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 +1,121 @@
+package microsoft
+
+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 TestMicrosoftCrawler_DiscoverAndFetch(t *testing.T) {
+	mockClient := &http.Client{
+		Transport: roundTripFunc(func(r *http.Request) *http.Response {
+			if strings.Contains(r.URL.Path, "/job/1789001") {
+				body := `{
+					"operationResult": {
+						"result": {
+							"jobId": "1789001",
+							"title": "Software Engineer II - Azure Core",
+							"description": "<p>Azure Core is building the backbone of cloud compute.</p>",
+							"responsibilities": "<li>Design and implement scalable distributed services.</li>",
+							"qualifications": "<li>Bachelor degree and 2+ years of software development experience.</li>",
+							"properties": {
+								"primaryLocation": "Redmond, WA, United States"
+							}
+						}
+					}
+				}`
+				return &http.Response{
+					StatusCode: http.StatusOK,
+					Body:       io.NopCloser(bytes.NewBufferString(body)),
+					Header:     make(http.Header),
+				}
+			}
+
+			if strings.HasSuffix(r.URL.Path, "/search") || strings.Contains(r.URL.RawQuery, "pgSz") {
+				body := `{
+					"operationResult": {
+						"result": {
+							"totalJobs": 1,
+							"jobs": [
+								{
+									"jobId": "1789001",
+									"title": "Software Engineer II - Azure Core",
+									"properties": {
+										"primaryLocation": "Redmond, WA, United States",
+										"discipline": "Software Engineering"
+									}
+								}
+							]
+						}
+					}
+				}`
+				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 := NewMicrosoftCrawlerWithClient(mockClient)
+	if !c.CanHandle("microsoft") || !c.CanHandle("azure") {
+		t.Errorf("CanHandle should support microsoft and azure")
+	}
+
+	discovered, err := c.DiscoverJobs(context.Background(), "https://gcsservices.careers.microsoft.com/search/api/v1/search?q=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 != "microsoft:1789001" {
+		t.Errorf("unexpected canonical ID: %s", discovered[0].CanonicalID)
+	}
+
+	extracted, err := c.FetchJobDetail(context.Background(), "https://jobs.careers.microsoft.com/global/en/share/1789001")
+	if err != nil {
+		t.Fatalf("FetchJobDetail failed: %v", err)
+	}
+
+	if extracted.Title != "Software Engineer II - Azure Core" {
+		t.Errorf("unexpected title: %s", extracted.Title)
+	}
+	if extracted.Company != "Microsoft" {
+		t.Errorf("unexpected company: %s", extracted.Company)
+	}
+	if extracted.Location != "Redmond, WA, United States" {
+		t.Errorf("unexpected location: %s", extracted.Location)
+	}
+}
+
+func TestMicrosoftCrawler_LiveIntegration(t *testing.T) {
+	crawlertest.RunLiveIntegration(t, crawlertest.LiveTestConfig{
+		PlatformName:            "Microsoft",
+		Crawler:                 NewMicrosoftCrawler(),
+		TargetURL:               "https://gcsservices.careers.microsoft.com/search/api/v1/search?q=Software+Engineer&lc=Redmond%2C+Washington%2C+United+States&l=en_us",
+		ExpectedCanonicalPrefix: "microsoft:",
+		ExpectedCompany:         "Microsoft",
+	})
+}
+