Add hashes for each book

Check if the uploaded book is already in the library

lib/database/books.go

@@ -1,6 +1,8 @@
 package database
 
 import (
+	log "github.com/cihub/seelog"
+
 	"strings"
 	"time"
 
@@ -20,6 +22,7 @@ type Book struct {
 	Lang        string `pg:"type:varchar(3)"`
 	Isbn        string `pg:"type:varchar(13)"`
 	FileSize    int    `pg:"type:integer"`
+	FileHash    []byte
 	Cover       bool      `pg:",notnull"`
 	Active      bool      `pg:",notnull"`
 	UploadDate  time.Time `pg:"type:timestamp"`
@@ -165,6 +168,18 @@ func (db *pgDB) IsBookActive(id string) bool {
 	return active[0]
 }
 
+// ExistsBookHash checks if the given hash matches the hash of any book in the lbirary
+func (db *pgDB) ExistsBookHash(hash []byte) bool {
+	count, err := db.sql.Model(&Book{}).
+		Where("file_hash = ?", hash).
+		Count()
+	if err != nil {
+		log.Warnf("There was an error looking for the hash: %v", err)
+		return false
+	}
+	return count > 0
+}
+
 type columnq struct {
 	column string
 	value  string

lib/database/database.go

@@ -17,6 +17,7 @@ type DB interface {
 	UpdateBook(id string, data map[string]interface{}) error
 	ActiveBook(id string) error
 	IsBookActive(id string) bool
+	ExistsBookHash(hash []byte) bool
 	AddUser(name string, pass string) error
 	AddRawUser(name string, hpass []byte, salt []byte, role string) error
 	SetAdminUser(name string, pass string) error
@@ -149,6 +150,7 @@ ALTER TABLE books ALTER COLUMN active SET DEFAULT false ;
 CREATE INDEX IF NOT EXISTS books_lang_idx ON books (lang);
 CREATE INDEX IF NOT EXISTS books_isbn_idx ON books (isbn);
 CREATE INDEX IF NOT EXISTS books_active_idx ON books (active);
+CREATE INDEX IF NOT EXISTS books_file_hash_idx ON books (file_hash);
 
 
 

lib/database/ro.go

@@ -45,6 +45,10 @@ func (db *roDB) IsBookActive(id string) bool {
 	return db.db.IsBookActive(id)
 }
 
+func (db *roDB) ExistsBookHash(hash []byte) bool {
+	return db.db.ExistsBookHash(hash)
+}
+
 func (db *roDB) AddUser(name string, pass string) error {
 	return errors.New("RO database")
 }

lib/upload.go

@@ -1,13 +1,14 @@
 package trantor
 
 import (
-	"net/http"
-
 	log "github.com/cihub/seelog"
 
 	"crypto/rand"
+	"crypto/sha256"
 	"encoding/base64"
+	"io"
 	"mime/multipart"
+	"net/http"
 
 	"github.com/meskio/epubgo"
 	"gitlab.com/trantor/trantor/lib/database"
@@ -41,6 +42,19 @@ func uploadWorker(database database.DB, store storage.Store) {
 func (req uploadRequest) processFile(db database.DB, store storage.Store) {
 	defer req.file.Close()
 
+	h := sha256.New()
+	if _, err := io.Copy(h, req.file); err != nil {
+		log.Warn("Can't read uploaded file ", req.header.Filename, ": ", err)
+		db.UpdateSubmission(req.id, "There was a problem with the uploaded book, try again later.", nil)
+		return
+	}
+	fileHash := h.Sum(nil)
+	if db.ExistsBookHash(fileHash) {
+		log.Info("Uploaded file ", req.header.Filename, " already in the library.")
+		db.UpdateSubmission(req.id, "The uploaded book is already in the library (it could be in the moderation queue)", nil)
+		return
+	}
+
 	epub, err := epubgo.Load(req.file, req.header.Size)
 	if err != nil {
 		log.Warn("Not valid epub uploaded file ", req.header.Filename, ": ", err)
@@ -61,6 +75,7 @@ func (req uploadRequest) processFile(db database.DB, store storage.Store) {
 	}
 
 	book.FileSize = int(size)
+	book.FileHash = fileHash
 	book.Cover = GetCover(epub, book.ID, store)
 
 	err = db.AddBook(book)