Skip to content

DeepL API wrapper translates only changes between revisions of big texts

Notifications You must be signed in to change notification settings

Halvanhelv/deepl_diff

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

1 Commit
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

DeepLDiff

DeepL API wrapper helps to translate only changes between revisions of long texts.

DeepLDiff based on GoogleTranslateDiff

Use case

Assume your project contains a significant amount of products descriptions which:

  • Require retranslation each time user edits them.
  • Have a lot of equal parts (like return policy).
  • Change frequently.

If your user changes a single word within the long description, you will be charged for the retranslation of the whole text.

Much better approach is to try to translate every repeated structural element (sentence) in your texts array just once to save money. This gem helps to make it done.

Installation

Add this line to your application's Gemfile:

gem 'deepl_diff'

And then execute:

$ bundle

Or install it yourself as:

$ gem install deepl_diff

Usage

require "deepl_diff"

# This dependencies are not included, as you might need to roll your own cache based on different store
require "redis"
require "connection_pool"
require "redis-namespace"
require "ratelimit" # Optional, if you will use

# Setup https://github.com/wikiti/deepl-rb
DeepL.configure do |config|
  config.auth_key = 'your-api-token'
  config.host = 'https://api-free.deepl.com' # Default value is 'https://api.deepl.com'
  config.version = 'v1' # Default value is 'v2'
end

# I always use pool for redis
pool = ConnectionPool.new(size: 10, timeout: 5) { Redis.new }

# Pass DeepL for DeepLDiff
DeepLDiff.api = DeepL

DeepLDiff.cache_store =
  DeepLDiff::RedisCacheStore.new(pool, timeout: 7.days, namespace: "t")

# Optional
DeepLDiff.rate_limiter =
  DeepLDiff::RedisRateLimiter.new(
    pool, threshold: 8000, interval: 60, namespace: "t"
  )

DeepLDiff.translate("test translations", from: "en", to: "es")

How it works

  • Text nodes are extracted from HTML.
  • Every text node is split into sentences (using punkt-segmenter gem).
  • Cache is checked for the presence of each sentence (using language couple and a hash of string).
  • Missing sentences are translated via API and cached.
  • Original HTML is recombined from translations and cache data.

NOTE: if :from is not specified or equal to nil, then the DeepL API will be called twice, the first time a sample of text up to 100 characters long will be transferred to determine the language, and the second time the entire text will be transferred. Try to specify :from explicitly

Input

::translate can receive string, array or deep hash and will return the same, but translated.

DeepLDiff.translate("test", from: "en", to: "es")
DeepLDiff.translate(%w[test language], from: "en", to: "es")
DeepLDiff.translate(
  { title: "test", values: { type: "frequent" } }, from: "en", to: "es"
)

See DeepLDiff::Linearizer for details.

HTML

You can pass HTML as like as plain text:

DeepLDiff.translate("<b>Black</b>", from: "en", to: "es")

Very long texts

DeepL API has a limitation: query can not be longer than approximately 128 KB. If your text is really that long, multiple queries will be used to translate it automatically.

Development

After checking out the repo, run bin/setup to install dependencies. Then, run rake spec to run the tests. You can also run bin/console for an interactive prompt that will allow you to experiment.

To install this gem onto your local machine, run bundle exec rake install. To release a new version, update the version number in version.rb, and then run bundle exec rake release, which will create a git tag for the version, push git commits and tags, and push the .gem file to rubygems.org.

Contributing

Bug reports and pull requests are welcome on GitHub at https://github.com/wikiti/deepl-rb.

About

DeepL API wrapper translates only changes between revisions of big texts

Topics

Resources

Stars

Watchers

Forks

Packages

No packages published

Languages