Changes

Apis/cv-matcher-api/Services/JobTextExtractor.cs

@@ -0,0 +1,51 @@
+using System.Net;
+using System.Text.RegularExpressions;
+using CvMatcher.Models.Settings;
+using Api.Services.Contracts;
+using Microsoft.Extensions.Options;
+
+namespace Api.Services;
+
+public sealed class JobTextExtractor : IJobTextExtractor
+{
+    private readonly HttpClient _http;
+    private readonly MatcherSettings _settings;
+
+    public JobTextExtractor(HttpClient http, IOptions<MatcherSettings> options)
+    {
+        _http = http;
+        _settings = options.Value;
+        _http.Timeout = TimeSpan.FromSeconds(25);
+        _http.DefaultRequestHeaders.UserAgent.ParseAdd("MyAi.ro CV Matcher/1.0");
+    }
+
+    public async Task<string> ExtractAsync(string? jobUrl, string? jobDescription, CancellationToken ct)
+    {
+        var pasted = Normalize(jobDescription ?? string.Empty);
+        if (!string.IsNullOrWhiteSpace(pasted)) return Limit(pasted);
+
+        if (string.IsNullOrWhiteSpace(jobUrl)) return string.Empty;
+        if (!Uri.TryCreate(jobUrl, UriKind.Absolute, out var uri) || uri.Scheme is not ("http" or "https"))
+        {
+            throw new InvalidOperationException("Invalid job URL.");
+        }
+
+        var html = await _http.GetStringAsync(uri, ct);
+        html = Regex.Replace(html, "<script[\\s\\S]*?</script>", " ", RegexOptions.IgnoreCase);
+        html = Regex.Replace(html, "<style[\\s\\S]*?</style>", " ", RegexOptions.IgnoreCase);
+        html = Regex.Replace(html, "<[^>]+>", " ");
+        return Limit(Normalize(WebUtility.HtmlDecode(html)));
+    }
+
+    private string Limit(string value)
+    {
+        var max = Math.Max(4000, _settings.MaxJobTextChars);
+        return value.Length <= max ? value : value[..max];
+    }
+
+    private static string Normalize(string value)
+    {
+        if (string.IsNullOrWhiteSpace(value)) return string.Empty;
+        return string.Join(' ', value.Split((char[]?)null, StringSplitOptions.RemoveEmptyEntries)).Trim();
+    }
+}