⌨️ sed, awk, curl and > on a Mac · sed, awk, curl และ > บนแมค · 在 Mac 上用 sed、awk、curl 和 >

1 > keeps the answer in a file · > เก็บคำตอบไว้ในไฟล์ · > 把答案留在文件里

⌨️ TYPE THESE · one line at a time

# > makes the file. If it was there already, the old one is gone.
echo "Hello, Chiang Mai" > hello.txt
cat hello.txt

# >> adds a line. It never throws anything away.
echo "Second line" >> hello.txt
cat hello.txt

# One > again, and both old lines are gone. No warning.
echo "This wipes it" > hello.txt
cat hello.txt

🖥️ WHAT IT PRINTED

Hello, Chiang Mai
Hello, Chiang Mai
Second line
This wipes it

2 Errors come out of a second pipe · ข้อผิดพลาดออกมาอีกท่อหนึ่ง · 报错走的是第二根管子

⌨️ TYPE THESE

# Make one real file, so one name works and one does not.
printf 'one\ntwo\n' > hello.txt

# cat reads both names. The good text goes left, the complaint goes right.
cat hello.txt missing.txt > found.txt 2> problems.txt

cat found.txt
cat problems.txt

# 2>&1 means: send the errors down the same pipe as the good output.
cat hello.txt missing.txt > both.txt 2>&1
cat both.txt

🖥️ WHAT IT PRINTED · found.txt, then problems.txt, then both.txt

one
two
cat: missing.txt: No such file or directory
one
two
cat: missing.txt: No such file or directory

3 | joins two programs together · | ต่อสองโปรแกรมเข้าด้วยกัน · | 把两个程序接起来

⌨️ TYPE THESE

# tr swaps one character for another. Here: every space becomes a new line.
echo "one two three" | tr ' ' '\n'

# Now add sort. Three words, one per line, in order.
echo "banana apple cherry" | tr ' ' '\n' | sort

# And wc -l counts the lines it was given.
echo "banana apple cherry" | tr ' ' '\n' | sort | wc -l

🖥️ WHAT IT PRINTED

one
two
three
apple
banana
cherry
3

4 One real price, on your screen · ราคาจริงหนึ่งราคา บนจอของคุณ · 一个真实价格，出现在你屏幕上

⌨️ TYPE THIS

# -s means silent. Without it curl draws a progress bar over your answer.
curl -s "https://api.bitkub.com/api/v3/market/ticker?sym=BTC_THB"

# The same line again, kept in a file this time.
curl -s "https://api.bitkub.com/api/v3/market/ticker?sym=BTC_THB" > price.json
cat price.json

🖥️ WHAT IT PRINTED · the real answer, 27 September 2026

[{"symbol":"BTC_THB","base_volume":"23.16745634","high_24_hr":"2824252","highest_bid":"2818749.1","last":"2818749.09","low_24_hr":"2799000","lowest_ask":"2819502.68","percent_change":"0.59","quote_volume":"65094106.59"}]

5 The five flags worth learning · ห้าแฟล็กที่ควรจำ · 值得记住的五个参数

⌨️ TYPE THESE · ask for a coin that does not exist

# Bitkub says error 11. But curl says everything went fine.
curl -s "https://api.bitkub.com/api/v3/market/ticker?sym=NOPE_THB"
echo "exit was $?"

# A page that is really missing. -f makes curl admit it.
curl -sS -f "https://api.bitkub.com/api/v3/market/nothing-here"
echo "exit was $?"

# -w prints one fact you choose. Here: the status number.
curl -s -o /dev/null -w '%{http_code}\n' "https://api.bitkub.com/api/v3/market/ticker?sym=BTC_THB"
curl -s -o /dev/null -w '%{http_code}\n' "https://api.bitkub.com/api/v3/market/nothing-here"

🖥️ WHAT IT PRINTED

{"error":11}
exit was 0
curl: (22) The requested URL returned error: 404
exit was 22
200
404

6 Swap one word for another · เปลี่ยนคำหนึ่งเป็นอีกคำ · 把一个词换成另一个

⌨️ TYPE THESE

# Change one thing. The rest of the line is untouched.
echo "I teach in Chiang Mai" | sed 's/Chiang Mai/Lamphun/'

# Without g, sed changes the first one on the line and stops.
echo "one one one" | sed 's/one/two/'

# With g, it keeps going to the end of the line. g is for global.
echo "one one one" | sed 's/one/two/g'

🖥️ WHAT IT PRINTED

I teach in Lamphun
two one one
two two two

7 Pull the price out of the JSON · ดึงราคาออกมาจาก JSON · 把价格从 JSON 里抠出来

⌨️ TYPE THESE

# Way one, in two steps. Cut off the front, then cut off the back.
# .* means "any amount of anything", so this eats everything before the price.
sed 's/.*"last":"//' price.json | sed 's/".*//'

# Way two, in one step. -E turns on groups. The ( ) marks what to keep,
# and \1 means "put group one back here". Everything else is thrown away.
sed -E 's/.*"last":"([0-9.]+)".*/\1/' price.json

🖥️ WHAT IT PRINTED · both ways, same answer

2818749.09
2818749.09

8 Whole lines, and the -i trap · ทั้งบรรทัด และกับดักของ -i · 整行，以及 -i 这个坑

⌨️ TYPE THESE

printf 'first\nsecond\nthird\nfourth\n' > lines.txt

# -n tells sed to print nothing on its own. Then p prints what you picked.
sed -n '2p' lines.txt

# Two numbers with a comma is a range.
sed -n '2,3p' lines.txt

# Slashes around a word match it. d deletes those lines.
sed '/second/d' lines.txt

# The portable way to change a file: write a new one, then rename it.
# && means "only if that worked", so a failed sed cannot wipe your file.
sed 's/first/FIRST/' lines.txt > lines.new && mv lines.new lines.txt
head -1 lines.txt

🖥️ WHAT IT PRINTED

second
second
third
first
third
fourth
FIRST

9 $1 is the first word · $1 คือคำแรก · $1 就是第一个词

⌨️ TYPE THESE

# The braces hold what awk should do with each line.
echo "Nari teaches English in Chiang Mai" | awk '{print $1}'

# A comma between them puts a space between them.
echo "Nari teaches English in Chiang Mai" | awk '{print $1, $NF}'

# NF counts the pieces. Six words, so NF is 6.
echo "Nari teaches English in Chiang Mai" | awk '{print NF}'

🖥️ WHAT IT PRINTED

Nari
Nari Mai
6

10 A table, a filter, and a total · ตาราง ตัวกรอง และผลรวม · 一张表、一个筛选、一个合计

⌨️ TYPE THESE

# Three students. A name, a subject, and a number of lessons.
printf 'Nari,English,12\nSom,Maths,9\nKai,English,15\n' > class.csv

# -F, says the columns are split by commas, not spaces.
awk -F, '{print $1}' class.csv

# A test in front of the braces. Only matching lines get printed.
awk -F, '$2 == "English" {print $1, $3}' class.csv

# total starts at zero on its own. awk does that for you.
awk -F, '{total = total + $3} END {print "total", total}' class.csv

# NR is the row number, so at the END it is the number of rows.
# printf lets you choose the shape. %.1f means one digit after the dot.
awk -F, '{total += $3} END {printf "average %.1f\n", total / NR}' class.csv

🖥️ WHAT IT PRINTED

Nari
Som
Kai
Nari 12
Kai 15
total 36
average 12.0

11 A price log that grows every day · บันทึกราคาที่ยาวขึ้นทุกวัน · 一个每天变长的价格记录

📄 NEW FILE · log_price.sh

# log_price.sh — one price, one dated line, appended to a log.
# Run it today and tomorrow. The file remembers both.

# A backslash at the end of a line means "this command carries on below".
# Nothing may follow it, not even a space, or the shell ends the line there.
curl -sS -f "https://api.bitkub.com/api/v3/market/ticker?sym=BTC_THB" \
  | sed -E 's/.*"last":"([0-9.]+)".*/\1/' \
  | awk -v day="$(date +%F)" '{printf "%s,%.2f\n", day, $1}' \
  >> price_log.csv

cat price_log.csv

🖥️ WHAT IT PRINTED · one run, 27 September 2026

2026-09-27,2818749.09

12 Try first. The answers are folded away. · ลองเองก่อน เฉลยพับซ่อนไว้ · 先自己试。答案折起来放着。

# Only the label changed. The shape of the rule is the same.
sed -E 's/.*"high_24_hr":"([0-9.]+)".*/\1/' price.json

2824252

# two_coins.sh — the same four lines, run once per coin.

# Start clean, so running this twice does not double the file.
rm -f two.csv

# sym holds BTC on the first pass, then ETH on the second.
for sym in BTC ETH; do
  curl -sS -f "https://api.bitkub.com/api/v3/market/ticker?sym=${sym}_THB" \
    | sed -E 's/.*"last":"([0-9.]+)".*/\1/' \
    | awk -v coin="$sym" '{printf "%s,%.2f\n", coin, $1}' \
    >> two.csv
done

cat two.csv

BTC,2818749.09
ETH,90098.44

# -? in the group means "a minus sign here is allowed, but not required".
sed -E 's/.*"percent_change":"(-?[0-9.]+)".*/\1/' price.json \
  | awk '{ if ($1 > 0) print "up", $1 "%"; else if ($1 < 0) print "down", -$1 "%"; else print "flat" }'

up 0.59%

# log_once.sh — the same log, but safe to run all day.

TODAY=$(date +%F)
LOG=price_log.csv

# touch makes an empty file if there is none, so grep has something to read.
touch "$LOG"

# -q means say nothing, just answer with the exit number. The ! flips it,
# so the body runs only when today was NOT found.
if ! grep -q "^$TODAY," "$LOG"; then
  curl -sS -f "https://api.bitkub.com/api/v3/market/ticker?sym=BTC_THB" \
    | sed -E 's/.*"last":"([0-9.]+)".*/\1/' \
    | awk -v day="$TODAY" '{printf "%s,%.2f\n", day, $1}' \
    >> "$LOG"
  echo "added $TODAY"
else
  echo "$TODAY is already there"
fi

# Three days of made-up log, so the answer is easy to check by eye.
printf '2026-09-25,2805205.52\n2026-09-26,2819755.67\n2026-09-27,2799000.00\n' > log.csv

# The test is in front, so this only runs on a line that beats the record.
# best starts empty, which counts as zero, so line one always wins.
awk -F, '$2 > best { best = $2; day = $1 } END { printf "%s %.2f\n", day, best }' log.csv

2026-09-26 2819755.67

# -t, sets the separator. -k2 picks column two. -n compares as numbers.
# Without -n, sort puts 100 before 9, because it reads them as words.
echo "cheapest:"
sort -t, -k2 -n log.csv | head -1

echo "dearest:"
sort -t, -k2 -n log.csv | tail -1

cheapest:
2026-09-27,2799000.00
dearest:
2026-09-26,2819755.67

# tee is named after a T pipe. What comes in goes two ways at once:
# into the file, and on down the pipe to your screen.
curl -sS -f "https://api.bitkub.com/api/v3/market/ticker?sym=BTC_THB" \
  | sed -E 's/.*"last":"([0-9.]+)".*/\1/' \
  | tee latest.txt

cat latest.txt

2818749.09
2818749.09