season: html unescape description and episode names

StalkR · Aug 3, 2024 · 271bc70 · 271bc70
1 parent ba11501
commit 271bc70
Show file tree

Hide file tree

Showing 4 changed files with 13 additions and 12 deletions.
diff --git a/season.go b/season.go
@@ -3,8 +3,8 @@ package imdb
 import (
 	"errors"
 	"fmt"
+	"html"
 	"io"
-	"io/ioutil"
 	"net/http"
 	"regexp"
 	"strconv"
@@ -43,7 +43,7 @@ func NewSeason(c *http.Client, id string, season int) (*Season, error) {
 		}
 		return nil, fmt.Errorf("imdb: status not ok: %v", resp.Status)
 	}
-	page, err := ioutil.ReadAll(resp.Body)
+	page, err := io.ReadAll(resp.Body)
 	if err != nil {
 		return nil, err
 	}
@@ -69,7 +69,7 @@ func (s *Season) Parse(c *http.Client, page []byte) error {
 
 	for _, ep := range episodesMatch {
 		n, _ := strconv.Atoi(string(ep[3]))
-		episode := Episode{SeasonNumber: s.SeasonNumber, EpisodeNumber: n, ImageURL: string(ep[1]), ID: string(ep[2]), Name: string(ep[4])}
+		episode := Episode{SeasonNumber: s.SeasonNumber, EpisodeNumber: n, ImageURL: string(ep[1]), ID: string(ep[2]), Name: html.UnescapeString(string(ep[4]))}
 		s.Episodes = append(s.Episodes, episode)
 	}
 

diff --git a/season_test.go b/season_test.go
@@ -18,7 +18,7 @@ func TestSeason(t *testing.T) {
 						EpisodeNumber: 1,
 						ImageURL:      "https://m.media-amazon.com/images/M/MV5BMWIwZWExNGUtNTJjNC00N2Q2LWFjYjItMzhhNzc2NmNlMzBhXkEyXkFqcGc@._V1_QL75_UY281_CR8,0,500,281_.jpg",
 						ID:            "tt1000552",
-						Name:          "I Fuckin&#x27; Miss Cory and Trevor",
+						Name:          "I Fuckin' Miss Cory and Trevor",
 					},
 					{
 						SeasonNumber:  7,
@@ -53,7 +53,7 @@ func TestSeason(t *testing.T) {
 						EpisodeNumber: 6,
 						ImageURL:      "https://m.media-amazon.com/images/M/MV5BZDI4Nzc3MTMtNjM0OC00ZDJiLTgzMzctNTZjZGExOTM4MGY3XkEyXkFqcGc@._V1_QL75_UY281_CR11,0,500,281_.jpg",
 						ID:            "tt1026371",
-						Name:          "We Can&#x27;t Call People Without Wings Angels So We Call Them Friends",
+						Name:          "We Can't Call People Without Wings Angels So We Call Them Friends",
 					},
 					{
 						SeasonNumber:  7,
@@ -111,7 +111,7 @@ func TestSeason(t *testing.T) {
 						EpisodeNumber: 3,
 						ImageURL:      "https://m.media-amazon.com/images/M/MV5BMmE4MDVhODctZGZmZi00MTdiLTkyZjUtYzBmNWZlYWU5ZDVlXkEyXkFqcGc@._V1_QL75_UX500_CR0,40,500,281_.jpg",
 						ID:            "tt0732924",
-						Name:          "The Fuckin&#x27; Way She Goes",
+						Name:          "The Fuckin' Way She Goes",
 					},
 					{
 						SeasonNumber:  5,
@@ -125,14 +125,14 @@ func TestSeason(t *testing.T) {
 						EpisodeNumber: 5,
 						ImageURL:      "https://m.media-amazon.com/images/M/MV5BMjU3Y2IzMDMtYjFkZC00NTdhLWI1YTQtMTRhNGM2MDY2MWJhXkEyXkFqcGdeQXVyMTQ2ODE0NDA@._V1_QL75_UX500_CR0,52,500,281_.jpg",
 						ID:            "tt0732913",
-						Name:          "Mr. Lahey Is a Fuckin&#x27; Drunk, and He Always Will Be",
+						Name:          "Mr. Lahey Is a Fuckin' Drunk, and He Always Will Be",
 					},
 					{
 						SeasonNumber:  5,
 						EpisodeNumber: 6,
 						ImageURL:      "https://m.media-amazon.com/images/M/MV5BNTE1MmU3MjItZWE4ZC00NjkwLTllZDAtMjM4YTlmZDM2MDRlXkEyXkFqcGdeQXVyMTQ2ODE0NDA@._V1_QL75_UX500_CR0,46,500,281_.jpg",
 						ID:            "tt0732904",
-						Name:          "Don&#x27;t Cross the Shitline",
+						Name:          "Don't Cross the Shitline",
 					},
 					{
 						SeasonNumber:  5,
@@ -146,7 +146,7 @@ func TestSeason(t *testing.T) {
 						EpisodeNumber: 8,
 						ImageURL:      "https://m.media-amazon.com/images/M/MV5BYTg1ZGI2YmItNDJiZC00NjJiLWE2ZTEtNTc2MGJlNzM3YjRkL2ltYWdlL2ltYWdlXkEyXkFqcGdeQXVyMTQ2ODE0NDA@._V1_QL75_UX500_CR0,46,500,281_.jpg",
 						ID:            "tt0732905",
-						Name:          "Dressed All Over &amp; Zesty Mordant",
+						Name:          "Dressed All Over & Zesty Mordant",
 					},
 					{
 						SeasonNumber:  5,

diff --git a/title.go b/title.go
@@ -4,6 +4,7 @@ import (
 	"encoding/json"
 	"errors"
 	"fmt"
+	"html"
 	"io"
 	"net/http"
 	"regexp"
@@ -369,7 +370,7 @@ func (t *Title) Parse(page []byte) error {
 		}
 	}
 
-	t.Description = v.Description
+	t.Description = html.UnescapeString(v.Description)
 
 	s = titlePosterRE.FindSubmatch(page)
 	if s != nil {

diff --git a/title_test.go b/title_test.go
@@ -269,7 +269,7 @@ func TestTitle(t *testing.T) {
 				Genres:        []string{"Crime", "Drama", "Romance"},
 				Languages:     []string{"Korean"},
 				Nationalities: []string{"South Korea", "Japan"},
-				Description:   "A transient young man breaks into empty homes to partake of the vacationing residents&apos; lives for a few days.",
+				Description:   "A transient young man breaks into empty homes to partake of the vacationing residents' lives for a few days.",
 				Poster:        Media{ID: "rm880057600", TitleID: "tt0423866", URL: "https://www.imdb.com/title/tt0423866/mediaviewer/rm880057600", ContentURL: "https://m.media-amazon.com/images/M/MV5BMTM1ODIwNzM5OV5BMl5BanBnXkFtZTcwNjk5MDkyMQ@@._V1_.jpg"},
 			},
 		},
@@ -309,7 +309,7 @@ func TestTitle(t *testing.T) {
 				Genres:        []string{"Comedy", "Crime"},
 				Languages:     []string{"English"},
 				Nationalities: []string{"Canada"},
-				Description:   "When Randy and Lahey rip the boys off on buying Rush tickets, Ricky &quot;borrows&quot; guitarist Alex Lifeson for access to the concert.",
+				Description:   "When Randy and Lahey rip the boys off on buying Rush tickets, Ricky \"borrows\" guitarist Alex Lifeson for access to the concert.",
 				Poster:        Media{ID: "rm1259893761", TitleID: "tt0732901", URL: "https://www.imdb.com/title/tt0732901/mediaviewer/rm1259893761", ContentURL: "https://m.media-amazon.com/images/M/MV5BNjY4YjJhZDQtM2U3ZS00ZThkLWIyZmMtZTY3MmU3N2QwOTVmXkEyXkFqcGc@._V1_.jpg"},
 				Season:        3,
 				Episode:       5,