Theodor-Springmann-Stiftung/hamann-ausgabe-core
1
0
Fork 0
mirror of https://github.com/Theodor-Springmann-Stiftung/hamann-ausgabe-core.git synced 2025-10-31 02:05:33 +00:00
Code Issues Packages Projects Releases Wiki Activity

Added generic collections of objects

Browse Source
This commit is contained in:
schnulller
2022-06-22 14:57:09 +02:00
parent ba3d63ffba
commit 83b4e5327a
3 changed files with 31 additions and 12 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
HaWeb/XMLParser/IXMLRoot.cs
View File

@@ -17,8 +17,15 @@ public interface IXMLRoot {
// Key: the key under which the element(s) will be files
// xPath: the (absolute) XPath to the element(s)
// KeyFunc: How to extrect an identifier for the single element in the collection
// LookupsFunc: Function to generate metadata fields for the object, which will then in turn be a possibility to seach and filter without parsing
// Searchable: Will the element be indexed for full-text-search?
public abstract (string Key, string xPath, Func<XElement, string?> KeyFunc, bool Searchable)[]? XPathCollection { get; }
public abstract (
string Key,
string xPath,
Func<XElement, string?> KeyFunc,
// Func<XElement, Dictionary<string, string[]>> LookupsFunc,
bool Searchable
)[]? XPathCollection { get; }
// Determines child objects to be collected
// (deprecated see collections above; only used internally)

21
HaWeb/XMLParser/XMLService.cs
View File

@@ -61,15 +61,22 @@ public class XMLService : IXMLService {
public void SetInProduction(XDocument document) {
if (document == null || _Roots == null) return;
var ret = new ConcurrentDictionary<string, ConcurrentDictionary<string, CollectedItem>>();
Parallel.ForEach(_Roots, (root) => {
int numProcs = Environment.ProcessorCount;
int concurrencyLevel = numProcs * 2;
int startingSize = 2909;
int startingSizeAllCollections = 23;
var ret = new ConcurrentDictionary<string, ConcurrentDictionary<string, CollectedItem>>(concurrencyLevel, startingSizeAllCollections);
// Note Parallelization brings almost nothing to the table (on a laptop) here and below.
// Parallel.ForEach(_Roots, (root) => {
foreach (var root in _Roots) {
if (root.Value.XPathCollection != null)
foreach (var coll in root.Value.XPathCollection) {
var elem = document.XPathSelectElements(coll.xPath);
if (elem != null && elem.Any()) {
if (!ret.ContainsKey(coll.Key))
ret[coll.Key] = new ConcurrentDictionary<string, CollectedItem>();
foreach(var e in elem) {
ret[coll.Key] = new ConcurrentDictionary<string, CollectedItem>(concurrencyLevel, startingSize);
Parallel.ForEach(elem, (e) => {
// foreach(var e in elem) {
var k = coll.KeyFunc(e);
if (k != null) {
var searchtext = coll.Searchable ?
@@ -77,10 +84,12 @@ public class XMLService : IXMLService {
null;
ret[coll.Key][k] = new CollectedItem(k, e, root.Value, coll.Key, searchtext);
}
}
// }
});
}
}
});
}
// });
_collectedProduction = ret.ToDictionary(x => x.Key, y => y.Value.ToDictionary(z => z.Key, f => f.Value, null), null);
}