Theodor-Springmann-Stiftung/hamann-ausgabe-core
1
0
Fork 0
mirror of https://github.com/Theodor-Springmann-Stiftung/hamann-ausgabe-core.git synced 2025-10-29 09:15:33 +00:00
Code Issues Packages Projects Releases Wiki Activity

Implemented Fulltext search across tags and linebreaks for letters

Browse Source
This commit is contained in:
schnulller
2022-06-19 00:01:14 +02:00
parent 078c4b75b8
commit 4ef0e260e0
24 changed files with 510 additions and 148 deletions
Download Patch File Download Diff File Expand all files Collapse all files

47
HaWeb/FileHelpers/HaDocumentWrapper.cs
View File

@@ -2,6 +2,11 @@ namespace HaWeb.FileHelpers;
using HaDocument.Interfaces;
using Microsoft.AspNetCore.Mvc.ModelBinding;
using Microsoft.Extensions.FileProviders;
using System.Collections.Concurrent;
using System.Threading.Tasks;
using HaXMLReader.Interfaces;
using HaWeb.SearchHelpers;
using System.Text;
public class HaDocumentWrapper : IHaDocumentWrappper {
private ILibrary Library;
@@ -10,6 +15,8 @@ public class HaDocumentWrapper : IHaDocumentWrappper {
public int StartYear { get; private set; }
public int EndYear { get; private set; }
public List<SearchHelpers.SeachableItem>? SearchableLetters { get; private set; }
public HaDocumentWrapper(IXMLProvider xmlProvider, IConfiguration configuration) {
_xmlProvider = xmlProvider;
StartYear = configuration.GetValue<int>("AvailableStartYear");
@@ -41,9 +48,45 @@ public class HaDocumentWrapper : IHaDocumentWrappper {
return null;
}
var searchableletters = new ConcurrentBag<SearchHelpers.SeachableItem>();
var letters = Library.Letters.Values;
Parallel.ForEach(letters, letter => {
var o = new SearchHelpers.SeachableItem(letter.Index, _prepareSearch(letter));
searchableletters.Add(o);
});
this.SearchableLetters = searchableletters.ToList();
return Library;
}
public List<(string Index, List<(string Page, string Line, string Preview)> Results)>? SearchLetters(string searchword, IReaderService reader) {
if (SearchableLetters == null) return null;
var res = new ConcurrentBag<(string Index, List<(string Page, string Line, string preview)> Results)>();
var sw = StringHelpers.NormalizeWhiteSpace(searchword.Trim());
Parallel.ForEach(SearchableLetters, (letter) => {
var state = new SearchState(sw);
var rd = reader.RequestStringReader(letter.SearchText);
var parser = new HaWeb.HTMLParser.LineXMLHelper<SearchState>(state, rd, new StringBuilder(), null, null, null, SearchRules.TextRules, SearchRules.WhitespaceRules);
rd.Read();
if (state.Results != null)
res.Add((
letter.Index,
state.Results.Select(x => (
x.Page,
x.Line,
parser.Lines != null ?
parser.Lines
.Where(y => y.Page == x.Page && y.Line == x.Line)
.Select(x => x.Text)
.FirstOrDefault(string.Empty)
: ""
)).ToList()));
});
return res.ToList();
}
public ILibrary GetLibrary() {
return Library;
}
@@ -57,4 +100,8 @@ public class HaDocumentWrapper : IHaDocumentWrappper {
}
}
}
private string _prepareSearch(HaDocument.Interfaces.ISearchable objecttoseach) {
return SearchHelpers.StringHelpers.NormalizeWhiteSpace(objecttoseach.Element, ' ', false);
}
}

2
HaWeb/FileHelpers/IHaDocumentWrapper.cs
View File

@@ -1,9 +1,11 @@
namespace HaWeb.FileHelpers;
using HaDocument.Interfaces;
using Microsoft.AspNetCore.Mvc.ModelBinding;
using HaXMLReader.Interfaces;
public interface IHaDocumentWrappper {
public ILibrary ResetLibrary();
public ILibrary? SetLibrary(string filepath, ModelStateDictionary ModelState);
public ILibrary GetLibrary();
public List<(string Index, List<(string Page, string Line, string Preview)> Results)>? SearchLetters(string searchword, IReaderService reader);
}