|
| 1 | +package antch |
| 2 | + |
| 3 | +import ( |
| 4 | + "io/ioutil" |
| 5 | + "net/http" |
| 6 | + "net/http/httptest" |
| 7 | + "net/url" |
| 8 | + "testing" |
| 9 | +) |
| 10 | + |
| 11 | +func TestCrawlerBasic(t *testing.T) { |
| 12 | + ts := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) { |
| 13 | + w.Write([]byte("ok")) |
| 14 | + })) |
| 15 | + defer ts.Close() |
| 16 | + |
| 17 | + tc := NewCrawler() |
| 18 | + // The custom spider handler. |
| 19 | + tc.Handle("*", HandlerFunc(func(c chan<- Item, resp *http.Response) { |
| 20 | + b, err := ioutil.ReadAll(resp.Body) |
| 21 | + if err != nil { |
| 22 | + t.Fatalf("ReadAll failed: %v", err) |
| 23 | + } |
| 24 | + c <- string(b) |
| 25 | + })) |
| 26 | + |
| 27 | + c := make(chan Item) |
| 28 | + // The custom pipeline handler. |
| 29 | + tc.UsePipeline(func(_ PipelineHandler) PipelineHandler { |
| 30 | + return PipelineHandlerFunc(func(v Item) { |
| 31 | + c <- v |
| 32 | + }) |
| 33 | + }) |
| 34 | + |
| 35 | + tc.StartURLs([]string{ts.URL}) |
| 36 | + // Waiting to receive a value from crawler tc. |
| 37 | + if g, e := (<-c).(string), "ok"; g != e { |
| 38 | + t.Errorf("expected %s; got %s", e, g) |
| 39 | + } |
| 40 | +} |
| 41 | + |
| 42 | +func TestCrawlerSpiderMux(t *testing.T) { |
| 43 | + var serveFakes = []struct { |
| 44 | + host string |
| 45 | + path string |
| 46 | + code int |
| 47 | + }{ |
| 48 | + {"example.com", "/", 200}, |
| 49 | + {"example.com", "/search", 201}, |
| 50 | + {"localhost", "/", 200}, |
| 51 | + } |
| 52 | + |
| 53 | + var spiderMuxTests = []struct { |
| 54 | + pattern string |
| 55 | + code int |
| 56 | + }{ |
| 57 | + {"example.com", 200}, |
| 58 | + {"example.com/search", 201}, |
| 59 | + {"localhost", 200}, |
| 60 | + } |
| 61 | + |
| 62 | + var tc = NewCrawler() |
| 63 | + tc.Handle("*", HandlerFunc(func(c chan<- Item, res *http.Response) { |
| 64 | + c <- 0 |
| 65 | + })) |
| 66 | + for _, e := range spiderMuxTests { |
| 67 | + tc.Handle(e.pattern, HandlerFunc(func(c chan<- Item, res *http.Response) { |
| 68 | + c <- res.StatusCode |
| 69 | + })) |
| 70 | + } |
| 71 | + for _, fake := range serveFakes { |
| 72 | + r := &http.Request{ |
| 73 | + Method: "GET", |
| 74 | + Host: fake.host, |
| 75 | + URL: &url.URL{ |
| 76 | + Path: fake.path, |
| 77 | + }, |
| 78 | + } |
| 79 | + res := &http.Response{ |
| 80 | + Request: r, |
| 81 | + StatusCode: fake.code, |
| 82 | + } |
| 83 | + h, _ := tc.Handler(res) |
| 84 | + c := make(chan Item, 1) |
| 85 | + h.ServeSpider(c, res) |
| 86 | + if code := (<-c).(int); code != fake.code { |
| 87 | + t.Errorf("%s expected %d; got %d", fake.host+fake.path, fake.code, code) |
| 88 | + } |
| 89 | + } |
| 90 | +} |
0 commit comments