translate(GetBaseQuery(BaseQueryType.FullMultiple));
var sqlIds = translate(GetBaseQuery(BaseQueryType.Ids));
return ProcessQuery(sqlFull, new PagingSqlQuery(sqlIds), true);
@@ -265,7 +278,7 @@ namespace Umbraco.Core.Persistence.Repositories
public override IContent GetByVersion(Guid versionId)
{
var sql = GetBaseQuery(false);
var sql = GetBaseQuery(BaseQueryType.FullSingle);
sql.Where("cmsContentVersion.VersionId = @VersionId", new { VersionId = versionId });
sql.OrderByDescending<ContentVersionDto>(x => x.VersionDate, SqlSyntax);
@@ -674,7 +687,7 @@ namespace Umbraco.Core.Persistence.Repositories
// ORDER BY substring(path, 1, len(path) - charindex(',', reverse(path))), sortOrder
// but that's probably an overkill - sorting by level,sortOrder should be enough
var sqlFull = GetBaseQuery(BaseQueryType.Full);
var sqlFull = GetBaseQuery(BaseQueryType.FullMultiple);
var translatorFull = new SqlTranslator<IContent>(sqlFull, query);
var sqlIds = GetBaseQuery(BaseQueryType.Ids);
var translatorIds = new SqlTranslator<IContent>(sqlIds, query);
@@ -889,12 +902,12 @@ order by umbracoNode.{2}, umbracoNode.parentID, umbracoNode.sortOrder",
return base.GetDatabaseFieldNameForOrderBy(orderBy);
}
/// <summary>
/// This is the underlying method that processes most queries for this repository
/// </summary>
/// <param name="sqlFull">
/// The full SQL with the outer join to return all data required to create an IContent
/// The FullMultiple SQL without the outer join to return all data required to create an IContent excluding it's published state data which this will query separately
/// </param>
/// <param name="pagingSqlQuery">
/// The Id SQL without the outer join to just return all document ids - used to process the properties for the content item
@@ -904,8 +917,43 @@ order by umbracoNode.{2}, umbracoNode.parentID, umbracoNode.sortOrder",
private IEnumerable<IContent> ProcessQuery(Sql sqlFull, PagingSqlQuery pagingSqlQuery, bool withCache = false)
{
// fetch returns a list so it's ok to iterate it in this method
var dtos = Database.Fetch<DocumentDto, ContentVersionDto, ContentDto, NodeDto, DocumentPublishedReadOnlyDto>(sqlFull);
var dtos = Database.Fetch<DocumentDto, ContentVersionDto, ContentDto, NodeDto>(sqlFull);
if (dtos.Count == 0) return Enumerable.Empty<IContent>();
//Go and get all of the published version data separately for this data, this is because when we are querying
//for multiple content items we don't include the outer join to fetch this data in the same query because
//it is insanely slow. Instead we just fetch the published version data separately in one query.
//we need to parse the original SQL statement and reduce the columns to just cmsDocument.nodeId so that we can use
// the statement to go get the published data for all of the items by using an inner join
var parsedOriginalSql = "SELECT cmsDocument.nodeId " + sqlFull.SQL.Substring(sqlFull.SQL.IndexOf("FROM", StringComparison.Ordinal));
//now remove everything from an Orderby clause and beyond
if (parsedOriginalSql.InvariantContains("ORDER BY "))
{
parsedOriginalSql = parsedOriginalSql.Substring(0, parsedOriginalSql.LastIndexOf("ORDER BY ", StringComparison.Ordinal));
}
var publishedSql = new Sql(@"SELECT *
FROM cmsDocument AS doc2
INNER JOIN
(" + parsedOriginalSql + @") as docData
ON doc2.nodeId = docData.nodeId
WHERE doc2.published = 1
ORDER BY doc2.nodeId
", sqlFull.Arguments);
//go and get the published version data, we do a Query here and not a Fetch so we are
//not allocating a whole list to memory just to allocate another list in memory since
//we are assigning this data to a keyed collection for fast lookup below
var publishedData = Database.Query<DocumentPublishedReadOnlyDto>(publishedSql);
var publishedDataCollection = new DocumentPublishedReadOnlyDtoCollection();
foreach (var publishedDto in publishedData)
{
//double check that there's no corrupt db data, there should only be a single published item
if (publishedDataCollection.Contains(publishedDto.NodeId) == false)
publishedDataCollection.Add(publishedDto);
}
var content = new IContent[dtos.Count];
var defs = new List<DocumentDefinition>();
@@ -919,6 +967,8 @@ order by umbracoNode.{2}, umbracoNode.parentID, umbracoNode.sortOrder",
for (var i = 0; i < dtos.Count; i++)
{
var dto = dtos[i];
DocumentPublishedReadOnlyDto publishedDto;
publishedDataCollection.TryGetValue(dto.NodeId, out publishedDto);
// if the cache contains the published version, use it
if (withCache)
@@ -946,7 +996,7 @@ order by umbracoNode.{2}, umbracoNode.parentID, umbracoNode.sortOrder",
contentTypes[dto.ContentVersionDto.ContentDto.ContentTypeId] = contentType;
}
content[i] = ContentFactory.BuildEntity(dto, contentType);
content[i] = ContentFactory.BuildEntity(dto, contentType, publishedDto);
// need template
if (dto.TemplateId.HasValue && dto.TemplateId.Value > 0)
@@ -1055,7 +1105,7 @@ order by umbracoNode.{2}, umbracoNode.parentID, umbracoNode.sortOrder",
return currentName;
}
/// <summary>
/// Dispose disposable properties
/// </summary>
@@ -1070,5 +1120,26 @@ order by umbracoNode.{2}, umbracoNode.parentID, umbracoNode.sortOrder",
_contentPreviewRepository.Dispose();
_contentXmlRepository.Dispose();
}
/// <summary>
/// A keyed collection for fast lookup when retrieving a separate list of published version data
/// </summary>
private class DocumentPublishedReadOnlyDtoCollection : KeyedCollection<int, DocumentPublishedReadOnlyDto>
{
protected override int GetKeyForItem(DocumentPublishedReadOnlyDto item)
{
return item.NodeId;
}
public bool TryGetValue(int key, out DocumentPublishedReadOnlyDto val)
{
if (Dictionary == null)
{
val = null;
return false;
}
return Dictionary.TryGetValue(key, out val);
}
}
}
}

2
src/Umbraco.Core/Persistence/Repositories/MediaRepository.cs
View File

@@ -100,7 +100,7 @@ namespace Umbraco.Core.Persistence.Repositories
protected override Sql GetBaseQuery(bool isCount)
{
return GetBaseQuery(isCount ? BaseQueryType.Count : BaseQueryType.Full);
return GetBaseQuery(isCount ? BaseQueryType.Count : BaseQueryType.FullSingle);
}
protected override string GetBaseWhereClause()

2
src/Umbraco.Core/Persistence/Repositories/MemberRepository.cs
View File

@@ -129,7 +129,7 @@ namespace Umbraco.Core.Persistence.Repositories
protected override Sql GetBaseQuery(bool isCount)
{
return GetBaseQuery(isCount ? BaseQueryType.Count : BaseQueryType.Full);
return GetBaseQuery(isCount ? BaseQueryType.Count : BaseQueryType.FullSingle);
}
protected override string GetBaseWhereClause()

3
src/Umbraco.Core/Persistence/Repositories/VersionableRepositoryBase.cs
View File

@@ -442,7 +442,7 @@ namespace Umbraco.Core.Persistence.Repositories
// Get base query for returning IDs
var sqlBaseIds = GetBaseQuery(BaseQueryType.Ids);
// Get base query for returning all data
var sqlBaseFull = GetBaseQuery(BaseQueryType.Full);
var sqlBaseFull = GetBaseQuery(BaseQueryType.FullMultiple);
if (query == null) query = new Query<TEntity>();
var translatorIds = new SqlTranslator<TEntity>(sqlBaseIds, query);
@@ -773,7 +773,6 @@ ORDER BY contentNodeId, propertytypeid
/// <returns></returns>
protected abstract Sql GetBaseQuery(BaseQueryType queryType);
internal class DocumentDefinition
{
/// <summary>