anas/divan
1
0
Fork 0
Code Issues 12 Pull Requests Actions Packages Projects Releases Wiki Activity

#401 code for parsing soha library pages

Browse Source
This commit is contained in:
Hamid Reza Mohammadi 2023-07-26 10:09:00 +03:30
parent 373f4e4ec6
commit 9ad3be2c43
1 changed files with 70 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

73
RMuseum/Services/Implementation/PDFLibraryService-Partials/PDFLibraryService-SohaLibrary.cs
View File

@ -13,6 +13,7 @@ using System.Collections.Generic;
using System.Linq;
using System.Net.Http;
using System.Security.Policy;
using System.Text.RegularExpressions;
using System.Threading.Tasks;
namespace RMuseum.Services.Implementation
{
@ -89,6 +90,16 @@ namespace RMuseum.Services.Implementation
}
}
if (html.IndexOf("/item/download/") == -1)
{
job.EndTime = DateTime.Now;
job.Status = ImportJobStatus.Failed;
job.Exception = $"/item/download/ not found in html source.";
context.Update(job);
await context.SaveChangesAsync();
return;
}
List<RTagValue> meta = new List<RTagValue>();
int idxStart;
int idx = html.IndexOf("branch-link");
@ -121,9 +132,6 @@ namespace RMuseum.Services.Implementation
await context.SaveChangesAsync();
return;
}
idxStart = html.IndexOf(">", idx);
if (idxStart != -1)
{
@ -137,6 +145,65 @@ namespace RMuseum.Services.Implementation
}
}
idxStart = html.IndexOf("width-150");
while(idxStart != -1)
{
idxStart = html.IndexOf(">", idxStart);
if (idxStart == -1) break;
int idxEnd = html.IndexOf("<", idxStart);
if (idxEnd == -1) break;
string tagName = html.Substring(idxStart + 1, idxEnd - idxStart - 1).ToPersianNumbers().ApplyCorrectYeKe();
idxStart = html.IndexOf("value-name", idxEnd);
if (idxStart == -1) break;
idxStart = html.IndexOf(">", idxStart);
if (idxStart == -1) break;
idxEnd = html.IndexOf("</span>", idxStart);
if (idxEnd == -1) break;
string tagValue = html.Substring(idxStart + 1, idxEnd - idxStart - 1).ToPersianNumbers().ApplyCorrectYeKe();
tagValue = Regex.Replace(tagValue, "<.*?>", string.Empty);
if (tagName == "نویسنده")
{
model.AuthorsLine = tagValue;
tagName = "Author";
}
if(tagName == "مترجم")
{
model.TranslatorsLine = tagValue;
model.IsTranslation = true;
tagName = "Translator";
}
if(tagName == "زبان")
{
model.Language = tagValue;
tagName = "Language";
}
if (tagName == "شماره جلد")
{
if (int.TryParse(tagValue, out int v))
{
model.VolumeOrder = v;
}
tagName = "Volume";
}
meta.Add
(
await TagHandler.PrepareAttribute(context, tagName, tagValue, 1)
);
idxStart = html.IndexOf("width-150", idxEnd);
}
idx = html.IndexOf("/item/download/");
int idxQuote = html.IndexOf('"', idx);
string downloadUrl = html.Substring(idx, idxQuote - idx - 1);
downloadUrl = "https://sohalibrary.com" + downloadUrl;
}
catch (Exception e)
{