Merge main branches

1 files changed, 210 insertions, 119 deletions

diff --git a/main.go b/main.go
index 00c3779..9985c1e 100644
--- a/main.go
+++ b/main.go
@@ -7,12 +7,21 @@ import (
 	"io"
 	"log"
 	"net/http"
+	"net/url"
 	"strconv"
 	"strings"
 	"time"
 
 	_ "github.com/ncruces/go-sqlite3/driver"
 	_ "github.com/ncruces/go-sqlite3/embed"
+	"github.com/PuerkitoBio/goquery"
+)
+
+const (
+	artistFetchDelay = 250 * time.Millisecond
+	pageFetchDelay   = 500 * time.Millisecond
+	maxPages         = 1000
+	itemsPerPage     = 1000
 )
 
 // RSS structures
@@ -50,6 +59,44 @@ type TitleAuthor struct {
 	Author string
 }
 
+func extractUsername(trackURL string) (string, error) {
+	u, err := url.Parse(trackURL)
+	if err != nil {
+		return "", err
+	}
+
+	parts := strings.Split(strings.Trim(u.Path, "/"), "/")
+	if len(parts) < 1 {
+		return "", fmt.Errorf("invalid URL format: %s", trackURL)
+	}
+
+	return parts[0], nil
+}
+
+func fetchArtistName(trackURL string) (string, error) {
+	resp, err := http.Get(trackURL)
+	if err != nil {
+		return "", err
+	}
+	defer resp.Body.Close()
+
+	if resp.StatusCode != 200 {
+		return "", fmt.Errorf("HTTP %d", resp.StatusCode)
+	}
+
+	doc, err := goquery.NewDocumentFromReader(resp.Body)
+	if err != nil {
+		return "", err
+	}
+
+	artistName := strings.TrimSpace(strings.TrimPrefix(doc.Find("#item_user a").Text(), "By "))
+	if artistName == "" {
+		return "", fmt.Errorf("artist name not found")
+	}
+
+	return artistName, nil
+}
+
 func parseYearWeek(date string) (YearWeek, error) {
 	parts := strings.Split(date, "/")
 	if len(parts) != 2 {
@@ -95,149 +142,145 @@ func isValidXMLByte(b byte) bool {
 	return b == 0x09 || b == 0x0A || b == 0x0D || b >= 0x20
 }
 
-func main() {
-	fmt.Println("Program started")
-	// Create SQLite database
-	db, err := sql.Open("sqlite3", "weeklybeats.db")
+func fetchRSSPage(year, page int) (*RSS, error) {
+	url := fmt.Sprintf("https://weeklybeats.com/music/rss?limit=%d&year=%d&page=%d", itemsPerPage, year, page)
+
+	resp, err := http.Get(url)
 	if err != nil {
-		log.Fatal("Failed to open database:", err)
+		return nil, fmt.Errorf("failed to fetch: %w", err)
 	}
-	defer db.Close()
+	defer resp.Body.Close()
 
-	// Create table
-	createTable := `
-	CREATE TABLE IF NOT EXISTS tracks (
-		id INTEGER PRIMARY KEY AUTOINCREMENT,
-		title TEXT NOT NULL,
-		link TEXT NOT NULL,
-		author TEXT,
-		week INTEGER,
-		year INTEGER,
-		url TEXT,
-		UNIQUE(author, week, year)
-	);`
+	if resp.StatusCode != 200 {
+		return nil, fmt.Errorf("received status %d", resp.StatusCode)
+	}
 
-	_, err = db.Exec(createTable)
+	body, err := io.ReadAll(resp.Body)
 	if err != nil {
-		log.Fatal("Failed to create table:", err)
+		return nil, fmt.Errorf("failed to read response: %w", err)
 	}
 
-	fmt.Println("Starting to scrape Weekly Beats RSS feed...")
+	cleanedBody := filterInvalidXMLBytes(body)
 
-	// Prepare insert statement
-	insertStmt, err := db.Prepare(`
-		INSERT OR IGNORE INTO tracks (title, link, author, week, year, url) 
-		VALUES (?, ?, ?, ?, ?, ?)
-	`)
+	var rss RSS
+	err = xml.Unmarshal(cleanedBody, &rss)
 	if err != nil {
-		log.Fatal("Failed to prepare insert statement:", err)
+		return nil, fmt.Errorf("failed to parse XML: %w", err)
 	}
-	defer insertStmt.Close()
 
-	year := 2012
+	return &rss, nil
+}
+
+func getArtistNameFromTitle(title string) (string, error) {
+	titleAuthor, err := parseTitle(title)
+	if err != nil {
+		return "", err
+	}
+	return titleAuthor.Author, nil
+}
+
+func resolveArtistName(item Item, cache map[string]string) (string, error) {
+	username, err := extractUsername(item.Link)
+	if err != nil {
+		log.Printf("Failed to extract username from '%s': %v", item.Link, err)
+		return getArtistNameFromTitle(item.Title)
+	}
+
+	// Try cache first
+	if artistName := cache[username]; artistName != "" {
+		return artistName, nil
+	}
+
+	// Not in cache, fetch from track page
+	artistName, err := fetchArtistName(item.Link)
+	if err != nil {
+		log.Printf("Failed to fetch artist for '%s': %v, using RSS fallback", username, err)
+		return getArtistNameFromTitle(item.Title)
+	}
+
+	cache[username] = artistName
+	fmt.Printf("Cached artist name '%s' for username '%s'\n", artistName, username)
+	time.Sleep(artistFetchDelay)
+
+	return artistName, nil
+}
+
+func insertItems(insertStmt *sql.Stmt, items []Item, cache map[string]string) int {
+	inserted := 0
+	for _, item := range items {
+		artistName, err := resolveArtistName(item, cache)
+		if err != nil {
+			log.Printf("Failed to resolve artist for item '%s': %v", item.Title, err)
+			continue
+		}
+
+		titleAuthor, err := parseTitle(item.Title)
+		if err != nil {
+			log.Printf("Failed to parse title for item '%s': %v", item.Title, err)
+			continue
+		}
+
+		yearWeek, err := parseYearWeek(item.Category)
+		if err != nil {
+			log.Printf("Failed to parse date for item '%s': %v", item.Title, err)
+			continue
+		}
+
+		_, err = insertStmt.Exec(titleAuthor.Title, item.Link, artistName, yearWeek.Week, yearWeek.Year, item.Enclosure.URL)
+		if err != nil {
+			log.Printf("Failed to insert item '%s': %v", item.Title, err)
+			continue
+		}
+		inserted++
+	}
+	return inserted
+}
+
+func scrapeYear(year int, insertStmt *sql.Stmt, cache map[string]string) int {
 	totalItems := 0
-	for year <= 2024 {
-		page := 1
-
-		for {
-			fmt.Printf("Fetching page %d for year %d...\n", page, year)
-
-			// Construct URL with page parameter
-			url := fmt.Sprintf("https://weeklybeats.com/music/rss?limit=1000&year=%d&page=%d", year, page)
-
-			// Fetch RSS feed
-			resp, err := http.Get(url)
-			if err != nil {
-				log.Printf("Failed to fetch page %d: %v", page, err)
-				break
-			}
-
-			if resp.StatusCode != 200 {
-				fmt.Printf("Received status %d for page %d, stopping\n", resp.StatusCode, page)
-				resp.Body.Close()
-				break
-			}
-
-			body, err := io.ReadAll(resp.Body)
-			resp.Body.Close()
-
-			if err != nil {
-				log.Printf("Failed to read response body for page %d: %v", page, err)
-				break
-			}
-
-			cleanedBody := filterInvalidXMLBytes(body)
-
-			// Parse XML
-			var rss RSS
-			err = xml.Unmarshal(cleanedBody, &rss)
-			if err != nil {
-				log.Printf("Failed to parse XML for page %d: %v", page, err)
-				break
-			}
-
-			// Check if we got any items
-			if len(rss.Channel.Items) == 0 {
-				fmt.Printf("No items found on page %d, stopping\n", page)
-				break
-			}
-
-			fmt.Printf("Found %d items on page %d\n", len(rss.Channel.Items), page)
-
-			// Insert items into database
-			pageItems := 0
-			for _, item := range rss.Channel.Items {
-				title, err := parseTitle(item.Title)
-				if err != nil {
-					log.Printf("Failed to parse title for item '%s': %v", title, err)
-					continue
-				}
-
-				yearWeek, err := parseYearWeek(item.Category)
-				if err != nil {
-					log.Printf("Failed to parse date for item '%s': %v", title, err)
-					continue
-				}
-
-				_, err = insertStmt.Exec(title, item.Link, item.Author, yearWeek.Week, yearWeek.Year, item.Enclosure.URL)
-				if err != nil {
-					log.Printf("Failed to insert item '%s': %v", item.Title, err)
-					continue
-				}
-				pageItems++
-			}
-
-			fmt.Printf("Inserted %d new items from page %d\n", pageItems, page)
-			totalItems += pageItems
-
-			// Be respectful to the server
-			time.Sleep(500 * time.Millisecond)
-
-			page++
-
-			// Safety check to prevent infinite loops
-			if page > 1000 {
-				fmt.Println("Reached maximum page limit (1000), stopping")
-				break
-			}
+	page := 1
+
+	for {
+		fmt.Printf("Fetching page %d for year %d...\n", page, year)
+
+		rss, err := fetchRSSPage(year, page)
+		if err != nil {
+			log.Printf("Failed to fetch page %d: %v", page, err)
+			break
 		}
 
-		year += 2
+		if len(rss.Channel.Items) == 0 {
+			fmt.Printf("No items found on page %d, stopping\n", page)
+			break
+		}
+
+		fmt.Printf("Found %d items on page %d\n", len(rss.Channel.Items), page)
+
+		pageItems := insertItems(insertStmt, rss.Channel.Items, cache)
+		fmt.Printf("Inserted %d new items from page %d\n", pageItems, page)
+		totalItems += pageItems
+
+		time.Sleep(pageFetchDelay)
+		page++
+
+		if page > maxPages {
+			fmt.Printf("Reached maximum page limit (%d), stopping\n", maxPages)
+			break
+		}
 	}
 
-	// Print summary
-	fmt.Printf("\nScraping complete! Total items processed: %d\n", totalItems)
+	return totalItems
+}
 
-	// Query and display some statistics
+func printStatistics(db *sql.DB) {
 	var count int
-	err = db.QueryRow("SELECT COUNT(*) FROM tracks").Scan(&count)
+	err := db.QueryRow("SELECT COUNT(*) FROM tracks").Scan(&count)
 	if err != nil {
 		log.Printf("Failed to count records: %v", err)
 	} else {
 		fmt.Printf("Total records in database: %d\n", count)
 	}
 
-	// Show sample of data
 	fmt.Println("\nSample of stored data:")
 	rows, err := db.Query("SELECT title, author, link FROM tracks LIMIT 5")
 	if err != nil {
@@ -256,3 +299,51 @@ func main() {
 		fmt.Printf("- %s by %s (%s)\n", title, author, link)
 	}
 }
+
+func main() {
+	fmt.Println("Program started")
+
+	db, err := sql.Open("sqlite3", "weeklybeats.db")
+	if err != nil {
+		log.Fatal("Failed to open database:", err)
+	}
+	defer db.Close()
+
+	createTable := `
+	CREATE TABLE IF NOT EXISTS tracks (
+		id INTEGER PRIMARY KEY AUTOINCREMENT,
+		title TEXT NOT NULL,
+		link TEXT NOT NULL,
+		author TEXT,
+		week INTEGER,
+		year INTEGER,
+		url TEXT,
+		UNIQUE(author, week, year)
+	);`
+
+	_, err = db.Exec(createTable)
+	if err != nil {
+		log.Fatal("Failed to create table:", err)
+	}
+
+	fmt.Println("Starting to scrape Weekly Beats RSS feed...")
+
+	insertStmt, err := db.Prepare(`
+		INSERT OR IGNORE INTO tracks (title, link, author, week, year, url)
+		VALUES (?, ?, ?, ?, ?, ?)
+	`)
+	if err != nil {
+		log.Fatal("Failed to prepare insert statement:", err)
+	}
+	defer insertStmt.Close()
+
+	cache := make(map[string]string)
+
+	totalItems := 0
+	for year := 2012; year <= 2024; year += 2 {
+		totalItems += scrapeYear(year, insertStmt, cache)
+	}
+
+	fmt.Printf("\nScraping complete! Total items processed: %d\n", totalItems)
+	printStatistics(db)
+}