feat: implement multiple new job crawlers and add corresponding test suites

2026-08-25 00:12:52 -07:00
parent 0de312edbd
commit 55d4f4a0d9
26 changed files with 4045 additions and 15 deletions
@@ -0,0 +1,119 @@
+package bytedance
+
+import (
+	"bytes"
+	"context"
+	"io"
+	"net/http"
+	"strings"
+	"testing"
+
+	"github.com/hireme/hireme/internal/crawler/crawlertest"
+)
+
+type roundTripFunc func(req *http.Request) *http.Response
+
+func (f roundTripFunc) RoundTrip(req *http.Request) (*http.Response, error) {
+	return f(req), nil
+}
+
+func TestByteDanceCrawler_DiscoverAndFetch(t *testing.T) {
+	mockClient := &http.Client{
+		Transport: roundTripFunc(func(r *http.Request) *http.Response {
+			if strings.Contains(r.URL.Path, "/search/job/posts") {
+				body := `{
+					"code": 0,
+					"data": {
+						"count": 1,
+						"job_post_list": [
+							{
+								"id": "7198234123",
+								"title": "Software Engineer - Distributed Storage (Seattle/Bellevue)",
+								"city_info": {
+									"name": "Bellevue",
+									"en_name": "Bellevue, WA"
+								}
+							}
+						]
+					}
+				}`
+				return &http.Response{
+					StatusCode: http.StatusOK,
+					Body:       io.NopCloser(bytes.NewBufferString(body)),
+					Header:     make(http.Header),
+				}
+			}
+
+			if strings.Contains(r.URL.Path, "/job/posts/7198234123") {
+				body := `{
+					"code": 0,
+					"data": {
+						"job_post": {
+							"id": "7198234123",
+							"title": "Software Engineer - Distributed Storage (Seattle/Bellevue)",
+							"description": "<p>Build high-throughput storage engines for TikTok recommendation algorithms.</p>",
+							"requirement": "<li>BS/MS/PhD in CS and 3+ years experience with C++/Go.</li>",
+							"city_info": {
+								"en_name": "Bellevue, WA"
+							}
+						}
+					}
+				}`
+				return &http.Response{
+					StatusCode: http.StatusOK,
+					Body:       io.NopCloser(bytes.NewBufferString(body)),
+					Header:     make(http.Header),
+				}
+			}
+
+			return &http.Response{
+				StatusCode: http.StatusNotFound,
+				Body:       io.NopCloser(bytes.NewBufferString("not found")),
+				Header:     make(http.Header),
+			}
+		}),
+	}
+
+	c := NewByteDanceCrawlerWithClient(mockClient)
+	if !c.CanHandle("bytedance") || !c.CanHandle("tiktok") {
+		t.Errorf("CanHandle should support bytedance and tiktok")
+	}
+
+	discovered, err := c.DiscoverJobs(context.Background(), "https://jobs.bytedance.com/api/v1/search/job/posts?keyword=Software+Engineer")
+	if err != nil {
+		t.Fatalf("DiscoverJobs failed: %v", err)
+	}
+
+	if len(discovered) != 1 {
+		t.Fatalf("expected 1 discovered job, got %d", len(discovered))
+	}
+
+	if discovered[0].CanonicalID != "bytedance:7198234123" {
+		t.Errorf("unexpected canonical ID: %s", discovered[0].CanonicalID)
+	}
+
+	extracted, err := c.FetchJobDetail(context.Background(), "https://jobs.bytedance.com/en/position/7198234123/detail")
+	if err != nil {
+		t.Fatalf("FetchJobDetail failed: %v", err)
+	}
+
+	if extracted.Title != "Software Engineer - Distributed Storage (Seattle/Bellevue)" {
+		t.Errorf("unexpected title: %s", extracted.Title)
+	}
+	if extracted.Company != "ByteDance" {
+		t.Errorf("unexpected company: %s", extracted.Company)
+	}
+	if extracted.Location != "Bellevue, WA" {
+		t.Errorf("unexpected location: %s", extracted.Location)
+	}
+}
+
+func TestByteDanceCrawler_LiveIntegration(t *testing.T) {
+	crawlertest.RunLiveIntegration(t, crawlertest.LiveTestConfig{
+		PlatformName:            "ByteDance",
+		Crawler:                 NewByteDanceCrawler(),
+		TargetURL:               "https://jobs.bytedance.com/api/v1/search/job/posts?keyword=Software+Engineer",
+		ExpectedCanonicalPrefix: "bytedance:",
+		ExpectedCompany:         "ByteDance",
+	})
+}