From 2d3cd6584dedce45ea709d1757a28ce7537f3472 Mon Sep 17 00:00:00 2001 From: Philipp Tanlak Date: Mon, 30 Oct 2023 19:02:50 +0100 Subject: Refactor to prepare for builtin JS functions --- modules/followlinks/followlinks_test.go | 124 +++++++++++++++++--------------- 1 file changed, 66 insertions(+), 58 deletions(-) (limited to 'modules/followlinks/followlinks_test.go') diff --git a/modules/followlinks/followlinks_test.go b/modules/followlinks/followlinks_test.go index f3eb4fe..af186f9 100644 --- a/modules/followlinks/followlinks_test.go +++ b/modules/followlinks/followlinks_test.go @@ -20,24 +20,26 @@ func TestFollowLinks(t *testing.T) { var urls []string var mu sync.Mutex - scraper := flyscrape.NewScraper() - scraper.LoadModule(&starturl.Module{URL: "http://www.example.com/foo/bar"}) - scraper.LoadModule(&followlinks.Module{}) - - scraper.LoadModule(hook.Module{ - AdaptTransportFn: func(rt http.RoundTripper) http.RoundTripper { - return flyscrape.MockTransport(200, ` + mods := []flyscrape.Module{ + &starturl.Module{URL: "http://www.example.com/foo/bar"}, + &followlinks.Module{}, + hook.Module{ + AdaptTransportFn: func(rt http.RoundTripper) http.RoundTripper { + return flyscrape.MockTransport(200, ` Baz Baz Google`) + }, + ReceiveResponseFn: func(r *flyscrape.Response) { + mu.Lock() + urls = append(urls, r.Request.URL) + mu.Unlock() + }, }, - ReceiveResponseFn: func(r *flyscrape.Response) { - mu.Lock() - urls = append(urls, r.Request.URL) - mu.Unlock() - }, - }) + } + scraper := flyscrape.NewScraper() + scraper.Modules = mods scraper.Run() require.Len(t, urls, 5) @@ -52,28 +54,30 @@ func TestFollowSelector(t *testing.T) { var urls []string var mu sync.Mutex - scraper := flyscrape.NewScraper() - scraper.LoadModule(&starturl.Module{URL: "http://www.example.com/foo/bar"}) - scraper.LoadModule(&followlinks.Module{ - Follow: []string{".next a[href]"}, - }) - - scraper.LoadModule(hook.Module{ - AdaptTransportFn: func(rt http.RoundTripper) http.RoundTripper { - return flyscrape.MockTransport(200, ` + mods := []flyscrape.Module{ + &starturl.Module{URL: "http://www.example.com/foo/bar"}, + &followlinks.Module{ + Follow: []string{".next a[href]"}, + }, + hook.Module{ + AdaptTransportFn: func(rt http.RoundTripper) http.RoundTripper { + return flyscrape.MockTransport(200, ` Baz Baz `) + }, + ReceiveResponseFn: func(r *flyscrape.Response) { + mu.Lock() + urls = append(urls, r.Request.URL) + mu.Unlock() + }, }, - ReceiveResponseFn: func(r *flyscrape.Response) { - mu.Lock() - urls = append(urls, r.Request.URL) - mu.Unlock() - }, - }) + } + scraper := flyscrape.NewScraper() + scraper.Modules = mods scraper.Run() require.Len(t, urls, 2) @@ -85,26 +89,28 @@ func TestFollowDataAttr(t *testing.T) { var urls []string var mu sync.Mutex - scraper := flyscrape.NewScraper() - scraper.LoadModule(&starturl.Module{URL: "http://www.example.com/foo/bar"}) - scraper.LoadModule(&followlinks.Module{ - Follow: []string{"[data-url]"}, - }) - - scraper.LoadModule(hook.Module{ - AdaptTransportFn: func(rt http.RoundTripper) http.RoundTripper { - return flyscrape.MockTransport(200, ` + mods := []flyscrape.Module{ + &starturl.Module{URL: "http://www.example.com/foo/bar"}, + &followlinks.Module{ + Follow: []string{"[data-url]"}, + }, + hook.Module{ + AdaptTransportFn: func(rt http.RoundTripper) http.RoundTripper { + return flyscrape.MockTransport(200, ` Baz Baz
Google
`) + }, + ReceiveResponseFn: func(r *flyscrape.Response) { + mu.Lock() + urls = append(urls, r.Request.URL) + mu.Unlock() + }, }, - ReceiveResponseFn: func(r *flyscrape.Response) { - mu.Lock() - urls = append(urls, r.Request.URL) - mu.Unlock() - }, - }) + } + scraper := flyscrape.NewScraper() + scraper.Modules = mods scraper.Run() require.Len(t, urls, 2) @@ -116,26 +122,28 @@ func TestFollowMultiple(t *testing.T) { var urls []string var mu sync.Mutex - scraper := flyscrape.NewScraper() - scraper.LoadModule(&starturl.Module{URL: "http://www.example.com/foo/bar"}) - scraper.LoadModule(&followlinks.Module{ - Follow: []string{"a.prev", "a.next"}, - }) - - scraper.LoadModule(hook.Module{ - AdaptTransportFn: func(rt http.RoundTripper) http.RoundTripper { - return flyscrape.MockTransport(200, ` + mods := []flyscrape.Module{ + &starturl.Module{URL: "http://www.example.com/foo/bar"}, + &followlinks.Module{ + Follow: []string{"a.prev", "a.next"}, + }, + hook.Module{ + AdaptTransportFn: func(rt http.RoundTripper) http.RoundTripper { + return flyscrape.MockTransport(200, ` Baz `) + }, + ReceiveResponseFn: func(r *flyscrape.Response) { + mu.Lock() + urls = append(urls, r.Request.URL) + mu.Unlock() + }, }, - ReceiveResponseFn: func(r *flyscrape.Response) { - mu.Lock() - urls = append(urls, r.Request.URL) - mu.Unlock() - }, - }) + } + scraper := flyscrape.NewScraper() + scraper.Modules = mods scraper.Run() require.Len(t, urls, 3) -- cgit v1.2.3