awk関数、printfによる書式設定、レポート生成

awkで関数とprintfが重要な理由

awkでユーザー関数を定義する

function name(param1, param2,    local1, local2) {
    # body
    return value
}

#!/usr/bin/env bash
# Demonstrate a simple awk user function
echo '10 3
7 0
100 4' | awk '
function divide(a, b,    result) {
    if (b == 0) return "ERR"
    result = a / b
    return result
}
{
    print $1 "/" $2 " = " divide($1, $2)
}
'

文字列処理を行う関数

#!/usr/bin/env bash
# trim() strips leading/trailing spaces; used to normalise CSV-like data
printf '  alice , 90 \n bob , 85 \n  carol , 92 \n' | awk -F',' '
function trim(s) {
    gsub(/^[ \t]+|[ \t]+$/, "", s)
    return s
}
{
    name  = trim($1)
    score = trim($2)
    print name, score
}
'

awkのprintf入門

#!/usr/bin/env bash
# printf format specifiers in awk
echo '' | awk '
BEGIN {
    printf "%s\n",        "plain string"
    printf "%10s\n",     "right-pad"
    printf "%-10s|\n",   "left-pad"
    printf "%8.2f\n",    3.14159
    printf "%08d\n",     42
    printf "%e\n",       123456.789
}
'

printfでレポートのヘッダーを作成する

#!/usr/bin/env bash
# Print a report header in BEGIN, data rows per-record, total in END
printf 'alice 9200 12\nbob 8750 10\ncarol 10400 14\n' | awk '
BEGIN {
    printf "%-10s %10s %6s %12s\n", "Name", "Salary", "Months", "Annual"
    printf "%s\n", "----------------------------------------------"
    total = 0
}
{
    annual = $2 * $3
    total += annual
    printf "%-10s %10d %6d %12d\n", $1, $2, $3, annual
}
END {
    printf "%s\n", "----------------------------------------------"
    printf "%-10s %10s %6s %12d\n", "TOTAL", "", "", total
}
'

書式設定ロジックにユーザー関数を使う

#!/usr/bin/env bash
# human_bytes() converts raw byte counts to KB/MB/GB strings
printf 'var.log 1024\naccess.log 2097152\ncore.dump 1073741824\ntiny.txt 512\n' | awk '
function human_bytes(n,    s) {
    if (n >= 1073741824) { s = sprintf("%.1f GB", n/1073741824) }
    else if (n >= 1048576) { s = sprintf("%.1f MB", n/1048576) }
    else if (n >= 1024)    { s = sprintf("%.1f KB", n/1024) }
    else                   { s = sprintf("%d B",    n) }
    return s
}
BEGIN { printf "%-20s %10s\n", "File", "Size"; print "------------------------------" }
{ printf "%-20s %10s\n", $1, human_bytes($2) }
'

印刷前に配列へデータを蓄積する

#!/usr/bin/env bash
# Summarise an nginx-style access log by HTTP status code
printf '127.0.0.1 200 1234\n127.0.0.1 404 512\n10.0.0.2 200 8900\n10.0.0.3 500 256\n10.0.0.4 200 4096\n10.0.0.5 404 300\n' | awk '
{
    status  = $2
    bytes   = $3
    count[status]++
    total_bytes[status] += bytes
}
END {
    printf "%-8s %8s %14s\n", "Status", "Requests", "Total Bytes"
    print  "-----------------------------------"
    for (s in count)
        printf "%-8s %8d %14d\n", s, count[s], total_bytes[s]
}
'

awkの再帰関数

#!/usr/bin/env bash
# Recursive repeat() to build separator lines dynamically
awk '
function repeat(s, n,    acc) {
    if (n <= 0) return ""
    acc = s repeat(s, n-1)
    return acc
}
BEGIN {
    header = sprintf("%-15s %10s %10s", "Metric", "Current", "Target")
    sep    = repeat("-", length(header))
    print sep
    print header
    print sep
    printf "%-15s %10.1f %10.1f\n", "CPU %",     72.4, 60.0
    printf "%-15s %10.1f %10.1f\n", "Mem GB",    14.2, 16.0
    printf "%-15s %10d %10d\n",   "Open FDs", 1024,  800
    print sep
}
' /dev/null

awk内でファイルまたはパイプラインにprintfする

#!/usr/bin/env bash
# Route records into per-status files using print redirection
tmpdir=$(mktemp -d)
trap 'rm -rf "$tmpdir"' EXIT

printf '200 /index.html\n404 /missing\n200 /api/data\n500 /crash\n404 /lost\n' | awk -v out="$tmpdir" '
{
    file = out "/" $1 ".log"
    printf "%-6s %s\n", $1, $2 > file
}
END {
    close(file)  # flush all open handles
}
'

echo "=== 200 ==="
cat "$tmpdir/200.log"
echo "=== 404 ==="
cat "$tmpdir/404.log"

awkでCSVレポートを生成する

#!/usr/bin/env bash
# Emit a valid CSV report from space-separated input
printf 'alice engineer 95000\nbob "sales,mgr" 82000\ncarol developer 110000\n' | awk '
function csv_field(s,    safe) {
    safe = s
    gsub(/"/, "\"\"" , safe)   # double any embedded quotes
    return "\"" safe "\""       # wrap in quotes
}
BEGIN { print "Name,Role,Salary" }
{
    printf "%s,%s,%d\n", csv_field($1), csv_field($2), $3
}
'

最初から最後まで：完全なアクセスログレポート

#!/usr/bin/env bash
# Full access log report: method breakdown with human-readable bytes
printf 'GET /index 200 4096\nPOST /api 201 512\nGET /img 200 102400\nDELETE /item 204 0\nPOST /login 401 256\nGET /style 200 8192\n' | awk '
function human_bytes(n,    s) {
    if (n >= 1048576) s = sprintf("%.1f MB", n/1048576)
    else if (n >= 1024) s = sprintf("%.1f KB", n/1024)
    else s = sprintf("%d B", n)
    return s
}
{
    method = $1
    size   = $4
    count[method]++
    bytes[method] += size
    grand_count++
    grand_bytes += size
}
END {
    fmt = "%-10s %8s %15s\n"
    sep = "----------------------------------"
    printf fmt, "Method", "Requests", "Bytes"
    print sep
    for (m in count)
        printf "%-10s %8d %15s\n", m, count[m], human_bytes(bytes[m])
    print sep
    printf "%-10s %8d %15s\n", "TOTAL", grand_count, human_bytes(grand_bytes)
}
'

理解度チェック：awk関数のローカル変数

レッスンのまとめ：awk関数、printf、レポート生成