podcast-buddy 0.1.1 → 0.2.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 9821fea9e037f8483457b4e8fc90e453a81f7ae90505b67b537683705eb57733
-  data.tar.gz: e0ef8feff9e801347a99beda8432ab46d92498aa9795408afaaf9167317cce83
+  metadata.gz: 6ebed9ffec7e1ce399ed7b64d332087e433b8348e84b1af2fc8a13a959163802
+  data.tar.gz: b3d9fe281bf8259bcc9b65ef466fe517885e3855fd95da0672ab93f7a08f80f0
 SHA512:
-  metadata.gz: 8b80d6502926859d82ed679ddcddf6dad8bdbfd6507e8936652861dac44f03b6a92d7fc06c06a93cc7886b0e2a9d5c126b72cd618c32015c3c61f77eee01819d
-  data.tar.gz: aa35a927a14fd5e48df5ec6a1d09bb5a262bfc06f60b3c75f9dfe3dcf7e672f4074d4d7bf77c96da09bcdfebd1dd608cf1245cd59cc7e605462a8a32ea716004
+  metadata.gz: b53e802df3a99224404789e1c9acb061b2d4dcc2ef57927ba0b2a0cf3a0aaf62fa8e115d148c27a62407c44851853b39fc2955d2e4748a2ecf869fa4cb926c43
+  data.tar.gz: 230641523d5daf007ac70ea44d41d2dad49a12c28741c26b78e348b1e573e5ffe24c707fd76c505ba64dee25e783c9c51722cb8bf4272410aaab41cef4c54fae

data/CHANGELOG.md

@@ -1,5 +1,24 @@
 ## [Unreleased]
 
+## [0.2.1] - 2024-08-13
+
+* Fixes typo in command-line help banner to reference proper podcast_buddy command
+* Updates README with new options
+
+## [0.2.0] - 2024-08-13
+
+* 9edc409 - Refactor to make use of Async
+  * Now properly generates show notes on exit
+* 7d8db13 - Update show notes prompt to cite sources
+* c3ef97d - Allow configuring the whisper model to use
+* e7cb278 - Adds session naming to store files in specific directories
+* 39e9e78 - Overwrite summary on every periodic update
+* 0518b56 - Update topic extraction prompt to fix 'Empty response' topics
+* b72ba8f - Use brew list -1 to /dev/null to quiet initial outputs when already setup
+* 98a2484 - Remove no longer needed commented code in Signal.trap now that we're Ansync
+* 62cf6ea - Update periodic summarize/topic calls to use new PodcastBuddy.current_transcript method, and update other current_* methods to re-use the memoized versions
+
+
 ## [0.1.1] - 2024-08-09
 
 - Fixes typo in topic extraction prompt ([7e07e](https://github.com/codenamev/podcast-buddy/commit/7e07e307135c95cb4bb68dadc354f0b2519c7721))

data/README.md

@@ -44,9 +44,17 @@ Once you're done, simply `ctrl-c` to wrap things up.
 
 ### Helpful files logged during your session with Buddy
 
-1. Your full transcript is stored in `tmp/transcript.log`
-2. A summarization of the discussion and topics are stored in `tmp/discussion.log`
-3. The raw whisper logs are stored in `tmp/whisper.log`
+1. Your full transcript is stored in `tmp/transcripti-%Y-%m-%d.log`.
+2. A summarization of the discussion is stored in `tmp/summary-%Y-%m-%d.log`.
+3. A list of topics extracted from the discussion is stored in `tmp/topics-%Y-%m-%d.log`.
+4. The Show Notes are stored in `tmp/show-notes-%Y-%m-%d.log`.
+5. The raw whisper logs are stored in `tmp/whisper.log`.
+
+### Options
+
+**debug mode**: `podcast_buddy --debug` – shows verbose logging
+**custom whisper model**: `podcast_buddy --whisper base.en` – use any of [these available models](https://github.com/ggerganov/whisper.cpp/blob/master/models/download-ggml-model.sh#L28-L49).
+**custom session**: `podcast_buddy --name "Ruby Rogues 08-15-2024"` – saves files to a new `tmp/Ruby Rogues 08-15-2024/` directory.
 
 ## Development
 

data/exe/podcast_buddy

@@ -2,16 +2,42 @@
 
 require_relative "../lib/podcast_buddy"
 require "optparse"
+require "rainbow"
+require "timeout"
+require "fileutils"
 
 options = {}
 OptionParser.new do |opts|
-  opts.banner = "Usage: podcast-buddy [options]"
+  opts.banner = "Usage: podcast_buddy [options]"
 
   opts.on("--debug", "Run in debug mode") do |v|
     options[:debug] = v
   end
+
+  opts.on("-w", "--whisper MODEL", "Use specific whisper model (default: small.en)") do |v|
+    options[:whisper_model] = v
+  end
+
+  opts.on("-n", "--name NAME", "A name for the session to label all log files") do |v|
+    options[:name] = v
+  end
 end.parse!
 
+def to_human(text, label = :info)
+  case label.to_sym
+  when :info
+    Rainbow(text).blue
+  when :wait
+    Rainbow(text).yellow
+  when :input
+    Rainbow(text).black.bg(:yellow)
+  when :success
+    Rainbow(text).green
+  else
+    text
+  end
+end
+
 PodcastBuddy.logger.formatter = proc do |severity, datetime, progname, msg|
   if severity.to_s == "INFO"
     "#{msg}\n"
@@ -21,212 +47,300 @@ PodcastBuddy.logger.formatter = proc do |severity, datetime, progname, msg|
 end
 
 if options[:debug]
-  PodcastBuddy.logger.info "Turning on debug mode..."
+  PodcastBuddy.logger.info to_human("Turning on debug mode...", :info)
   PodcastBuddy.logger.level = Logger::DEBUG
 else
   PodcastBuddy.logger.level = Logger::INFO
 end
 
+if options[:whisper_model]
+  PodcastBuddy.whisper_model = options[:whisper_model]
+  PodcastBuddy.logger.info to_human("Using whisper model: #{options[:whisper_model]}", :info)
+end
+
+if options[:name]
+  base_path = "#{PodcastBuddy.root}/tmp/#{options[:name]}"
+  FileUtils.mkdir_p base_path
+  PodcastBuddy.session = options[:name]
+  PodcastBuddy.logger.info to_human("Using custom session name: #{options[:name]}", :info)
+  PodcastBuddy.logger.info to_human("  Saving files to: #{PodcastBuddy.session}", :info)
+end
+
 PodcastBuddy.logger.info "Setting up dependencies..."
 PodcastBuddy.setup
 PodcastBuddy.logger.info "Setup complete."
 
 # OpenAI API client setup
-raise "Please set an OPENAI_ACCESS_TOKEN environment variable." if ENV["OPENAI_ACCESS_TOKEN"].to_s.strip.squeeze.empty?
-
-client = OpenAI::Client.new(access_token: ENV["OPENAI_ACCESS_TOKEN"], log_errors: true)
-
-# Signals for question detection
-question_signal = PodcastBuddy::PodSignal.new
-end_signal = PodcastBuddy::PodSignal.new
+# Raises error if no OPENAI_ACCESS_TOKEN env is set
+PodcastBuddy.openai_client
 
 # Initialize variables
 @latest_transcription = ""
-@question_transcription = ""
-@full_transcription = ""
+@full_transcript = ""
 @listening_for_question = false
-@shutdown_flag = false
+@shutdown = false
 @threads = []
 @transcription_queue = Queue.new
+@question_transcript = Queue.new
 
 # Method to extract topics and summarize
-def extract_topics_and_summarize(client, text)
-  current_discussion = File.exist?(PodcastBuddy.discussion_log_file) ? File.read(PodcastBuddy.discussion_log_file) : ""
-  current_discussion.rstrip.squeeze!
-  messages = []
-  if current_discussion.length > 1
-    messages << {role: "user", content: "Given the prior summary and topics:\n---\n#{current_discussion}\n---"}
-    messages << {role: "user", content: "Continue the running summary and list of topics using the following discussion:\n---#{text}\n---"}
-  else
-    messages << {role: "user", content: "Extract topics and summarize the following discussion:\n---#{text}\n---"}
+def extract_topics_and_summarize(text)
+  Sync do |parent|
+    parent.async { update_topics(text) }
+    parent.async { update_summary(text) }
   end
-  response = client.chat(parameters: {
-    model: "gpt-4o",
-    messages: messages,
-    max_tokens: 150
-  })
-  response.dig("choices", 0, "message", "content").strip
 end
 
-def answer_question(client, question)
-  latest_context = `tail -n 25 #{PodcastBuddy.discussion_log_file}`
-  previous_discussion = @full_transcription.split("\n").last(25)
-  PodcastBuddy.logger.info "Answering question:\n#{question}"
-  PodcastBuddy.logger.debug "Context:\n---#{latest_context}\n---\nPrevious discussion:\n---#{previous_discussion}\n---\nAnswering question:\n---#{question}\n---"
-  response = client.chat(parameters: {
-    model: "gpt-4o",
-    messages: [
-      {role: "system", content: "You are a kind and helpful podcast assistant helping to answer questions as them come up during a recording. Keep the answer succinct and without speculation."},
-      {role: "user", content: "Context:\n#{latest_context}\n\nPrevious Discussion:\n#{previous_discussion}\n\nQuestion:\n#{question}\n\nAnswer:"}
-    ],
-    max_tokens: 150
-  })
-  response.dig("choices", 0, "message", "content").strip
+def update_topics(text)
+  Sync do
+    PodcastBuddy.logger.debug "Looking for topics related to: #{text}"
+    response = PodcastBuddy.openai_client.chat(parameters: {
+      model: "gpt-4o",
+      messages: [
+        {role: "system", content: "You are a kind and helpful podcast assistant diligantly extracting topics related to the discussion on the show. Your goal is to prepare this information in an optimal way for participants to ask questions in the context of what is being discussed, and optimally for use in the closing Show Notes."},
+        {role: "user", content: "Extract topics for the following Discussion:\n---\n#{text}\n---\n\nExclude any of these topics that have been previously discussed:\n---\n#{PodcastBuddy.current_topics}\n---\n\nIf there are no new topics to add, return NONE."}
+      ],
+      max_tokens: 250
+    })
+    new_topics = response.dig("choices", 0, "message", "content").gsub("NONE", "").strip
+    PodcastBuddy.logger.debug "Updating topics: #{new_topics.split("\n").join(", ")}"
+    PodcastBuddy.add_to_topics(new_topics)
+  end
+end
+
+def update_summary(text)
+  Sync do
+    response = PodcastBuddy.openai_client.chat(parameters: {
+      model: "gpt-4o",
+      messages: [
+        {role: "system", content: "You are a kind and helpful podcast assistant diligantly keeping track of the overall discussion on the show. Your goal is to prepare a concise summary of the overall discussion."},
+        {role: "user", content: "Previous Summary:\n---\n#{PodcastBuddy.current_summary}\n---\n\nLatest Discussion:\n---\n#{text}\n---"},
+        {role: "user", content: "Use the Previous Summary, and Latest Discussion, to combine into a single cohesive summary.\nSummary: "}
+      ],
+      max_tokens: 250
+    })
+    new_summary = response.dig("choices", 0, "message", "content").strip
+    PodcastBuddy.logger.debug "Updating summary: #{new_summary}"
+    PodcastBuddy.update_summary(new_summary)
+  end
+end
+
+def answer_question
+  Async do
+    latest_context = "#{PodcastBuddy.current_summary}\nTopics discussed recently:\n---\n#{PodcastBuddy.current_topics.split("\n").last(10)}\n---\n"
+    previous_discussion = @full_transcript.split("\n").last(25)
+    question = ""
+    question << @question_transcript.pop until @question_transcript.empty?
+    PodcastBuddy.logger.info "Answering question:\n#{question}"
+    PodcastBuddy.logger.debug "Context:\n---#{latest_context}\n---\nPrevious discussion:\n---#{previous_discussion}\n---\nAnswering question:\n---#{question}\n---"
+    response = PodcastBuddy.openai_client.chat(parameters: {
+      model: "gpt-4o",
+      messages: [
+        {role: "system", content: "You are a kind and helpful podcast assistant helping to answer questions as they come up during a recording. Keep the answer succinct and without speculation. Be playful, witty, and kind.  If a question is not asked try to add to the conversation as it best relates to the current discussion. Remember, this is a live recording, so there's no need to analyze and report on what is being asked."},
+        {role: "user", content: "Context:\n#{latest_context}\n\nPrevious Discussion:\n#{previous_discussion}\n\nQuestion:\n#{question}\n\nAnswer:"}
+      ],
+      max_tokens: 150
+    })
+    answer = response.dig("choices", 0, "message", "content").strip
+    PodcastBuddy.logger.debug "Answer: #{answer}"
+    text_to_speech(answer)
+    PodcastBuddy.logger.debug("Answer converted to speech: #{PodcastBuddy.answer_audio_file_path}")
+    play_answer
+  end
 end
 
 # Method to convert text to speech
-def text_to_speech(client, text)
-  response = client.audio.speech(parameters: {
+def text_to_speech(text)
+  response = PodcastBuddy.openai_client.audio.speech(parameters: {
     model: "tts-1",
     input: text,
     voice: "onyx",
     response_format: "mp3",
     speed: 1.0
   })
-  File.binwrite("response.mp3", response)
+  File.binwrite(PodcastBuddy.answer_audio_file_path, response)
+end
+
+def play_answer
+  PodcastBuddy.logger.debug("Playing answer...")
+  system("afplay #{PodcastBuddy.answer_audio_file_path}")
 end
 
 # Method to handle audio stream processing
-def process_audio_stream(client)
-  Thread.new do
-    whisper_command = "./whisper.cpp/stream -m ./whisper.cpp/models/ggml-#{PodcastBuddy.whisper_model}.bin -t 4 --step 0 --length 5000 --keep 500 --vad-thold 0.60 --audio-ctx 0 --keep-context -c 1"
+def process_audio_stream
+  Sync do |parent|
     PodcastBuddy.logger.info "Buffering audio..."
-    Open3.popen3(whisper_command) do |stdin, stdout, stderr, thread|
-      stderr_thread = Thread.new do
-        stderr.each { |line| PodcastBuddy.whisper_logger.error line }
-      end
-      stdout_thread = Thread.new do
-        stdout.each { |line| PodcastBuddy.whisper_logger.debug line }
-      end
+    Open3.popen3(PodcastBuddy.whisper_command) do |stdin, stdout, stderr, _thread|
+      error_log_task = parent.async { stderr.each { |line| PodcastBuddy.whisper_logger.error line } }
+      output_log_task = parent.async { stdout.each { |line| PodcastBuddy.whisper_logger.debug line } }
 
       while (transcription = stdout.gets)
-        # Cleanup captured text from whisper
-        transcription_text = (transcription.scan(/\[[0-9]+.*[0-9]+\]\s?(.*)\n/) || [p])[0].to_s
-        PodcastBuddy.logger.debug "Received transcription: #{transcription_text}"
-        transcription_text = transcription_text.gsub("[BLANK_AUDIO]", "")
-        transcription_text = transcription_text.gsub(/\A\["\s?/, "")
-        transcription_text = transcription_text.gsub(/"\]\Z/, "")
-        break if @shutdown_flag
-
-        PodcastBuddy.logger.debug "Filtered: #{transcription_text}"
-        next if transcription_text.to_s.strip.squeeze.empty?
-
-        @full_transcription += transcription_text
-
-        if @listening_for_question
-          @question_transcription += transcription_text
-          PodcastBuddy.logger.info "Heard Question: #{transcription_text}"
-        else
-          @transcription_queue << transcription_text
-          PodcastBuddy.logger.info "Heard: #{transcription_text}"
-          PodcastBuddy.update_transcript(transcription_text)
-        end
+        PodcastBuddy.logger.debug("Shutdown: process_audio_stream...") and break if @shutdown
+        PodcastBuddy.logger.debug("Still listening...")
+
+        last_transcription_task = parent.async { process_transcription(transcription) }
       end
 
-      # Close streams and join stderr thread on shutdown
-      stderr_thread.join
-      stdout_thread.join
-      stdin.close
-      stdout.close
-      stderr.close
+      error_log_task.wait
+      output_log_task.wait
+      last_transcription_task.wait
     end
   end
 end
 
+def process_transcription(line)
+  # Cleanup captured text from whisper
+  transcription_text = (line.scan(/\[[0-9]+.*[0-9]+\]\s?(.*)\n/) || [p])[0].to_s
+  PodcastBuddy.logger.debug "Received transcription: #{transcription_text}" if transcription_text
+  transcription_text = transcription_text.gsub("[BLANK_AUDIO]", "")
+  transcription_text = transcription_text.gsub(/\A\["\s?/, "")
+  transcription_text = transcription_text.gsub(/"\]\Z/, "")
+
+  return if transcription_text.to_s.strip.squeeze.empty?
+
+  @full_transcript += transcription_text
+
+  if @listening_for_question
+    PodcastBuddy.logger.info "Heard Question: #{transcription_text}"
+    @question_transcript << transcription_text
+  else
+    PodcastBuddy.logger.info "Heard: #{transcription_text}"
+    PodcastBuddy.update_transcript(transcription_text)
+    @transcription_queue << transcription_text
+  end
+end
+
 # Method to generate show notes
-def generate_show_notes(client, transcription)
-  summary = extract_topics_and_summarize(client, transcription)
-  File.open("show_notes.md", "w") do |file|
-    file.puts "# Show Notes"
-    file.puts summary
+def generate_show_notes
+  return if PodcastBuddy.current_transcript.strip.squeeze.empty?
+
+  PodcastBuddy.logger.info "Generting show notes..."
+  response = PodcastBuddy.openai_client.chat(parameters: {
+    model: "gpt-4o",
+    messages: [
+      {role: "system", content: "You are a kind and helpful podcast assistant helping to take notes for the show, and extract useful information being discussed for listeners."},
+      {role: "user", content: "Transcript:\n---\n#{PodcastBuddy.current_transcript}\n---\n\nTopics:\n---\n#{PodcastBuddy.current_topics}\n---\n\nUse the above transcript and topics to create Show Notes in markdown that outline the discussion.  Extract a breif summary that describes the overall conversation, the people involved and their roles, and sentiment of the topics discussed.  Follow the summary with a list of helpful links to any libraries, products, or other resources related to the discussion. Cite sources." }
+    ],
+    max_tokens: 150
+  })
+  show_notes = response.dig("choices", 0, "message", "content").strip
+  File.open(PodcastBuddy.show_notes_file, "w") do |file|
+    file.puts show_notes
   end
+
+  PodcastBuddy.logger.info to_human("Show notes saved to: #{PodcastBuddy.show_notes_file}", :success)
 end
 
 # Periodically summarize latest transcription
-def periodic_summarization(client, interval = 15)
-  Thread.new do
+def periodic_summarization(interval = 15)
+  Sync do
     loop do
-      break if @shutdown_flag
+      PodcastBuddy.logger.debug("Shutdown: periodic_summarization...") and break if @shutdown
 
       sleep interval
-      summarize_latest(client)
+      summarize_latest
     rescue => e
       PodcastBuddy.logger.warn "[summarization] periodic summarization failed: #{e.message}"
     end
   end
 end
 
-def summarize_latest(client = OpenAI::Client.new(access_token: ENV["OPENAI_ACCESS_TOKEN"], log_errors: true))
+def summarize_latest
   latest_transcriptions = []
   latest_transcriptions << @transcription_queue.pop until @transcription_queue.empty?
   return if latest_transcriptions.join.strip.squeeze.empty?
 
   PodcastBuddy.logger.debug "[periodic summarization] Latest transcript: #{latest_transcriptions.join}"
-  summary = extract_topics_and_summarize(client, latest_transcriptions.join)
-  File.write(PodcastBuddy.discussion_log_file, summary)
+  extract_topics_and_summarize(latest_transcriptions.join)
 end
 
-# Handle Ctrl-C to generate show notes
-Signal.trap("INT") do
-  puts "\nShutting down streams..."
-  @shutdown_flag = true
-  # Wait for all threads to finish
-  @threads.compact.each(&:join)
+def wait_for_question_end
+  Sync do |parent|
+    loop do
+      PodcastBuddy.logger.debug("Shutdown: wait_for_question_end...") and break if @shutdown
 
-  # Will have to re-think this.  Can't syncronize from a trap (Faraday)
-  # Also cannot log from trap
-  #puts "\nGenerating show notes..."
-  #generate_show_notes(client, @full_transcription)
-  exit
-end
+      # The wait_for_question_start will signal this
+      sleep 0.1 and next if !@listening_for_question
+
+      input = ""
+      Timeout.timeout(5) do
+        input = gets
+        PodcastBuddy.logger.debug("Input received...") if input.include?("\n")
+        next unless input.to_s.include?("\n")
+      rescue Timeout::Error
+        # Let the loop continue so that it can catch @shutdown periodically, or
+        # capture the signal
+        PodcastBuddy.logger.debug("Input timeout...")
+        next
+      end
 
-# Setup signal subscriptions
-question_signal.subscribe do
-  PodcastBuddy.logger.debug "Question signal received"
-  @listening_for_question = true
-  @question_transcription.clear
-  summarize_latest
+      if input.empty?
+        next
+      else
+        PodcastBuddy.logger.info "End of question signal.  Generating answer..."
+        @listening_for_question = false
+        answer_question.wait
+        PodcastBuddy.logger.info Rainbow("Press ").blue + Rainbow("Enter").black.bg(:yellow) + Rainbow(" to signal a question start...").blue
+        break
+      end
+    end
+  end
 end
 
-end_signal.subscribe do
-  PodcastBuddy.logger.info "End of question signal received"
-  @listening_for_question = false
+def wait_for_question_start
+  Sync do |parent|
+    PodcastBuddy.logger.info Rainbow("Press ").blue + Rainbow("Enter").black.bg(:yellow) + Rainbow(" to signal a question start...").blue
+    loop do
+      PodcastBuddy.logger.debug("Shutdown: wait_for_question...") and break if @shutdown
 
-  PodcastBuddy.logger.info "Waiting for question to complete (enter return again)..."
-  loop do
-    PodcastBuddy.logger.debug "Checking for question..."
-    sleep 0.3
-    break if @question_transcription.to_s.length > 0 || @listening_for_question == true
-  end
-  # Guard against interruptions
-  if !@question_transcription.empty?
-    response_text = answer_question(client, @question_transcription)
-    text_to_speech(client, response_text)
-    system("afplay response.mp3")
+      PodcastBuddy.logger.debug("Waiting for input...")
+      input = ""
+      Timeout.timeout(5) do
+        input = gets
+        PodcastBuddy.logger.debug("Input received...") if input.include?("\n")
+        @question_transcript.clear
+        @listening_for_question = true if input.include?("\n")
+      rescue Timeout::Error
+        # Let the loop continue so that it can catch @shutdown periodically, or
+        # capture the signal
+        next
+      end
+
+      next unless @listening_for_question
+
+      PodcastBuddy.logger.info to_human("🎙️ Listening for quesiton. Press ", :wait) + to_human("Enter", :input) + to_human(" to signal the end of the question...", :wait)
+      wait_for_question_end
+    end
   end
 end
 
+
+# Handle Ctrl-C to generate show notes
+Signal.trap("INT") do
+  puts to_human("\nShutting down streams...", :wait)
+  @shutdown = true
+end
+
 # Start audio stream processing
-@threads << process_audio_stream(client)
-# Start periodic summarization
-@threads << periodic_summarization(client, 60)
-
-# Main loop to wait for question and end signals
-loop do
-  PodcastBuddy.logger.info "Press Enter to signal a question start..."
-  gets
-  question_signal.trigger
-  PodcastBuddy.logger.info "Press Enter to signal the end of the question..."
-  gets
-  end_signal.trigger
+Async do |task|
+  audio_input_task = task.async { process_audio_stream }
+  periodic_summarization_task = task.async { periodic_summarization(60) }
+  question_listener_task = task.async { wait_for_question_start }
+
+  # Allow recording up to 2 hours, but hard-timeout so that we don't become
+  # zombies.
+  task.with_timeout(60*60*2) do
+    # Wait for shutdown sub-tasks to complete
+    task.yield until @shutdown
+  end
+
+  PodcastBuddy.logger.info "Waiting for all tasks to complete before exiting..."
+  audio_input_task.wait
+  PodcastBuddy.logger.debug "Audio input task complete."
+  question_listener_task.wait
+  PodcastBuddy.logger.debug "Question listener task complete."
+  periodic_summarization_task.wait
+  PodcastBuddy.logger.debug "Periodic Summarization task complete."
+
+  generate_show_notes
 end

data/lib/podcast_buddy/system_dependency.rb

@@ -15,12 +15,30 @@ module PodcastBuddy
       system_dependency.install unless system_dependency.installed?
     end
 
+    def self.resolve_whisper_model(model)
+      return if model_downloaded?(model)
+      download_model(model)
+    end
+
+    def self.model_downloaded?(model)
+      File.exist?(File.join(PodcastBuddy.root, "whisper.cpp", "models", "ggml-#{model}.bin"))
+    end
+
+    def self.download_model(model)
+      originating_directory = Dir.pwd
+      Dir.chdir("whisper.cpp")
+      PodcastBuddy.logger.info "Downloading GGML model: #{PodcastBuddy.whisper_model}"
+      PodcastBuddy.logger.info `bash ./models/download-ggml-model.sh #{PodcastBuddy.whisper_model}`
+    ensure
+      Dir.chdir(originating_directory)
+    end
+
     def installed?
       PodcastBuddy.logger.info "Checking for system dependency: #{name}..."
       if name.to_s == "whisper"
         Dir.exist?("whisper.cpp")
       else
-        system("brew list #{name}") || system("type -a #{name}")
+        system("brew list -1 #{name} > /dev/null") || system("type -a #{name}")
       end
     end
 

data/lib/podcast_buddy/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module PodcastBuddy
-  VERSION = "0.1.1"
+  VERSION = "0.2.1"
 end

data/lib/podcast_buddy.rb

@@ -1,10 +1,11 @@
-# frozen_string_literal: true
-
 require "net/http"
 require "uri"
 require "json"
 require "open3"
 require "logger"
+require "async"
+require "async/http/faraday"
+require "rainbow"
 
 require "bundler/inline"
 
@@ -19,6 +20,14 @@ module PodcastBuddy
     Dir.pwd
   end
 
+  def self.session=(name)
+    @session = "#{root}/tmp/#{name}"
+  end
+
+  def self.session
+    @session ||= "#{root}/tmp"
+  end
+
   def self.logger
     @logger ||= Logger.new($stdout, level: Logger::DEBUG)
   end
@@ -27,24 +36,92 @@ module PodcastBuddy
     @logger = logger
   end
 
+  def self.whisper_model=(model)
+    @whisper_model = model
+  end
+
   def self.whisper_model
-    "small.en"
+    @whisper_model ||= "small.en"
+  end
+
+  def self.whisper_command
+    "./whisper.cpp/stream -m ./whisper.cpp/models/ggml-#{PodcastBuddy.whisper_model}.bin -t 4 --step 0 --length 5000 --keep 500 --vad-thold 0.60 --audio-ctx 0 --keep-context -c 1"
+  end
+
+  def self.whisper_logger=(file_path)
+    @whisper_logger ||= Logger.new(file_path, "daily")
   end
 
   def self.whisper_logger
-    @whisper_logger ||= Logger.new("tmp/whisper.log", "daily")
+    @whisper_logger ||= Logger.new("#{session}/whisper.log", "daily")
   end
 
-  def self.discussion_log_file
-    @discussion_log_file ||= "#{root}/tmp/discussion-#{Time.new.strftime("%Y-%m-%d")}.log"
+  def self.topics_log_file=(log_file_path)
+    @topics_log_file = log_file_path
+  end
+
+  def self.topics_log_file
+    @topics_log_file ||= "#{session}/topics-#{Time.new.strftime("%Y-%m-%d")}.log"
+  end
+
+  def self.summary_log_file=(log_file_path)
+    @summary_log_file = log_file_path
+  end
+
+  def self.summary_log_file
+    @summary_log_file ||= "#{session}/summary-#{Time.new.strftime("%Y-%m-%d")}.log"
+  end
+
+  def self.transcript_file=(file_path)
+    @transcript_file = file_path
   end
 
   def self.transcript_file
-    @transcript_file ||= "#{root}/tmp/transcript-#{Time.new.strftime("%Y-%m-%d")}.log"
+    @transcript_file ||= "#{session}/transcript-#{Time.new.strftime("%Y-%m-%d")}.log"
+  end
+
+  def self.show_notes_file=(file_path)
+    @show_notes_file = file_path
+  end
+
+  def self.show_notes_file
+    @show_notes_file ||= "#{session}/show-notes-#{Time.new.strftime("%Y-%m-%d")}.md"
+  end
+
+  def self.current_transcript
+    @current_transcript ||= File.exist?(transcript_file) ? File.read(transcript_file) : ""
   end
 
   def self.update_transcript(text)
     File.open(transcript_file, "a") { |f| f.puts text }
+    current_transcript << text
+  end
+
+  def self.current_summary
+    @current_summary ||= File.exist?(summary_log_file) ? File.read(summary_log_file) : ""
+  end
+
+  def self.update_summary(text)
+    @current_summary = text
+    File.write(summary_log_file, text)
+    @current_summary
+  end
+
+  def self.current_topics
+    @current_topics ||= File.exist?(topics_log_file) ? File.read(topics_log_file) : ""
+  end
+
+  def self.add_to_topics(topics)
+    File.open(topics_log_file, "a") { |f| f.puts topics }
+    current_topics << topics
+  end
+
+  def self.answer_audio_file_path=(file_path)
+    @answer_audio_file_path = file_path
+  end
+
+  def self.answer_audio_file_path
+    @answer_audio_file_path ||= "response.mp3"
   end
 
   def self.setup
@@ -58,5 +135,16 @@ module PodcastBuddy
     SystemDependency.auto_install!(:git)
     SystemDependency.auto_install!(:sdl2)
     SystemDependency.auto_install!(:whisper)
+    SystemDependency.resolve_whisper_model(whisper_model)
+  end
+
+  def self.openai_client=(client)
+    @openai_client = client
+  end
+
+  def self.openai_client
+    raise Error, "Please set an OPENAI_ACCESS_TOKEN environment variable." if ENV["OPENAI_ACCESS_TOKEN"].to_s.strip.squeeze.empty?
+
+    @openai_client ||= OpenAI::Client.new(access_token: ENV["OPENAI_ACCESS_TOKEN"], log_errors: true)
   end
 end

metadata

@@ -1,15 +1,57 @@
 --- !ruby/object:Gem::Specification
 name: podcast-buddy
 version: !ruby/object:Gem::Version
-  version: 0.1.1
+  version: 0.2.1
 platform: ruby
 authors:
 - Valentino Stoll
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2024-08-09 00:00:00.000000000 Z
-dependencies: []
+date: 2024-08-14 00:00:00.000000000 Z
+dependencies:
+- !ruby/object:Gem::Dependency
+  name: async
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: '0'
+  type: :runtime
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: '0'
+- !ruby/object:Gem::Dependency
+  name: async-http-faraday
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: '0'
+  type: :runtime
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: '0'
+- !ruby/object:Gem::Dependency
+  name: rainbow
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: '0'
+  type: :runtime
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - ">="
+      - !ruby/object:Gem::Version
+        version: '0'
 description: A simple Ruby command-line tool for dropping in an AI buddy to your podcast.
 email:
 - v@codenamev.com