Add RBS type signatures

.github/workflows/ruby.yml

@@ -33,3 +33,7 @@ jobs:
         bundler-cache: true # runs 'bundle install' and caches installed gems automatically
     - name: Run tests
       run: bundle exec rake test
+    - name: Install RBS
+      run: gem install rbs
+    - name: Validate RBS types
+      run: rbs -I sig validate

sig/classifier.rbs

@@ -0,0 +1,2 @@
+module Classifier
+end

sig/classifier/bayes.rbs

@@ -0,0 +1,26 @@
+module Classifier
+  class Bayes
+    @categories: Hash[Symbol, Hash[Symbol, Integer]]
+    @total_words: Integer
+    @category_counts: Hash[Symbol, Integer]
+    @category_word_count: Hash[Symbol, Integer]
+
+    def initialize: (*_ToS categories) -> void
+
+    def train: (_ToS category, String text) -> void
+
+    def untrain: (_ToS category, String text) -> void
+
+    def classifications: (String text) -> Hash[String, Float]
+
+    def classify: (String text) -> String
+
+    def categories: () -> Array[String]
+
+    def add_category: (_ToS category) -> Hash[Symbol, Integer]
+
+    alias append_category add_category
+
+    def remove_category: (_ToS category) -> void
+  end
+end

sig/classifier/extensions/object.rbs

@@ -0,0 +1,3 @@
+class Object
+  def prepare_category_name: () -> Symbol
+end

sig/classifier/extensions/string.rbs

@@ -0,0 +1,25 @@
+class String
+  CORPUS_SKIP_WORDS: Set[String]
+
+  def without_punctuation: () -> String
+
+  def word_hash: () -> Hash[Symbol, Integer]
+
+  def clean_word_hash: () -> Hash[Symbol, Integer]
+
+  def summary: (?Integer count, ?String separator) -> String
+
+  def paragraph_summary: (?Integer count, ?String separator) -> String
+
+  def split_sentences: () -> Array[String]
+
+  def split_paragraphs: () -> Array[String]
+
+  private
+
+  def word_hash_for_words: (Array[String] words) -> Hash[Symbol, Integer]
+
+  def word_hash_for_symbols: (Array[String] words) -> Hash[Symbol, Integer]
+
+  def perform_lsi: (Array[String] chunks, Integer count, String separator) -> String
+end

sig/classifier/extensions/vector.rbs

@@ -0,0 +1,23 @@
+class Array[unchecked out Elem]
+  def sum_with_identity: (?Float identity) ?{ (Elem) -> Numeric } -> Float
+end
+
+module VectorExtensions
+  def magnitude: () -> Float
+
+  def normalize: () -> Vector[Rational]
+end
+
+class Vector[out Elem]
+  include VectorExtensions
+end
+
+class Matrix[out Elem]
+  def self.diag: (Array[Numeric] diagonal_elements) -> Matrix[Numeric]
+
+  def trans: () -> Matrix[Elem]
+
+  def SV_decomp: (?Integer max_sweeps) -> [Matrix[Numeric], Matrix[Numeric], Array[Float]]
+
+  def []=: (Integer row_index, Integer col_index, Numeric value) -> Numeric
+end

sig/classifier/lsi.rbs

@@ -0,0 +1,57 @@
+module Classifier
+  class LSI
+    @auto_rebuild: bool
+    @word_list: WordList
+    @items: Hash[untyped, ContentNode]
+    @version: Integer
+    @built_at_version: Integer
+
+    attr_reader word_list: WordList
+    attr_accessor auto_rebuild: bool
+
+    def self.gsl_available: () -> bool
+    def self.gsl_available=: (bool value) -> bool
+
+    def initialize: (?Hash[Symbol, untyped] options) -> void
+
+    def needs_rebuild?: () -> bool
+
+    def add_item: (untyped item, *untyped categories) ?{ (untyped) -> String } -> void
+
+    def <<: (untyped item) -> void
+
+    def categories_for: (untyped item) -> Array[untyped]
+
+    def remove_item: (untyped item) -> void
+
+    def items: () -> Array[untyped]
+
+    def build_index: (?Float cutoff) -> void
+
+    def highest_relative_content: (?Integer max_chunks) -> Array[untyped]
+
+    def proximity_array_for_content: (untyped doc) ?{ (untyped) -> String } -> Array[[untyped, Float]]
+
+    def proximity_norms_for_content: (untyped doc) ?{ (untyped) -> String } -> Array[[untyped, Float]]
+
+    def search: (String string, ?Integer max_nearest) -> Array[untyped]
+
+    def find_related: (untyped doc, ?Integer max_nearest) ?{ (untyped) -> String } -> Array[untyped]
+
+    def classify: (untyped doc, ?Float cutoff) ?{ (untyped) -> String } -> untyped
+
+    def vote: (untyped doc, ?Float cutoff) ?{ (untyped) -> String } -> Hash[untyped, Float]
+
+    def classify_with_confidence: (untyped doc, ?Float cutoff) ?{ (untyped) -> String } -> [untyped, Float?]
+
+    def highest_ranked_stems: (untyped doc, ?Integer count) -> Array[Symbol]
+
+    private
+
+    def build_reduced_matrix: (untyped matrix, ?Float cutoff) -> untyped
+
+    def node_for_content: (untyped item) ?{ (untyped) -> String } -> ContentNode
+
+    def make_word_list: () -> void
+  end
+end

sig/classifier/lsi/content_node.rbs

@@ -0,0 +1,22 @@
+module Classifier
+  class ContentNode
+    @word_hash: Hash[Symbol, Integer]
+    @categories: Array[untyped]
+
+    attr_accessor raw_vector: untyped
+    attr_accessor raw_norm: untyped
+    attr_accessor lsi_vector: untyped
+    attr_accessor lsi_norm: untyped
+    attr_accessor categories: Array[untyped]
+
+    attr_reader word_hash: Hash[Symbol, Integer]
+
+    def initialize: (Hash[Symbol, Integer] word_frequencies, *untyped categories) -> void
+
+    def search_vector: () -> untyped
+
+    def search_norm: () -> untyped
+
+    def raw_vector_with: (WordList word_list) -> untyped
+  end
+end

sig/classifier/lsi/word_list.rbs

@@ -0,0 +1,15 @@
+module Classifier
+  class WordList
+    @location_table: Hash[Symbol, Integer]
+
+    def initialize: () -> void
+
+    def add_word: (Symbol word) -> Integer?
+
+    def []: (Symbol lookup) -> Integer?
+
+    def word_for_index: (Integer ind) -> Symbol?
+
+    def size: () -> Integer
+  end
+end