Shell scripting: parsing command output safely

Need to extract specific fields from command output in shell scripts without brittle text parsing.

Safe parsing patterns:

# Use built-in field splitting
# Get specific columns from output
docker ps --format '{{.Names}}\t{{.Status}}' | while IFS=

Anti-patterns:
Don't parse ls output (use glob patterns instead)
Don't use regex for JSON/XML (use jq/xmlstarlet)
Don't use grep | awk | sed chains when awk alone suffices

\t' read -r name status; do
  echo "Container: $name -> $status"
done

# Parse key=value output
while IFS='=' read -r key value; do
  case "$key" in
    hostname) HOSTNAME="$value" ;;
    port) PORT="$value" ;;
  esac
done < config.env

# JSON parsing with jq (always prefer for JSON)
curl -s api.example.com/users | jq -r '.[] | "\(.name)\t\(.email)"'

# Get specific JSON field
VERSION=$(jq -r '.version' package.json)

# Conditional with jq
jq -e '.status == "active"' response.json && echo "Active"

# awk for columnar data
df -h | awk 'NR>1 && $5+0 > 80 {print $6, $5}'
# Skip header, find > 80% used partitions

# cut for delimited fields
echo "user:password:1000:1000:User:/home/user:/bin/bash" | cut -d: -f1,3
# Output: user:1000

# Parameter expansion (no external tools needed)
path="/home/user/documents/file.tar.gz"
echo "${path##*/}"    # file.tar.gz (basename)
echo "${path%/*}"     # /home/user/documents (dirname)
echo "${path%%.*}"    # /home/user/documents/file (remove all extensions)
echo "${path%.gz}"    # /home/user/documents/file.tar (remove last extension)

Anti-patterns:

• Don't parse ls output (use glob patterns instead)
• Don't use regex for JSON/XML (use jq/xmlstarlet)
• Don't use grep | awk | sed chains when awk alone suffices

Proper field splitting and structured output formats (JSON, TSV) are more reliable than brittle regex on unstructured text.

Shell scripts that process command output