Update collation

Files.com does not always return a UTF-8 encoded path. This leads to
mysql errors because the corresponding field in the database are UTF-8
formatted. This change modifies the field such that encodings other than
UTF-8 can be compared with the field.

Closes: SET-600
Signed-off-by: Nicolas Bock <nicolas.bock@canonical.com>

pkg/common/db/conn.go

@@ -4,6 +4,7 @@ import (
 	"fmt"
 
 	"github.com/canonical/athena-core/pkg/config"
+	log "github.com/sirupsen/logrus"
 	"gorm.io/driver/mysql"
 	"gorm.io/driver/sqlite"
 	"gorm.io/gorm"
@@ -25,5 +26,16 @@ func GetDBConn(cfg *config.Config) (*gorm.DB, error) {
 	}
 
 	dbInstance.AutoMigrate(File{}, Report{}, Script{})
+
+	switch cfg.Db.Dialect {
+	case "sqlite":
+		log.Debugln("Will not change collation")
+	case "mysql":
+		log.Debugln("Changing collation to UTF-8")
+		err = dbInstance.Exec("ALTER TABLE files MODIFY Path VARCHAR(10240) CHARACTER SET utf8 COLLATE utf8_general_ci").Error
+		if err != nil {
+			log.Errorln("Could not change collation of files table")
+		}
+	}
 	return dbInstance, nil
 }

pkg/monitor/monitor.go

@@ -62,6 +62,13 @@ func (m *Monitor) GetLatestFiles(dirs []string, duration time.Duration) ([]db.Fi
 	}
 
 	for _, file := range files {
+		// var count int64
+		// m.Db.Raw("SELECT COUNT(*) FROM files WHERE path COLLATE utf8_general_ci = ?", file.Path).Scan(&count)
+		// if count == 0 {
+		// 	m.Db.Exec("INSERT INTO files (path) VALUES (?)", file.Path)
+		// } else {
+		// 	m.Db.Raw("SELECT * FROM files WHERE path COLLATE utf8_general_ci = ?", file.Path).Scan(&file)
+		// }
 		m.Db.Where(db.File{Path: file.Path}).FirstOrCreate(&file)
 	}