classify_failed_servers.sh

#!/usr/bin/env bash

# Exit immediately if any command fails
set -e

# --- CONFIGURATION ---
FAILED_LIST="failed_servers.txt"
ENV_DIR="./environments"
OUTPUT_FILE="classified_failed_servers.txt"
TEMP_FILE="matched_raw.tmp"

# --- VALIDATION ---
if [ ! -f "$FAILED_LIST" ]; then
    echo "Error: Failed server list '$FAILED_LIST' not found." >&2
    exit 1
fi

if [ ! -d "$ENV_DIR" ]; then
    echo "Error: Environments directory '$ENV_DIR' not found." >&2
    exit 1
fi

echo "=========================================="
echo " Starting YAML Server Classification"
echo "=========================================="

# Clean up any leftover temp or output files from previous runs
rm -f "$TEMP_FILE" "$OUTPUT_FILE"
echo -n "Processing environment files... "

# --- PROCESSING ---
for env_file in "$ENV_DIR"/*.yml "$ENV_DIR"/*.yaml; do
    # Ensure the glob actually found files
    [ -e "$env_file" ] || continue
    [ -d "$env_file" ] && continue
    
    env_name=$(basename "$env_file")

    # 1. Clean the YAML lines (strip spaces, hyphens, colons, and structural keys)
    # 2. Grep for exact matches against the failed list
    # 3. Append the environment filename to the matched host
    awk '
    {
        gsub(/^[ \t-]+/, ""); # Strip leading spaces and hyphens
        gsub(/:.*/, "");      # Strip trailing colons and anything after them
        gsub(/\047/, "");     # Safely strip single quotes using octal code
        gsub(/\042/, "");     # Safely strip double quotes using octal code
        
        # Skip empty lines or top-level YAML structural keys
        if ($0 == "" || $0 == "all" || $0 == "hosts" || $0 == "hostnames") {
            next;
        }
        print $0
    }' "$env_file" | \
    grep -Fxf "$FAILED_LIST" | \
    awk -v fname="$env_name" '{print $0 " , " fname}' >> "$TEMP_FILE" || true
done

echo "Done."

# --- SORTING & CLEANUP ---
if [ -f "$TEMP_FILE" ] && [ -s "$TEMP_FILE" ]; then
    echo "Sorting and saving results to '$OUTPUT_FILE'..."
    
    # Sort naturally (-V) based on the server name
    sort -V "$TEMP_FILE" -o "$OUTPUT_FILE"
    rm -f "$TEMP_FILE"
    
    echo "=========================================="
    echo " Success! Process complete."
    echo " Total classified lines: $(wc -l < "$OUTPUT_FILE")"
    echo "=========================================="
else
    echo "=========================================="
    echo " Finished: No matching servers found."
    echo "=========================================="
    rm -f "$TEMP_FILE"
fi