Merge pull request #38 from tphoney/cinema_paradiso_improve_matching

(fix) improve tv, cinemaparadiso matching

TODO

@@ -7,6 +7,8 @@
 ## bugs
 
 - improve cinema-paradiso movie scrape, many search results are the same page. wasted processing
+- mandalorian is not showing up on amazon tv search
+- vikings, Once Upon a Time in Wonderland,What We Do in the Shadows is not showing up on cinema-paradiso tv search
 
 ## done
 

cinemaparadiso/cinemaparadiso.go

@@ -167,7 +167,7 @@ func searchTVShow(plexTVShow *types.PlexTVShow, tvSearchResult chan<- types.Sear
 	urlEncodedTitle := url.QueryEscape(plexTVShow.Title)
 	result.PlexTVShow = *plexTVShow
 	result.SearchURL = cinemaparadisoSearchURL + "?form-search-field=" + urlEncodedTitle
-	rawData, err := makeRequest(result.SearchURL, http.MethodPost, fmt.Sprintf("form-search-field=%s", urlEncodedTitle))
+	rawData, err := makeRequest(result.SearchURL, http.MethodGet, "")
 	if err != nil {
 		fmt.Println("searchTVShow: Error making web request:", err)
 		tvSearchResult <- result
@@ -342,9 +342,20 @@ func makeRequest(urlEncodedTitle, method, content string) (rawResponse string, e
 		fmt.Println("Error reading response body:", err)
 		return rawResponse, err
 	}
+	if resp.StatusCode != http.StatusOK {
+		fmt.Println("Error response code:", resp.StatusCode, urlEncodedTitle)
+	}
 	rawData := string(body)
-	// write the raw data to a file
-	// os.WriteFile("search.html", body, 0644)
+
+	//nolint
+	// bla := strings.Split(urlEncodedTitle, "=")
+	// if len(bla) > 1 {
+	// 	err = os.WriteFile(fmt.Sprintf("%s.html", bla[1]), body, 0644)
+	// 	if err != nil {
+	// 		fmt.Println("Error writing file:", bla[1], err)
+	// 	}
+	// }
+
 	return rawData, nil
 }
 

cinemaparadiso/cinemaparadiso_test.go

@@ -106,12 +106,12 @@ func TestSearchCinemaParadisoTV(t *testing.T) {
 		t.Skip("ACCEPTANCE TEST: PLEX environment variables not set")
 	}
 	show := types.PlexTVShow{
-		// Title: "Friends",
-		// Year:  "1994",
-		// Title: "Charmed",
-		// Year:  "1998",
-		Title: "Adventure Time",
-		Year:  "2010",
+		// Title:             "Friends",
+		// FirstEpisodeAired: time.Date(1994, time.September, 22, 0, 0, 0, 0, time.UTC),
+		// LastEpisodeAired:  time.Date(2004, time.May, 6, 0, 0, 0, 0, time.UTC),
+		Title:             "Stargate Origins",
+		FirstEpisodeAired: time.Date(2018, time.February, 15, 0, 0, 0, 0, time.UTC),
+		LastEpisodeAired:  time.Date(2018, time.March, 8, 0, 0, 0, 0, time.UTC),
 	}
 	ch := make(chan types.SearchResults, 1)
 	searchTVShow(&show, ch)

types/types.go

@@ -25,6 +25,7 @@ type SearchResults struct {
 	SearchURL          string
 	Matches4k          int
 	MatchesBluray      int
+	MatchesDVD         int
 	MovieSearchResults []MovieSearchResult
 	TVSearchResults    []TVSearchResult
 	MusicSearchResults []MusicArtistSearchResult

utils/utils.go

@@ -33,9 +33,12 @@ func MarkBestMatchTV(search *types.SearchResults) types.SearchResults {
 	lastEpisodeBoundry := search.PlexTVShow.LastEpisodeAired.Year() + 1
 	for i := range search.TVSearchResults {
 		resultYear := YearToDate(search.TVSearchResults[i].FirstAiredYear)
-		if strings.EqualFold(search.TVSearchResults[i].FoundTitle, search.PlexTVShow.Title) &&
-			resultYear.Year() >= firstEpisodeBoundry && resultYear.Year() <= lastEpisodeBoundry {
+		if matchTVShow(search.PlexTVShow.Title, search.TVSearchResults[i].FoundTitle,
+			resultYear.Year(), firstEpisodeBoundry, lastEpisodeBoundry) {
 			search.TVSearchResults[i].BestMatch = true
+			if slices.Contains(search.TVSearchResults[i].Format, types.DiskDVD) {
+				search.MatchesDVD++
+			}
 			if slices.Contains(search.TVSearchResults[i].Format, types.DiskBluray) {
 				search.MatchesBluray++
 			}
@@ -47,6 +50,28 @@ func MarkBestMatchTV(search *types.SearchResults) types.SearchResults {
 	return *search
 }
 
+func matchTVShow(plexTitle, foundTitle string, foundYear, lowerBound, upperBound int) bool {
+	plexTitle = strings.ToLower(plexTitle)
+	foundTitle = strings.ToLower(foundTitle)
+	remove := []string{"the"}
+	for _, word := range remove {
+		plexTitle = strings.ReplaceAll(plexTitle, word, "")
+		foundTitle = strings.ReplaceAll(foundTitle, word, "")
+	}
+	// remove colons
+	plexTitle = strings.ReplaceAll(plexTitle, ":", "")
+	foundTitle = strings.ReplaceAll(foundTitle, ":", "")
+	// trim whitespace
+	plexTitle = strings.TrimSpace(plexTitle)
+	foundTitle = strings.TrimSpace(foundTitle)
+
+	if strings.EqualFold(plexTitle, foundTitle) &&
+		foundYear >= lowerBound && foundYear <= upperBound {
+		return true
+	}
+	return false
+}
+
 func YearToDate(yearString string) time.Time {
 	year, err := strconv.Atoi(yearString)
 	if err != nil {

utils/utils_test.go

@@ -211,3 +211,48 @@ func TestWitinOneYear(t *testing.T) {
 		t.Errorf("Expected %v, but got %v", expectedResult, result)
 	}
 }
+
+func Test_matchTVShow(t *testing.T) {
+	type args struct {
+		plexTitle  string
+		foundTitle string
+		foundYear  int
+		lowerBound int
+		upperBound int
+	}
+	tests := []struct {
+		name string
+		args args
+		want bool
+	}{
+		{
+			name: "colons in title",
+			args: args{
+				plexTitle:  "Stargate origins",
+				foundTitle: "Stargate: Origins",
+				foundYear:  2018,
+				lowerBound: 2018,
+				upperBound: 2018,
+			},
+			want: true,
+		},
+		{
+			name: "Peter Serafinowicz Show",
+			args: args{
+				plexTitle:  "The Peter Serafinowicz Show",
+				foundTitle: "Peter Serafinowicz Show",
+				foundYear:  2008,
+				lowerBound: 2007,
+				upperBound: 2009,
+			},
+			want: true,
+		},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			if got := matchTVShow(tt.args.plexTitle, tt.args.foundTitle, tt.args.foundYear, tt.args.lowerBound, tt.args.upperBound); got != tt.want {
+				t.Errorf("matchTVShow() = %v, want %v", got, tt.want)
+			}
+		})
+	}
+}

web/tv/tv.go

@@ -96,7 +96,7 @@ func ProgressBarHTML(w http.ResponseWriter, _ *http.Request) {
 
 func renderTVTable(searchResults []types.SearchResults) (tableRows string) {
 	searchResults = filterTVSearchResults(searchResults)
-	tableRows = `<thead><tr><th data-sort="string"><strong>Plex Title</strong></th><th data-sort="int"><strong>Blu-ray</strong></th><th data-sort="int"><strong>4K-ray</strong></th><th><strong>Disc</strong></th></tr></thead><tbody>` //nolint: lll
+	tableRows = `<thead><tr><th data-sort="string"><strong>Plex Title</strong></th><th data-sort="int"><strong>DVD</strong></th><th data-sort="int"><strong>Blu-ray</strong></th><th data-sort="int"><strong>4K-ray</strong></th><th><strong>Disc</strong></th></tr></thead><tbody>` //nolint: lll
 	for i := range searchResults {
 		// build up plex season / resolution row
 		plexSeasonsString := ""
@@ -106,10 +106,10 @@ func renderTVTable(searchResults []types.SearchResults) (tableRows string) {
 		}
 		plexSeasonsString = plexSeasonsString[:len(plexSeasonsString)-1] // remove trailing comma
 		tableRows += fmt.Sprintf(
-			`<tr><td><a href=%q target="_blank">%s [%v]:<br>%s</a></td><td>%d</td><td>%d</td>`,
+			`<tr><td><a href=%q target="_blank">%s [%v]:<br>%s</a></td><td>%d</td><td>%d</td><td>%d</td>`,
 			searchResults[i].SearchURL, searchResults[i].PlexTVShow.Title, searchResults[i].PlexTVShow.Year, plexSeasonsString,
-			searchResults[i].MatchesBluray, searchResults[i].Matches4k)
-		if (searchResults[i].MatchesBluray + searchResults[i].Matches4k) > 0 {
+			searchResults[i].MatchesDVD, searchResults[i].MatchesBluray, searchResults[i].Matches4k)
+		if (searchResults[i].MatchesDVD + searchResults[i].MatchesBluray + searchResults[i].Matches4k) > 0 {
 			tableRows += "<td>"
 			for j := range searchResults[i].TVSearchResults {
 				if searchResults[i].TVSearchResults[j].BestMatch {