Update core/visual_engine.py

core/visual_engine.py

@@ -1,18 +1,26 @@
 # core/visual_engine.py
-from PIL import Image, ImageDraw, ImageFont, ImageOps # Pillow should be >= 10.1.0
+from PIL import Image, ImageDraw, ImageFont, ImageOps
 # --- MONKEY PATCH FOR Image.ANTIALIAS ---
 # This is applied at module load time.
+# Attempting to make MoviePy's internal calls to Image.ANTIALIAS work with Pillow 10+
 try:
     if hasattr(Image, 'Resampling') and hasattr(Image.Resampling, 'LANCZOS'): # Pillow 9+
         if not hasattr(Image, 'ANTIALIAS'):
             Image.ANTIALIAS = Image.Resampling.LANCZOS
-            print("INFO: Monkey-patched PIL.Image.ANTIALIAS with Image.Resampling.LANCZOS.")
+            print("INFO: Monkey-patched PIL.Image.ANTIALIAS with Image.Resampling.LANCZOS for MoviePy compatibility.")
     elif hasattr(Image, 'LANCZOS'): # Pillow 8 used Image.LANCZOS directly
          if not hasattr(Image, 'ANTIALIAS'):
-            Image.ANTIALIAS = Image.LANCZOS
-            print("INFO: Monkey-patched PIL.Image.ANTIALIAS with Image.LANCZOS.")
+            Image.ANTIALIAS = Image.LANCZOS # Pillow 8 had Image.LANCZOS
+            print("INFO: Monkey-patched PIL.Image.ANTIALIAS with Image.LANCZOS for MoviePy compatibility.")
+    else:
+        # This case means Pillow is too old, or an unexpected version.
+        # If ANTIALIAS is already present, no action needed.
+        if not hasattr(Image, 'ANTIALIAS'):
+             print("WARNING: Pillow version does not have common Resampling attributes or ANTIALIAS. Video effects might fail if MoviePy relies on ANTIALIAS.")
 except AttributeError:
-    print("WARNING: Could not determine Pillow version or attributes for ANTIALIAS monkey-patch. Video effects might fail.")
+    print("WARNING: Could not perform ANTIALIAS monkey-patch due to AttributeError (Image module might be incomplete).")
+except Exception as e_mp:
+    print(f"WARNING: An unexpected error occurred during ANTIALIAS monkey-patch: {e_mp}")
 # --- END MONKEY PATCH ---
 
 from moviepy.editor import (ImageClip, concatenate_videoclips, TextClip, 
@@ -29,60 +37,112 @@ import subprocess
 import logging
 
 logger = logging.getLogger(__name__)
-logger.setLevel(logging.INFO)
+logger.setLevel(logging.INFO) # Ensure logs from this module are visible
 
 ELEVENLABS_CLIENT_IMPORTED = False
-ElevenLabsAPIClient = None; Voice = None; VoiceSettings = None      
+ElevenLabsAPIClient = None 
+Voice = None              
+VoiceSettings = None      
 try:
     from elevenlabs.client import ElevenLabs as ImportedElevenLabsClient 
     from elevenlabs import Voice as ImportedVoice, VoiceSettings as ImportedVoiceSettings
-    ElevenLabsAPIClient = ImportedElevenLabsClient; Voice = ImportedVoice; VoiceSettings = ImportedVoiceSettings
-    ELEVENLABS_CLIENT_IMPORTED = True; logger.info("Successfully imported ElevenLabs client components.")
-except ImportError as e: logger.warning(f"ElevenLabs client import failed: {e}. Audio disabled.")
-except Exception as e: logger.warning(f"General ElevenLabs import error: {e}. Audio disabled.")
+    ElevenLabsAPIClient = ImportedElevenLabsClient 
+    Voice = ImportedVoice
+    VoiceSettings = ImportedVoiceSettings
+    ELEVENLABS_CLIENT_IMPORTED = True
+    logger.info("Successfully imported ElevenLabs client components (SDK v1.x.x pattern).")
+except ImportError as e_eleven:
+    logger.warning(f"Could not import ElevenLabs client components: {e_eleven}. ElevenLabs audio will be disabled.")
+except Exception as e_gen_eleven: 
+    logger.warning(f"General error importing ElevenLabs: {e_gen_eleven}. ElevenLabs audio will be disabled.")
 
 
 class VisualEngine:
     def __init__(self, output_dir="temp_cinegen_media", default_elevenlabs_voice_id="Rachel"):
-        self.output_dir = output_dir; os.makedirs(self.output_dir, exist_ok=True)
-        self.font_filename = "arial.ttf"; self.font_path_in_container = f"/usr/local/share/fonts/truetype/mycustomfonts/{self.font_filename}"
-        self.font_size_pil = 20; self.video_overlay_font_size = 30; self.video_overlay_font_color = 'white'; self.video_overlay_font = 'Liberation-Sans-Bold' # Or 'Arial-Bold'
-        try: self.font = ImageFont.truetype(self.font_path_in_container, self.font_size_pil); logger.info(f"Placeholder font loaded: {self.font_path_in_container}.")
-        except IOError: logger.warning(f"Placeholder font '{self.font_path_in_container}' fail. Default."); self.font = ImageFont.load_default(); self.font_size_pil = 10
+        self.output_dir = output_dir
+        os.makedirs(self.output_dir, exist_ok=True)
         
+        self.font_filename = "arial.ttf" 
+        self.font_path_in_container = f"/usr/local/share/fonts/truetype/mycustomfonts/{self.font_filename}"
+        self.font_size_pil = 20 
+        self.video_overlay_font_size = 30 
+        self.video_overlay_font_color = 'white'
+        self.video_overlay_font = 'Liberation-Sans-Bold' # More likely to be found by ImageMagick on Linux
+
+        try:
+            self.font = ImageFont.truetype(self.font_path_in_container, self.font_size_pil)
+            logger.info(f"Placeholder font loaded: {self.font_path_in_container}.")
+        except IOError:
+            logger.warning(f"Placeholder font '{self.font_path_in_container}' not found. Using default.")
+            self.font = ImageFont.load_default()
+            self.font_size_pil = 10
+
         self.openai_api_key = None; self.USE_AI_IMAGE_GENERATION = False
-        self.dalle_model = "dall-e-3"; self.image_size_dalle3 = "1792x1024"; self.video_frame_size = (1280, 720) 
-        self.elevenlabs_api_key = None; self.USE_ELEVENLABS = False; self.elevenlabs_client = None 
+        self.dalle_model = "dall-e-3"; self.image_size_dalle3 = "1792x1024" 
+        self.video_frame_size = (1280, 720) 
+
+        self.elevenlabs_api_key = None; self.USE_ELEVENLABS = False
+        self.elevenlabs_client = None 
         self.elevenlabs_voice_id = default_elevenlabs_voice_id
         if VoiceSettings and ELEVENLABS_CLIENT_IMPORTED: 
-            self.elevenlabs_voice_settings = VoiceSettings(stability=0.60, similarity_boost=0.80, style=0.15, use_speaker_boost=True)
-        else: self.elevenlabs_voice_settings = None
+            self.elevenlabs_voice_settings = VoiceSettings(
+                stability=0.60, similarity_boost=0.80, 
+                style=0.15, use_speaker_boost=True
+            )
+        else:
+            self.elevenlabs_voice_settings = None
+
         self.pexels_api_key = None; self.USE_PEXELS = False
         logger.info("VisualEngine initialized.")
 
-    def set_openai_api_key(self,k): self.openai_api_key=k; self.USE_AI_IMAGE_GENERATION=bool(k); logger.info(f"DALL-E {self.dalle_model} {'Ready.' if k else 'Disabled.'}")
+    def set_openai_api_key(self,k): 
+        self.openai_api_key=k; self.USE_AI_IMAGE_GENERATION=bool(k)
+        logger.info(f"DALL-E ({self.dalle_model}) {'Ready.' if k else 'Disabled (no API key).'}")
+
     def set_elevenlabs_api_key(self,api_key, voice_id_from_secret=None):
         self.elevenlabs_api_key=api_key
-        if voice_id_from_secret: self.elevenlabs_voice_id = voice_id_from_secret; logger.info(f"ElevenLabs Voice ID set from config: {self.elevenlabs_voice_id}")
+        if voice_id_from_secret: 
+            self.elevenlabs_voice_id = voice_id_from_secret
+            logger.info(f"ElevenLabs Voice ID set from config: {self.elevenlabs_voice_id}")
+        
         if api_key and ELEVENLABS_CLIENT_IMPORTED and ElevenLabsAPIClient: 
             try: 
                 self.elevenlabs_client = ElevenLabsAPIClient(api_key=api_key) 
-                if self.elevenlabs_client: self.USE_ELEVENLABS=True; logger.info(f"ElevenLabs Client Ready (Voice: {self.elevenlabs_voice_id}).")
-                else: self.USE_ELEVENLABS=False; logger.warning("ElevenLabs client is None post-init.")
-            except Exception as e: logger.error(f"Error initializing ElevenLabs client: {e}. Disabled.", exc_info=True); self.USE_ELEVENLABS=False; self.elevenlabs_client = None
-        else: self.USE_ELEVENLABS=False; self.elevenlabs_client = None; 
-              
-    def set_pexels_api_key(self,k): self.pexels_api_key=k; self.USE_PEXELS=bool(k); logger.info(f"Pexels Search {'Ready.' if k else 'Disabled.'}")
+                if self.elevenlabs_client: 
+                    self.USE_ELEVENLABS=True
+                    logger.info(f"ElevenLabs Client Ready (Using Voice ID: {self.elevenlabs_voice_id}).")
+                else: 
+                    self.USE_ELEVENLABS=False; logger.warning("ElevenLabs client is None after initialization attempt.")
+            except Exception as e: 
+                logger.error(f"Error initializing ElevenLabs client: {e}. ElevenLabs Disabled.", exc_info=True); 
+                self.USE_ELEVENLABS=False; self.elevenlabs_client = None
+        else: 
+            self.USE_ELEVENLABS=False; self.elevenlabs_client = None
+            if not (ELEVENLABS_CLIENT_IMPORTED and ElevenLabsAPIClient):
+                pass # Already logged at import time if client class itself failed to import
+            else: # Client class imported, but API key was not provided
+                logger.info("ElevenLabs API Key not provided. ElevenLabs Disabled.")
+            
+    def set_pexels_api_key(self,k):
+        self.pexels_api_key=k; self.USE_PEXELS=bool(k)
+        logger.info(f"Pexels Search {'Ready.' if k else 'Disabled (no API key).'}")
     
-    def _get_text_dimensions(self,text_content,font_obj): # Same
+    def _get_text_dimensions(self,text_content,font_obj):
         if not text_content: return 0,self.font_size_pil 
         try:
-            if hasattr(font_obj,'getbbox'): bbox=font_obj.getbbox(text_content);w=bbox[2]-bbox[0];h=bbox[3]-bbox[1];return w, h if h > 0 else self.font_size_pil
-            elif hasattr(font_obj,'getsize'): w,h=font_obj.getsize(text_content);return w, h if h > 0 else self.font_size_pil
-            else: return int(len(text_content)*self.font_size_pil*0.6),int(self.font_size_pil*1.2 if self.font_size_pil*1.2>0 else self.font_size_pil)
-        except: return int(len(text_content)*self.font_size_pil*0.6),int(self.font_size_pil*1.2)
+            if hasattr(font_obj,'getbbox'): 
+                bbox=font_obj.getbbox(text_content);w=bbox[2]-bbox[0];h=bbox[3]-bbox[1]
+                return w, h if h > 0 else self.font_size_pil
+            elif hasattr(font_obj,'getsize'): 
+                w,h=font_obj.getsize(text_content)
+                return w, h if h > 0 else self.font_size_pil
+            else: 
+                return int(len(text_content)*self.font_size_pil*0.6),int(self.font_size_pil*1.2 if self.font_size_pil*1.2>0 else self.font_size_pil)
+        except Exception as e:
+            logger.warning(f"Error in _get_text_dimensions for '{text_content[:20]}...': {e}")
+            return int(len(text_content)*self.font_size_pil*0.6),int(self.font_size_pil*1.2) # Fallback
     
-    def _create_placeholder_image_content(self,text_description,filename,size=None): # Same
+    def _create_placeholder_image_content(self,text_description,filename,size=None):
         if size is None: size = self.video_frame_size
         img=Image.new('RGB',size,color=(20,20,40));d=ImageDraw.Draw(img);padding=25;max_w=size[0]-(2*padding);lines=[];
         if not text_description: text_description="(Placeholder: No prompt text)"
@@ -91,12 +151,19 @@ class VisualEngine:
             test_line=current_line+word+" "; 
             if self._get_text_dimensions(test_line,self.font)[0] <= max_w: current_line=test_line
             else:
-                if current_line: lines.append(current_line.strip()); current_line=word+" "
-        if current_line: lines.append(current_line.strip())
-        if not lines: lines.append("(Text error or too long for placeholder)")
+                if current_line: lines.append(current_line.strip()); 
+                current_line=word+" "
+        if current_line.strip(): lines.append(current_line.strip()) # Add last line
+        if not lines and text_description: lines.append(text_description[:max_w//int(self.font_size_pil*0.6 +1)]+"..." if text_description else "(Text too long)") # Handle single very long word
+        elif not lines: lines.append("(Placeholder Text Error)")
+
         _,single_line_h=self._get_text_dimensions("Ay",self.font); single_line_h = single_line_h if single_line_h > 0 else self.font_size_pil + 2
-        max_lines_to_display=min(len(lines),(size[1]-(2*padding))//(single_line_h+2))
+        
+        max_lines_to_display=min(len(lines),(size[1]-(2*padding))//(single_line_h+2)) if single_line_h > 0 else 1
+        if max_lines_to_display <=0: max_lines_to_display = 1 # Ensure at least one line can be attempted
+
         y_text=padding + (size[1]-(2*padding) - max_lines_to_display*(single_line_h+2))/2.0
+
         for i in range(max_lines_to_display):
             line_content=lines[i];line_w,_=self._get_text_dimensions(line_content,self.font);x_text=(size[0]-line_w)/2.0
             d.text((x_text,y_text),line_content,font=self.font,fill=(200,200,180));y_text+=single_line_h+2
@@ -105,7 +172,7 @@ class VisualEngine:
         try:img.save(filepath);return filepath
         except Exception as e:logger.error(f"Saving placeholder image {filepath}: {e}", exc_info=True);return None
 
-    def _search_pexels_image(self, query, output_filename_base): # Same
+    def _search_pexels_image(self, query, output_filename_base):
         if not self.USE_PEXELS or not self.pexels_api_key: return None
         headers = {"Authorization": self.pexels_api_key}; params = {"query": query, "per_page": 1, "orientation": "landscape", "size": "large"}
         pexels_filename = output_filename_base.replace(".png", f"_pexels_{random.randint(1000,9999)}.jpg")
@@ -120,11 +187,11 @@ class VisualEngine:
                 img_data = Image.open(io.BytesIO(image_response.content))
                 if img_data.mode != 'RGB': img_data = img_data.convert('RGB')
                 img_data.save(filepath); logger.info(f"Pexels image saved: {filepath}"); return filepath
-            else: logger.info(f"No photos on Pexels for query: '{effective_query}'")
+            else: logger.info(f"No photos found on Pexels for query: '{effective_query}'")
         except Exception as e: logger.error(f"Pexels search/download for query '{query}': {e}", exc_info=True)
         return None
 
-    def generate_image_visual(self, image_prompt_text, scene_data, scene_identifier_filename): # Same
+    def generate_image_visual(self, image_prompt_text, scene_data, scene_identifier_filename):
         filepath = os.path.join(self.output_dir, scene_identifier_filename)
         if self.USE_AI_IMAGE_GENERATION and self.openai_api_key:
             max_retries = 2
@@ -155,32 +222,33 @@ class VisualEngine:
         else: 
             return self._create_placeholder_image_content(image_prompt_text, scene_identifier_filename)
 
-    def generate_narration_audio(self, text_to_narrate, output_filename="narration_overall.mp3"): # Corrected client call
+    def generate_narration_audio(self, text_to_narrate, output_filename="narration_overall.mp3"):
         if not self.USE_ELEVENLABS or not self.elevenlabs_client or not text_to_narrate:
-            logger.info("ElevenLabs conditions not met. Skipping audio generation.")
+            logger.info("ElevenLabs conditions not met (API key, client init, or text). Skipping audio.")
             return None
         
         audio_filepath = os.path.join(self.output_dir, output_filename)
         try:
             logger.info(f"Generating ElevenLabs audio (Voice ID: {self.elevenlabs_voice_id}) for: {text_to_narrate[:70]}...")
             
-            # Use the client's text_to_speech.stream() method
+            # Using client.text_to_speech.stream() for newer SDKs (e.g., elevenlabs >= 1.0)
             if hasattr(self.elevenlabs_client, 'text_to_speech') and hasattr(self.elevenlabs_client.text_to_speech, 'stream'):
                 logger.info("Using elevenlabs_client.text_to_speech.stream()")
                 audio_data_iterator = self.elevenlabs_client.text_to_speech.stream(
                     text=text_to_narrate,
-                    voice_id=str(self.elevenlabs_voice_id), # Must be string ID for this method
-                    model_id="eleven_multilingual_v2", 
-                    # voice_settings=self.elevenlabs_voice_settings # Pass VoiceSettings if API supports it here
+                    voice_id=str(self.elevenlabs_voice_id), 
+                    model_id="eleven_multilingual_v2",
+                    # voice_settings=self.elevenlabs_voice_settings # Pass VoiceSettings object if supported by stream
                 )
-            # Fallback if .generate() with Voice object is preferred by a slightly different client version
+            # Fallback to direct .generate() if text_to_speech.stream isn't there 
+            # AND Voice/VoiceSettings were successfully imported
             elif hasattr(self.elevenlabs_client, 'generate') and Voice and self.elevenlabs_voice_settings:
                  logger.info("Using elevenlabs_client.generate() with Voice object as fallback.")
                  voice_param = Voice(voice_id=str(self.elevenlabs_voice_id), settings=self.elevenlabs_voice_settings)
                  audio_data_iterator = self.elevenlabs_client.generate(
                     text=text_to_narrate, voice=voice_param, model="eleven_multilingual_v2")
             else:
-                logger.error("No recognized audio generation method found on ElevenLabs client (tried text_to_speech.stream and generate).")
+                logger.error("No recognized audio generation method found on ElevenLabs client or Voice/VoiceSettings not imported.")
                 return None
             
             with open(audio_filepath, "wb") as f:
@@ -189,7 +257,7 @@ class VisualEngine:
             logger.info(f"ElevenLabs audio saved: {audio_filepath}")
             return audio_filepath
         except AttributeError as ae:
-             logger.error(f"AttributeError with ElevenLabs client: {ae}. SDK method/structure might be different.", exc_info=True)
+             logger.error(f"AttributeError with ElevenLabs client: {ae}. SDK method might be different.", exc_info=True)
         except Exception as e:
             logger.error(f"Error generating ElevenLabs audio: {e}", exc_info=True)
         return None
@@ -207,7 +275,6 @@ class VisualEngine:
                 if pil_img.mode != 'RGB': pil_img = pil_img.convert('RGB')
                 
                 img_copy = pil_img.copy()
-                # Using modern Resampling.LANCZOS
                 img_copy.thumbnail(self.video_frame_size, Image.Resampling.LANCZOS) 
                 
                 canvas = Image.new('RGB', self.video_frame_size, (random.randint(0,5), random.randint(0,5), random.randint(0,5))) 
@@ -217,24 +284,17 @@ class VisualEngine:
                 
                 img_clip_base = ImageClip(frame_np).set_duration(duration_per_image)
 
-                # Ken Burns Effect (vfx.resize) - THIS IS THE LINE THAT CAUSED ANTIALIAS ERROR
-                # If ANTIALIAS error persists even with monkey patch and correct Pillow/MoviePy versions,
-                # this specific effect might need to be disabled or re-implemented differently.
-                # For now, we assume the monkey patch or correct versions will fix it.
+                # Ken Burns Effect (vfx.resize)
+                # The monkey patch at the top of this file should address Image.ANTIALIAS
+                img_clip = img_clip_base # Start with base
                 try:
-                    end_scale = random.uniform(1.03, 1.08)
-                    img_clip = img_clip_base.fx(vfx.resize, lambda t: 1 + (end_scale - 1) * (t / duration_per_image))
+                    end_scale = random.uniform(1.03, 1.08); 
+                    img_clip = img_clip_base.fx(vfx.resize, lambda t: 1+(end_scale-1)*(t/duration_per_image))
                     img_clip = img_clip.set_position('center')
-                except AttributeError as e_alias: # Specifically catch ANTIALIAS here if it happens
-                    if 'ANTIALIAS' in str(e_alias):
-                        logger.error(f"ANTIALIAS error during vfx.resize for {img_path}. Disabling Ken Burns for this clip. Error: {e_alias}")
-                        img_clip = img_clip_base # Use the clip without the resize effect
-                    else:
-                        raise # Re-raise other AttributeErrors
-                except Exception as e_fx: # Catch other errors from fx
-                    logger.error(f"Error applying vfx.resize for {img_path}. Using base clip. Error: {e_fx}")
-                    img_clip = img_clip_base
-
+                except AttributeError as e_alias: 
+                    if 'ANTIALIAS' in str(e_alias): logger.error(f"ANTIALIAS error in vfx.resize for {img_path}. Ken Burns disabled for this clip. Error: {e_alias}")
+                    else: raise # Re-raise other AttributeErrors
+                except Exception as e_fx: logger.error(f"Error in vfx.resize for {img_path}: {e_fx}. Ken Burns disabled for this clip.")
 
                 if key_action:
                     txt_clip = TextClip(f"Scene {scene_num}\n{key_action}", fontsize=self.video_overlay_font_size, 
@@ -258,8 +318,12 @@ class VisualEngine:
             if overall_narration_path and os.path.exists(overall_narration_path):
                 try:
                     narration_audio_clip = AudioFileClip(overall_narration_path)
+                    # Adjust video duration to match audio if audio is shorter.
+                    # If audio is longer, MoviePy's set_audio will truncate audio to video length.
                     if narration_audio_clip.duration < final_video_clip_obj.duration:
+                        logger.info(f"Narration shorter than visuals. Trimming video to {narration_audio_clip.duration}s.")
                         final_video_clip_obj = final_video_clip_obj.subclip(0, narration_audio_clip.duration)
+                    
                     final_video_clip_obj = final_video_clip_obj.set_audio(narration_audio_clip); logger.info("Overall narration added.")
                 except Exception as e: logger.error(f"Adding overall narration: {e}", exc_info=True)