cohere-transcribe 0.1.2 → 0.1.3

data/lib/cohere/transcribe/types.rb

@@ -1,10 +1,13 @@
 # frozen_string_literal: true
 
+require_relative "constants"
+require_relative "internal/utf8"
+
 module Cohere
   module Transcribe
     module TypesSupport
-      DEFAULT_ASR_MODEL_ID = "CohereLabs/cohere-transcribe-arabic-07-2026"
-      OUTPUT_FORMATS = %w[txt srt vtt json].freeze
+      DEFAULT_ASR_MODEL_ID = Cohere::Transcribe::DEFAULT_ASR_MODEL_ID
+      OUTPUT_FORMATS = Cohere::Transcribe::OUTPUT_FORMATS
       EXISTING_POLICIES = %w[error overwrite skip].freeze
 
       module KeywordOnlyConstructor
@@ -47,10 +50,21 @@ module Cohere
         immutable(array)
       end
 
-      def path(value)
+      def reference(value)
+        source = value.respond_to?(:to_path) ? value.to_path : value
+        return immutable(value) unless source.is_a?(String)
+
+        text = Internal::UTF8.normalize(source)
+        text ? text.freeze : immutable(source)
+      end
+
+      def path(value, field: "path")
         source = value.respond_to?(:to_path) ? value.to_path : value
-        pathname = Pathname.new(source)
-        text = pathname.to_s
+        raise TypeError, "#{field} must resolve to a text path" unless source.is_a?(String)
+
+        text = Internal::UTF8.normalize(source)
+        raise ArgumentError, "#{field} must contain valid UTF-8" unless text
+
         absolute = text.start_with?("/")
         prefix = if text.start_with?("//") && !text.start_with?("///")
                    "//"
@@ -84,9 +98,9 @@ module Cohere
 
         super(
           formats: formats,
-          output_dir: output_dir.nil? ? nil : TypesSupport.path(output_dir),
+          output_dir: output_dir.nil? ? nil : TypesSupport.path(output_dir, field: "output_dir"),
           existing: TypesSupport.immutable(existing),
-          profile_json: profile_json.nil? ? nil : TypesSupport.path(profile_json)
+          profile_json: profile_json.nil? ? nil : TypesSupport.path(profile_json, field: "profile_json")
         )
       end
     end
@@ -183,10 +197,10 @@ module Cohere
         publication: nil
       )
         super(
-          model: TypesSupport.immutable(model),
-          model_revision: TypesSupport.immutable(model_revision),
-          adapter: TypesSupport.immutable(adapter),
-          adapter_revision: TypesSupport.immutable(adapter_revision),
+          model: TypesSupport.reference(model),
+          model_revision: TypesSupport.reference(model_revision),
+          adapter: TypesSupport.reference(adapter),
+          adapter_revision: TypesSupport.reference(adapter_revision),
           language: TypesSupport.immutable(language),
           text_only: text_only,
           recursive: recursive,

data/lib/cohere/transcribe/version.rb

@@ -2,6 +2,6 @@
 
 module Cohere
   module Transcribe
-    VERSION = "0.1.2"
+    VERSION = "0.1.3"
   end
 end

data/sig/cohere/transcribe.rbs

@@ -5,6 +5,7 @@ module Cohere
     DEFAULT_ASR_MODEL_ID: String
     DEFAULT_ASR_MODEL_REVISION: String
     SAMPLE_RATE: Integer
+    OUTPUT_FORMATS: Array[output_format]
 
     type path_input = String | _ToPath
     type audio_input = path_input | Array[path_input]

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: cohere-transcribe
 version: !ruby/object:Gem::Version
-  version: 0.1.2
+  version: 0.1.3
 platform: ruby
 authors:
 - Ali Hamdi Ali Fadel
@@ -88,6 +88,7 @@ files:
 - THIRD_PARTY_NOTICES.md
 - exe/cohere-transcribe
 - exe/cohere-transcribe-doctor
+- ext/cohere_transcribe_native/.gitignore
 - ext/cohere_transcribe_native/CMakeLists.txt
 - ext/cohere_transcribe_native/README.md
 - ext/cohere_transcribe_native/audio_abi.cpp
@@ -130,6 +131,9 @@ files:
 - lib/cohere/transcribe/gguf_writer.rb
 - lib/cohere/transcribe/hub.rb
 - lib/cohere/transcribe/input.rb
+- lib/cohere/transcribe/internal/interruptible_native_call.rb
+- lib/cohere/transcribe/internal/session_ownership.rb
+- lib/cohere/transcribe/internal/utf8.rb
 - lib/cohere/transcribe/licenses/crispasr.txt
 - lib/cohere/transcribe/loader.rb
 - lib/cohere/transcribe/model_identity.rb