generate unique visitor key from date + ip + user agent

api/collect.go

@@ -1,7 +1,9 @@
 package api
 
 import (
+	"database/sql"
 	"encoding/base64"
+	"log"
 	"net/http"
 	"strings"
 	"time"
@@ -42,13 +44,18 @@ func CollectHandler(w http.ResponseWriter, r *http.Request) {
 	defer stmt.Close()
 	err := stmt.QueryRow(page.Hostname, page.Path).Scan(&page.ID)
 	if err != nil {
+		if err == sql.ErrNoRows {
 			page.Save(db.Conn)
+		} else {
+			log.Fatal(err)
+		}
 	}
 
 	// find or insert visitor.
-	// TODO: Mask IP Address
+	now := time.Now()
+	ip := getRequestIp(r)
 	visitor := models.Visitor{
-		IpAddress:        getRequestIp(r),
+		IpAddress:        ip,
 		BrowserLanguage:  q.Get("l"),
 		ScreenResolution: q.Get("sr"),
 		DeviceOS:         ua.OS(),
@@ -60,21 +67,26 @@ func CollectHandler(w http.ResponseWriter, r *http.Request) {
 	visitor.BrowserName = parseMajorMinor(visitor.BrowserName)
 
 	// query by unique visitor key
-	visitor.GenerateKey()
+	visitor.Key = visitor.GenerateKey(now.Format("2006-01-02"), visitor.IpAddress, r.UserAgent())
+
 	stmt, _ = db.Conn.Prepare("SELECT v.id FROM visitors v WHERE v.visitor_key = ? LIMIT 1")
 	defer stmt.Close()
 	err = stmt.QueryRow(visitor.Key).Scan(&visitor.ID)
 	if err != nil {
-		visitor.Save(db.Conn)
+		if err == sql.ErrNoRows {
+			err = visitor.Save(db.Conn)
+			checkError(err)
+		} else {
+			log.Fatal(err)
+		}
 	}
 
-	now := time.Now().Format("2006-01-02 15:04:05")
 	pageview := models.Pageview{
 		PageID:          page.ID,
 		VisitorID:       visitor.ID,
 		ReferrerUrl:     q.Get("ru"),
 		ReferrerKeyword: q.Get("rk"),
-		Timestamp:       now,
+		Timestamp:       now.Format("2006-01-02 15:04:05"),
 	}
 
 	// only store referrer URL if not coming from own site

db/seed.go

@@ -99,6 +99,7 @@ func Seed(n int) {
 
 		// print a dot as progress indicator
 		fmt.Print(".")
+		date := randomDateBeforeNow()
 
 		// create or find visitor
 		visitor := models.Visitor{
@@ -110,7 +111,8 @@ func Seed(n int) {
 			ScreenResolution: randSliceElement(screenResolutions),
 			Country:          randomdata.Country(randomdata.TwoCharCountry),
 		}
-		visitor.Key = visitor.GenerateKey()
+		dummyUserAgent := visitor.BrowserName + visitor.BrowserVersion + visitor.DeviceOS
+		visitor.Key = visitor.GenerateKey(date.Format("2006-01-02"), visitor.IpAddress, dummyUserAgent)
 
 		err := stmtVisitor.QueryRow(visitor.Key).Scan(&visitor.ID)
 		if err != nil {
@@ -118,7 +120,6 @@ func Seed(n int) {
 		}
 
 		// generate random timestamp
-		date := randomDateBeforeNow()
 		timestamp := fmt.Sprintf("%s %d:%d:%d", date.Format("2006-01-02"), randInt(10, 24), randInt(10, 60), randInt(10, 60))
 
 		pv := models.Pageview{
@@ -144,7 +145,7 @@ func Seed(n int) {
 func randomDate() time.Time {
 	now := time.Now()
 	month := months[randInt(0, len(months))]
-	t := time.Date(now.Year(), month, randInt(1, 31), randInt(0, 23), randInt(0, 59), randInt(0, 59), 0, time.UTC)
+	t := time.Date(randInt(now.Year()-1, now.Year()), month, randInt(1, 31), randInt(0, 23), randInt(0, 59), randInt(0, 59), 0, time.UTC)
 	return t
 }
 

models/visitor.go

@@ -34,7 +34,6 @@ func (v *Visitor) Save(conn *sql.DB) error {
 		return err
 	}
 	defer stmt.Close()
-
 	result, err := stmt.Exec(
 		v.Key,
 		v.IpAddress,
@@ -53,8 +52,8 @@ func (v *Visitor) Save(conn *sql.DB) error {
 	return err
 }
 
-// GenerateKey generates the "unique" visitor key
-func (v *Visitor) GenerateKey() string {
-	byteKey := md5.Sum([]byte(v.IpAddress + v.DeviceOS + v.BrowserName + v.ScreenResolution))
+// GenerateKey generates the "unique" visitor key from date, user agent + screen resolution
+func (v *Visitor) GenerateKey(date string, ipAddress string, userAgent string) string {
+	byteKey := md5.Sum([]byte(date + ipAddress + userAgent))
 	return hex.EncodeToString(byteKey[:])
 }