imdbs/imdb-scraper_test.go

package imdbs

import "testing"

func TestBuildScrapingURL(t *testing.T) {
	data := map[string]string{
		"tt2911666":  "https://www.imdb.com/title/tt2911666/",
		"tt10366206": "https://www.imdb.com/title/tt10366206/",
	}

	for d, expectedResult := range data {
		result := buildScrapingURL(d)
		if result != expectedResult {
			t.Errorf("\ngot:    %q\nwanted: %q\nfor:    %q", result, expectedResult, d)
		}
	}
}

func TestConvertIMDbRuntimeIntoMinutes(t *testing.T) {
	data := map[string]int64{
		"PT41M":    41,
		"PT2H":     120,
		"PT2H1M":   121,
		" PT2H1M ": 0,
		"PT12H1M ": 0,
		"PT121M":   0,
		"":         0,
	}

	for d, expectedResult := range data {
		result := convertIMDbRuntimeIntoMinutes(d)
		if result != expectedResult {
			t.Errorf("\ngot:    %d\nwanted: %d\nfor:    %q", result, expectedResult, d)
		}
	}
}

func TestGetIMDbIDFromQuery(t *testing.T) {
	data := map[string]string{
		"tt2861424":                                           "tt2861424",
		" tt2861424 ":                                         "tt2861424",
		"https://www.imdb.com/title/tt2861424":                "tt2861424",
		"https://www.imdb.com/title/tt2861424/":               "tt2861424",
		"https://www.imdb.com/title/tt2861424/?ref_=vp_vi_tt": "tt2861424",
		"https://m.imdb.com/title/tt2861424":                  "tt2861424",
		"https://m.imdb.com/title/tt2861424/":                 "tt2861424",
		"https://m.imdb.com/title/tt2861424/?ref_=vp_vi_tt":   "tt2861424",
		"https://www.google.com/title/tt2861424/":             "tt2861424",
		"https://www.imdb.com/":                               "",
		"":                                                    "",
	}

	for d, expectedResult := range data {
		result, _ := GetIMDbIDFromQuery(d)
		if result != expectedResult {
			t.Errorf("\ngot:    %q\nwanted: %q\nfor:    %q", result, expectedResult, d)
		}
	}
}

func TestExtractYearFromIMDbTitle(t *testing.T) {
	type tResult struct {
		Data           string
		Title          string
		ExpectedResult int64
	}

	data := []tResult{
		{
			Data:           "John Wick: Kapitel 4 (2023) - IMDb",
			ExpectedResult: 2023,
		},
		{
			Data:           "Matrix (1999) - IMDb",
			ExpectedResult: 1999,
		},
		{
			Data:           "Thurgood (Fernsehfilm 2011) - IMDb",
			ExpectedResult: 2011,
		},
		{
			Data:           "Pretty/Handsome (Fernsehfilm 2008) - IMDb",
			ExpectedResult: 2008,
		},
		{
			Data:           "Red Planet: Deleted Scenes (Video 2000) - IMDb",
			ExpectedResult: 2000,
		},
		{
			Data:           "Last Night in Soho: Deleted Scenes (Video 2022) - IMDb",
			ExpectedResult: 2022,
		},
		{
			Data:           "Eine schrecklich nette Familie (Fernsehserie 1987–1997) - IMDb",
			ExpectedResult: 1987,
		},
		{
			Data:           "Rick and Morty (Fernsehserie 2013– ) - IMDb",
			ExpectedResult: 2013,
		},
		{
			Data:           "Unser Kosmos: Die Reise geht weiter (Miniserie 2014) - IMDb",
			ExpectedResult: 2014,
		},
	}

	for _, r := range data {
		result := extractYearFromIMDbTitle(r.Data)
		if result != r.ExpectedResult {
			t.Errorf("\ngot:    %d\nwanted: %d\nfor:    %q", result, r.ExpectedResult, r.Data)
		}
	}
}

func TestIsValidIMDbID(t *testing.T) {
	data := map[string]bool{
		"tt0000000":             true,
		"tt99999999":            true,
		"tt2911666":             true,
		"tt10366206":            true,
		"tt0944947":             true,
		"tt11737520":            true,
		"tt291166":              false, // too short
		"tt103662060":           false, // too long (i hear that one a lot)
		"ttt1036620":            false, // invalid characters
		"tt2911A66":             false, // invalid characters
		" tt2911666 ":           false, // not trimmed
		"https://www.imdb.com/": false, // wtf is this
		"":                      false, // rly?!
	}

	for d, expectedResult := range data {
		result := IsValidIMDbID(d)
		if result != expectedResult {
			t.Errorf("\ngot:    %t\nwanted: %t\nfor:    %q", result, expectedResult, d)
		}
	}
}

func TestScrap(t *testing.T) {
	data := map[string]IMDbEntry{
		"tt0000000": {},
		"tt2911666": {
			AlternateName: "",
			IMDbID:        "tt2911666",
			Rating:        7.4,
			RuntimeInMins: 101,
			Title:         "John Wick",
			Type:          "Movie",
			Year:          2014,
		},
		"tt7798634": {
			AlternateName: "Ready or Not - Auf die Plätze, fertig, tot",
			IMDbID:        "tt7798634",
			Rating:        6.9,
			RuntimeInMins: 95,
			Title:         "Ready or Not",
			Type:          "Movie",
			Year:          2019,
		},
		"tt0087803": {
			AlternateName: "1984",
			IMDbID:        "tt0087803",
			Rating:        7.1,
			RuntimeInMins: 113,
			Title:         "Nineteen Eighty-Four",
			Type:          "Movie",
			Year:          1984,
		},
		"tt8579674": {
			AlternateName: "",
			IMDbID:        "tt8579674",
			Rating:        8.2,
			RuntimeInMins: 119,
			Title:         "1917",
			Type:          "Movie",
			Year:          2019,
		},
		"tt2861424": {
			AlternateName: "",
			IMDbID:        "tt2861424",
			Rating:        9.1,
			RuntimeInMins: 0,
			Title:         "Rick and Morty",
			Type:          "Series",
			Year:          2013,
		},
		"tt0092400": {
			AlternateName: "Eine schrecklich nette Familie",
			IMDbID:        "tt0092400",
			Rating:        8.1,
			RuntimeInMins: 0,
			Title:         "Married with Children",
			Type:          "Series",
			Year:          1987,
		},
		"tt7366338": {
			AlternateName: "",
			IMDbID:        "tt7366338",
			Rating:        9.3,
			RuntimeInMins: 0,
			Title:         "Chernobyl",
			Type:          "Series",
			Year:          2019,
		},
		"tt0366551": {
			AlternateName: "Harold & Kumar",
			IMDbID:        "tt0366551",
			Rating:        7.0,
			RuntimeInMins: 88,
			Title:         "Harold & Kumar Go to White Castle",
			Type:          "Movie",
			Year:          2004,
		},
	}

	for d, expectedResult := range data {
		result, _ := Scrap(d)
		if result != expectedResult {
			t.Errorf("\ngot:    %+v\nwanted: %+v\nfor:    %q", result, expectedResult, d)
		}
	}
}