Merge pull request #112 from ajcwebdev/refactor

Remove Ollama from Docker Image and Add RSS Retry Logic

Author: ajcwebdev

.github/Dockerfile

@@ -78,39 +78,39 @@ RUN chmod +x /usr/src/app/docker-entrypoint.sh
 # 2) Setup Ollama with models
 # ---------------------------------------------------
 
-# 1. Use the Ollama image as a base
-FROM ollama/ollama:latest AS ollama
-
-# 2. Set working directory to Ollama config directory
-WORKDIR /root/.ollama
-
-# 3. Start Ollama server and pull models
-RUN ollama serve & \
-    sleep 10 && \
-    ollama pull qwen2.5:0.5b && \
-    ollama ls && \
-    echo "Listing /root/.ollama after qwen2.5:0.5b pull:" && \
-    ls -lh /root/.ollama || true && \
-    echo "Listing /root/.ollama/models after qwen2.5:0.5b pull:" && \
-    ls -lh /root/.ollama/models || true && \
-    pkill ollama
+# # 1. Use the Ollama image as a base
+# FROM ollama/ollama:latest AS ollama
+
+# # 2. Set working directory to Ollama config directory
+# WORKDIR /root/.ollama
+
+# # 3. Start Ollama server and pull models
+# RUN ollama serve & \
+#     sleep 10 && \
+#     ollama pull qwen2.5:0.5b && \
+#     ollama ls && \
+#     echo "Listing /root/.ollama after qwen2.5:0.5b pull:" && \
+#     ls -lh /root/.ollama || true && \
+#     echo "Listing /root/.ollama/models after qwen2.5:0.5b pull:" && \
+#     ls -lh /root/.ollama/models || true && \
+#     pkill ollama
 
 # ---------------------------------------------------
 # 3) Final stage combining everything
 # ---------------------------------------------------
 
 FROM base
 
-# 1. Copy Ollama binary
-COPY --from=ollama /bin/ollama /usr/local/bin/ollama
+# # 1. Copy Ollama binary
+# COPY --from=ollama /bin/ollama /usr/local/bin/ollama
 
-# 2. Copy pre-downloaded models
-COPY --from=ollama /root/.ollama /root/.ollama
+# # 2. Copy pre-downloaded models
+# COPY --from=ollama /root/.ollama /root/.ollama
 
-RUN echo "Listing /root/.ollama in final stage:" && \
-    ls -lh /root/.ollama || true && \
-    echo "Listing /root/.ollama/models in final stage:" && \
-    ls -lh /root/.ollama/models || true
+# RUN echo "Listing /root/.ollama in final stage:" && \
+#     ls -lh /root/.ollama || true && \
+#     echo "Listing /root/.ollama/models in final stage:" && \
+#     ls -lh /root/.ollama/models || true
 
 # Set environment variables for Whisper
 ENV WHISPER_FORCE_CPU=1
@@ -122,9 +122,9 @@ RUN mkdir -p /usr/src/app/content
 # 4. Set proper permissions for the entire app directory including content
 RUN chown -R node:node /usr/src/app \
  && chmod -R 755 /usr/src/app \
- && chmod 777 /usr/src/app/content \
- && chown -R node:node /root/.ollama \
- && chmod -R 755 /root/.ollama
+ && chmod 777 /usr/src/app/content
+#  && chown -R node:node /root/.ollama \
+#  && chmod -R 755 /root/.ollama
 
 # Switch to non-root user
 USER node

.github/docker-entrypoint.sh

@@ -17,15 +17,15 @@ log_error() {
     exit 1
 }
 
-echo "Debug: (docker-entrypoint.sh) Checking /root/.ollama before starting Ollama..."
-ls -lR /root/.ollama || true
+# echo "Debug: (docker-entrypoint.sh) Checking /root/.ollama before starting Ollama..."
+# ls -lR /root/.ollama || true
 
-# Start Ollama server in the background
-echo "Starting Ollama server..."
-ollama serve &
+# # Start Ollama server in the background
+# echo "Starting Ollama server..."
+# ollama serve &
 
-# Wait for Ollama server to start
-sleep 5
+# # Wait for Ollama server to start
+# sleep 5
 
 # If first argument is "serve", then start the server.
 if [ "$1" = "serve" ]; then

src/commander.ts

@@ -15,9 +15,10 @@ import { argv, exit } from 'node:process'
 import { fileURLToPath } from 'node:url'
 import { Command } from 'commander'
 import { selectPrompts } from './process-steps/04-select-prompt'
-import { validateProcessAction, validateLLM, validateTranscription, processAction } from './utils/validate-option'
-import { l, err, logCompletionSeparator } from './utils/logging'
+import { processAction, validateCLIOptions } from './utils/validate-option'
+import { l, err, logSeparator } from './utils/logging'
 import { envVarsMap } from './utils/globals/llms'
+
 import type { ProcessingOptions } from './utils/types/process'
 
 // Initialize the command-line interface using Commander.js
@@ -67,12 +68,7 @@ program
   .option('--printPrompt <sections...>', 'Print the prompt sections without processing')
   .option('--customPrompt <filePath>', 'Use a custom prompt from a markdown file')
   .option('--saveAudio', 'Do not delete intermediary files after processing')
-  // Added options to override environment variables from CLI
-  /**
-   * Additional CLI options to allow passing API keys from the command line,
-   * overriding .env values if they exist. This way, if the .env is missing
-   * a key, the user can supply it via the CLI.
-   */
+  // Options to override environment variables from CLI
   .option('--openaiApiKey <key>', 'Specify OpenAI API key (overrides .env variable)')
   .option('--anthropicApiKey <key>', 'Specify Anthropic API key (overrides .env variable)')
   .option('--deepgramApiKey <key>', 'Specify Deepgram API key (overrides .env variable)')
@@ -84,21 +80,6 @@ program
   .option('--togetherApiKey <key>', 'Specify Together API key (overrides .env variable)')
   .option('--fireworksApiKey <key>', 'Specify Fireworks API key (overrides .env variable)')
   .option('--groqApiKey <key>', 'Specify Groq API key (overrides .env variable)')
-  // Add examples and additional help text
-  .addHelpText(
-    'after',
-    `
-        Examples:
-          $ autoshow --video "https://www.youtube.com/watch?v=..."
-          $ autoshow --playlist "https://www.youtube.com/playlist?list=..."
-          $ autoshow --channel "https://www.youtube.com/channel/..."
-          $ autoshow --file "content/audio.mp3"
-          $ autoshow --rss "https://feeds.transistor.fm/fsjam-podcast/"
-
-        Documentation: https://github.com/ajcwebdev/autoshow#readme
-        Report Issues: https://github.com/ajcwebdev/autoshow/issues
-        `
-  )
 
 /**
  * Main action for the program.
@@ -125,19 +106,13 @@ program.action(async (options: ProcessingOptions) => {
     exit(0)
   }
 
-  // 1) Validate which action was chosen
-  const action = validateProcessAction(options, "action")
-
-  // 2) Validate LLM
-  const llmServices = validateLLM(options)
-
-  // 3) Validate transcription
-  const transcriptServices = validateTranscription(options)
+  // Validate action, LLM, and transcription inputs
+  const { action, llmServices, transcriptServices } = validateCLIOptions(options)
 
   try {
     // Helper to handle all action processing logic. If successful, log and exit.
     await processAction(action, options, llmServices, transcriptServices)
-    logCompletionSeparator(action)
+    logSeparator({ type: 'completion', descriptor: action })
     exit(0)
   } catch (error) {
     err(`Error processing ${action}:`, (error as Error).message)

src/llms/chatgpt.ts

@@ -3,7 +3,7 @@
 import { env } from 'node:process'
 import { OpenAI } from 'openai'
 import { GPT_MODELS } from '../utils/globals/llms'
-import { err, logAPIResults } from '../utils/logging'
+import { err, logLLMCost } from '../utils/logging'
 import type { ChatGPTModelType } from '../utils/types/llms'
 
 /**
@@ -43,7 +43,7 @@ export const callChatGPT = async (
 
     const content = firstChoice.message.content
 
-    logAPIResults({
+    logLLMCost({
       modelName: actualModel,
       stopReason: firstChoice.finish_reason ?? 'unknown',
       tokenUsage: {

src/llms/claude.ts

@@ -3,9 +3,19 @@
 import { env } from 'node:process'
 import { Anthropic } from '@anthropic-ai/sdk'
 import { CLAUDE_MODELS } from '../utils/globals/llms'
-import { err, logAPIResults } from '../utils/logging'
+import { err, logLLMCost } from '../utils/logging'
 import type { ClaudeModelType } from '../utils/types/llms'
 
+/**
+ * Extracts text content from the API response
+ * @param content - The content returned by the API
+ * @returns The extracted text content, or null if no text content is found
+ */
+interface ContentBlock {
+  type: string;
+  text?: string;
+}
+
 /**
  * Main function to call Claude API.
  * @param {string} prompt - The prompt or instructions to process.
@@ -42,7 +52,7 @@ export const callClaude = async (
       throw new Error('No text content generated from the API')
     }
 
-    logAPIResults({
+    logLLMCost({
       modelName: actualModel,
       stopReason: response.stop_reason ?? 'unknown',
       tokenUsage: {
@@ -59,18 +69,13 @@ export const callClaude = async (
   }
 }
 
-/**
- * Extracts text content from the API response
- * @param content - The content returned by the API
- * @returns The extracted text content, or null if no text content is found
- */
-function extractTextContent(content: any[]): string | null {
+function extractTextContent(content: ContentBlock[]): string | null {
   for (const block of content) {
     if (typeof block === 'object' && block !== null && 'type' in block) {
       if (block.type === 'text' && 'text' in block) {
-        return block.text
+        return block.text ?? null;
       }
     }
   }
-  return null
+  return null;
 }

src/llms/cohere.ts

@@ -3,7 +3,7 @@
 import { env } from 'node:process'
 import { CohereClient } from 'cohere-ai'
 import { COHERE_MODELS } from '../utils/globals/llms'
-import { err, logAPIResults } from '../utils/logging'
+import { err, logLLMCost } from '../utils/logging'
 import type { CohereModelType } from '../utils/types/llms'
 
 /**
@@ -43,7 +43,7 @@ export const callCohere = async (
 
     const { inputTokens, outputTokens } = meta?.tokens ?? {}
 
-    logAPIResults({
+    logLLMCost({
       modelName: actualModel,
       stopReason: finishReason ?? 'unknown',
       tokenUsage: {

src/llms/fireworks.ts

@@ -2,7 +2,7 @@
 
 import { env } from 'node:process'
 import { FIREWORKS_MODELS } from '../utils/globals/llms'
-import { err, logAPIResults } from '../utils/logging'
+import { err, logLLMCost } from '../utils/logging'
 import type { FireworksModelType, FireworksResponse } from '../utils/types/llms'
 
 /**
@@ -59,7 +59,7 @@ export const callFireworks = async (
       throw new Error('No content generated from the Fireworks API')
     }
 
-    logAPIResults({
+    logLLMCost({
       modelName: modelKey,
       stopReason: data.choices[0]?.finish_reason ?? 'unknown',
       tokenUsage: {

src/llms/gemini.ts

@@ -3,7 +3,7 @@
 import { env } from 'node:process'
 import { GoogleGenerativeAI } from "@google/generative-ai"
 import { GEMINI_MODELS } from '../utils/globals/llms'
-import { err, logAPIResults } from '../utils/logging'
+import { err, logLLMCost } from '../utils/logging'
 import type { GeminiModelType } from '../utils/types/llms'
 
 /**
@@ -46,7 +46,7 @@ export const callGemini = async (
       const { usageMetadata } = response
       const { promptTokenCount, candidatesTokenCount, totalTokenCount } = usageMetadata ?? {}
 
-      logAPIResults({
+      logLLMCost({
         modelName: actualModel,
         stopReason: 'complete',
         tokenUsage: {

src/llms/groq.ts

@@ -2,7 +2,7 @@
 
 import { env } from 'node:process'
 import { GROQ_MODELS } from '../utils/globals/llms'
-import { err, logAPIResults } from '../utils/logging'
+import { err, logLLMCost } from '../utils/logging'
 import type { GroqModelType, GroqChatCompletionResponse } from '../utils/types/llms'
 
 /**
@@ -58,7 +58,7 @@ export const callGroq = async (
       throw new Error('No content generated from the Groq API')
     }
 
-    logAPIResults({
+    logLLMCost({
       modelName: modelKey,
       stopReason: data.choices[0]?.finish_reason ?? 'unknown',
       tokenUsage: {

src/llms/mistral.ts

@@ -3,7 +3,7 @@
 import { env } from 'node:process'
 import { Mistral } from '@mistralai/mistralai'
 import { MISTRAL_MODELS } from '../utils/globals/llms'
-import { err, logAPIResults } from '../utils/logging'
+import { err, logLLMCost } from '../utils/logging'
 import type { MistralModelType } from '../utils/types/llms'
 
 /**
@@ -46,7 +46,7 @@ export const callMistral = async (
     const content = firstChoice.message.content
     const contentString = Array.isArray(content) ? content.join('') : content
 
-    logAPIResults({
+    logLLMCost({
       modelName: actualModel,
       stopReason: firstChoice.finishReason ?? 'unknown',
       tokenUsage: {

src/llms/ollama.ts

@@ -2,7 +2,7 @@
 
 import { env } from 'node:process'
 import { OLLAMA_MODELS } from '../utils/globals/llms'
-import { l, err, logAPIResults } from '../utils/logging'
+import { l, err, logLLMCost } from '../utils/logging'
 import { checkOllamaServerAndModel } from '../utils/validate-option'
 import type { OllamaModelType, OllamaResponse } from '../utils/types/llms'
 
@@ -68,7 +68,7 @@ export const callOllama = async (
     const totalPromptTokens = data.prompt_eval_count ?? 0
     const totalCompletionTokens = data.eval_count ?? 0
 
-    logAPIResults({
+    logLLMCost({
       modelName: modelKey,
       stopReason: 'stop',
       tokenUsage: {

src/llms/together.ts

@@ -2,7 +2,7 @@
 
 import { env } from 'node:process'
 import { TOGETHER_MODELS } from '../utils/globals/llms'
-import { err, logAPIResults } from '../utils/logging'
+import { err, logLLMCost } from '../utils/logging'
 import type { TogetherModelType, TogetherResponse } from '../utils/types/llms'
 
 /**
@@ -59,7 +59,7 @@ export const callTogether = async (
       throw new Error('No content generated from the Together AI API')
     }
 
-    logAPIResults({
+    logLLMCost({
       modelName: modelKey,
       stopReason: data.choices[0]?.finish_reason ?? 'unknown',
       tokenUsage: {