chore: improve get_value's performance

[nvh0412]

Reducing object allocations and also improve time complexity of get_value

1. There is 1 object allocation each time we look up the key
2. O(1) lookup after initial build

The previous code would loop through all key value every time in ENV. The pre-built hash is over 2,500x faster than the alternatives.

Benchmark

require 'benchmark'

# Setup
ENV.clear
1000.times { |i| ENV["KEY_#{i}"] = "value_#{i}" }
ENV["TARGET_KEY"] = "target_value"
key = "TARGET_KEY"

# Method 1: Pre-built hash
env_lowercase = ENV.each_with_object({}) { |(k, v), h| h[k.downcase] = v }

# Method 2: Direct iteration
def find_case_insensitive(hash, key)
  key_down = key.downcase
  hash.each do |k, v|
    return v if k.downcase == key_down
  end
  nil
end

Benchmark.bm do |x|
  x.report("Pre-built hash:") do
    10000.times { env_lowercase[key.downcase] }
  end
  
  x.report("Direct iteration:") do
    10000.times { find_case_insensitive(ENV, key) }
  end
  
  x.report("Current detect:") do
    10000.times { ENV.detect { |k, _| k.downcase == key.downcase }&.last }
  end
end

Result:

[nvh0412]

Hi @laserlemon, I’m open to optimizing has_key?(key) by using a shared case-insensitive key hash if you think it’s worth it. In my project, the get_value(key) method is the hot path, so optimizing that alone is sufficient for me.