From a8333b8afac072dc4b6736fa231148f250f3cb9f Mon Sep 17 00:00:00 2001
From: sentriz <senan@senan.xyz>
Date: Sun, 10 Dec 2023 18:44:59 +0000
Subject: [PATCH] consolidate external text trimming

---
 lastfm/client.go                              | 20 -------------------
 server/ctrlsubsonic/handlers_by_tags.go       |  4 ++--
 server/ctrlsubsonic/spec/construct_podcast.go | 13 ++----------
 server/ctrlsubsonic/spec/spec.go              | 17 ++++++++++++++++
 4 files changed, 21 insertions(+), 33 deletions(-)

diff --git a/lastfm/client.go b/lastfm/client.go
index d29df73..cc38eaa 100644
--- a/lastfm/client.go
+++ b/lastfm/client.go
@@ -8,10 +8,8 @@ import (
 	"fmt"
 	"net/http"
 	"net/url"
-	"regexp"
 	"sort"
 	"strconv"
-	"strings"
 	"time"
 
 	"github.com/andybalholm/cascadia"
@@ -60,9 +58,6 @@ func (c *Client) ArtistGetInfo(artistName string) (Artist, error) {
 		return Artist{}, fmt.Errorf("make request: %w", err)
 	}
 
-	resp.Artist.Bio.Summary = cleanLicenceText(resp.Artist.Bio.Summary)
-	resp.Artist.Bio.Content = cleanLicenceText(resp.Artist.Bio.Content)
-
 	return resp.Artist, nil
 }
 
@@ -83,9 +78,6 @@ func (c *Client) AlbumGetInfo(artistName, albumName string) (Album, error) {
 		return Album{}, fmt.Errorf("make request: %w", err)
 	}
 
-	resp.Album.Wiki.Summary = cleanLicenceText(resp.Album.Wiki.Summary)
-	resp.Album.Wiki.Content = cleanLicenceText(resp.Album.Wiki.Content)
-
 	return resp.Album, nil
 }
 
@@ -333,15 +325,3 @@ func GetParamSignature(params url.Values, secret string) string {
 	hash := md5.Sum([]byte(toHash))
 	return hex.EncodeToString(hash[:])
 }
-
-var doublePuncExpr = regexp.MustCompile(`\.\s+\.\s+`)
-var licenceExpr = regexp.MustCompile(`(?i)user-contributed text.*`)
-
-func cleanLicenceText(text string) string {
-	text = licenceExpr.ReplaceAllString(text, "")
-	text = doublePuncExpr.ReplaceAllString(text, ". ")
-	text = strings.ReplaceAll(text, " .", ".")
-	text = strings.Join(strings.Fields(text), " ")
-	text = strings.TrimSpace(text)
-	return text
-}
diff --git a/server/ctrlsubsonic/handlers_by_tags.go b/server/ctrlsubsonic/handlers_by_tags.go
index 1f45247..a4d8274 100644
--- a/server/ctrlsubsonic/handlers_by_tags.go
+++ b/server/ctrlsubsonic/handlers_by_tags.go
@@ -333,7 +333,7 @@ func (c *Controller) ServeGetArtistInfoTwo(r *http.Request) *spec.Response {
 		return sub
 	}
 
-	sub.ArtistInfoTwo.Biography = info.Biography
+	sub.ArtistInfoTwo.Biography = spec.CleanExternalText(info.Biography)
 	sub.ArtistInfoTwo.MusicBrainzID = info.MusicBrainzID
 	sub.ArtistInfoTwo.LastFMURL = info.LastFMURL
 
@@ -409,7 +409,7 @@ func (c *Controller) ServeGetAlbumInfoTwo(r *http.Request) *spec.Response {
 		return sub
 	}
 
-	sub.AlbumInfo.Notes = info.Notes
+	sub.AlbumInfo.Notes = spec.CleanExternalText(info.Notes)
 	sub.AlbumInfo.MusicBrainzID = info.MusicBrainzID
 	sub.AlbumInfo.LastFMURL = info.LastFMURL
 
diff --git a/server/ctrlsubsonic/spec/construct_podcast.go b/server/ctrlsubsonic/spec/construct_podcast.go
index 70e51ce..bff2555 100644
--- a/server/ctrlsubsonic/spec/construct_podcast.go
+++ b/server/ctrlsubsonic/spec/construct_podcast.go
@@ -2,19 +2,14 @@ package spec
 
 import (
 	"go.senan.xyz/gonic/db"
-	"jaytaylor.com/html2text"
 )
 
 func NewPodcastChannel(p *db.Podcast) *PodcastChannel {
-	desc, err := html2text.FromString(p.Description, html2text.Options{TextOnly: true})
-	if err != nil {
-		desc = ""
-	}
 	ret := &PodcastChannel{
 		ID:               p.SID(),
 		OriginalImageURL: p.ImageURL,
 		Title:            p.Title,
-		Description:      desc,
+		Description:      CleanExternalText(p.Description),
 		URL:              p.URL,
 		CoverArt:         p.SID(),
 		Status:           "skipped",
@@ -30,17 +25,13 @@ func NewPodcastEpisode(pe *db.PodcastEpisode) *PodcastEpisode {
 	if pe == nil {
 		return nil
 	}
-	desc, err := html2text.FromString(pe.Description, html2text.Options{TextOnly: true})
-	if err != nil {
-		desc = ""
-	}
 	r := &PodcastEpisode{
 		ID:          pe.SID(),
 		StreamID:    pe.SID(),
 		ContentType: pe.MIME(),
 		ChannelID:   pe.PodcastSID(),
 		Title:       pe.Title,
-		Description: desc,
+		Description: CleanExternalText(pe.Description),
 		Status:      string(pe.Status),
 		CoverArt:    pe.PodcastSID(),
 		PublishDate: *pe.PublishDate,
diff --git a/server/ctrlsubsonic/spec/spec.go b/server/ctrlsubsonic/spec/spec.go
index 4a33caa..5a0df70 100644
--- a/server/ctrlsubsonic/spec/spec.go
+++ b/server/ctrlsubsonic/spec/spec.go
@@ -2,11 +2,13 @@ package spec
 
 import (
 	"fmt"
+	"regexp"
 	"strings"
 	"time"
 
 	"go.senan.xyz/gonic"
 	"go.senan.xyz/gonic/server/ctrlsubsonic/specid"
+	"jaytaylor.com/html2text"
 )
 
 // https://web.archive.org/web/20220707025402/https://www.subsonic.org/pages/api.jsp
@@ -472,3 +474,18 @@ func formatRating(rating float64) string {
 func formatExt(ext string) string {
 	return strings.TrimPrefix(ext, ".")
 }
+
+var doublePuncExpr = regexp.MustCompile(`\.\s+\.\s+`)
+var licenceExpr = regexp.MustCompile(`(?i)\buser-contributed text.*`)
+var readMoreExpr = regexp.MustCompile(`(?i)\bread more on.*`)
+
+func CleanExternalText(text string) string {
+	text, _ = html2text.FromString(text, html2text.Options{TextOnly: true})
+	text = licenceExpr.ReplaceAllString(text, "")
+	text = readMoreExpr.ReplaceAllString(text, "")
+	text = doublePuncExpr.ReplaceAllString(text, ". ")
+	text = strings.ReplaceAll(text, " .", ".")
+	text = strings.Join(strings.Fields(text), " ")
+	text = strings.TrimSpace(text)
+	return text
+}