From bd7c6d1ee3e5bc95f726f78b0aea2283fcaaf677 Mon Sep 17 00:00:00 2001
From: Jacob Casper <dev@jacobcasper.com>
Date: Thu, 16 Apr 2020 18:27:47 -0500
Subject: [PATCH] Add a background scraper to pull in related artists

---
 main.go                                  |  3 ++
 migrations/004-create_scraped_artist.sql |  5 ++
 scrape/graph/graph.go                    | 67 ++++++++++++++++++++++++
 3 files changed, 75 insertions(+)
 create mode 100644 migrations/004-create_scraped_artist.sql
 create mode 100644 scrape/graph/graph.go

diff --git a/main.go b/main.go
index b59a627..702fd48 100644
--- a/main.go
+++ b/main.go
@@ -4,6 +4,7 @@ import (
 	"git.jacobcasper.com/brackets/env"
 	"git.jacobcasper.com/brackets/routes/artist"
 	"git.jacobcasper.com/brackets/routes/genre"
+	"git.jacobcasper.com/brackets/scrape/graph"
 	_ "github.com/mattn/go-sqlite3"
 	"log"
 	"net/http"
@@ -21,5 +22,7 @@ func main() {
 
 	http.HandleFunc("/genre", genre.Index(env))
 
+	go graph.Scrape(env)
+
 	log.Fatal(http.ListenAndServe(":8080", nil))
 }
diff --git a/migrations/004-create_scraped_artist.sql b/migrations/004-create_scraped_artist.sql
new file mode 100644
index 0000000..bc51487
--- /dev/null
+++ b/migrations/004-create_scraped_artist.sql
@@ -0,0 +1,5 @@
+CREATE TABLE SCRAPED_ARTIST (
+  ARTIST_ID TEXT UNIQUE NOT NULL,
+  SCRAPED BOOLEAN NOT NULL DEFAULT 0,
+  FOREIGN KEY(ARTIST_ID) REFERENCES ARTIST(ID)
+);
diff --git a/scrape/graph/graph.go b/scrape/graph/graph.go
new file mode 100644
index 0000000..7a31924
--- /dev/null
+++ b/scrape/graph/graph.go
@@ -0,0 +1,67 @@
+package graph
+
+import (
+	"git.jacobcasper.com/brackets/env"
+	"github.com/zmb3/spotify"
+	"log"
+	"net/http"
+	"net/url"
+	"time"
+)
+
+func Scrape(env *env.Env) {
+infinite:
+	for {
+		time.Sleep(time.Second * 5)
+		rows, err := env.Db.Db.Query(`
+SELECT ID
+FROM ARTIST
+WHERE ID NOT IN (
+  SELECT ARTIST_ID
+	FROM SCRAPED_ARTIST
+	WHERE SCRAPED == 1
+)`)
+		if err != nil {
+			log.Print(err)
+			continue infinite
+		}
+		defer rows.Close()
+
+		var artistId string
+		for rows.Next() {
+			if err := rows.Scan(&artistId); err != nil {
+				log.Print(err)
+				continue infinite
+			}
+
+			artists, err := env.C.GetRelatedArtists(spotify.ID(artistId))
+			if err != nil {
+				log.Print(err)
+				continue infinite
+			}
+
+			success := true
+		postArtists:
+			for _, artist := range artists {
+				resp, err := http.PostForm("http://localhost:8080/artist/add", url.Values{"id": {string(artist.ID)}})
+				if err != nil {
+					log.Print(err)
+					success = false
+					continue postArtists
+				}
+				if resp.StatusCode != http.StatusCreated {
+					success = false
+				}
+			}
+
+			if success {
+				env.Db.Mu.Lock()
+				env.Db.Db.Exec(`
+REPLACE INTO SCRAPED_ARTIST (ARTIST_ID, SCRAPED)
+VALUES (?, 1)`,
+					string(artistId))
+				env.Db.Mu.Unlock()
+			}
+		}
+	}
+}
-- 
2.20.1