active_harness 0.3.3 → 0.3.5

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 4297f5db3bd99d90d12679afcc742998ed815606ee25e3c0490187370fad8e2d
-  data.tar.gz: c2955dc8a5391fcf13fc3752ba490222ea8d55f7127cc90f8aee93937b1bac91
+  metadata.gz: 348fea391f00283a45f57b119081ec4c820559e5df34651f988f3d5c68538d03
+  data.tar.gz: a5be81902bc6ff447ec64e434a08a25bc9ea193f8d89706cc6e3ec1a7514007e
 SHA512:
-  metadata.gz: 731130020cff7ed11514659b2cfc7b9dd6aed0e6a935e9ff198ebf695b44065a7e1a139da1824178b289c1aa13d75e8aaba5b32373b781b4f9639a61d9c5135e
-  data.tar.gz: 6240251c55447e9a56d3c87d704b441d447ad3c8604b09651c2dfbb9577e20c11165bb4fba740bdf9c0cb386adca2c0ab52a80485c7410d56d1285fddf86ff61
+  metadata.gz: 1bd229ceed23615fbb1d59de9d4a8747722eb19849791d9911b218ee474672b6fdf3779f6bb6449af55150b6a8ad99bc5437e106789dc4f85550cc97077c7cca
+  data.tar.gz: 6011181b558cfb445f8479e27dc19af4086daa38c994c2d57d61c66c15228638b8a18dee842524c1e590d855b029235eae80a9929e8385874876972d1aea4bb9

data/lib/active_harness/providers/audio/openai.rb

@@ -27,7 +27,18 @@ module ActiveHarness
         # Synchronous — this endpoint has no job/polling API. Unlike OpenRouter's
         # transcription endpoint, OpenAI's is multipart/form-data only (no
         # base64/JSON request mode).
-        def call(model:, audio_data:, audio_format:, language: nil, **_)
+        # @param response_format [String] "json" (default, plain text) or,
+        #   for gpt-4o-transcribe-diarize, "diarized_json" — returns per-segment
+        #   speaker/text/start/end instead of just plain text.
+        # @param chunking_strategy [String, Hash] required by
+        #   gpt-4o-transcribe-diarize for any audio over 30 seconds; "auto" is
+        #   the simple/common value, or a voice-activity-detection config hash.
+        # @param known_speaker_names [Array<String>] up to 4 labels, paired
+        #   positionally with known_speaker_references, for gpt-4o-transcribe-diarize.
+        # @param known_speaker_references [Array<String>] up to 4 short (2-10s)
+        #   reference clips as base64 data URLs, e.g. "data:audio/wav;base64,...".
+        def call(model:, audio_data:, audio_format:, language: nil, response_format: nil,
+                 chunking_strategy: nil, known_speaker_names: nil, known_speaker_references: nil, **_)
           content_type = CONTENT_TYPES[audio_format]
           unless content_type
             raise Errors::InvalidRequestError,
@@ -36,9 +47,16 @@ module ActiveHarness
 
           boundary = SecureRandom.hex(16)
           fields   = { "model" => model }
-          fields["language"] = language if language
+          fields["language"]          = language if language
+          fields["response_format"]   = response_format if response_format
+          fields["chunking_strategy"] = chunking_strategy if chunking_strategy
 
-          body = build_multipart_body(boundary, fields, audio_data, "audio.#{audio_format}", content_type)
+          array_fields = {
+            "known_speaker_names[]"      => known_speaker_names,
+            "known_speaker_references[]" => known_speaker_references
+          }
+
+          body = build_multipart_body(boundary, fields, audio_data, "audio.#{audio_format}", content_type, array_fields: array_fields)
           headers = {
             "Content-Type"  => "multipart/form-data; boundary=#{boundary}",
             "Authorization" => "Bearer #{api_key}"
@@ -48,15 +66,24 @@ module ActiveHarness
           data = parse!(raw)
           handle_error!(data)
 
-          text = data["text"]
-          raise Errors::ProviderError, "No transcription text in response: #{data.keys}" if text.nil?
+          # diarized_json returns a structured { segments: [...] } shape (each
+          # with speaker/text/start/end) — hand the whole thing back as JSON
+          # text so `format :json` on the Request subclass parses it into a
+          # Hash. Plain "json"/unset keeps the existing behavior: just the
+          # flat text.
+          content =
+            if response_format == "diarized_json"
+              data.to_json
+            else
+              data["text"].tap { |t| raise Errors::ProviderError, "No transcription text in response: #{data.keys}" if t.nil? }
+            end
 
-          { content: text, provider: :openai, model: model, usage: extract_transcription_usage(data) }
+          { content: content, provider: :openai, model: model, usage: extract_transcription_usage(data) }
         end
 
         private
 
-        def build_multipart_body(boundary, fields, file_data, filename, content_type)
+        def build_multipart_body(boundary, fields, file_data, filename, content_type, array_fields: {})
           body = +""
           fields.each do |name, value|
             body << "--#{boundary}\r\n"
@@ -64,6 +91,14 @@ module ActiveHarness
             body << "#{value}\r\n"
           end
 
+          array_fields.each do |name, values|
+            Array(values).each do |value|
+              body << "--#{boundary}\r\n"
+              body << "Content-Disposition: form-data; name=\"#{name}\"\r\n\r\n"
+              body << "#{value}\r\n"
+            end
+          end
+
           body << "--#{boundary}\r\n"
           body << "Content-Disposition: form-data; name=\"file\"; filename=\"#{filename}\"\r\n"
           body << "Content-Type: #{content_type}\r\n\r\n"

data/lib/active_harness/providers/audio/openrouter.rb

@@ -13,11 +13,22 @@ module ActiveHarness
         # @param audio_data   [String]  raw binary audio bytes
         # @param audio_format [String]  "wav", "mp3", "flac", "m4a", "ogg", "webm", "aac"
         # @param language     [String]  ISO-639-1 code, e.g. "en" (optional — auto-detected if omitted)
+        # @param response_format [String] "json" (default, plain text) or "verbose_json"
+        #   — the latter returns per-segment start/end timestamps (and a "speaker"
+        #   index per segment when the provider supports diarization).
+        # @param timestamp_granularities [Array<String>] e.g. ["segment", "word"] —
+        #   only meaningful with response_format: "verbose_json".
+        # @param provider_options [Hash] passed through as-is to the request body's
+        #   `provider.options` — this is how provider-specific features (e.g. Azure
+        #   speaker diarization) are enabled: { azure: { diarization: { enabled: true } } }.
+        #   Only specific provider/model combos support this; most will just ignore
+        #   an option they don't recognize, but some may 400 on it.
         #
         # Synchronous call — OpenRouter's transcription endpoint has no job/polling
         # API. Upstream providers time out after ~60s per request, so long audio
         # should be split into shorter chunks by the caller before transcribing.
-        def call(model:, audio_data:, audio_format:, language: nil, **_)
+        def call(model:, audio_data:, audio_format:, language: nil, response_format: nil,
+                 timestamp_granularities: nil, provider_options: nil, **_)
           headers = {
             "Content-Type"  => "application/json",
             "Authorization" => "Bearer #{api_key}"
@@ -29,16 +40,27 @@ module ActiveHarness
             model: model,
             input_audio: { data: Base64.strict_encode64(audio_data), format: audio_format }
           }
-          body[:language] = language if language
+          body[:language]                 = language if language
+          body[:response_format]          = response_format if response_format
+          body[:timestamp_granularities]  = timestamp_granularities if timestamp_granularities
+          body[:provider]                 = { options: provider_options } if provider_options
 
           raw  = post_json(URI(ENDPOINT), headers: headers, body: body, timeout: 90)
           data = parse!(raw)
           handle_error!(data)
 
-          text = data["text"]
-          raise Errors::ProviderError, "No transcription text in response: #{data.keys}" if text.nil?
+          # verbose_json returns a richer structure (language/duration/segments with
+          # timestamps and, when supported, speaker) — hand the whole thing back as
+          # JSON text so `format :json` on the Request subclass parses it into a Hash.
+          # Plain "json"/unset keeps the existing behavior: just the flat text.
+          content =
+            if response_format == "verbose_json"
+              data.to_json
+            else
+              data["text"].tap { |t| raise Errors::ProviderError, "No transcription text in response: #{data.keys}" if t.nil? }
+            end
 
-          { content: text, provider: :openrouter, model: model, usage: extract_transcription_usage(data) }
+          { content: content, provider: :openrouter, model: model, usage: extract_transcription_usage(data) }
         end
 
         private

data/lib/active_harness/request/models.rb

@@ -134,17 +134,25 @@ module ActiveHarness
       @models = []
     end
 
-    def use(provider:, model:, temperature: nil, name: nil, size: nil, quality: nil, retry_attempts: nil, retry_delay: nil, questions: nil)
+    def use(provider:, model:, temperature: nil, name: nil, size: nil, quality: nil, retry_attempts: nil, retry_delay: nil,
+            questions: nil, response_format: nil, timestamp_granularities: nil, provider_options: nil,
+            chunking_strategy: nil, known_speaker_names: nil, known_speaker_references: nil)
       @models << {
-        provider:       provider,
-        model:          model,
-        temperature:    temperature,
-        name:           name,
-        size:           size,
-        quality:        quality,
-        retry_attempts: retry_attempts,
-        retry_delay:    retry_delay,
-        questions:      questions
+        provider:                 provider,
+        model:                    model,
+        temperature:              temperature,
+        name:                     name,
+        size:                     size,
+        quality:                  quality,
+        retry_attempts:           retry_attempts,
+        retry_delay:              retry_delay,
+        questions:                questions,
+        response_format:          response_format,
+        timestamp_granularities:  timestamp_granularities,
+        provider_options:         provider_options,
+        chunking_strategy:        chunking_strategy,
+        known_speaker_names:      known_speaker_names,
+        known_speaker_references: known_speaker_references
       }.compact
     end
 

data/lib/active_harness/request/providers.rb

@@ -100,7 +100,13 @@ module ActiveHarness
 
       opts = { model: entry[:model], audio_data: File.binread(path), audio_format: format }
       language = entry[:language] || @config[:transcribe_language]
-      opts[:language] = language if language
+      opts[:language]                 = language if language
+      opts[:response_format]          = entry[:response_format] if entry[:response_format]
+      opts[:timestamp_granularities]  = entry[:timestamp_granularities] if entry[:timestamp_granularities]
+      opts[:provider_options]         = entry[:provider_options] if entry[:provider_options]
+      opts[:chunking_strategy]        = entry[:chunking_strategy] if entry[:chunking_strategy]
+      opts[:known_speaker_names]      = entry[:known_speaker_names] if entry[:known_speaker_names]
+      opts[:known_speaker_references] = entry[:known_speaker_references] if entry[:known_speaker_references]
       factory.call.call(**opts)
     end
 

data/lib/active_harness/request.rb

@@ -140,6 +140,7 @@ module ActiveHarness
         model:          entry[:model],
         error:          error.message,
         error_code:     error.respond_to?(:error_code) ? error.error_code : nil,
+        error_metadata: error.respond_to?(:metadata) ? error.metadata : nil,
         execution_time: elapsed
       }
     end

data/lib/active_harness.rb

@@ -35,7 +35,7 @@ require_relative "active_harness/pipeline"
 require_relative "active_harness/railtie" if defined?(Rails::Railtie)
 
 module ActiveHarness
-  VERSION = "0.3.3"
+  VERSION = "0.3.5"
 
   class << self
     # Configure ActiveHarness.

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: active_harness
 version: !ruby/object:Gem::Version
-  version: 0.3.3
+  version: 0.3.5
 platform: ruby
 authors:
 - the-teacher
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2026-09-25 00:00:00.000000000 Z
+date: 2026-10-03 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: concurrent-ruby