dedupe blob file data

database.go

@@ -17,6 +17,7 @@ package main
 
 import (
 	"bytes"
+	"crypto/sha512"
 	"database/sql"
 	"encoding/json"
 	"fmt"
@@ -509,8 +510,20 @@ func savefile(name string, desc string, url string, media string, local bool, da
 	return fileid, err
 }
 
+func hashfiledata(data []byte) string {
+	h := sha512.New512_256()
+	h.Write(data)
+	return fmt.Sprintf("%x", h.Sum(nil))
+}
+
 func savefileandxid(name string, desc string, url string, media string, local bool, data []byte) (int64, string, error) {
-	xid := xfiltrate()
+	var xid string
+	if local {
+		hash := hashfiledata(data)
+		row := stmtCheckFileData.QueryRow(hash)
+		err := row.Scan(&xid)
+		if err == sql.ErrNoRows {
+			xid = xfiltrate()
 			switch media {
 			case "image/png":
 				xid += ".png"
@@ -521,21 +534,24 @@ func savefileandxid(name string, desc string, url string, media string, local bo
 			case "text/plain":
 				xid += ".txt"
 			}
+			_, err = stmtSaveFileData.Exec(xid, media, hash, data)
+			if err != nil {
+				return 0, "", err
+			}
+		} else if err != nil {
+			log.Printf("error checking file hash: %s", err)
+			return 0, "", err
+		}
 		if url == "" {
 			url = fmt.Sprintf("https://%s/d/%s", serverName, xid)
 		}
+	}
 
 	res, err := stmtSaveFile.Exec(xid, name, desc, url, media, local)
 	if err != nil {
 		return 0, "", err
 	}
 	fileid, _ := res.LastInsertId()
-	if local {
-		_, err = stmtSaveFileData.Exec(xid, media, data)
-		if err != nil {
-			return 0, "", err
-		}
-	}
 	return fileid, xid, nil
 }
 
@@ -890,6 +906,7 @@ var stmtHonksFromLongAgo *sql.Stmt
 var stmtHonksByHonker, stmtSaveHonk, stmtUserByName, stmtUserByNumber *sql.Stmt
 var stmtEventHonks, stmtOneBonk, stmtFindZonk, stmtFindXonk, stmtSaveDonk *sql.Stmt
 var stmtFindFile, stmtGetFileData, stmtSaveFileData, stmtSaveFile *sql.Stmt
+var stmtCheckFileData *sql.Stmt
 var stmtAddDoover, stmtGetDoovers, stmtLoadDoover, stmtZapDoover, stmtOneHonker *sql.Stmt
 var stmtUntagged, stmtDeleteHonk, stmtDeleteDonks, stmtDeleteOnts, stmtSaveZonker *sql.Stmt
 var stmtGetZonkers, stmtRecentHonkers, stmtGetXonker, stmtSaveXonker, stmtDeleteXonker *sql.Stmt
@@ -951,7 +968,8 @@ func prepareStatements(db *sql.DB) {
 	stmtDeleteDonks = preparetodie(db, "delete from donks where honkid = ?")
 	stmtSaveFile = preparetodie(db, "insert into filemeta (xid, name, description, url, media, local) values (?, ?, ?, ?, ?, ?)")
 	blobdb := openblobdb()
-	stmtSaveFileData = preparetodie(blobdb, "insert into filedata (xid, media, content) values (?, ?, ?)")
+	stmtSaveFileData = preparetodie(blobdb, "insert into filedata (xid, media, hash, content) values (?, ?, ?, ?)")
+	stmtCheckFileData = preparetodie(blobdb, "select xid from filedata where hash = ?")
 	stmtGetFileData = preparetodie(blobdb, "select media, content from filedata where xid = ?")
 	stmtFindXonk = preparetodie(db, "select honkid from honks where userid = ? and xid = ?")
 	stmtFindFile = preparetodie(db, "select fileid, xid from filemeta where url = ? and local = 1")

docs/changelog.txt

@@ -2,6 +2,8 @@ changelog
 
 === next
 
++ Dedupe blob file data.
+
 - Custom lingo for those who don't like honking.
 
 + Better support for rich text bios.

upgradedb.go

@@ -23,7 +23,7 @@ import (
 	"time"
 )
 
-var myVersion = 39
+var myVersion = 40
 
 type dbexecer interface {
 	Exec(query string, args ...interface{}) (sql.Result, error)
@@ -168,6 +168,39 @@ func upgradedb() {
 		doordie(db, "update config set value = 39 where key = 'dbversion'")
 		fallthrough
 	case 39:
+		blobdb := openblobdb()
+		doordie(blobdb, "alter table filedata add column hash text")
+		doordie(blobdb, "create index idx_filehash on filedata(hash)")
+		rows, err := blobdb.Query("select xid, content from filedata")
+		if err != nil {
+			log.Fatal(err)
+		}
+		m := make(map[string]string)
+		for rows.Next() {
+			var xid string
+			var data sql.RawBytes
+			err := rows.Scan(&xid, &data)
+			if err != nil {
+				log.Fatal(err)
+			}
+			hash := hashfiledata(data)
+			m[xid] = hash
+		}
+		rows.Close()
+		tx, err := blobdb.Begin()
+		if err != nil {
+			log.Fatal(err)
+		}
+		for xid, hash := range m {
+			doordie(tx, "update filedata set hash = ? where xid = ?", hash, xid)
+		}
+		err = tx.Commit()
+		if err != nil {
+			log.Fatal(err)
+		}
+		doordie(db, "update config set value = 40 where key = 'dbversion'")
+		fallthrough
+	case 40:
 
 	default:
 		log.Fatalf("can't upgrade unknown version %d", dbversion)

util.go

@@ -172,7 +172,7 @@ func initblobdb() {
 		log.Print(err)
 		return
 	}
-	_, err = blobdb.Exec("create table filedata (xid text, media text, content blob)")
+	_, err = blobdb.Exec("create table filedata (xid text, media text, hash text, content blob)")
 	if err != nil {
 		log.Print(err)
 		return
@@ -182,6 +182,11 @@ func initblobdb() {
 		log.Print(err)
 		return
 	}
+	_, err = blobdb.Exec("create index idx_filehash on filedata(hash)")
+	if err != nil {
+		log.Print(err)
+		return
+	}
 	blobdb.Close()
 }