Reading Streams Safely with read

1. read assigns fields from an input record

printf 'api staging 3\n' |
while read -r service environment replicas; do
  printf 'service=%s env=%s replicas=%s\n'     "$service" "$environment" "$replicas"
done

This is appropriate only when whitespace-separated fields are truly the data format.

2. IFS= read -r preserves a line

while IFS= read -r line; do
  printf 'line=<%s>\n' "$line"
done < input.txt

3. Set IFS when the format has a simple delimiter

record='api:staging:3'
IFS=: read -r service environment replicas <<<"$record"

printf '%s | %s | %s\n' "$service" "$environment" "$replicas"

This works when the format guarantees that fields themselves cannot contain unescaped colons.

4. Complex formats need format-aware parsers

Real CSV can contain quoted commas and embedded newlines. JSON and YAML have nested syntax and escaping. Do not recreate those parsers with read, cut, or regular expressions.

5. The final variable receives remaining fields

line='INFO api deployment completed successfully'
read -r level service message <<<"$line"

printf 'level=%s\n' "$level"
printf 'service=%s\n' "$service"
printf 'message=%s\n' "$message"

6. NUL is the safe filename-record delimiter

Unix filenames cannot contain the NUL byte, so NUL-delimited streams can represent arbitrary filenames without ambiguity.

find ./artifacts -type f -print0 |
while IFS= read -r -d '' file; do
  printf 'file=%q\n' "$file"
done

7. Pipeline-fed read loops can lose variable changes

count=0
printf '%s\n' a b c |
while IFS= read -r line; do
  ((count += 1))
done
printf 'count=%d\n' "$count"

The loop commonly runs in a subshell. Use process substitution when state must survive:

count=0
while IFS= read -r line; do
  ((count += 1))
done < <(printf '%s\n' a b c)
printf 'count=%d\n' "$count"

8. read -u consumes a dedicated descriptor

exec 3< services.txt
while IFS= read -r -u 3 service; do
  printf 'service=%s\n' "$service"
done
exec 3<&-

This preserves stdin for another purpose and makes input ownership explicit.

9. read can wait with a timeout

if IFS= read -r -t 5 response; then
  printf 'received=%s\n' "$response"
else
  printf 'input timeout\n' >&2
fi

Timeouts are useful for controlled interactive or pipe-based workflows. CI automation should generally avoid human prompts.

10. read -a assigns fields into an array

line='api staging 3 enabled'
read -r -a fields <<<"$line"

printf 'count=%d\n' "${#fields[@]}"
printf 'service=%s\n' "${fields[0]}"

11. Handle a final unterminated line if your contract requires it

while IFS= read -r line || [[ -n $line ]]; do
  printf '%s\n' "$line"
done < input.txt

12. Hands-on lab: hostile filenames

mkdir -p "$HOME/devops-academy/bash/chapter05/lesson03/files"
cd "$HOME/devops-academy/bash/chapter05/lesson03"

printf 'a\n' > "files/normal.txt"
printf 'b\n' > "files/space name.txt"
printf 'c\n' > $'files/tab\tname.txt'

count=0
while IFS= read -r -d '' file; do
  ((count += 1))
  printf 'path=%q bytes=%s\n' "$file" "$(wc -c < "$file")"
done < <(find files -type f -print0)

printf 'count=%d\n' "$count"

13. Knowledge check

14. Summary

Safe stream ingestion begins with a record contract. Use IFS= read -r for lines, explicit delimiters for simple records, NUL for filenames, and a real parser for structured formats.

15. Further reading

• GNU Bash Reference Manual — read.
• GNU Bash Reference Manual — Word Splitting.
• GNU findutils manual — -print0.
• ShellCheck documentation — safe read loops.