rosbag

package module
v0.0.0-...-5a3e0be Latest Latest
Warning

This package is not in the latest version of its module.

 Go to latest Published: Aug 26, 2021 License: Apache-2.0 Imports: 15 Imported by: 0

README

go-rosbag

PkgGoDev CI

Go Rosbag parser. Designed to provide ease of use, speed, and streamability.

Usage

View Messages as map[string]interface{}
package main

import (
	"fmt"
	"io"
	"os"

	"github.com/rovechkin1/go-rosbag"
)

func main() {
	f, _ := os.Open("example.bag")
	defer f.Close()

	decoder := rosbag.NewDecoder(f)
	for {
		record, err := decoder.Read()
		if err == io.EOF {
			break
		}

		switch record := record.(type) {
		case *rosbag.RecordMessageData:
			data := make(map[string]interface{})
			// After ViewAs gets called, the variable "data" will be used to map
			// the underlying buffer to Go data structures, and they'll SHARE the
			// same underlying buffer.
			_ = record.ViewAs(data)
			fmt.Println(data)
		}
    
    		// Mark the current record is no longer used, the underlying buffer can be reused.
		record.Close()
	}
}
View Messages as structs
package main

import (
	"fmt"
	"io"
	"os"

	"github.com/rovechkin1/go-rosbag"
)

type Pixel struct {
	R uint8	`rosbag:"r"`
	G uint8	`rosbag:"g"`
	B uint8	`rosbag:"b"`
}

func main() {
	f, _ := os.Open("example.bag")
	defer f.Close()

	decoder := rosbag.NewDecoder(f)
	for {
		record, err := decoder.Read()
		if err == io.EOF {
			break
		}

		switch record := record.(type) {
		case *rosbag.RecordMessageData:
			if record.ConnectionHeader().Topic == "/pixel" {
				var pixel Pixel
				_ = record.ViewAs(&pixel)
				fmt.Println(pixel)
			}
		}
    
    		// Mark the current record is no longer used, the underlying buffer can be reused.
		record.Close()
	}
}

Data Type Mapping

Primitive Types
ROS Type Go Type
bool bool
int8 int8
byte int8
uint8 uint8
char uint8
int16 int16
uint16 uint16
int32 uint32
int64 uint64
float32 float32
float64 float64
string string
time time.Time
duration time.Duration
Array Handling

Both fixed-length and variable-length arrays are mapped to Go slices. For example, uint8[] with a length of 3 and uint8[3] will be mapped to []uint8 in Go.

Benchmark

Hardware specs:

  • Model: MacBook Pro (15-inch, 2017)
  • Processor: 2.8 GHz Quad-Core Intel Core i7
  • Memory: 16 GB 2133 MHz LPDDR3
  • Storage: 256GB PCIe-based onboard SSD

The following benchmark result (source) was taken by parsing a 696 MB rosbag file from webviz. The bag contains frames and metadata from a car.

Time taken: 147 ms
Throughput: 4.59 GB/s
Memory usage: 33.60 MB

For a reference, time cat <bag> > /dev/null takes 118 ms. This means, go-rosbag only adds 29 ms to the total runtime!

Examples

Please see the examples directory within this repository.

Code Coverage

Currently, the unit tests only cover the happy path, and it's not extensive as I want it to be. I'm still working on adding coverage during my free time.

Alternatives

There are two known alternatives to go-rosbag (at least that I know of):

Ros Message Format

link

Contributing

Any contribution is welcomed! I will review pull requests as soon as possible.

License

go-rosbag is licensed under Apache license version 2.0. See LICENSE file.

Documentation

Overview

rosbag implements Rosbag Format Version 2.0, http://wiki.ros.org/Bags/Format/2.0. Currently, this package only implements the decoder.

Index

Constants

This section is empty.

Variables

This section is empty.

Functions

This section is empty.

Types

type Compression 

type Compression string
const (
	CompressionNone Compression = "none"
	CompressionBZ2  Compression = "bz2"
	CompressionLZ4  Compression = "lz4"
)

type ConnectionHeader 

type ConnectionHeader struct {
	Topic             string
	Type              string
	MD5Sum            string
	MessageDefinition MessageDefinition
}

type Decoder 

type Decoder struct {
	// contains filtered or unexported fields
}

func NewDecoder 

func NewDecoder(r io.Reader) *Decoder

func (*Decoder) Read 

func (decoder *Decoder) Read() (Record, error)

Read returns the next record in the rosbag. Next might will return nil record and error at the beginning to mark that the rosbag format version is supported. When, it reaches EOF, Next returns io.EOF error.

type MessageDefinition 

type MessageDefinition struct {
	Type   string
	Fields []*MessageFieldDefinition
}

MessageDefinition is defined here, http://wiki.ros.org/msg

type MessageFieldDefinition 

type MessageFieldDefinition struct {
	Type    MessageFieldType
	Name    string
	IsArray bool
	// ArraySize is only used when the field is a fixed-size array. If it's a slice, ArraySize is -1
	ArraySize int
	// Value is an optional field. It's only being used for constants
	Value interface{}
	// MsgType is only being used when type is complex. This defines the custom
	// message type.
	MsgType *MessageDefinition
}

type MessageFieldType 

type MessageFieldType uint8
const (
	MessageFieldTypeBool MessageFieldType = iota + 1
	MessageFieldTypeInt8
	MessageFieldTypeUint8
	MessageFieldTypeInt16
	MessageFieldTypeUint16
	MessageFieldTypeInt32
	MessageFieldTypeUint32
	MessageFieldTypeInt64
	MessageFieldTypeUint64
	MessageFieldTypeFloat32
	MessageFieldTypeFloat64
	MessageFieldTypeString
	MessageFieldTypeTime
	MessageFieldTypeDuration
	MessageFieldTypeComplex
)

type Op 

type Op uint8
const (
	// OpInvalid is an extension from the standard. This Op marks an invalid Op.
	OpInvalid     Op = 0x00
	OpBagHeader   Op = 0x03
	OpChunk       Op = 0x05
	OpConnection  Op = 0x07
	OpMessageData Op = 0x02
	OpIndexData   Op = 0x04
	OpChunkInfo   Op = 0x06
)

type Record 

type Record interface {
	// Op parses the header, and lookup for the op field
	Op() (Op, error)
	// Header returns raw header byte slice. It contains metadata for the record
	Header() []byte
	// Data returns raw data byte slice. It contains record specific data
	Data() []byte
	// Close closes the underlying memory allocation for this record so that
	// the allocated memory can be reused
	Close()
	// prints contents
	String() (string, error)
}

type RecordBagHeader 

type RecordBagHeader struct {
	*RecordBase
}

RecordBagHeader occurs once in the file as the first record.

func (*RecordBagHeader) ChunkCount 

func (record *RecordBagHeader) ChunkCount() (uint32, error)

ChunkCount parses Header to get a number of chunk records in the file

func (*RecordBagHeader) ConnCount 

func (record *RecordBagHeader) ConnCount() (uint32, error)

ConnCount parses Header to get a number of unique connections in the file

func (*RecordBagHeader) IndexPos 

func (record *RecordBagHeader) IndexPos() (uint64, error)

IndexPos parses Header to get an offset of first record after the chunk section

func (*RecordBagHeader) String 

func (record *RecordBagHeader) String() (string, error)

type RecordBase 

type RecordBase struct {
	// Raw contains: <header_len><header><data_len><data>
	Raw []byte
	// HeaderLen contains length in bytes of the header
	HeaderLen uint32
	// DataLen contains length in bytes of the data
	DataLen uint32
	// contains filtered or unexported fields
}

func (*RecordBase) Close 

func (record *RecordBase) Close()

func (*RecordBase) Data 

func (record *RecordBase) Data() []byte

func (*RecordBase) Header 

func (record *RecordBase) Header() []byte

func (*RecordBase) Op 

func (record *RecordBase) Op() (Op, error)

type RecordChunk 

type RecordChunk struct {
	*RecordBase
}

RecordChunk is a record that contains one or more RecordConnection and/or RecordMessageData.

func (*RecordChunk) Compression 

func (record *RecordChunk) Compression() (Compression, error)

Compression parses Header to get the compression algorithm that's used for the underlying chunk data. The supported compression values are "none", "lz4", and "bz2".

func (*RecordChunk) Size 

func (record *RecordChunk) Size() (uint32, error)

Size parses Header to get the size in bytes of the uncompressed chunk

func (*RecordChunk) String 

func (record *RecordChunk) String() (string, error)

type RecordChunkInfo 

type RecordChunkInfo struct {
	*RecordBase
}

RecordChunkInfo contains metadata about Chunks

func (*RecordChunkInfo) ChunkPos 

func (record *RecordChunkInfo) ChunkPos() (uint64, error)

ChunkPos parses Header to get the offset of the chunk record

func (*RecordChunkInfo) Count 

func (record *RecordChunkInfo) Count() (uint32, error)

Count parses Header to get the number of connections in the chunk

func (*RecordChunkInfo) EndTime 

func (record *RecordChunkInfo) EndTime() (time.Time, error)

EndTime parses Header to get the timestamp of latest message in the chunk

func (*RecordChunkInfo) StartTime 

func (record *RecordChunkInfo) StartTime() (time.Time, error)

StartTime parses Header to get the timestamp of earliest message in the chunk

func (*RecordChunkInfo) String 

func (record *RecordChunkInfo) String() (string, error)

func (*RecordChunkInfo) Ver 

func (record *RecordChunkInfo) Ver() (uint32, error)

Ver parses Header to get the version of this ChunkInfo record

type RecordConnection 

type RecordConnection struct {
	*RecordBase
}

RecordConnection is a record that contains metadata about message data. Some of the metadata are used to encode/decode message data.

func (*RecordConnection) Conn 

func (record *RecordConnection) Conn() (uint32, error)

Conn parses Header to get the unique connection ID within a bag

func (*RecordConnection) ConnectionHeader 

func (record *RecordConnection) ConnectionHeader() (*ConnectionHeader, error)

ConnectionHeader reads the underlying data and decode it to ConnectionHeader

func (*RecordConnection) String 

func (record *RecordConnection) String() (string, error)

func (*RecordConnection) Topic 

func (record *RecordConnection) Topic() (string, error)

Topic parses Header to get the topic

type RecordIndexData 

type RecordIndexData struct {
	*RecordBase
}

func (*RecordIndexData) Conn 

func (record *RecordIndexData) Conn() (uint32, error)

Conn parses Header to get the unique connection ID within a bag

func (*RecordIndexData) Count 

func (record *RecordIndexData) Count() (uint32, error)

Count parses Header to get the number of messages on conn in the preceding chunk

func (*RecordIndexData) String 

func (record *RecordIndexData) String() (string, error)

func (*RecordIndexData) Ver 

func (record *RecordIndexData) Ver() (uint32, error)

Ver parses Header to get the version of this Index record

type RecordMessageData 

type RecordMessageData struct {
	*RecordBase
	// contains filtered or unexported fields
}

RecordMessageData contains the serialized message data in the ROS serialization format.

func (*RecordMessageData) Conn 

func (record *RecordMessageData) Conn() (uint32, error)

Conn parses Header to get the unique connection ID within a bag

func (*RecordMessageData) ConnectionHeader 

func (record *RecordMessageData) ConnectionHeader() *ConnectionHeader

ConnectionHeader returns the parsed ROS connection header.

func (*RecordMessageData) String 

func (record *RecordMessageData) String() (string, error)

func (*RecordMessageData) Time 

func (record *RecordMessageData) Time() (time.Time, error)

Time parses Header to get the timestamp when this message was recorded. Note that this timestamp is different from the message header timestamp from ROS. The main difference is that this timestamp is the retrieved time NOT sent time.

func (*RecordMessageData) ViewAs 

func (record *RecordMessageData) ViewAs(v interface{}) error

ViewAs views the underlying raw data in the given v format. When possible, View will convert raw data without making a copy. With no copy, decoding large arrays become really fast! But, this also means that any data types that are reference based can't be used after this Record is closed.

So, if the data is absolutely needed after reading this record, you MUST NOT CLOSE this record so that the underlying raw data is not overwritten by other records.

type Rosbag 

type Rosbag struct {
	Version Version
	Record  []Record
}

type Version 

type Version struct {
	Major uint
	Minor uint
}

func (*Version) String 

func (version *Version) String() string

Source Files

View all Source files

Directories

Path Synopsis
examples
logging

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL
go.dev uses cookies from Google to deliver and enhance the quality of its services and to analyze traffic. Learn more.