w4c-workflows-api/Services/Nodes/Executors/HttpRequestNodeExecutor.Sending.cs

using System.Net;
using System.Text.Json.Nodes;
using w4c_workflows.Models.Credentials;
using w4c_workflows.Models.Nodes;
using w4c_workflows.Services.Credentials;
using w4c_workflows.Services.Nodes.Binary;
using w4c_workflows.Services.Security;

namespace w4c_workflows.Services.Nodes.Executors;

// Sending half of the HTTP Request node: one logical request with hand-followed,
// per-hop-vetted redirects. See HttpRequestNodeExecutor.cs for the contract.

public sealed partial class HttpRequestNodeExecutor
{
    private sealed record PageResult(
        int Status,
        string Body,
        byte[] Bytes,
        JsonNode? Parsed,
        string? ContentType,
        string? SuggestedFileName,
        NodeExecutionOutcome? Failure);

    /// <summary>
    /// Issues one logical request, following redirects by hand so every hop is
    /// egress-vetted. Credentials/body are dropped on a cross-host redirect.
    /// </summary>
    private async Task<PageResult> SendPageAsync(
        NodeExecutionContext context,
        HttpClient client,
        CancellationTokenSource timeout,
        int timeoutMs,
        Uri initialUri,
        HttpMethod method,
        bool sendBody,
        JsonNode? body,
        IReadOnlyDictionary<string, string> headerOverrides,
        bool sendCredential,
        bool hasCredential,
        CredentialData? credential,
        bool followRedirects,
        int maxRedirects,
        CancellationToken ct)
    {
        var currentUri = initialUri;
        var currentMethod = method;
        var currentHost = initialUri.DnsSafeHost;
        var bodyEnabled = sendBody;
        var credEnabled = sendCredential;
        var redirectsFollowed = 0;
        HttpResponseMessage? response = null;

        while (true)
        {
            var verdict = await _egress.AuthorizeAsync(currentUri, ct);
            if (!verdict.Allowed)
            {
                response?.Dispose();
                return PageFailure(verdict.Reason ?? "blocked by the egress policy", "egress_blocked");
            }

            if (!_quota.TryReserveRequest(context.State, out var requestsUsed))
            {
                response?.Dispose();
                return PageFailure(
                    $"outbound request quota exceeded ({_quota.MaxRequestsPerRun} per run)",
                    "quota_exceeded",
                    $"this run has already issued {requestsUsed - 1} requests");
            }

            using var request = new HttpRequestMessage(currentMethod, currentUri);
            ApplyHeaders(context, request, headerOverrides);

            if (credEnabled && hasCredential)
            {
                var authError = CredentialInjector.Apply(credential!, _credentialTypes, request);
                if (authError != null)
                {
                    response?.Dispose();
                    return PageFailure(authError, "missing_credential");
                }
            }

            if (bodyEnabled)
            {
                var bodyError = await ApplyBodyAsync(context, body, request, ct);
                if (bodyError != null)
                {
                    response?.Dispose();
                    return PageFailure(bodyError, "invalid_parameter");
                }
            }

            try
            {
                // Headers-only completion so the body is streamed under the size
                // cap below instead of being buffered by the transport first.
                response = await client.SendAsync(
                    request, HttpCompletionOption.ResponseHeadersRead, timeout.Token);
            }
            catch (OperationCanceledException) when (ct.IsCancellationRequested)
            {
                throw;
            }
            catch (OperationCanceledException)
            {
                return PageFailure($"request timed out after {timeoutMs} ms", "timeout");
            }
            catch (HttpRequestException ex)
            {
                return PageFailure(ex.Message, "request_failed");
            }

            var location = response.Headers.Location;
            if (!followRedirects || location == null || !IsRedirect(response.StatusCode))
                break;

            if (redirectsFollowed >= maxRedirects)
            {
                response.Dispose();
                return PageFailure(
                    $"exceeded the maximum of {maxRedirects} redirects", "too_many_redirects");
            }

            if (!TryResolveRedirect(currentUri, location, out var nextUri))
            {
                response.Dispose();
                return PageFailure(
                    $"invalid redirect location '{location}'", "invalid_redirect");
            }

            // 307/308 preserve method and body; the others downgrade to GET.
            if (response.StatusCode != HttpStatusCode.TemporaryRedirect
                && response.StatusCode != HttpStatusCode.PermanentRedirect
                && currentMethod != HttpMethod.Head)
            {
                currentMethod = HttpMethod.Get;
                bodyEnabled = false;
            }

            var nextHost = nextUri.DnsSafeHost;
            if (!string.Equals(nextHost, currentHost, StringComparison.OrdinalIgnoreCase))
                credEnabled = false;

            response.Dispose();
            response = null;
            currentUri = nextUri;
            currentHost = nextHost;
            redirectsFollowed++;
        }

        using (response!)
        {
            var status = (int)response.StatusCode;
            var read = await HttpBodyReader.ReadAsync(response, _quota.MaxResponseBytes, timeout.Token);
            if (read.TooLarge)
                return PageFailure(
                    $"response exceeded the {_quota.MaxResponseBytes}-byte limit", "response_too_large");

            var bodyText = read.Body!;
            var contentType = response.Content.Headers.ContentType?.MediaType;
            var fileName = ResolveFileName(response, currentUri);
            return new PageResult(
                status, bodyText, read.Bytes!, NodeValueAccess.TryParseJson(bodyText), contentType, fileName, null);
        }
    }

    private static PageResult PageFailure(string message, string? code, string? description = null)
        => new(
            0, string.Empty, Array.Empty<byte>(), null, null, null,
            NodeExecutionOutcome.Failed(message, code, description));

    // ------------------------------------------------------------------ redirects

    private static bool IsRedirect(HttpStatusCode status)
        => status is HttpStatusCode.MovedPermanently   // 301
            or HttpStatusCode.Found                    // 302
            or HttpStatusCode.SeeOther                 // 303
            or HttpStatusCode.TemporaryRedirect        // 307
            or HttpStatusCode.PermanentRedirect;       // 308

    /// <summary>Resolves a Location header against the current URL and keeps it HTTP(S).</summary>
    private static bool TryResolveRedirect(Uri current, Uri location, out Uri next)
    {
        next = null!;
        try
        {
            next = location.IsAbsoluteUri ? location : new Uri(current, location);
            return next.Scheme is "http" or "https";
        }
        catch (UriFormatException)
        {
            return false;
        }
    }
}