tiktoken_ruby 0.0.1-x86_64-linux → 0.0.3-x86_64-linux

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: ff6fd122cc2065464bcf5a11fbda3ed6eab2255c505ba97840fd4ca56a70e7e0
-  data.tar.gz: 5273ce2b34a40134636e3ab7bcc96ee6479d45346a2fcf196e2495a277bdd8e1
+  metadata.gz: 52f5e991c44d3f916a956b6e0142af8e9157307672c7bd537c370a17100eaaa3
+  data.tar.gz: 26701f56607b59d0c7d3f0a56f98c4bf5fe970bfd01b5143668726fed4512ddf
 SHA512:
-  metadata.gz: e6778020a5790bac306ff9340c24a16e03bfeba8b624572e8eb43394638c9fe74ab90c8a3e9717140c8234e69582287ab59af1e99738bd4ee691c27076d02e09
-  data.tar.gz: aed1277c3259e9b35982d0f1fe4f470cafc5bd0384db978fe568b1449516ecff786e15f2dc41aad6e01d27de4d32e6c59d1b681c0cbc482c76f8c94e6fece79d
+  metadata.gz: 8fa65091cc2df7b2bc6e8ff74dbee70a8ba05527e8fc2fe220acffb3c8c22651da71874f6b68829becf29a5268124bd8907a3b5e60faa5624adda74066a64225
+  data.tar.gz: c7abf9c984d945e38e4e711e1e7854a4bb65233b2ce526178750149782a054fd97014898b3c3502db1ac5a1e8cb308c4bc8816cda07d94ece1b9bde07d0d425c

data/Gemfile

@@ -14,3 +14,5 @@ gem "rspec", "~> 3.0"
 
 gem "standard", "~> 1.3"
 gem 'pry', '~> 0.14.2'
+
+gem "yard-doctest", "~> 0.1.17"

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    tiktoken_ruby (0.0.1)
+    tiktoken_ruby (0.0.3)
 
 GEM
   remote: https://rubygems.org/
@@ -12,6 +12,7 @@ GEM
     json (2.6.3)
     language_server-protocol (3.17.0.3)
     method_source (1.0.0)
+    minitest (5.18.0)
     parallel (1.22.1)
     parser (3.2.1.1)
       ast (~> 2.4.1)
@@ -59,6 +60,12 @@ GEM
       rubocop (= 1.48.1)
       rubocop-performance (= 1.16.0)
     unicode-display_width (2.4.2)
+    webrick (1.7.0)
+    yard (0.9.28)
+      webrick (~> 1.7.0)
+    yard-doctest (0.1.17)
+      minitest
+      yard
 
 PLATFORMS
   arm64-darwin-22
@@ -72,6 +79,7 @@ DEPENDENCIES
   rspec (~> 3.0)
   standard (~> 1.3)
   tiktoken_ruby!
+  yard-doctest (~> 0.1.17)
 
 BUNDLED WITH
    2.4.6

data/README.md

@@ -1,3 +1,4 @@
+[![Gem Version](https://badge.fury.io/rb/tiktoken_ruby.svg)](https://badge.fury.io/rb/tiktoken_ruby)
 # tiktoken_ruby
 
 [Tiktoken](https://github.com/openai/tiktoken) is BPE tokenizer from OpenAI used with their GPT models.
@@ -5,19 +6,32 @@ This is a wrapper around it aimed primarily at enabling accurate counts of GPT m
 
 ## Installation
 
-TODO: Replace `UPDATE_WITH_YOUR_GEM_NAME_PRIOR_TO_RELEASE_TO_RUBYGEMS_ORG` with your gem name right after releasing it to RubyGems.org. Please do not do it earlier due to security reasons. Alternatively, replace this section with instructions to install your gem from git if you don't plan to release to RubyGems.org.
-
 Install the gem and add to the application's Gemfile by executing:
 
-    $ bundle add UPDATE_WITH_YOUR_GEM_NAME_PRIOR_TO_RELEASE_TO_RUBYGEMS_ORG
+    $ bundle add tiktoken_ruby
 
 If bundler is not being used to manage dependencies, install the gem by executing:
 
-    $ gem install UPDATE_WITH_YOUR_GEM_NAME_PRIOR_TO_RELEASE_TO_RUBYGEMS_ORG
+    $ gem install tiktoken_ruby
 
 ## Usage
+Usage should be very similar to the python library. here's a simple example
+
+Encode and decode text
+```ruby
+require 'tiktoken_ruby'
+
+enc = Tiktoken.get_encoding("cl100k_base")
+enc.decode(enc.encode("hello world")) #=> "hello world"
+```
+
+Encoders can also be retrieved by model name
+```ruby
+require 'tiktoken_ruby'
 
-TODO: Write usage instructions here
+enc = Tiktoken.encoding_for_model("gpt-4")
+enc.encode("hello world").length #=> 2
+```
 
 ## Development
 
@@ -27,7 +41,7 @@ To install this gem onto your local machine, run `bundle exec rake install`. To
 
 ## Contributing
 
-Bug reports and pull requests are welcome on GitHub at https://github.com/[USERNAME]/tiktoken_ruby.
+Bug reports and pull requests are welcome on GitHub at https://github.com/iapark/tiktoken_ruby.
 
 ## License
 

data/doctest_helper.rb

@@ -0,0 +1 @@
+require 'lib/tiktoken_ruby.rb'

data/lib/tiktoken_ruby/encoding.rb

@@ -1,23 +1,51 @@
 # frozen_string_literal: true
 
-class Tiktoken::Encoding 
-    def self.method_missing(method)
-        Tiktoken::Encoding.new(Tiktoken::BpeFactory.send(method))
+class Tiktoken::Encoding
+    attr_reader :name
+
+    # This returns a new Tiktoken::Encoding instance for the requested encoding
+    # @param encoding [Symbol] The name of the encoding to load
+    # @return [Tiktoken::Encoding] The encoding instance
+    def self.for_name(encoding)
+        Tiktoken::Encoding.new(Tiktoken::BpeFactory.send(encoding.to_sym), encoding.to_sym)
     end
 
-    def initialize(ext_base_bpe)
-        @ext_base_bpe = ext_base_bpe
+    # This returns a Tiktoken::Encoding instance for the requested encoding
+    # It will reuse an existing encoding if it's already been loaded
+    # @param encoding [Symbol] The name of the encoding to load
+    # @return [Tiktoken::Encoding] The encoding instance
+    def self.for_name_cached(encoding)
+        @encodings ||= {}
+        @encodings[encoding.to_sym] ||= Tiktoken::Encoding.for_name(encoding)
     end
 
+    # Encodes the text as a list of integer tokens. This encoding will encode special non text tokens
+    # basically it's unescaped
+    # @param text [String] The text to encode
+    # @return [Array<Integer>] The encoded tokens
     def encode_ordinary(text)
         @ext_base_bpe.encode_ordinary(text)
     end
 
+    # Encodes the text as a list of integer tokens. This encoding will treat special non text tokens
+    # as text unless they're in the allowed_special array. It's basically like the text was escaped
+    # @param text [String] The text to encode
+    # @param allowed_special [Array<String>] An array of special tokens to allow
+    # @return [Array<Integer>] The encoded tokens
     def encode(text, allowed_special: [])
         @ext_base_bpe.encode(text, allowed_special)
     end
 
+    # Decodes the tokens back into text
+    # @param tokens [Array<Integer>] The tokens to decode
+    # @return [String] The decoded text
     def decode(tokens)
         @ext_base_bpe.decode(tokens)
     end
+
+    private
+    def initialize(ext_base_bpe, name)
+        @ext_base_bpe = ext_base_bpe
+        @name = name
+    end
 end

data/lib/tiktoken_ruby/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Tiktoken
-  VERSION = "0.0.1"
+  VERSION = "0.0.3"
 end

data/lib/tiktoken_ruby.rb

@@ -2,8 +2,114 @@
 
 require_relative "tiktoken_ruby/version"
 require_relative "tiktoken_ruby/encoding.rb"
-require_relative "tiktoken_ruby/tiktoken_ruby"
+
+begin
+  RUBY_VERSION =~ /(\d+\.\d+)/
+  require_relative "tiktoken_ruby/#{$1}/tiktoken_ruby"
+rescue LoadError
+  require_relative "tiktoken_ruby/tiktoken_ruby"
+end
 
 module Tiktoken
-  class Error < StandardError; end
+  class << self
+
+    # Returns an encoding by name. If the encoding is not already loaded it will be loaded, but otherwise
+    # it will reuse the instance of that type that was previous loaded
+    # @param name [Symbol|String] The name of the encoding to load
+    # @return [Tiktoken::Encoding] The encoding instance
+    # @example Encode and decode text
+    #   enc = Tiktoken.get_encoding("cl100k_base")
+    #   enc.decode(enc.encode("hello world")) #=> "hello world"
+    def get_encoding(name)
+      name = name.to_sym
+      return nil unless SUPPORTED_ENCODINGS.include?(name)
+
+      Tiktoken::Encoding.for_name_cached(name)
+    end
+
+    # Gets the encoding for an OpenAI model
+    # @param model_name [Symbol|String] The name of the model to get the encoding for
+    # @return [Tiktoken::Encoding] The encoding instance
+    # @example Count tokens for text
+    #   enc = Tiktoken.encoding_for_model("gpt-4")
+    #   enc.encode("hello world").length #=> 2
+    def encoding_for_model(model_name)
+      for prefix in PREFIX_MODELS
+        if model_name.to_s.start_with?("#{prefix}-")
+          model_name = prefix
+          break
+        end
+      end
+
+      encoding_name = MODEL_TO_ENCODING_NAME[model_name.to_sym]
+      return nil unless encoding_name
+
+      get_encoding(encoding_name)
+    end
+
+    # Lists all the encodings that are supported
+    # @return [Array<Symbol>] The list of supported encodings
+    def list_encoding_names
+      SUPPORTED_ENCODINGS
+    end
+
+    # Lists all the models that are supported
+    # @return [Array<Symbol>] The list of supported models
+    def list_model_names
+      MODEL_TO_ENCODING_NAME.keys
+    end
+
+    private
+
+    SUPPORTED_ENCODINGS = [
+      :r50k_base,
+      :p50k_base,
+      :p50k_edit,
+      :cl100k_base,
+    ]
+
+    # taken from the python library here https://github.com/openai/tiktoken/blob/main/tiktoken/model.py#L13-L53
+    # that is also MIT licensed but by OpenAI
+    MODEL_TO_ENCODING_NAME = {
+      "gpt-4": "cl100k_base",
+      "gpt-3.5-turbo": "cl100k_base",
+      # text
+      "text-davinci-003": "p50k_base",
+      "text-davinci-002": "p50k_base",
+      "text-davinci-001": "r50k_base",
+      "text-curie-001": "r50k_base",
+      "text-babbage-001": "r50k_base",
+      "text-ada-001": "r50k_base",
+      "davinci": "r50k_base",
+      "curie": "r50k_base",
+      "babbage": "r50k_base",
+      "ada": "r50k_base",
+      # code
+      "code-davinci-002": "p50k_base",
+      "code-davinci-001": "p50k_base",
+      "code-cushman-002": "p50k_base",
+      "code-cushman-001": "p50k_base",
+      "davinci-codex": "p50k_base",
+      "cushman-codex": "p50k_base",
+      # edit
+      "text-davinci-edit-001": "p50k_edit",
+      "code-davinci-edit-001": "p50k_edit",
+      # embeddings
+      "text-embedding-ada-002": "cl100k_base",
+      # old embeddings
+      "text-similarity-davinci-001": "r50k_base",
+      "text-similarity-curie-001": "r50k_base",
+      "text-similarity-babbage-001": "r50k_base",
+      "text-similarity-ada-001": "r50k_base",
+      "text-search-davinci-doc-001": "r50k_base",
+      "text-search-curie-doc-001": "r50k_base",
+      "text-search-babbage-doc-001": "r50k_base",
+      "text-search-ada-doc-001": "r50k_base",
+      "code-search-babbage-code-001": "r50k_base",
+      "code-search-ada-code-001": "r50k_base",
+    }
+
+    # these are models that have a versioned models that are otherwise identical
+    PREFIX_MODELS = ["gpt-4", "gpt-3.5-turbo"]
+  end
 end

metadata

@@ -1,16 +1,18 @@
 --- !ruby/object:Gem::Specification
 name: tiktoken_ruby
 version: !ruby/object:Gem::Version
-  version: 0.0.1
+  version: 0.0.3
 platform: x86_64-linux
 authors:
 - IAPark
 autorequire: 
 bindir: exe
 cert_chain: []
-date: 2023-03-19 00:00:00.000000000 Z
+date: 2023-03-21 00:00:00.000000000 Z
 dependencies: []
-description: Unofficial Ruby wrapper for Tiktoken by way of the unofficial rust bindings
+description: An unofficial Ruby wrapper for Tiktoken, a BPE tokenizer written by and
+  used by OpenAI. It can be used to count the number of tokens in text before sending
+  it to OpenAI APIs.
 email:
 - isaac.a.park@gmail.com
 executables: []
@@ -24,6 +26,7 @@ files:
 - LICENSE.txt
 - README.md
 - Rakefile
+- doctest_helper.rb
 - lib/tiktoken_ruby.rb
 - lib/tiktoken_ruby/2.7/tiktoken_ruby.so
 - lib/tiktoken_ruby/3.0/tiktoken_ruby.so
@@ -38,6 +41,7 @@ licenses:
 metadata:
   homepage_uri: https://github.com/IAPark/tiktoken_ruby
   source_code_uri: https://github.com/IAPark/tiktoken_ruby
+  documentation_uri: https://rubydoc.info/github/IAPark/tiktoken_ruby/main
 post_install_message: 
 rdoc_options: []
 require_paths:
@@ -54,7 +58,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
   requirements:
   - - ">="
     - !ruby/object:Gem::Version
-      version: 3.3.11
+      version: 3.1.0
 requirements: []
 rubygems_version: 3.4.4
 signing_key: