count-k8s-objects.sh - Count the number of Kubernetes objects in a directory and error if there are too many

count-k8s-objects.sh

#!/bin/bash

# This script recursively counts the number of Kubernetes objects in YAML and JSON files within a directory.
# It uses `yq` to parse YAML and JSON, prints the object count for each file,
# and also prints the total count for each directory. It errors if the total object count exceeds 2000.
# It skips directories that start with "."

set -euo pipefail

# --- Configuration ---
readonly MAX_OBJECTS=2000

# --- Functions ---

# Usage: count_objects_in_file <file>
# Counts the number of Kubernetes objects in the specified file.
count_objects_in_file() {
  local file="$1"

  yq -eN '[[select(documentIndex >= 0)] | length]' "$file" | yq -e '.[] as $n ireduce (0; . + $n)'
}

# Usage: process_directory <directory> [indent]
# Processes a directory, counts objects in files, prints file and directory totals.
process_directory() {
    local directory="$1"
    local indent="${2:-""}" # Default indent to empty string if not provided
    local dir_total_objects=0
    local file_count
    local subdir_total

    # Iterate over each subdirectory
    while IFS= read -r -d $'\0' subdir; do
        # Skip directories that start with "." using find's -not -path option
        subdir_total=$(process_directory "$subdir" "${indent}  ") # Capture output from recursive call
        dir_total_objects=$((dir_total_objects + subdir_total))
    done < <(find "$directory" -mindepth 1 -maxdepth 1 -type d -not -path '*/.*' -print0)

    # Iterate over each file in the directory
    while IFS= read -r -d $'\0' file; do
        file_count=$(count_objects_in_file "$file")
        # Safely print the file count with indentation
        >&2 echo "${file} (${file_count})"
        dir_total_objects=$((dir_total_objects + file_count))
    done < <(find "$directory" -maxdepth 1 -type f \( -name "*.yaml" -o -name "*.yml" -o -name "*.json" \) -print0)

    # Print object count for this directory
    echo "${dir_total_objects}"
}

# Usage: validate_count <count>
# Checks if the object count exceeds the maximum allowed.
validate_count() {
    local count="$1"

    if (( count > MAX_OBJECTS )); then
        echo "Error: Too many Kubernetes objects found ($count). Maximum allowed is $MAX_OBJECTS." >&2
        return 1
    fi

    return 0
}

# --- Main ---

main() {
    local directory="$1"
    local total_objects

    if [[ -z "$directory" ]]; then
        echo "Usage: $0 <directory>" >&2
        return 1
    fi

    if [[ ! -d "$directory" ]]; then
        echo "Error: Directory '$directory' does not exist." >&2
        return 1
    fi

    # Process the directory, which updates the global total_objects
    total_objects=$(process_directory "$directory")

    if ! validate_count "$total_objects"; then
        return 1
    fi

    echo "Total number of Kubernetes objects: $total_objects"
    return 0
}

# --- Script Execution ---

main "$@"

$ git clone https://github.com/config-sync-examples/crontab-crs
$ cd crontab-crs
$ ./count-k8s-objects.sh .
./configs/crs.yaml (13060)
./configs/ns.yaml (1)
./root-sync-cr.yaml (1)
./config-management-multi.yaml (1)
Error: Too many Kubernetes objects found (13063). Maximum allowed is 2000.