summarize-meeting 0.2.1 → 1.1.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: bd8ba0fbf2d8b0610cebeccbd7906283f9b39a357dc594026dd31c4853488b92
-  data.tar.gz: 6ab995e6472dc2b9cd749a99f8f5ecb634de40baf8645a66e2b2a761676b235e
+  metadata.gz: 44c7e54486689e6bfa88f4b9ee7e0e82cf90220d64becacc16755ebdd8977dca
+  data.tar.gz: 0a905c1c9bc06bbe8f5cc825410aa3771597680526f11c6a7775afb9aed2d5b1
 SHA512:
-  metadata.gz: 90d848336e888922379d35d17b54fcd7312e5671005f1731ac93921e724c0b17c013c4230615e0842a771afc824af15280d18db1104ee404ffb9a8c8356545d7
-  data.tar.gz: ba360026742a884fb68cd7c2f5f079b1502da67da3ed9747f30585ff109ca59ddbb00869d18e796ad4ee38d439e4f063ced2a65609ec74f5d6f9b8844344ff28
+  metadata.gz: 2975c64d3571e15303d164014182e477c387fef9bac96aa9bdd5e7b3088eff238f8379e4305b9ec9442faf12a13bd2a1f63d5a7188ceca74dc7ccd42c9e8e475
+  data.tar.gz: 9f48334f5e4df0969de1ddc3055c677f1e2e7df2ea2bda38624e6cfcb5175ba8fa655092bf6bb02080f96b43f827827e1feb1a78fa181cb6b49142c3ecf0aefc

data/bin/summarize-meeting

@@ -44,9 +44,10 @@ def main
   end
 
   transcript_file = ARGV[0]
-  transcript = File.read(transcript_file)
+  transcript = File.read(transcript_file, encoding: "UTF-8")
 
   meeting = SummarizeMeeting::Meeting.new(transcript)
+
   summary = meeting.summarize
   summary_file_name = if options[:output_file]
     options[:output_file]

data/lib/summarize-meeting/ai.rb

@@ -1,11 +1,17 @@
-require "openai"
-
 module SummarizeMeeting
   module Ai
+    class OpenAiError < StandardError; end
+    MAX_TOTAL_TOKENS = 4096
+    WORDS_PER_TOKEN = 0.75
+
     @@access_token = ENV["OPENAI_KEY"]
     @@organization_id = ENV["OPENAI_ORG"]
 
     def self.client
+      @client ||= new_client(access_token: access_token, organization_id: organization_id)
+    end
+
+    def self.new_client(access_token:, organization_id:)
       OpenAI::Client.new(access_token: access_token, organization_id: organization_id)
     end
 
@@ -24,5 +30,27 @@ module SummarizeMeeting
     def self.organization_id=(id)
       @@organization_id = id
     end
+
+    def self.calculate_token_word_count(token_count)
+      (token_count * WORDS_PER_TOKEN.to_f).ceil
+    end
+
+    def self.calculate_word_token_count(word_count)
+      (word_count / WORDS_PER_TOKEN.to_f).ceil
+    end
+
+    def self.chat(messages, client: self.client)
+      parameters = {
+        model: "gpt-3.5-turbo",
+        messages: messages,
+      }
+      response = client.chat(parameters: parameters)
+      content = response.dig("choices", 0, "message", "content")
+      if !content
+        raise OpenAiError, "No response from OpenAI"
+      else
+        content
+      end
+    end
   end
 end

data/lib/summarize-meeting/meeting.rb

@@ -4,6 +4,8 @@ require "openai"
 
 module SummarizeMeeting
   class Meeting
+    RESPONSE_RESERVE_TOKENS = 500
+
     LINE_SUMMARY_PROMPT_TEMPLATE = [
       {
         role: "system",
@@ -60,17 +62,14 @@ module SummarizeMeeting
     attr_reader :transcript
 
     def summarize
-
       # Step 1. Split the transcript into lines.
-      lines = transcript.split("\n")
+      lines = transcript.lines
 
       # Step 2. Calculate the maximum chunk size in words.
-      max_total_tokens = 4000
-      response_token_reserve = 500
-      template_tokens = LINE_SUMMARY_PROMPT_TEMPLATE.map { |line| line[:content].split.size }.sum
-      max_chunk_tokens = max_total_tokens - response_token_reserve - template_tokens
-      words_per_token = 0.7
-      max_chunk_word_count = max_chunk_tokens * words_per_token
+      template_word_count = LINE_SUMMARY_PROMPT_TEMPLATE.map { |line| line[:content].split.size }.sum
+      template_token_count = SummarizeMeeting::Ai.calculate_word_token_count(template_word_count)
+      max_chunk_token_count = SummarizeMeeting::Ai::MAX_TOTAL_TOKENS - RESPONSE_RESERVE_TOKENS - template_token_count
+      max_chunk_word_count = SummarizeMeeting::Ai.calculate_token_word_count(max_chunk_token_count)
 
       # Step 3. Split the transcript into equally sized chunks.
       chunks = split_lines_into_equal_size_chunks(lines, max_chunk_word_count)
@@ -86,27 +85,14 @@ module SummarizeMeeting
       consolidated_template = CONSOLIDATED_SUMMARY_PROMPT_TEMPLATE
       prompt = Mustache.render(consolidated_template.to_json, { notes: previous_chunks_summary.to_json })
       messages = JSON.parse(prompt)
-      response = SummarizeMeeting::Ai.client.chat(
-        parameters: {
-          model: "gpt-3.5-turbo",
-          messages: messages,
-        }
-      )
-      response.dig("choices", 0, "message", "content")
+      SummarizeMeeting::Ai.chat(messages: messages)
     end
 
     def summarize_chunk(chunk, chunk_index, chunk_count, previous_chunks_summary)
       template = LINE_SUMMARY_PROMPT_TEMPLATE
       prompt = Mustache.render(template.to_json, { chunkCount: chunk_count, chunkIndex: chunk_index + 1, chunk: chunk.join("\n").to_json })
       messages = JSON.parse(prompt)
-
-      response = SummarizeMeeting::Ai.client.chat(
-        parameters: {
-          model: "gpt-3.5-turbo",
-          messages: messages,
-        }
-      )
-      response.dig("choices", 0, "message", "content")
+      SummarizeMeeting::Ai.chat(messages: messages)
     end
 
     def split_lines_into_equal_size_chunks(lines, max_chunk_word_count)

data/lib/summarize-meeting/version.rb

@@ -1,3 +1,3 @@
 module SummarizeMeeting
-  VERSION = "0.2.1"
+  VERSION = "1.1.0"
 end

data/lib/summarize-meeting.rb

@@ -3,4 +3,4 @@ require_relative "./summarize-meeting/ai"
 require_relative "./summarize-meeting/meeting"
 
 module SummarizeMeeting
-end
+end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: summarize-meeting
 version: !ruby/object:Gem::Version
-  version: 0.2.1
+  version: 1.1.0
 platform: ruby
 authors:
 - Sean Devine
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2023-03-03 00:00:00.000000000 Z
+date: 2023-03-04 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: optparse