summarize-meeting 0.1.2 → 0.2.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 70d706fa7630923e2599b27fc3916a16523785a915a659a3b81d9ff47853b6b2
-  data.tar.gz: 98bc7a33992281e1476702fb927f3f41f2b81a5f8b8ebd1c264d37ac236af87e
+  metadata.gz: 3c14fc93afabfd456a2c3778dffccf9d2a66e77f2d47ab98b864986ad0e10cd2
+  data.tar.gz: 68ba17378b48f84ad029ef1d3d6e1575110739bb4c8f5c40e25a759350722cba
 SHA512:
-  metadata.gz: 31e4d53b5969ba8cc4341045f05888b68d6812ff4480d8b2d4f850533bd76f30665a4a92c29872c2731a040e0d1d0b25aa9554fe9f72046cde0d22b5a7fbb902
-  data.tar.gz: 75035c790d2448947a98a51b2fb19f23c6e2d394095a9090bc5242009443db0b7ffd4ac23327d5fc9b24999a16fe6ad2367007be5096693106a8cf04701e41b7
+  metadata.gz: 40a944d8e2289d02368c1f05c98ef3b26612f40c01c0aa761c110e7f46664b8411f11c7881152b32cc4aa3d1c4db20d726906fdc00ef69c5d362779cd5300e40
+  data.tar.gz: 6477d716ebf27e2b6b7a8a5e15c4ce2ec205e7f089b277002fb5f1ab5273d1e486f2248625571038475fac55e91d81fa77378094615727d0bccae970251b1f74

data/bin/summarize-meeting

@@ -1,8 +1,7 @@
 #!/usr/bin/env ruby
 
 require "optparse"
-
-require_relative "../lib/meeting"
+require "summarize-meeting"
 
 def main
   options = {}
@@ -19,8 +18,8 @@ def main
       options[:output_file] = file
     end
 
-    if ENV["OPENAI_KEY"]
-      options[:openai_key] = ENV["OPENAI_KEY"]
+    if ENV["OPENAI_API_KEY"]
+      options[:openai_key] = ENV["OPENAI_API_KEY"]
     end
 
     opts.on("-k", "--openai-key KEY", "The OpenAI API key to use") do |key|
@@ -36,8 +35,8 @@ def main
     end
   end.parse!
 
-  Ai.access_token = options[:openai_key] if options[:openai_key]
-  Ai.organization_id = options[:openai_org] if options[:openai_org]
+  SummarizeMeeting::Ai.access_token = options[:openai_key] if options[:openai_key]
+  SummarizeMeeting::Ai.organization_id = options[:openai_org] if options[:openai_org]
 
   if ARGV.length != 1
     puts "Error: You must specify a transcript file to summarize."
@@ -47,7 +46,7 @@ def main
   transcript_file = ARGV[0]
   transcript = File.read(transcript_file)
 
-  meeting = Meeting.new(transcript)
+  meeting = SummarizeMeeting::Meeting.new(transcript)
   summary = meeting.summarize
   summary_file_name = if options[:output_file]
     options[:output_file]
@@ -59,6 +58,6 @@ def main
   File.write(summary_file_name, summary)
 end
 
-if __FILE__ == $0
+if __FILE__.to_s.end_with?("summarize-meeting") && $0.to_s.end_with?("summarize-meeting")
   main
 end

data/lib/summarize-meeting/ai.rb

@@ -0,0 +1,28 @@
+require "openai"
+
+module SummarizeMeeting
+  module Ai
+    @@access_token = ENV["OPENAI_KEY"]
+    @@organization_id = ENV["OPENAI_ORG"]
+
+    def self.client
+      OpenAI::Client.new(access_token: access_token, organization_id: organization_id)
+    end
+
+    def self.access_token
+      @@access_token
+    end
+
+    def self.organization_id
+      @@organization_id
+    end
+
+    def self.access_token=(token)
+      @@access_token = token
+    end
+
+    def self.organization_id=(id)
+      @@organization_id = id
+    end
+  end
+end

data/lib/summarize-meeting/meeting.rb

@@ -0,0 +1,130 @@
+require "json"
+require "mustache"
+require "openai"
+
+module SummarizeMeeting
+  class Meeting
+    LINE_SUMMARY_PROMPT_TEMPLATE = [
+      {
+        role: "system",
+        content: "You are an assistant summarizing a meeting.",
+      },
+      {
+        role: "system",
+        content: "The transcript of the meeting is split into {{chunkCount}} chunks. This is the {{chunkIndex}} chunk.",
+      },
+      {
+        role: "assistant",
+        content: "Please provide me with the next chunk of the transcript.",
+      },
+      {
+        role: "user",
+        content: "{{chunk}}",
+      }
+    ]
+
+    CONSOLIDATED_SUMMARY_PROMPT_TEMPLATE = [
+      {
+        role: "system",
+        content: "You are an assistant summarizing a meeting.",
+      },
+      {
+        role: "system",
+        content: "Notes about the meeting have been compiled.",
+      },
+      {
+        role: "system",
+        content: <<~CONTENT
+          Your job is to write a thorough summary of the meeting.
+          The summary should start with a brief overview of the meeting.
+          The summary should be detailed and should extract any action items that were discussed.
+          The summary should be organized into sections with headings and bullet points.
+          The summary should include a list of attendees.
+          The order of the sections should be overview, attendees, action items, and detailed notes by topic.
+        CONTENT
+      },
+      {
+        role: "assistant",
+        content: "Please provide me with notes from the meeting.",
+      },
+      {
+        role: "user",
+        content: "{{notes}}",
+      }
+    ]
+
+    def initialize(transcript)
+      @transcript = transcript
+    end
+
+    attr_reader :transcript
+
+    def summarize
+
+      # Step 1. Split the transcript into lines.
+      lines = transcript.split("\n")
+
+      # 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
+
+      # Step 3. Split the transcript into equally sized chunks.
+      chunks = split_lines_into_equal_size_chunks(lines, max_chunk_word_count)
+
+      # Step 4. Summarize each chunk.
+      previous_chunks_summary = ""
+      chunks.each_with_index do |chunk, chunk_index|
+        chunk_summary = summarize_chunk(chunk, chunk_index, chunks.size, previous_chunks_summary)
+        previous_chunks_summary += chunk_summary
+      end
+
+      # Step 5. Write a consolidated summary.
+      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")
+    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")
+    end
+
+    def split_lines_into_equal_size_chunks(lines, max_chunk_word_count)
+      chunks = []
+      chunk = []
+      chunk_word_count = 0
+      lines.each do |line|
+        line_word_count = line.split.size
+        if chunk_word_count + line_word_count > max_chunk_word_count
+          chunks << chunk
+          chunk = []
+          chunk_word_count = 0
+        end
+        chunk << line
+        chunk_word_count += line_word_count
+      end
+      chunks << chunk
+      chunks
+    end
+  end
+end

data/lib/summarize-meeting/version.rb

@@ -0,0 +1,3 @@
+module SummarizeMeeting
+  VERSION = "0.2.0"
+end

data/lib/summarize-meeting.rb

@@ -0,0 +1,6 @@
+require_relative "./summarize-meeting/version"
+require_relative "./summarize-meeting/ai"
+require_relative "./summarize-meeting/meeting"
+
+module SummarizeMeeting
+end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: summarize-meeting
 version: !ruby/object:Gem::Version
-  version: 0.1.2
+  version: 0.2.0
 platform: ruby
 authors:
 - Sean Devine
@@ -133,8 +133,10 @@ files:
 - LICENSE.txt
 - README.md
 - bin/summarize-meeting
-- lib/ai.rb
-- lib/meeting.rb
+- lib/summarize-meeting.rb
+- lib/summarize-meeting/ai.rb
+- lib/summarize-meeting/meeting.rb
+- lib/summarize-meeting/version.rb
 homepage: 
 licenses:
 - MIT
@@ -142,7 +144,7 @@ metadata: {}
 post_install_message: 
 rdoc_options: []
 require_paths:
-- "."
+- lib
 required_ruby_version: !ruby/object:Gem::Requirement
   requirements:
   - - ">="

data/lib/ai.rb

@@ -1,26 +0,0 @@
-require "openai"
-
-module Ai
-  @@access_token = ENV["OPENAI_KEY"]
-  @@organization_id = ENV["OPENAI_ORG"]
-
-  def self.client
-    OpenAI::Client.new(access_token: access_token, organization_id: organization_id)
-  end
-
-  def self.access_token
-    @@access_token
-  end
-
-  def self.organization_id
-    @@organization_id
-  end
-
-  def self.access_token=(token)
-    @@access_token = token
-  end
-
-  def self.organization_id=(id)
-    @@organization_id = id
-  end
-end

data/lib/meeting.rb

@@ -1,130 +0,0 @@
-require "json"
-require "mustache"
-require "openai"
-
-require_relative "./ai"
-
-class Meeting
-  LINE_SUMMARY_PROMPT_TEMPLATE = [
-    {
-      role: "system",
-      content: "You are an assistant summarizing a meeting.",
-    },
-    {
-      role: "system",
-      content: "The transcript of the meeting is split into {{chunkCount}} chunks. This is the {{chunkIndex}} chunk.",
-    },
-    {
-      role: "assistant",
-      content: "Please provide me with the next chunk of the transcript.",
-    },
-    {
-      role: "user",
-      content: "{{chunk}}",
-    }
-  ]
-
-  CONSOLIDATED_SUMMARY_PROMPT_TEMPLATE = [
-    {
-      role: "system",
-      content: "You are an assistant summarizing a meeting.",
-    },
-    {
-      role: "system",
-      content: "Notes about the meeting have been compiled.",
-    },
-    {
-      role: "system",
-      content: <<~CONTENT
-        Your job is to write a thorough summary of the meeting.
-        The summary should start with a brief overview of the meeting.
-        The summary should be detailed and should extract any action items that were discussed.
-        The summary should be organized into sections with headings and bullet points.
-        The summary should include a list of attendees.
-        The order of the sections should be overview, attendees, action items, and detailed notes by topic.
-      CONTENT
-    },
-    {
-      role: "assistant",
-      content: "Please provide me with notes from the meeting.",
-    },
-    {
-      role: "user",
-      content: "{{notes}}",
-    }
-  ]
-
-  def initialize(transcript)
-    @transcript = transcript
-  end
-
-  attr_reader :transcript
-
-  def summarize
-
-    # Step 1. Split the transcript into lines.
-    lines = transcript.split("\n")
-
-    # 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
-
-    # Step 3. Split the transcript into equally sized chunks.
-    chunks = split_lines_into_equal_size_chunks(lines, max_chunk_word_count)
-
-    # Step 4. Summarize each chunk.
-    previous_chunks_summary = ""
-    chunks.each_with_index do |chunk, chunk_index|
-      chunk_summary = summarize_chunk(chunk, chunk_index, chunks.size, previous_chunks_summary)
-      previous_chunks_summary += chunk_summary
-    end
-
-    # Step 5. Write a consolidated summary.
-    consolidated_template = CONSOLIDATED_SUMMARY_PROMPT_TEMPLATE
-    prompt = Mustache.render(consolidated_template.to_json, { notes: previous_chunks_summary.to_json })
-    messages = JSON.parse(prompt)
-    response = Ai.client.chat(
-      parameters: {
-        model: "gpt-3.5-turbo",
-        messages: messages,
-      }
-    )
-    response.dig("choices", 0, "message", "content")
-  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 = Ai.client.chat(
-      parameters: {
-        model: "gpt-3.5-turbo",
-        messages: messages,
-      }
-    )
-    response.dig("choices", 0, "message", "content")
-  end
-
-  def split_lines_into_equal_size_chunks(lines, max_chunk_word_count)
-    chunks = []
-    chunk = []
-    chunk_word_count = 0
-    lines.each do |line|
-      line_word_count = line.split.size
-      if chunk_word_count + line_word_count > max_chunk_word_count
-        chunks << chunk
-        chunk = []
-        chunk_word_count = 0
-      end
-      chunk << line
-      chunk_word_count += line_word_count
-    end
-    chunks << chunk
-    chunks
-  end
-end