func updateSigs() (string, error) {
url, _, _ := config.UpdateOptions()
if url == "" {
return "Update is not available for this distribution of siegfried", nil
}
response, err := getHttp(url)
if err != nil {
return "", err
}
var u Update
if err := json.Unmarshal(response, &u); err != nil {
return "", err
}
version := config.Version()
if version[0] < u.Version[0] || (version[0] == u.Version[0] && version[1] < u.Version[1]) || // if the version is out of date
u.Version == [3]int{0, 0, 0} || u.Created == "" || u.Size == 0 || u.Path == "" { // or if the unmarshalling hasn't worked and we have blank values
return "Your version of siegfried is out of date; please install latest from http://www.itforarchivists.com/siegfried before continuing.", nil
}
s, err := siegfried.Load(config.Signature())
if err == nil {
if !s.Update(u.Created) {
return "You are already up to date!", nil
}
} else {
// this hairy bit of golang exception handling is thanks to Ross! :)
if _, err = os.Stat(config.Home()); err != nil {
if os.IsNotExist(err) {
err = os.MkdirAll(config.Home(), os.ModePerm)
if err != nil {
return "", fmt.Errorf("Siegfried: cannot create home directory %s, %v", config.Home(), err)
}
} else {
return "", fmt.Errorf("Siegfried: error opening directory %s, %v", config.Home(), err)
}
}
}
fmt.Println("... downloading latest signature file ...")
response, err = getHttp(u.Path)
if err != nil {
return "", fmt.Errorf("Siegfried: error retrieving %s.\nThis may be a network or firewall issue. See https://github.com/richardlehane/siegfried/wiki/Getting-started for manual instructions.\nSystem error: %v", config.SignatureBase(), err)
}
if len(response) != u.Size {
return "", fmt.Errorf("Siegfried: error retrieving %s; expecting %d bytes, got %d bytes", config.SignatureBase(), u.Size, len(response))
}
err = ioutil.WriteFile(config.Signature(), response, os.ModePerm)
if err != nil {
return "", fmt.Errorf("Siegfried: error writing to directory, %v", err)
}
fmt.Printf("... writing %s ...\n", config.Signature())
return "Your signature file has been updated", nil
}
func blameSig(i int) error {
if *inspectHome != config.Home() {
config.SetHome(*inspectHome)
}
s, err := siegfried.Load(config.Signature())
if err != nil {
return err
}
fmt.Println(s.Blame(i, *inspectCType, *inspectCName))
return nil
}
func inspectSig(t core.MatcherType) error {
if *inspectHome != config.Home() {
config.SetHome(*inspectHome)
}
s, err := siegfried.Load(config.Signature())
if err != nil {
return err
}
fmt.Print(s.Inspect(t))
return nil
}
func setHarvestOptions() {
if *harvestHome != config.Home() {
config.SetHome(*harvestHome)
}
if *harvestDroid != config.Droid() {
config.SetDroid(*harvestDroid)()
}
if *timeout != htimeout {
config.SetHarvestTimeout(*timeout)
}
if *throttlef > 0 {
config.SetHarvestThrottle(*throttlef)
}
}
func initSets() error {
// load all json files in the sets directory and add them to a single map
sets = make(map[string][]string)
wf := func(path string, info os.FileInfo, err error) error {
if err != nil {
return errors.New("error walking sets directory, must have a 'sets' directory in siegfried home: " + err.Error())
}
if info.IsDir() {
return nil
}
switch filepath.Ext(path) {
default:
return nil // ignore non json files
case ".json":
}
set := make(map[string][]string)
byts, err := ioutil.ReadFile(path)
if err != nil {
return errors.New("error loading " + path + " " + err.Error())
}
err = json.Unmarshal(byts, &set)
if err != nil {
return errors.New("error unmarshalling " + path + " " + err.Error())
}
for k, v := range set {
k = stripComment(k)
v = stripComments(v)
sort.Strings(v)
m, ok := sets[k]
if !ok {
sets[k] = v
} else {
// if we already have this key, add any new items in its list to the existing list
for _, w := range v {
idx := sort.SearchStrings(m, w)
if idx == len(m) || m[idx] != w {
m = append(m, w)
}
}
sort.Strings(m)
sets[k] = m
}
}
return nil
}
return filepath.Walk(filepath.Join(config.Home(), "sets"), wf)
}
func main() {
flag.Parse()
/*//UNCOMMENT TO RUN PROFILER
go func() {
log.Println(http.ListenAndServe("localhost:6060", nil))
}()*/
// configure home and signature if not default
if *home != config.Home() {
config.SetHome(*home)
}
if *sig != config.SignatureBase() {
config.SetSignature(*sig)
}
// handle -update
if *update {
msg, err := updateSigs()
if err != nil {
log.Fatalf("[FATAL] failed to update signature file, %v", err)
}
fmt.Println(msg)
return
}
// handle -hash error
hashT := getHash(*hashf)
if *hashf != "" && hashT < 0 {
log.Fatalf("[FATAL] invalid hash type; choose from %s", hashChoices)
}
// load and handle signature errors
s, err := siegfried.Load(config.Signature())
if err != nil {
log.Fatalf("[FATAL] error loading signature file, got: %v", err)
}
// handle -version
if *version {
version := config.Version()
fmt.Printf("siegfried %d.%d.%d\n%s", version[0], version[1], version[2], s)
return
}
// handle -fpr
if *fprflag {
log.Printf("FPR server started at %s. Use CTRL-C to quit.\n", config.Fpr())
serveFpr(config.Fpr(), s)
return
}
// check -multi
if *multi > maxMulti || *multi < 1 || (*archive && *multi > 1) {
log.Println("[WARN] -multi must be > 0 and =< 1024. If -z, -multi must be 1. Resetting -multi to 1")
*multi = 1
}
// start logger
lg, err := newLogger(*logf)
if err != nil {
log.Fatalln(err)
}
if config.Slow() || config.Debug() {
if *serve != "" || *fprflag {
log.Fatalln("[FATAL] debug and slow logging cannot be run in server mode")
}
}
// start throttle
if *throttlef != 0 {
throttle = time.NewTicker(*throttlef)
defer throttle.Stop()
}
// start the printer
lenCtxts := *multi
if lenCtxts == 1 {
lenCtxts = 8
}
ctxts := make(chan *context, lenCtxts)
go printer(ctxts, lg)
// set default writer
var w writer
switch {
case *csvo:
w = newCSV(os.Stdout)
case *jsono:
w = newJSON(os.Stdout)
case *droido:
w = newDroid(os.Stdout)
if len(s.Fields()) != 1 || len(s.Fields()[0]) != 7 {
close(ctxts)
log.Fatalln("[FATAL] DROID output is limited to signature files with a single PRONOM identifier")
}
default:
w = newYAML(os.Stdout)
}
// overrite writer with nil writer if logging is to stdout
if lg != nil && lg.w == os.Stdout {
w = logWriter{}
}
// setup default waitgroup
wg := &sync.WaitGroup{}
// setup context pool
setCtxPool(s, w, wg, hashT, *archive)
// handle -serve
if *serve != "" {
log.Printf("Starting server at %s. Use CTRL-C to quit.\n", *serve)
listen(*serve, s, ctxts)
return
}
// handle no file/directory argument
if flag.NArg() != 1 {
close(ctxts)
log.Fatalln("[FATAL] expecting a single file or directory argument")
}
w.writeHead(s, hashT)
// support reading list files from stdin
if flag.Arg(0) == "-" {
scanner := bufio.NewScanner(os.Stdin)
for scanner.Scan() {
info, err := os.Stat(scanner.Text())
if err != nil {
info, err = retryStat(scanner.Text(), err)
}
if err != nil || info.IsDir() {
ctx := getCtx(scanner.Text(), "", "", 0)
ctx.res <- results{fmt.Errorf("failed to identify %s (in scanning mode, inputs must all be files and not directories), got: %v", scanner.Text(), err), nil, nil}
ctx.wg.Add(1)
ctxts <- ctx
} else {
identifyFile(getCtx(scanner.Text(), "", info.ModTime().Format(time.RFC3339), info.Size()), ctxts, getCtx)
}
}
} else {
err = identify(ctxts, flag.Arg(0), "", *nr, getCtx)
}
wg.Wait()
close(ctxts)
w.writeTail()
// log time elapsed
if !lg.start.IsZero() {
fmt.Fprintf(lg.w, "%s %v\n", timeString, time.Since(lg.start))
}
if err != nil {
log.Fatal(err)
}
os.Exit(0)
}
errString = "[ERROR]"
warnString = "[WARN]"
timeString = "[TIME]"
)
// flags
var (
update = flag.Bool("update", false, "update or install the default signature file")
version = flag.Bool("version", false, "display version information")
logf = flag.String("log", "error", "log errors, warnings, debug or slow output, knowns or unknowns to stderr or stdout e.g. -log error,warn,unknown,stdout")
nr = flag.Bool("nr", false, "prevent automatic directory recursion")
csvo = flag.Bool("csv", false, "CSV output format")
jsono = flag.Bool("json", false, "JSON output format")
droido = flag.Bool("droid", false, "DROID CSV output format")
sig = flag.String("sig", config.SignatureBase(), "set the signature file")
home = flag.String("home", config.Home(), "override the default home directory")
serve = flag.String("serve", "", "start siegfried server e.g. -serve localhost:5138")
multi = flag.Int("multi", 1, "set number of parallel file ID processes")
archive = flag.Bool("z", false, "scan archive formats (zip, tar, gzip, warc, arc)")
hashf = flag.String("hash", "", "calculate file checksum with hash algorithm; options "+hashChoices)
throttlef = flag.Duration("throttle", 0, "set a time to wait between scanning files e.g. 50ms")
)
var (
throttle *time.Ticker
ctxPool *sync.Pool
)
type WalkError struct {
path string
err error
func getOptions() []config.Option {
if *home != config.Home() {
config.SetHome(*home)
}
if *inspectHome != config.Home() {
config.SetHome(*inspectHome)
}
opts := []config.Option{}
// build options
if *droid != config.Droid() {
opts = append(opts, config.SetDroid(*droid))
}
if *container != config.Container() {
opts = append(opts, config.SetContainer(*container))
}
if *mi != "" {
opts = append(opts, config.SetMIMEInfo(*mi))
}
if *fdd != "" {
opts = append(opts, config.SetLOC(*fdd))
}
if *locfdd {
opts = append(opts, config.SetLOC(""))
}
if *nopronom {
opts = append(opts, config.SetNoPRONOM())
}
if *name != "" {
opts = append(opts, config.SetName(*name))
}
if *details != config.Details() {
opts = append(opts, config.SetDetails(*details))
}
if *extend != "" {
opts = append(opts, config.SetExtend(expandSets(*extend)))
}
if *extendc != "" {
if *extend == "" {
fmt.Println(
`roy: warning! Unless the container extension only extends formats defined in
the DROID signature file you should also include a regular signature extension
(-extend) that includes a FileFormatCollection element describing the new formats.`)
}
opts = append(opts, config.SetExtendC(expandSets(*extendc)))
}
if *include != "" {
opts = append(opts, config.SetLimit(expandSets(*include)))
}
if *exclude != "" {
opts = append(opts, config.SetExclude(expandSets(*exclude)))
}
if *bof != 0 {
opts = append(opts, config.SetBOF(*bof))
}
if *eof != 0 {
opts = append(opts, config.SetEOF(*eof))
}
if *noeof {
opts = append(opts, config.SetNoEOF())
}
if *multi != "" {
opts = append(opts, config.SetMulti(strings.ToLower(*multi)))
}
if *nocontainer {
opts = append(opts, config.SetNoContainer())
}
if *notext {
opts = append(opts, config.SetNoText())
}
if *noname {
opts = append(opts, config.SetNoName())
}
if *nomime {
opts = append(opts, config.SetNoMIME())
}
if *noxml {
opts = append(opts, config.SetNoXML())
}
if *noriff {
opts = append(opts, config.SetNoRIFF())
}
if *noreports {
opts = append(opts, config.SetNoReports())
}
if *doubleup {
opts = append(opts, config.SetDoubleUp())
}
if *rng != config.Range() {
opts = append(opts, config.SetRange(*rng))
}
if *distance != config.Distance() {
opts = append(opts, config.SetDistance(*distance))
}
if *choices != config.Choices() {
opts = append(opts, config.SetChoices(*choices))
}
// inspect options
if *inspectMI != "" {
opts = append(opts, config.SetMIMEInfo(*inspectMI))
}
if *inspectFDD != "" {
opts = append(opts, config.SetLOC(*fdd))
}
if *inspectLOC {
opts = append(opts, config.SetLOC(""))
}
if *inspectInclude != "" {
opts = append(opts, config.SetLimit(expandSets(*inspectInclude)))
}
if *inspectExclude != "" {
opts = append(opts, config.SetExclude(expandSets(*inspectExclude)))
}
if *inspectExtend != "" {
opts = append(opts, config.SetExtend(expandSets(*inspectExtend)))
}
if *inspectExtendc != "" {
if *inspectExtend == "" {
fmt.Println(
`roy: warning! Unless the container extension only extends formats defined in
the DROID signature file you should also include a regular signature extension
(-extend) that includes a FileFormatCollection element describing the new formats.`)
}
opts = append(opts, config.SetExtendC(expandSets(*inspectExtendc)))
}
return opts
}
E.g. roy inspect -limit @pdfa priorities
-mi, -loc, -fdd
Specify particular MIME-info or LOC FDD signature files for inspecting
formats or viewing priorities.
-reports
Build from PRONOM reports files (rather than just using the DROID XML
file as input). A bit slower but can be more accurate for a small set
of formats like FLAC.
-home
Use a different siegfried home directory.
`
var (
// BUILD, ADD flag sets
build = flag.NewFlagSet("build | add", flag.ExitOnError)
home = build.String("home", config.Home(), "override the default home directory")
droid = build.String("droid", config.Droid(), "set name/path for DROID signature file")
mi = build.String("mi", "", "set name/path for MIMEInfo signature file")
fdd = build.String("fdd", "", "set name/path for LOC FDD signature file")
locfdd = build.Bool("loc", false, "build a LOC FDD signature file")
nopronom = build.Bool("nopronom", false, "don't include PRONOM sigs with LOC signature file")
container = build.String("container", config.Container(), "set name/path for Droid Container signature file")
name = build.String("name", "", "set identifier name")
details = build.String("details", config.Details(), "set identifier details")
extend = build.String("extend", "", "comma separated list of additional signatures")
extendc = build.String("extendc", "", "comma separated list of additional container signatures")
include = build.String("limit", "", "comma separated list of PRONOM signatures to include")
exclude = build.String("exclude", "", "comma separated list of PRONOM signatures to exclude")
bof = build.Int("bof", 0, "define a maximum BOF offset")
eof = build.Int("eof", 0, "define a maximum EOF offset")
noeof = build.Bool("noeof", false, "ignore EOF segments in signatures")