func (group *TailGroup) Refresh() {
d, _ := os.Getwd()
logs.Debug("pwd:", d)
matches, err := filepath.Glob(group.Glob)
if err != nil {
logs.Debug("Error in glob: ", err)
} else if matches == nil {
logs.Debug("Glob matched zero files: ", group.Glob)
} else if matches != nil {
logs.Debug("Glob matched %d files: ", len(matches), group.Glob)
for _, match := range matches {
info, err := os.Stat(match)
if err != nil {
logs.Debug("Can't stat: ", err)
} else if info.IsDir() {
logs.Debug("Ignoring directory: ", match)
} else {
if time.Since(info.ModTime()).Hours() >= 1 {
logs.Debug("Ignoring idle file: ", match)
group.deactivate(match)
} else {
logs.Debug("Tailing: ", match)
group.activate(match)
}
}
}
}
}
func (tail *Tail) Poll() {
size := 16384
buffer := make([]byte, size)
for {
len, err := tail.handle.Read(buffer)
if err == io.EOF {
fi, err := tail.Stat()
if err != nil {
logs.Warn("Can't stat %s", err)
} else if fi.Size() < tail.Offset() {
logs.Warn("File truncated, resetting...")
tail.SetOffset(0)
tail.WriteOffset()
tail.seek()
}
return
} else if err != nil {
logs.Debug("read error: ", err)
return
} else {
tail.Parser.Consume(buffer[0:len], &tail.offset)
tail.WriteOffset()
}
}
}
func NewTail(parser Parser, maxBackfill int64, path string, offsetPath string, offset int64) *Tail {
tail := new(Tail)
tail.Path = path
tail.offset = offset
tail.OffsetPath = offsetPath
tail.Parser = parser
tracker := watch.NewInotifyTracker()
w, err := tracker.NewWatcher()
if err != nil {
return nil
}
tail.Watcher = watch.NewInotifyFileWatcher(path, w)
tail.LoadOffset()
tail.maxBackfill = maxBackfill
handle, err := os.Open(path)
if err != nil {
logs.Debug("Can't open file: ", path)
return nil
} else {
tail.handle = handle
}
tail.seek()
return tail
}
func (tail *Tail) WriteOffset() {
path := path.Join(os.TempDir(), path.Base(tail.OffsetPath))
temp, err := os.Create(path)
if err != nil {
logs.Debug("Can't create tempfile:", err)
} else {
_, err := fmt.Fprintf(temp, "%d\n", tail.Offset())
if err != nil {
logs.Debug("Can't write to tempfile:", err)
temp.Close()
} else {
temp.Close()
err := os.Rename(path, tail.OffsetPath)
if err != nil {
logs.Debug("Rename failed:", err)
}
}
}
}
func NewRegexpParser(hostname string, group string, file string, output chan Record, pattern string, fields []FieldConfig, maxLineSize int64) Parser {
parser := new(RegexpParser)
parser.maxLineSize = int(maxLineSize)
parser.hostname = hostname
parser.file = file
parser.group = group
parser.output = output
parser.buffer = make([]byte, 0)
re, err := regexp.Compile(pattern)
if err != nil {
panic(err)
} else {
parser.compiled = re
for i, name := range re.SubexpNames() {
if name != "" {
found := false
for n, spec := range fields {
if spec.Name == "" {
spec.Name = spec.Alias
}
if name == spec.Name {
found = true
fields[n].Group = i
logs.Debug("setting group alias: %s, name: %s, group: %d", spec.Alias, spec.Name, spec.Group)
}
}
if !found {
var spec FieldConfig
spec.Group = i
spec.Alias = name
spec.Type = String
fields = append(fields, spec)
}
}
}
}
parser.fields = fields
for _, f := range parser.fields {
logs.Debug("p.f: alias: %s, name: %s, group: %d, type: %d", f.Alias, f.Name, f.Group, f.Type)
}
return parser
}
func (tail *Tail) LoadOffset() {
file, err := os.Open(tail.OffsetPath)
if err != nil {
tail.WriteOffset()
} else {
reader := bufio.NewReader(file)
str, err := reader.ReadString('\n')
if err != nil {
logs.Debug("Malformed offset file: ", err)
} else {
out, err := strconv.ParseInt(strings.TrimSpace(str), 10, 64)
if err != nil {
logs.Debug("Malformed offset file: ", err)
} else {
logs.Debug("Found offset: %d", out)
tail.SetOffset(out)
}
}
file.Close()
}
}
func (parser *RegexpParser) Consume(bytes []byte, counter *int64) {
parser.buffer = append(parser.buffer, bytes...)
logs.Debug("consuming %d bytes of %s", len(bytes), parser.file)
l := len(parser.buffer)
if l > parser.maxLineSize {
off := l - parser.maxLineSize
logs.Debug("chopping %d bytes off buffer (was: %d, max: %d)", off, l, parser.maxLineSize)
atomic.AddInt64(counter, int64(off))
parser.buffer = parser.buffer[off:]
}
for {
m := parser.compiled.FindSubmatchIndex(parser.buffer)
if m == nil {
return
}
hasher := sha1.New()
out := make(map[string]Column)
out["_offset"] = Column{Integer, Simple, atomic.LoadInt64(counter)}
out["_file"] = Column{String, Simple, parser.file}
out["_time"] = Column{Timestamp, Simple, StandardTimeProvider.Now().Unix()}
out["_group"] = Column{String, Simple, parser.group}
out["_hostname"] = Column{String, Simple, parser.hostname}
for _, spec := range parser.fields {
g := spec.Group
if g < 0 || g > len(m)/2 {
logs.Error("spec group out of range: alias: %s, name: %s, g: %d", spec.Alias, spec.Name, g)
panic(-1)
}
if m[g*2] == -1 {
continue
}
s := string(parser.buffer[m[g*2]:m[g*2+1]])
switch spec.Type {
case Timestamp:
t, err := time.Parse(spec.Format, s)
if err != nil {
logs.Warn("date parse error: %s", err)
} else {
if t.Year() == 0 {
now := StandardTimeProvider.Now()
adjusted := time.Date(now.Year(), t.Month(), t.Day(), t.Hour(), t.Minute(), t.Second(), t.Nanosecond(), t.Location())
if adjusted.After(now) {
adjusted = time.Date(now.Year()-1, t.Month(), t.Day(), t.Hour(), t.Minute(), t.Second(), t.Nanosecond(), t.Location())
}
t = adjusted
}
out[spec.Alias] = Column{Timestamp, spec.Treatment, t.Unix()}
}
case String:
if spec.Treatment == Tokens {
out[spec.Alias] = Column{Tokens, spec.Treatment, spec.Pattern.FindAllString(s, -1)}
} else if spec.Treatment == Hash {
hasher.Reset()
hasher.Write([]byte(spec.Salt))
hasher.Write([]byte(s))
sha := base64.URLEncoding.EncodeToString(hasher.Sum(nil))
out[spec.Alias] = Column{Tokens, spec.Treatment, sha}
} else {
out[spec.Alias] = Column{String, spec.Treatment, s}
}
case Integer:
n, err := strconv.ParseInt(s, 10, 64)
if err == nil {
out[spec.Alias] = Column{spec.Type, spec.Treatment, n}
}
case Double:
n, err := strconv.ParseFloat(s, 64)
if err == nil {
out[spec.Alias] = Column{spec.Type, spec.Treatment, n}
}
default:
panic(nil)
}
}
parser.output <- out
atomic.AddInt64(counter, int64(m[1]))
parser.buffer = parser.buffer[m[1]:]
}
logs.Debug("done with %s", parser.file)
}
func configFromMapping(mapping map[string]interface{}, hostname string) (*Config, error) {
b, _ := json.Marshal(mapping)
logs.Debug("mapping: %s", string(b))
var err error = nil
config := new(Config)
config.Sources = make([]SourceConfig, 0)
config.Destinations = make([]DestinationConfig, 0)
global, err := getMap(mapping, "global")
if err != nil {
return nil, fmt.Errorf("no global section in the config file")
}
config.OffsetDir, err = getString(global, "offset_dir")
if err != nil {
logs.Warn("no offset_dir specified")
config.MaxBackfillBytes = -1
}
config.MaxBackfillBytes, err = getInt64(global, "max_backfill_bytes")
if err != nil {
logs.Warn("no max_backfill_bytes, continuing with unlimited")
config.MaxBackfillBytes = -1
}
config.MaxLineSizeBytes, err = getInt64(global, "max_linesize_bytes")
if err != nil {
logs.Warn("no max_linesize_bytes, continuing with 32768")
config.MaxLineSizeBytes = 32768
}
sources, err := getMap(mapping, "sources")
if err != nil {
return nil, fmt.Errorf("no sources section in the config file")
}
for name, _ := range sources {
src, err := getMap(sources, name)
if err != nil {
logs.Warn("Invalid source: %s, continuing...", name)
continue
}
var source SourceConfig
source.Hostname = hostname
source.Fields = make([]FieldConfig, 0)
source.OffsetDir = config.OffsetDir
source.MaxBackfillBytes = config.MaxBackfillBytes
source.MaxLineSizeBytes = config.MaxLineSizeBytes
source.Name = name
source.Glob, err = getString(src, "glob")
if err != nil {
return nil, err
}
source.Pattern, err = getString(src, "pattern")
if err != nil {
source.Pattern = DefaultPattern
}
_, err = regexp.Compile(source.Pattern)
if err != nil {
logs.Warn("%s is not a valid regexp, continuing... (%s)", source.Pattern, err)
continue
}
fields, err := getMap(src, "fields")
for name, _ := range fields {
fld, err := getMap(fields, name)
if err != nil {
logs.Warn("%s is not a map, continuing... (%s)", name, err)
continue
}
var field FieldConfig
field.Alias = name
field.Name, err = getString(fld, "name")
if err != nil {
field.Name = field.Alias
}
field.Group, err = getInt(fld, "group")
s, err := getString(fld, "type")
if err != nil {
field.Type = String
} else {
field.Type, err = parseFieldType(s)
if err != nil {
logs.Warn("Invalid field type: %s, continuing... (error was %s)", s, err)
continue
}
}
logs.Info("found type %s", field.Type)
s, err = getString(fld, "treatment")
if err != nil {
field.Treatment = String
} else {
field.Treatment, err = parseFieldTreatment(s)
if err != nil {
logs.Warn("Invalid field treatment: %s, continuing... (error was %s)", s, err)
continue
}
}
logs.Info("found treatment %s", field.Treatment)
field.Salt, err = getString(fld, "salt")
field.Format, err = getString(fld, "format")
s, err = getString(fld, "pattern")
field.Pattern, err = regexp.Compile(s)
if err != nil {
logs.Warn("Invalid regex: %s, continuing... (error was %s)", s, err)
continue
}
source.Fields = append(source.Fields, field)
}
config.Sources = append(config.Sources, source)
}
destinations, err := getMap(mapping, "destinations")
if err != nil {
return nil, fmt.Errorf("no destinations section in the config file")
}
for name, _ := range destinations {
var dest DestinationConfig
urlString, err := getString(destinations, name)
u, err := url.Parse(urlString)
if err != nil {
logs.Warn("Invalid URL: %s, continuing... (error was %s)", urlString, err)
continue
}
logs.Info("Found destination: %s", urlString)
dest.Name = name
dest.Url = u
config.Destinations = append(config.Destinations, dest)
}
return config, nil
}
func main() {
flag.Parse()
runtime.GOMAXPROCS(*cpus)
// set the logger path
handle, err := os.OpenFile(*logFile, os.O_APPEND|os.O_CREATE|os.O_WRONLY, 0666)
if err != nil {
logs.Warn("Unable to open log file %s, using stderr: %s", *logFile, err)
} else {
logs.Logger = log.New(handle, "", log.LstdFlags|log.Lshortfile)
}
// Check whether we're in debug mode
if *debug {
logs.SetLevel(logs.DEBUG)
logs.Debug("logging at DEBUG")
} else {
logs.SetLevel(logs.INFO)
}
if *name == "unknown" {
*name, err = os.Hostname()
if err != nil {
logs.Warn("Unable to determine hostname: %s", err)
}
}
// Read the config files
config, err := dendrite.NewConfig(*configFile, *name)
if err != nil {
logs.Fatal("Can't read configuration: %s", err)
}
// Link up all of the objects
ch := make(chan dendrite.Record, 100)
logs.Debug("original %s", ch)
dests := config.CreateDestinations()
groups := config.CreateAllTailGroups(ch)
// If any of our destinations talk back, log it.
go func() {
reader := bufio.NewReader(dests.Reader())
for {
str, err := reader.ReadString('\n')
if err == io.EOF {
logs.Debug("eof")
time.Sleep(1 * time.Second)
} else if err != nil {
logs.Error("error reading: %s", err)
} else {
logs.Info("received: %s", str)
}
}
}()
// Do the event loop
finished := make(chan bool, 0)
go dests.Consume(ch, finished)
if *quitAfter >= 0 {
start := time.Now()
logs.Debug("starting the poll")
i := 0
for {
i++
if i%10 == 0 {
groups.Refresh()
}
groups.Poll()
if time.Now().Sub(start) >= time.Duration((*quitAfter)*float64(time.Second)) {
break
}
}
} else {
logs.Debug("starting the loop")
groups.Loop()
}
logs.Info("Closing...")
close(ch)
<-finished
logs.Info("Goodbye!")
}
