myAi/Jobs/cv-search-job/Services/HtmlJobSearcher.cs

using System.Text.RegularExpressions;
using System.Web;
using CvSearch.Models.Settings;
using Microsoft.Extensions.Logging;

namespace CvSearchJob.Services;

public sealed class HtmlJobSearcher
{
    private readonly HttpClient _http;
    private readonly ILogger<HtmlJobSearcher> _logger;

    public HtmlJobSearcher(HttpClient http, ILogger<HtmlJobSearcher> logger)
    {
        _http = http;
        _logger = logger;
        _http.Timeout = TimeSpan.FromSeconds(20);
        _http.DefaultRequestHeaders.UserAgent.ParseAdd("Mozilla/5.0 (compatible; MyAi.ro CV-Search/1.0)");
    }

    public async Task<IReadOnlyList<string>> SearchJobUrlsAsync(
        JobProviderConfig provider,
        IReadOnlyList<string> cvKeywords,
        CancellationToken ct)
    {
        var allKeywords = provider.InitialKeywords
            .Concat(cvKeywords)
            .Where(k => !string.IsNullOrWhiteSpace(k))
            .Distinct(StringComparer.OrdinalIgnoreCase)
            .ToList();

        if (allKeywords.Count == 0)
            return [];

        var keywordsEncoded = HttpUtility.UrlEncode(string.Join(" ", allKeywords));
        var searchUrl = provider.SearchUrlTemplate.Replace("{keywords}", keywordsEncoded);

        string html;
        try
        {
            html = await _http.GetStringAsync(searchUrl, ct);
        }
        catch (Exception ex)
        {
            _logger.LogWarning(ex, "Failed to fetch search results from {Provider} at {Url}", provider.Name, searchUrl);
            return [];
        }

        var baseUri = new Uri(searchUrl);
        var results = new List<string>();
        var seen = new HashSet<string>(StringComparer.OrdinalIgnoreCase);

        // Match all anchor tags capturing href and inner text
        var anchorPattern = new Regex(@"<a[^>]+href=[""']([^""']+)[""'][^>]*>(.*?)</a>",
            RegexOptions.IgnoreCase | RegexOptions.Singleline);

        foreach (Match match in anchorPattern.Matches(html))
        {
            if (results.Count >= provider.MaxResults) break;

            var href = match.Groups[1].Value.Trim();
            var anchorText = Regex.Replace(match.Groups[2].Value, "<[^>]+>", " ").Trim();

            if (!href.Contains(provider.JobLinkContains, StringComparison.OrdinalIgnoreCase))
                continue;

            // Stage 2: anchor text must contain at least one CV keyword
            if (!cvKeywords.Any(k => anchorText.Contains(k, StringComparison.OrdinalIgnoreCase)))
                continue;

            // Make absolute URL
            if (!Uri.TryCreate(href, UriKind.Absolute, out var absoluteUri))
            {
                if (!Uri.TryCreate(baseUri, href, out absoluteUri))
                    continue;
            }

            var url = absoluteUri.GetLeftPart(UriPartial.Path);
            if (seen.Add(url))
                results.Add(url);
        }

        _logger.LogInformation("Provider {Provider}: found {Count} job URLs", provider.Name, results.Count);
        return results;
    }
}