Files
hamkadr/src/JobsMedical.Web/Services/Scraping/DivarListingSource.cs
T

75 lines
2.9 KiB
C#
Raw Normal View History

using System.Text;
using System.Text.Json;
using JobsMedical.Web.Models;
namespace JobsMedical.Web.Services.Scraping;
/// <summary>
/// Best-effort Divar fetch: queries Divar's web-search JSON for each term and harvests post
/// titles + descriptions. Enabled + city + queries come from admin settings (DB). Divar's
/// private API shifts shape, so we walk JSON tolerantly and fail soft.
/// </summary>
public class DivarListingSource : IListingSource
{
private const string BaseUrl = "https://api.divar.ir/v8/web-search";
private readonly ScrapeHttpClients _clients;
private readonly ILogger<DivarListingSource> _log;
public DivarListingSource(ScrapeHttpClients clients, ILogger<DivarListingSource> log)
{
_clients = clients;
_log = log;
}
public string Name => "دیوار";
public async Task<IReadOnlyList<ScrapedItem>> FetchAsync(AppSetting s, CancellationToken ct = default)
{
var queries = AppSetting.SplitList(s.DivarQueries);
if (!s.DivarEnabled || queries.Count == 0) return Array.Empty<ScrapedItem>();
var city = string.IsNullOrWhiteSpace(s.DivarCity) ? "tehran" : s.DivarCity.Trim();
var client = _clients.For(s, s.DivarUseProxy);
var items = new List<ScrapedItem>();
foreach (var q in queries)
{
try
{
var url = $"{BaseUrl}/{city}/jobs?q={Uri.EscapeDataString(q)}";
var body = await client.GetStringAsync(url, ct);
using var doc = JsonDocument.Parse(body);
foreach (var text in Harvest(doc.RootElement).Take(25))
items.Add(new ScrapedItem("دیوار", text, "https://divar.ir"));
}
catch (Exception ex) { _log.LogWarning(ex, "Divar fetch failed for query {Query}", q); }
}
return items;
}
private static readonly string[] DescKeys =
{ "description", "middle_description_text", "subtitle", "bottom_description_text", "normal_text" };
private static IEnumerable<string> Harvest(JsonElement el)
{
if (el.ValueKind == JsonValueKind.Object)
{
if (el.TryGetProperty("title", out var t) && t.ValueKind == JsonValueKind.String)
{
var sb = new StringBuilder(t.GetString());
foreach (var k in DescKeys)
if (el.TryGetProperty(k, out var d) && d.ValueKind == JsonValueKind.String)
{ sb.Append(" — ").Append(d.GetString()); break; }
var text = sb.ToString().Trim();
if (text.Length >= 15) yield return text;
}
foreach (var p in el.EnumerateObject())
foreach (var x in Harvest(p.Value)) yield return x;
}
else if (el.ValueKind == JsonValueKind.Array)
{
foreach (var item in el.EnumerateArray())
foreach (var x in Harvest(item)) yield return x;
}
}
}