lattifai 0.4.5__py3-none-any.whl → 0.4.6__py3-none-any.whl

lattifai/workflows/gemini.py

@@ -24,16 +24,16 @@ class GeminiTranscriber:
     """
 
     # The specific Gem URL provided by the user
-    GEM_URL = 'https://gemini.google.com/gem/1870ly7xvW2hU_umtv-LedGsjywT0sQiN'
+    GEM_URL = "https://gemini.google.com/gem/1870ly7xvW2hU_umtv-LedGsjywT0sQiN"
 
     def __init__(self, api_key: Optional[str] = None):
         self.api_key = api_key
-        self.logger = setup_workflow_logger('gemini')
+        self.logger = setup_workflow_logger("gemini")
         self.prompt_loader = get_prompt_loader()
 
         if not self.api_key:
             self.logger.warning(
-                '⚠️ Gemini API key not provided. API key will be required when calling transcription methods.'
+                "⚠️ Gemini API key not provided. API key will be required when calling transcription methods."
             )
 
     async def __call__(self, youtube_url: str) -> str:
@@ -51,9 +51,9 @@ class GeminiTranscriber:
             Transcribed text
         """
         if not self.api_key:
-            raise ValueError('Gemini API key is required for transcription')
+            raise ValueError("Gemini API key is required for transcription")
 
-        self.logger.info(f'🎤 Starting Gemini transcription for: {youtube_url}')
+        self.logger.info(f"🎤 Starting Gemini transcription for: {youtube_url}")
 
         try:
             # Initialize client
@@ -63,11 +63,11 @@ class GeminiTranscriber:
             system_prompt = self.prompt_loader.get_gemini_transcription_prompt()
 
             # Generate transcription with extended thinking
-            self.logger.info('🔄 Sending request to Gemini 2.5 Pro...')
+            self.logger.info("🔄 Sending request to Gemini 2.5 Pro...")
             config = GenerateContentConfig(
                 system_instruction=system_prompt,
                 # Enable thinking by including it in response modalities
-                response_modalities=['TEXT'],
+                response_modalities=["TEXT"],
                 thinking_config=ThinkingConfig(
                     include_thoughts=False,
                     thinking_budget=-1,
@@ -76,25 +76,25 @@ class GeminiTranscriber:
             response = await asyncio.get_event_loop().run_in_executor(
                 None,
                 lambda: client.models.generate_content(
-                    model='gemini-2.5-pro',
-                    contents=Part.from_uri(file_uri=youtube_url, mime_type='video/*'),
+                    model="gemini-2.5-pro",
+                    contents=Part.from_uri(file_uri=youtube_url, mime_type="video/*"),
                     config=config,
                 ),
             )
 
             if not response.text:
-                raise RuntimeError('Empty response from Gemini API')
+                raise RuntimeError("Empty response from Gemini API")
 
             transcript = response.text.strip()
 
-            self.logger.info(f'✅ Transcription completed: {len(transcript)} characters')
+            self.logger.info(f"✅ Transcription completed: {len(transcript)} characters")
             return transcript
 
         except ImportError:
-            raise RuntimeError('Google GenAI SDK not installed. Please install with: pip install google-genai')
+            raise RuntimeError("Google GenAI SDK not installed. Please install with: pip install google-genai")
         except Exception as e:
-            self.logger.error(f'Gemini transcription failed: {str(e)}')
-            raise RuntimeError(f'Gemini transcription failed: {str(e)}')
+            self.logger.error(f"Gemini transcription failed: {str(e)}")
+            raise RuntimeError(f"Gemini transcription failed: {str(e)}")
 
     async def transcribe_file(self, media_file_path: str) -> str:
         """
@@ -107,9 +107,9 @@ class GeminiTranscriber:
             Transcribed text
         """
         if not self.api_key:
-            raise ValueError('Gemini API key is required for transcription')
+            raise ValueError("Gemini API key is required for transcription")
 
-        self.logger.info(f'🎤 Starting Gemini transcription for file: {media_file_path}')
+        self.logger.info(f"🎤 Starting Gemini transcription for file: {media_file_path}")
 
         try:
             # Initialize client
@@ -119,16 +119,16 @@ class GeminiTranscriber:
             system_prompt = self.prompt_loader.get_gemini_transcription_prompt()
 
             # Upload audio file
-            self.logger.info('📤 Uploading audio file to Gemini...')
+            self.logger.info("📤 Uploading audio file to Gemini...")
             media_file = client.files.upload(path=media_file_path)
 
             # Generate transcription with extended thinking
             # Note: For thinking mode, you may want to use 'gemini-2.0-flash-thinking-exp' or similar models
-            self.logger.info('🔄 Sending transcription request...')
+            self.logger.info("🔄 Sending transcription request...")
             config = GenerateContentConfig(
                 system_instruction=system_prompt,
                 # Enable thinking by including it in response modalities
-                response_modalities=['TEXT'],
+                response_modalities=["TEXT"],
                 thinking_config=ThinkingConfig(
                     include_thoughts=False,
                     thinking_budget=-1,
@@ -137,31 +137,31 @@ class GeminiTranscriber:
             response = await asyncio.get_event_loop().run_in_executor(
                 None,
                 lambda: client.models.generate_content(
-                    model='gemini-2.5-pro',
+                    model="gemini-2.5-pro",
                     contents=Part.from_uri(file_uri=media_file.uri, mime_type=media_file.mime_type),
                     config=config,
                 ),
             )
 
             if not response.text:
-                raise RuntimeError('Empty response from Gemini API')
+                raise RuntimeError("Empty response from Gemini API")
 
             transcript = response.text.strip()
 
-            self.logger.info(f'✅ Transcription completed: {len(transcript)} characters')
+            self.logger.info(f"✅ Transcription completed: {len(transcript)} characters")
             return transcript
 
         except ImportError:
-            raise RuntimeError('Google GenAI SDK not installed. Please install with: pip install google-genai')
+            raise RuntimeError("Google GenAI SDK not installed. Please install with: pip install google-genai")
         except Exception as e:
-            self.logger.error(f'Gemini transcription failed: {str(e)}')
-            raise RuntimeError(f'Gemini transcription failed: {str(e)}')
+            self.logger.error(f"Gemini transcription failed: {str(e)}")
+            raise RuntimeError(f"Gemini transcription failed: {str(e)}")
 
     def get_gem_info(self) -> dict:
         """Get information about the Gem being used"""
         return {
-            'gem_name': 'Audio Transcription Gem',
-            'gem_url': self.GEM_URL,
-            'model': 'Gemini 2.5 Pro',
-            'description': 'Specialized Gem for media content transcribe',
+            "gem_name": "Audio Transcription Gem",
+            "gem_url": self.GEM_URL,
+            "model": "Gemini 2.5 Pro",
+            "description": "Specialized Gem for media content transcribe",
         }

lattifai/workflows/prompts/__init__.py

@@ -26,16 +26,16 @@ class PromptLoader:
         Raises:
             FileNotFoundError: If prompt file doesn't exist
         """
-        prompt_path = self.prompts_dir / category / f'{name}.txt'
+        prompt_path = self.prompts_dir / category / f"{name}.txt"
 
         if not prompt_path.exists():
-            raise FileNotFoundError(f'Prompt not found: {prompt_path}')
+            raise FileNotFoundError(f"Prompt not found: {prompt_path}")
 
-        return prompt_path.read_text(encoding='utf-8').strip()
+        return prompt_path.read_text(encoding="utf-8").strip()
 
     def get_gemini_transcription_prompt(self) -> str:
         """Get the Gemini transcription Gem prompt"""
-        return self.load_prompt('gemini', 'transcription_gem')
+        return self.load_prompt("gemini", "transcription_gem")
 
 
 # Global instance