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 +1,111 @@
+package amazon
+
+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 TestAmazonCrawler_DiscoverAndFetch(t *testing.T) {
+	mockClient := &http.Client{
+		Transport: roundTripFunc(func(r *http.Request) *http.Response {
+			if strings.Contains(r.URL.Path, "search.json") {
+				body := `{
+					"hits": 1,
+					"jobs": [
+						{
+							"id": "2500123",
+							"title": "Software Development Engineer II - AWS",
+							"company_name": "Amazon Web Services, Inc.",
+							"city": "Seattle",
+							"state": "WA",
+							"country_code": "USA",
+							"job_path": "/en/jobs/2500123/software-development-engineer-ii-aws"
+						}
+					]
+				}`
+				return &http.Response{
+					StatusCode: http.StatusOK,
+					Body:       io.NopCloser(bytes.NewBufferString(body)),
+					Header:     make(http.Header),
+				}
+			}
+
+			if strings.Contains(r.URL.Path, "2500123.json") {
+				body := `{
+					"job": {
+						"id": "2500123",
+						"title": "Software Development Engineer II - AWS",
+						"company_name": "Amazon Web Services, Inc.",
+						"location": "Seattle, WA, USA",
+						"description": "<p>Come build the future of cloud computing.</p>",
+						"basic_qualifications": "<li>3+ years of non-internship professional software development experience</li>",
+						"preferred_qualifications": "<li>Experience with distributed systems and Go/Java/C++</li>"
+					}
+				}`
+				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 := NewAmazonCrawlerWithClient(mockClient)
+	if !c.CanHandle("amazon") || !c.CanHandle("aws") {
+		t.Errorf("CanHandle should support amazon and aws")
+	}
+
+	discovered, err := c.DiscoverJobs(context.Background(), "https://www.amazon.jobs/en/search.json?base_query=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 != "amazon:2500123" {
+		t.Errorf("unexpected canonical ID: %s", discovered[0].CanonicalID)
+	}
+
+	extracted, err := c.FetchJobDetail(context.Background(), "https://www.amazon.jobs/en/jobs/2500123")
+	if err != nil {
+		t.Fatalf("FetchJobDetail failed: %v", err)
+	}
+
+	if extracted.Title != "Software Development Engineer II - AWS" {
+		t.Errorf("unexpected title: %s", extracted.Title)
+	}
+	if extracted.Company != "Amazon Web Services, Inc." {
+		t.Errorf("unexpected company: %s", extracted.Company)
+	}
+}
+
+func TestAmazonCrawler_LiveIntegration(t *testing.T) {
+	crawlertest.RunLiveIntegration(t, crawlertest.LiveTestConfig{
+		PlatformName:            "Amazon",
+		Crawler:                 NewAmazonCrawler(),
+		TargetURL:               "https://www.amazon.jobs/en/search.json?base_query=Software+Development+Engineer&loc_query=Seattle%2C+WA%2C+USA",
+		ExpectedCanonicalPrefix: "amazon:",
+		ExpectedCompany:         "Amazon",
+	})
+}