SimplerLLM 0.3.3.5__tar.gz → 0.3.4__tar.gz

{simplerllm-0.3.3.5 → simplerllm-0.3.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: SimplerLLM
-Version: 0.3.3.5
+Version: 0.3.4
 Summary: An easy-to-use Library for interacting with language models.
 Home-page: https://github.com/hassancs91/SimplerLLM
 Author: Hasan Aboul Hasan
@@ -48,17 +48,10 @@ Requires-Dist: pydub>=0.25.1
 Requires-Dist: elevenlabs>=2.22.0
 Requires-Dist: moviepy>=1.0.3
 Requires-Dist: yt-dlp>=2023.3.4
-Provides-Extra: voice
-Requires-Dist: pygame>=2.5.0; extra == "voice"
-Requires-Dist: sounddevice>=0.4.6; extra == "voice"
-Requires-Dist: pynput>=1.7.6; extra == "voice"
-Provides-Extra: live-voice
-Requires-Dist: sounddevice>=0.4.6; extra == "live-voice"
-Requires-Dist: pynput>=1.7.6; extra == "live-voice"
-Provides-Extra: all
-Requires-Dist: pygame>=2.5.0; extra == "all"
-Requires-Dist: sounddevice>=0.4.6; extra == "all"
-Requires-Dist: pynput>=1.7.6; extra == "all"
+Provides-Extra: transformers
+Requires-Dist: transformers>=4.36.0; extra == "transformers"
+Requires-Dist: torch>=2.0.0; extra == "transformers"
+Requires-Dist: accelerate>=0.25.0; extra == "transformers"
 Dynamic: author
 Dynamic: author-email
 Dynamic: classifier

simplerllm-0.3.4/SimplerLLM/__init__.py

@@ -0,0 +1,128 @@
+# SimplerLLM - Simplified interface for LLMs and Voice APIs
+
+# Language module exports
+from .language import (
+    LLM,
+    LLMProvider,
+    ReliableLLM,
+    OpenAILLM,
+    GeminiLLM,
+    AnthropicLLM,
+    OllamaLLM,
+    DeepSeekLLM,
+    LLMJudge,
+    JudgeMode,
+    JudgeResult,
+    ProviderResponse,
+    ProviderEvaluation,
+    EvaluationReport,
+    LLMFeedbackLoop,
+    FeedbackResult,
+    IterationResult,
+    Critique,
+    EmbeddingsLLM,
+    EmbeddingsProvider,
+    OpenAIEmbeddings,
+    VoyageEmbeddings,
+    CohereEmbeddings,
+)
+
+# Voice module exports
+from .voice import (
+    # TTS
+    TTS,
+    TTSBase,
+    TTSProvider,
+    TTSResponse,
+    Voice,
+    TTSError,
+    TTSValidationError,
+    TTSProviderError,
+    TTSVoiceNotFoundError,
+    OpenAITTS,
+    OPENAI_VOICES,
+    OPENAI_MODELS,
+    OPENAI_FORMATS,
+    ELEVENLABS_MODELS,
+    ELEVENLABS_FORMATS,
+    # STT
+    STT,
+    STTProvider,
+    OpenAISTT,
+    STTFullResponse,
+)
+
+# Optional ElevenLabs TTS
+try:
+    from .voice import ElevenLabsTTS
+except (ImportError, TypeError):
+    ElevenLabsTTS = None
+
+# Image module exports
+from .image import (
+    ImageGenerator,
+    ImageProvider,
+    ImageSize,
+    OpenAIImageGenerator,
+    StabilityImageGenerator,
+    GoogleImageGenerator,
+    ImageGenerationResponse,
+)
+
+__all__ = [
+    # Language module
+    'LLM',
+    'LLMProvider',
+    'ReliableLLM',
+    'OpenAILLM',
+    'GeminiLLM',
+    'AnthropicLLM',
+    'OllamaLLM',
+    'DeepSeekLLM',
+    'LLMJudge',
+    'JudgeMode',
+    'JudgeResult',
+    'ProviderResponse',
+    'ProviderEvaluation',
+    'EvaluationReport',
+    'LLMFeedbackLoop',
+    'FeedbackResult',
+    'IterationResult',
+    'Critique',
+    # Embeddings
+    'EmbeddingsLLM',
+    'EmbeddingsProvider',
+    'OpenAIEmbeddings',
+    'VoyageEmbeddings',
+    'CohereEmbeddings',
+    # Voice module - TTS
+    'TTS',
+    'TTSBase',
+    'TTSProvider',
+    'TTSResponse',
+    'Voice',
+    'TTSError',
+    'TTSValidationError',
+    'TTSProviderError',
+    'TTSVoiceNotFoundError',
+    'OpenAITTS',
+    'ElevenLabsTTS',
+    'OPENAI_VOICES',
+    'OPENAI_MODELS',
+    'OPENAI_FORMATS',
+    'ELEVENLABS_MODELS',
+    'ELEVENLABS_FORMATS',
+    # Voice module - STT
+    'STT',
+    'STTProvider',
+    'OpenAISTT',
+    'STTFullResponse',
+    # Image module
+    'ImageGenerator',
+    'ImageProvider',
+    'ImageSize',
+    'OpenAIImageGenerator',
+    'StabilityImageGenerator',
+    'GoogleImageGenerator',
+    'ImageGenerationResponse',
+]

{simplerllm-0.3.3.5 → simplerllm-0.3.4}/SimplerLLM/image/generation/base.py

@@ -1,5 +1,4 @@
 from enum import Enum
-import os
 from SimplerLLM.utils.custom_verbose import verbose_print
 
 
@@ -16,10 +15,23 @@ class ImageProvider(Enum):
 
 class ImageSize(Enum):
     """User-friendly image size options that map to provider-specific dimensions."""
-    SQUARE = "square"
-    HORIZONTAL = "horizontal"
-    VERTICAL = "vertical"
-    PORTRAIT_3_4 = "portrait_3_4"  # 3:4 aspect ratio, closest to A4 paper
+    # Standard sizes
+    SQUARE = "square"           # 1:1
+    HORIZONTAL = "horizontal"   # 16:9
+    VERTICAL = "vertical"       # 9:16
+
+    # Portrait ratios
+    PORTRAIT_3_4 = "portrait_3_4"   # 3:4 aspect ratio, closest to A4 paper
+    PORTRAIT_2_3 = "portrait_2_3"   # 2:3 aspect ratio
+    PORTRAIT_4_5 = "portrait_4_5"   # 4:5 aspect ratio (Instagram portrait)
+
+    # Landscape ratios
+    LANDSCAPE_3_2 = "landscape_3_2" # 3:2 aspect ratio (classic photo)
+    LANDSCAPE_4_3 = "landscape_4_3" # 4:3 aspect ratio (standard monitor)
+    LANDSCAPE_5_4 = "landscape_5_4" # 5:4 aspect ratio
+
+    # Ultrawide
+    ULTRAWIDE = "ultrawide"         # 21:9 aspect ratio (cinematic)
 
 
 class ImageGenerator:
@@ -94,7 +106,7 @@ class ImageGenerator:
             from .wrappers.google_wrapper import GoogleImageGenerator
             return GoogleImageGenerator(
                 provider=provider,
-                model_name=model_name or "gemini-2.5-flash-image-preview",
+                model_name=model_name or "gemini-2.5-flash-image",
                 api_key=api_key,
                 verbose=verbose,
             )
@@ -161,31 +173,50 @@ class ImageGenerator:
 
         # Map sizes based on provider
         if provider == ImageProvider.OPENAI_DALL_E:
+            # OpenAI has limited size options, map to closest equivalent
             size_map = {
                 ImageSize.SQUARE: "1024x1024",
                 ImageSize.HORIZONTAL: "1792x1024",
                 ImageSize.VERTICAL: "1024x1792",
-                ImageSize.PORTRAIT_3_4: "768x1024",
+                ImageSize.PORTRAIT_3_4: "1024x1792",      # Map to vertical
+                ImageSize.PORTRAIT_2_3: "1024x1792",      # Map to vertical
+                ImageSize.PORTRAIT_4_5: "1024x1792",      # Map to vertical
+                ImageSize.LANDSCAPE_3_2: "1792x1024",     # Map to horizontal
+                ImageSize.LANDSCAPE_4_3: "1792x1024",     # Map to horizontal
+                ImageSize.LANDSCAPE_5_4: "1024x1024",     # Map to square (closest)
+                ImageSize.ULTRAWIDE: "1792x1024",         # Map to horizontal
             }
             return size_map.get(size, "1024x1024")
 
         elif provider == ImageProvider.STABILITY_AI:
-            # Stability AI uses aspect ratios instead of dimensions
+            # Stability AI uses aspect ratios
             size_map = {
                 ImageSize.SQUARE: "1:1",
                 ImageSize.HORIZONTAL: "16:9",
                 ImageSize.VERTICAL: "9:16",
                 ImageSize.PORTRAIT_3_4: "3:4",
+                ImageSize.PORTRAIT_2_3: "2:3",
+                ImageSize.PORTRAIT_4_5: "4:5",
+                ImageSize.LANDSCAPE_3_2: "3:2",
+                ImageSize.LANDSCAPE_4_3: "4:3",
+                ImageSize.LANDSCAPE_5_4: "5:4",
+                ImageSize.ULTRAWIDE: "21:9",
             }
             return size_map.get(size, "1:1")
 
         elif provider == ImageProvider.GOOGLE_GEMINI:
-            # Google Gemini uses aspect ratios like Stability AI
+            # Google Gemini supports: 1:1, 2:3, 3:2, 3:4, 4:3, 4:5, 5:4, 9:16, 16:9, 21:9
             size_map = {
                 ImageSize.SQUARE: "1:1",
                 ImageSize.HORIZONTAL: "16:9",
                 ImageSize.VERTICAL: "9:16",
                 ImageSize.PORTRAIT_3_4: "3:4",
+                ImageSize.PORTRAIT_2_3: "2:3",
+                ImageSize.PORTRAIT_4_5: "4:5",
+                ImageSize.LANDSCAPE_3_2: "3:2",
+                ImageSize.LANDSCAPE_4_3: "4:3",
+                ImageSize.LANDSCAPE_5_4: "5:4",
+                ImageSize.ULTRAWIDE: "21:9",
             }
             return size_map.get(size, "1:1")
 
@@ -196,6 +227,12 @@ class ImageGenerator:
                 ImageSize.HORIZONTAL: "2K",
                 ImageSize.VERTICAL: "2K",
                 ImageSize.PORTRAIT_3_4: "2K",
+                ImageSize.PORTRAIT_2_3: "2K",
+                ImageSize.PORTRAIT_4_5: "2K",
+                ImageSize.LANDSCAPE_3_2: "2K",
+                ImageSize.LANDSCAPE_4_3: "2K",
+                ImageSize.LANDSCAPE_5_4: "2K",
+                ImageSize.ULTRAWIDE: "2K",
             }
             return size_map.get(size, "2K")
 

{simplerllm-0.3.3.5 → simplerllm-0.3.4}/SimplerLLM/image/generation/providers/__init__.py

@@ -7,10 +7,12 @@ from .image_response_models import ImageGenerationResponse
 from . import openai_image
 from . import stability_image
 from . import google_image
+from . import seedream_image
 
 __all__ = [
     'ImageGenerationResponse',
     'openai_image',
     'stability_image',
     'google_image',
+    'seedream_image',
 ]

{simplerllm-0.3.3.5 → simplerllm-0.3.4}/SimplerLLM/image/generation/providers/google_image.py

@@ -33,6 +33,27 @@ def _get_image_mime_type(image_source):
     return 'image/jpeg'
 
 
+def _is_gemini_3_model(model_name: str) -> bool:
+    """
+    Check if the model is a Gemini 3+ model that supports advanced features.
+
+    Gemini 3+ models support:
+    - 4K resolution output
+    - Advanced reasoning/thinking mode
+    - Up to 14 reference images (6 object + 5 human)
+
+    Args:
+        model_name: The model name string
+
+    Returns:
+        bool: True if model is Gemini 3+
+    """
+    if not model_name:
+        return False
+    model_lower = model_name.lower()
+    return "gemini-3" in model_lower or "nano-banana-pro" in model_lower
+
+
 def _load_image_data(image_source):
     """
     Load image data from various source types.
@@ -130,9 +151,6 @@ def generate_image(
     if verbose:
         print(f"[Google Gemini] Generating image with model={model_name}, aspect_ratio={aspect_ratio}, resolution={resolution}")
 
-    # Always print which model is being used for verification
-    print(f"[Google Gemini API] Using model: {model_name}")
-
     retry_delay = RETRY_DELAY
 
     for attempt in range(MAX_RETRIES):
@@ -174,9 +192,9 @@ def generate_image(
             ]
 
             # Configure generation with both IMAGE and TEXT modalities
-            # image_size only supported by Gemini 3+ models
+            # image_size (4K resolution) only supported by Gemini 3+ models
             image_config_params = {"aspect_ratio": aspect_ratio}
-            if "gemini-3" in model_name.lower():
+            if _is_gemini_3_model(model_name):
                 image_config_params["image_size"] = resolution
 
             generate_content_config = types.GenerateContentConfig(
@@ -408,9 +426,6 @@ def edit_image(
         print(f"[Google Gemini] Editing image with model={model_name}, aspect_ratio={aspect_ratio}, resolution={resolution}")
         print(f"[Google Gemini] Edit prompt: {edit_prompt}")
 
-    # Always print which model is being used for verification
-    print(f"[Google Gemini API] Using model: {model_name}")
-
     retry_delay = RETRY_DELAY
 
     for attempt in range(MAX_RETRIES):
@@ -439,9 +454,9 @@ def edit_image(
             ]
 
             # Configure generation with lower temperature for consistency
-            # image_size only supported by Gemini 3+ models
+            # image_size (4K resolution) only supported by Gemini 3+ models
             image_config_params = {"aspect_ratio": aspect_ratio}
-            if "gemini-3" in model_name.lower():
+            if _is_gemini_3_model(model_name):
                 image_config_params["image_size"] = resolution
 
             generate_content_config = types.GenerateContentConfig(