godd

A Go implementation of the dd command-line utility for file copying and conversion.

Usage

godd [OPERAND]...

Operands

Operand	Description
if=FILE	Read from FILE, URL, or - (default: stdin)
of=FILE	Write to FILE (default: stdout)
bs=BYTES	Read and write up to BYTES bytes at a time
ibs=BYTES	Read up to BYTES bytes at a time (default: 512)
obs=BYTES	Write BYTES bytes at a time (default: 512)
cbs=BYTES	Convert BYTES bytes at a time
count=N	Copy only N input blocks
skip=N	Skip N ibs-sized input blocks
seek=N	Skip N obs-sized output blocks
conv=CONVS	Comma-separated list of conversions
iflag=FLAGS	Comma-separated list of input flags
oflag=FLAGS	Comma-separated list of output flags
status=LEVEL	Status output level

Conversions

Symbol	Description
lcase	Convert uppercase to lowercase
ucase	Convert lowercase to uppercase
ascii	Convert from EBCDIC to ASCII
ebcdic	Convert from ASCII to EBCDIC
ibm	Convert from ASCII to alternate EBCDIC
block	Pad newline-terminated records to cbs size
unblock	Replace trailing spaces with newlines
swab	Swap every pair of bytes
sparse	Seek rather than write all-NUL output blocks
sync	Pad incomplete input blocks to ibs size
excl	Fail if output file exists
nocreat	Do not create output file
notrunc	Do not truncate output file
noerror	Continue after read errors
fdatasync	Sync data before finishing
fsync	Sync data and metadata before finishing

Input Flags (iflag)

Symbol	Description
fullblock	Accumulate full ibs blocks before writing
nonblock	Use non-blocking I/O
noatime	Do not update access time
nofollow	Do not follow symlinks
direct	Use direct I/O (Linux)
directory	Fail unless file is a directory
dsync	Synchronized I/O for data
sync	Synchronized I/O for data and metadata
nocache	Drop file system cache after read (Linux)
noctty	Do not assign controlling terminal

Output Flags (oflag)

Symbol	Description
append	Append mode
direct	Use direct I/O (Linux)
directory	Fail unless file is a directory
dsync	Synchronized I/O for data
sync	Synchronized I/O for data and metadata
nonblock	Use non-blocking I/O
noatime	Do not update access time
nocache	Drop file system cache after write (Linux)
noctty	Do not assign controlling terminal
nofollow	Do not follow symlinks

Numeric Suffixes

Suffixes can be appended to numeric operands (bs, ibs, obs, cbs, count, skip, seek):

Suffix	Multiplier	Example
c	1	bs=1c = 1
w	2	bs=2w = 4
b	512	bs=4b = 2048
B	1	bs=1B = 1
kB	1000	bs=1kB = 1000
K	1024	bs=1K = 1024
MB	1000^2	bs=1MB = 1000000
M	1024^2	bs=1M = 1048576
GB	1000^3	bs=1GB = 1000000000
G	1024^3	bs=1G = 1073741824
TB	1000^4	bs=1TB = 1000000000000
T	1024^4	bs=1T = 1099511627776
PB	1000^5
P	1024^5
EB	1000^6
E	1024^6

Status Levels

Level	Description
(default)	Print final transfer statistics
none	Suppress all non-error output
noxfer	Suppress final transfer statistics
progress	Show periodic progress statistics
bar	Colorful progress bar with braille chars, percentage, throughput, ETA

Examples

# Copy a file
godd if=input.txt of=output.txt

# Convert to uppercase
godd if=input.txt of=output.txt conv=ucase

# Copy with progress
godd if=/dev/zero of=/tmp/out bs=1M count=100 status=progress

# Create a 1GB sparse file
godd if=/dev/zero of=sparse.img bs=1M count=0 seek=1024

# Read from stdin, write to stdout
cat input.txt | godd conv=lcase

# Use EBCDIC conversion
godd if=ebcdic.dat of=ascii.txt conv=ascii

# Stream an HTTP URL directly
godd if=https://example.com/image.iso of=image.iso bs=1M status=bar

if= accepts http:// and https:// URLs. The response body is streamed directly. Content-Length is used for progress bar when available. Combine with status=bar for live download progress.

## Benchmarks

`make bench` runs godd against system `dd` across several workloads.
Results on Linux 6.x, AMD Ryzen 7, NVMe SSD:

| Workload | `dd` | `godd` | Ratio |
|----------|------|--------|-------|
| 1M blocks, 1G total | 839 MB/s | 933 MB/s | **1.11x** |
| 64K blocks, 1G total | 1.6 GB/s | 1.5 GB/s | 0.94x |
| 512b blocks, 64M total | 346 MB/s | 275 MB/s | 0.79x |
| conv=lcase, 64M | 492 MB/s | 390 MB/s | 0.79x |

godd is competitive with GNU dd for bulk I/O and often faster with
larger block sizes. Small-block and conversion-heavy workloads favor
dd's C implementation, as expected.