▾ Example (Dictionary)

A preset dictionary can be used to improve the compression ratio. The downside to using a dictionary is that the compressor and decompressor must agree in advance what dictionary to use.

Code:

// The dictionary is a string of bytes. When compressing some input data,
// the compressor will attempt to substitute substrings with matches found
// in the dictionary. As such, the dictionary should only contain substrings
// that are expected to be found in the actual data stream.
const dict = `<?xml version="1.0"?>` + `<book>` + `<data>` + `<meta name="` + `" content="`

// The data to compress should (but is not required to) contain frequent
// substrings that match those in the dictionary.
const data = `<?xml version="1.0"?>
<book>
<meta name="title" content="The Go Programming Language"/>
<meta name="authors" content="Alan Donovan and Brian Kernighan"/>
<meta name="published" content="2015-10-26"/>
<meta name="isbn" content="978-0134190440"/>
<data>...</data>
</book>
`

var b bytes.Buffer

// Compress the data using the specially crafted dictionary.
zw, err := flate.NewWriterDict(&b, flate.DefaultCompression, []byte(dict))
if err != nil {
    log.Fatal(err)
}
if _, err := io.Copy(zw, strings.NewReader(data)); err != nil {
    log.Fatal(err)
}
if err := zw.Close(); err != nil {
    log.Fatal(err)
}

// The decompressor must use the same dictionary as the compressor.
// Otherwise, the input may appear as corrupted.
fmt.Println("Decompressed output using the dictionary:")
zr := flate.NewReaderDict(bytes.NewReader(b.Bytes()), []byte(dict))
if _, err := io.Copy(os.Stdout, zr); err != nil {
    log.Fatal(err)
}
if err := zr.Close(); err != nil {
    log.Fatal(err)
}

fmt.Println()

// Substitute all of the bytes in the dictionary with a '#' to visually
// demonstrate the approximate effectiveness of using a preset dictionary.
fmt.Println("Substrings matched by the dictionary are marked with #:")
hashDict := []byte(dict)
for i := range hashDict {
    hashDict[i] = '#'
}
zr = flate.NewReaderDict(&b, hashDict)
if _, err := io.Copy(os.Stdout, zr); err != nil {
    log.Fatal(err)
}
if err := zr.Close(); err != nil {
    log.Fatal(err)
}

Output:

Decompressed output using the dictionary:
<?xml version="1.0"?>
<book>
	<meta name="title" content="The Go Programming Language"/>
	<meta name="authors" content="Alan Donovan and Brian Kernighan"/>
	<meta name="published" content="2015-10-26"/>
	<meta name="isbn" content="978-0134190440"/>
	<data>...</data>
</book>

Substrings matched by the dictionary are marked with #:
#####################
######
	############title###########The Go Programming Language"/#
	############authors###########Alan Donovan and Brian Kernighan"/#
	############published###########2015-10-26"/#
	############isbn###########978-0134190440"/#
	######...</#####
</#####

▾ Example (Reset)

In performance critical applications, Reset can be used to discard the current compressor or decompressor state and reinitialize them quickly by taking advantage of previously allocated memory.

Code:

proverbs := []string{
    "Don't communicate by sharing memory, share memory by communicating.\n",
    "Concurrency is not parallelism.\n",
    "The bigger the interface, the weaker the abstraction.\n",
    "Documentation is for users.\n",
}

var r strings.Reader
var b bytes.Buffer
buf := make([]byte, 32<<10)

zw, err := flate.NewWriter(nil, flate.DefaultCompression)
if err != nil {
    log.Fatal(err)
}
zr := flate.NewReader(nil)

for _, s := range proverbs {
    r.Reset(s)
    b.Reset()

    // Reset the compressor and encode from some input stream.
    zw.Reset(&b)
    if _, err := io.CopyBuffer(zw, &r, buf); err != nil {
        log.Fatal(err)
    }
    if err := zw.Close(); err != nil {
        log.Fatal(err)
    }

    // Reset the decompressor and decode to some output stream.
    if err := zr.(flate.Resetter).Reset(&b, nil); err != nil {
        log.Fatal(err)
    }
    if _, err := io.CopyBuffer(os.Stdout, zr, buf); err != nil {
        log.Fatal(err)
    }
    if err := zr.Close(); err != nil {
        log.Fatal(err)
    }
}

Output:

Don't communicate by sharing memory, share memory by communicating.
Concurrency is not parallelism.
The bigger the interface, the weaker the abstraction.
Documentation is for users.

▾ Example (Synchronization)

DEFLATE is suitable for transmitting compressed data across the network.

Code:

var wg sync.WaitGroup
defer wg.Wait()

// Use io.Pipe to simulate a network connection.
// A real network application should take care to properly close the
// underlying connection.
rp, wp := io.Pipe()

// Start a goroutine to act as the transmitter.
wg.Add(1)
go func() {
    defer wg.Done()

    zw, err := flate.NewWriter(wp, flate.BestSpeed)
    if err != nil {
        log.Fatal(err)
    }

    b := make([]byte, 256)
    for _, m := range strings.Fields("A long time ago in a galaxy far, far away...") {
        // We use a simple framing format where the first byte is the
        // message length, followed the message itself.
        b[0] = uint8(copy(b[1:], m))

        if _, err := zw.Write(b[:1+len(m)]); err != nil {
            log.Fatal(err)
        }

        // Flush ensures that the receiver can read all data sent so far.
        if err := zw.Flush(); err != nil {
            log.Fatal(err)
        }
    }

    if err := zw.Close(); err != nil {
        log.Fatal(err)
    }
}()

// Start a goroutine to act as the receiver.
wg.Add(1)
go func() {
    defer wg.Done()

    zr := flate.NewReader(rp)

    b := make([]byte, 256)
    for {
        // Read the message length.
        // This is guaranteed to return for every corresponding
        // Flush and Close on the transmitter side.
        if _, err := io.ReadFull(zr, b[:1]); err != nil {
            if err == io.EOF {
                break // The transmitter closed the stream
            }
            log.Fatal(err)
        }

        // Read the message content.
        n := int(b[0])
        if _, err := io.ReadFull(zr, b[:n]); err != nil {
            log.Fatal(err)
        }

        fmt.Printf("Received %d bytes: %s\n", n, b[:n])
    }
    fmt.Println()

    if err := zr.Close(); err != nil {
        log.Fatal(err)
    }
}()

Output:

Received 1 bytes: A
Received 4 bytes: long
Received 4 bytes: time
Received 3 bytes: ago
Received 2 bytes: in
Received 1 bytes: a
Received 6 bytes: galaxy
Received 4 bytes: far,
Received 3 bytes: far
Received 7 bytes: away...

Internal call graph ▾

In the call graph viewer below, each node is a function belonging to this package and its children are the functions it calls—perhaps dynamically.

The root nodes are the entry points of the package: functions that may be called from outside the package. There may be non-exported or anonymous functions among them if they are called dynamically from another package.

Click a node to visit that function's source code. From there you can visit its callers by clicking its declaring func token.

Functions may be omitted if they were determined to be unreachable in the particular programs or tests that were analyzed.