Update core/visual_engine.py

core/visual_engine.py

@@ -10,13 +10,14 @@ import requests
 import io
 import time
 import random 
-import subprocess # For the dummy video fallback
+import subprocess # For dummy video fallback
 
 # --- ElevenLabs Import ---
+# These will be assigned if the import is successful
 ELEVENLABS_CLIENT_IMPORTED = False
-ElevenLabsAPIClient = None # Placeholder for the class
-Voice = None               # Placeholder for the class
-VoiceSettings = None       # Placeholder for the class
+ElevenLabsAPIClient = None 
+Voice = None              
+VoiceSettings = None      
 
 try:
     from elevenlabs.client import ElevenLabs as ImportedElevenLabsClient 
@@ -26,11 +27,11 @@ try:
     Voice = ImportedVoice
     VoiceSettings = ImportedVoiceSettings
     ELEVENLABS_CLIENT_IMPORTED = True
-    print("Successfully imported ElevenLabs client components (SDK v1.x.x pattern).")
+    print("INFO: Successfully imported ElevenLabs client components (SDK v1.x.x pattern).")
 except ImportError as e_eleven:
-    print(f"WARNING: Could not import ElevenLabs client components: {e_eleven}. ElevenLabs audio generation will be disabled.")
-except Exception as e_gen_eleven: # Catch any other general import error for elevenlabs
-    print(f"WARNING: General error importing ElevenLabs: {e_gen_eleven}. ElevenLabs audio generation will be disabled.")
+    print(f"WARNING: Could not import ElevenLabs client components: {e_eleven}. ElevenLabs audio will be disabled.")
+except Exception as e_gen_eleven: 
+    print(f"WARNING: General error importing ElevenLabs: {e_gen_eleven}. ElevenLabs audio will be disabled.")
 
 
 class VisualEngine:
@@ -43,66 +44,66 @@ class VisualEngine:
         self.font_size_pil = 20 
         self.video_overlay_font_size = 30 
         self.video_overlay_font_color = 'white'
-        self.video_overlay_font = 'Arial-Bold' 
+        self.video_overlay_font = 'Arial-Bold' # Relies on ImageMagick font discovery
 
         try:
             self.font = ImageFont.truetype(self.font_path_in_container, self.font_size_pil)
-            # print(f"Placeholder font loaded: {self.font_path_in_container}.") # Less verbose
+            print(f"INFO: Placeholder font loaded: {self.font_path_in_container}.")
         except IOError:
-            print(f"Warning: Placeholder font '{self.font_path_in_container}' not loaded. Using default.")
+            print(f"WARNING: 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)
+        # API Client States
+        self.openai_api_key = None; self.USE_AI_IMAGE_GENERATION = False
+        self.dalle_model = "dall-e-3"; self.image_size_dalle3 = "1792x1024" # Landscape orientation
+        self.video_frame_size = (1280, 720) # 16:9 standard HD
 
-        # ElevenLabs Client
-        self.elevenlabs_api_key = None
-        self.USE_ELEVENLABS = False
+        self.elevenlabs_api_key = None; self.USE_ELEVENLABS = False
         self.elevenlabs_client = None 
-        self.elevenlabs_voice_id = "Rachel" # Default, can be name or ID
+        self.elevenlabs_voice_id = "Rachel" # Default
         if VoiceSettings: # Check if VoiceSettings was successfully imported
             self.elevenlabs_voice_settings = VoiceSettings(
-                stability=0.65, similarity_boost=0.75, 
-                style=0.1, use_speaker_boost=True
+                stability=0.60, similarity_boost=0.80, # Adjusted for potentially more character
+                style=0.15, use_speaker_boost=True
             )
         else:
             self.elevenlabs_voice_settings = None
 
-        self.pexels_api_key = None
-        self.USE_PEXELS = False
+        self.pexels_api_key = None; self.USE_PEXELS = False
+        print("INFO: VisualEngine initialized.")
 
     def set_openai_api_key(self,k): 
-        self.openai_api_key=k
-        self.USE_AI_IMAGE_GENERATION=bool(k)
-        # print(f"DALL-E ({self.dalle_model}) {'Ready' if k else 'Disabled'}.")
+        self.openai_api_key=k; self.USE_AI_IMAGE_GENERATION=bool(k)
+        print(f"INFO: DALL-E ({self.dalle_model}) {'Ready.' if k else 'Disabled (no API key).'}")
 
     def set_elevenlabs_api_key(self,api_key):
         self.elevenlabs_api_key=api_key
         if api_key and ELEVENLABS_CLIENT_IMPORTED and ElevenLabsAPIClient: 
             try: 
                 self.elevenlabs_client = ElevenLabsAPIClient(api_key=api_key) 
-                # Optional: Test client (e.g., fetch voices) can be added here for robust init
-                # voices_test = self.elevenlabs_client.voices.get_all() # This makes an API call
-                # if voices_test and voices_test.voices: print("ElevenLabs client connected.")
-                self.USE_ELEVENLABS=True
-                # print("ElevenLabs Client Ready.")
+                # Minimal test: Check if client object was created.
+                # Fetching voices makes an API call, can do it later if needed.
+                if self.elevenlabs_client:
+                    self.USE_ELEVENLABS=True
+                    print("INFO: ElevenLabs Client Ready.")
+                else: # Should not happen if ElevenLabsAPIClient() doesn't raise error
+                    print("WARNING: ElevenLabs client is None after init. ElevenLabs Disabled.")
+                    self.USE_ELEVENLABS=False
             except Exception as e: 
-                print(f"Error initializing ElevenLabs client with API key: {e}. ElevenLabs Disabled."); 
+                print(f"ERROR: Initializing ElevenLabs client: {e}. ElevenLabs Disabled."); 
                 self.USE_ELEVENLABS=False; self.elevenlabs_client = None
         else: 
             self.USE_ELEVENLABS=False; self.elevenlabs_client = None
-            # if not ELEVENLABS_CLIENT_IMPORTED or not ElevenLabsAPIClient:
-                # print("ElevenLabs Client class was not imported. ElevenLabs Disabled.") # Already printed at import
-            # else:
-                # print("ElevenLabs API Key not provided. ElevenLabs Disabled.") # Less verbose
+            if not ELEVENLABS_CLIENT_IMPORTED or not ElevenLabsAPIClient:
+                # This message is already printed at import time if it fails
+                pass 
+            else:
+                print("INFO: ElevenLabs API Key not provided or client class not imported. ElevenLabs Disabled.")
             
     def set_pexels_api_key(self,k):
         self.pexels_api_key=k; self.USE_PEXELS=bool(k)
-        # print(f"Pexels {'Ready' if k else 'Disabled'}.")
+        print(f"INFO: Pexels Search {'Ready.' if k else 'Disabled (no API key).'}")
     
     def _get_text_dimensions(self,text_content,font_obj):
         if not text_content: return 0,self.font_size_pil 
@@ -113,12 +114,13 @@ class VisualEngine:
             elif hasattr(font_obj,'getsize'): 
                 w,h=font_obj.getsize(text_content)
                 return w, h if h > 0 else self.font_size_pil
-            else: # Fallback
+            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: # Generic fallback on error
+        except Exception: 
             return int(len(text_content)*self.font_size_pil*0.6),int(self.font_size_pil*1.2)
     
-    def _create_placeholder_image_content(self,text_description,filename,size=(1280,720)):
+    def _create_placeholder_image_content(self,text_description,filename,size=None):
+        if size is None: size = self.video_frame_size # Default to 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)"
         words=text_description.split();current_line=""
@@ -134,40 +136,39 @@ class VisualEngine:
         _,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 based on height
+        max_lines_to_display=min(len(lines),(size[1]-(2*padding))//(single_line_h+2))
         
         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
-            if i==6 and max_lines_to_display > 7: # Show ellipsis if more text
+            if i==6 and max_lines_to_display > 7: 
                 d.text((x_text,y_text),"...",font=self.font,fill=(200,200,180));break
         filepath=os.path.join(self.output_dir,filename)
         try:img.save(filepath);return filepath
-        except Exception as e:print(f"Error saving placeholder image {filepath}: {e}");return None
+        except Exception as e:print(f"ERROR: Saving placeholder image {filepath}: {e}");return None
 
     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"} # Get only 1 relevant image
-        pexels_filename = output_filename_base.replace(".png", f"_pexels_{random.randint(100,999)}.jpg")
+        params = {"query": query, "per_page": 1, "orientation": "landscape", "size": "large"}
+        pexels_filename = output_filename_base.replace(".png", f"_pexels_{random.randint(1000,9999)}.jpg")
         filepath = os.path.join(self.output_dir, pexels_filename)
         try:
-            print(f"Searching Pexels for: '{query}'")
-            effective_query = " ".join(query.split()[:5]) # Use first 5 words for Pexels query
+            print(f"INFO: Searching Pexels for: '{query}'")
+            effective_query = " ".join(query.split()[:5]) 
             params["query"] = effective_query
-
             response = requests.get("https://api.pexels.com/v1/search", headers=headers, params=params, timeout=20)
             response.raise_for_status(); data = response.json()
             if data.get("photos") and len(data["photos"]) > 0:
-                photo_url = data["photos"][0]["src"]["large2x"] # High quality
+                photo_url = data["photos"][0]["src"]["large2x"] 
                 image_response = requests.get(photo_url, timeout=60); image_response.raise_for_status()
                 img_data = Image.open(io.BytesIO(image_response.content))
                 if img_data.mode != 'RGB': img_data = img_data.convert('RGB')
-                img_data.save(filepath); print(f"Pexels image saved: {filepath}"); return filepath
-            else: print(f"No photos found on Pexels for query: '{effective_query}'")
-        except Exception as e: print(f"Pexels search/download error for query '{query}': {e}")
+                img_data.save(filepath); print(f"INFO: Pexels image saved: {filepath}"); return filepath
+            else: print(f"INFO: No photos found on Pexels for query: '{effective_query}'")
+        except Exception as e: print(f"ERROR: Pexels search/download for query '{query}': {e}")
         return None
 
     def generate_image_visual(self, image_prompt_text, scene_data, scene_identifier_filename):
@@ -176,57 +177,58 @@ class VisualEngine:
             max_retries = 2
             for attempt in range(max_retries):
                 try:
-                    print(f"Attempt {attempt+1}: DALL-E ({self.dalle_model}) for: {image_prompt_text[:120]}...")
-                    client = openai.OpenAI(api_key=self.openai_api_key, timeout=90.0)
+                    print(f"INFO: Attempt {attempt+1}: DALL-E ({self.dalle_model}) for: {image_prompt_text[:100]}...")
+                    client = openai.OpenAI(api_key=self.openai_api_key, timeout=90.0) 
                     response = client.images.generate(
                         model=self.dalle_model, prompt=image_prompt_text, n=1, 
                         size=self.image_size_dalle3, quality="hd", response_format="url", style="vivid"
                     )
                     image_url = response.data[0].url
                     revised_prompt = getattr(response.data[0], 'revised_prompt', None)
-                    if revised_prompt: print(f"DALL-E 3 revised_prompt: {revised_prompt[:100]}...")
+                    if revised_prompt: print(f"INFO: DALL-E 3 revised_prompt: {revised_prompt[:100]}...")
                     
                     image_response = requests.get(image_url, timeout=120)
                     image_response.raise_for_status()
                     img_data = Image.open(io.BytesIO(image_response.content))
                     if img_data.mode != 'RGB': img_data = img_data.convert('RGB')
                     
-                    img_data.save(filepath); print(f"AI Image (DALL-E) saved: {filepath}"); return filepath 
+                    img_data.save(filepath); print(f"INFO: AI Image (DALL-E) saved: {filepath}"); return filepath 
                 except openai.RateLimitError as e: 
-                    print(f"OpenAI Rate Limit: {e}. Retrying after {5*(attempt+1)}s...")
+                    print(f"WARNING: OpenAI Rate Limit: {e}. Retrying after {5*(attempt+1)}s...")
                     time.sleep(5 * (attempt + 1))
-                    if attempt == max_retries - 1: print("Max retries for RateLimitError."); break
+                    if attempt == max_retries - 1: print("ERROR: Max retries for RateLimitError."); break
                     else: continue
-                except openai.APIError as e: print(f"OpenAI API Error: {e}"); break 
-                except requests.exceptions.RequestException as e: print(f"Requests Error (DALL-E download): {e}"); break
-                except Exception as e: print(f"Generic error (DALL-E gen): {e}"); break
+                except openai.APIError as e: print(f"ERROR: OpenAI API Error: {e}"); break 
+                except requests.exceptions.RequestException as e: print(f"ERROR: Requests Error (DALL-E download): {e}"); break
+                except Exception as e: print(f"ERROR: Generic error (DALL-E gen): {e}"); break
             
-            print("DALL-E generation failed. Trying Pexels fallback...")
+            print("WARNING: DALL-E generation failed. Trying Pexels fallback...")
             pexels_query_text = scene_data.get('pexels_search_query_감독', 
                                              f"{scene_data.get('emotional_beat','')} {scene_data.get('setting_description','')}")
             pexels_path = self._search_pexels_image(pexels_query_text, scene_identifier_filename)
             if pexels_path: return pexels_path
             
-            print("Pexels also failed/disabled. Using placeholder.")
+            print("WARNING: Pexels also failed/disabled. Using placeholder.")
             return self._create_placeholder_image_content(
                 f"[AI/Pexels Failed] Original Prompt: {image_prompt_text[:100]}...", 
-                scene_identifier_filename, size=self.video_frame_size
+                scene_identifier_filename 
             )
         else: 
+            # print(f"INFO: AI image generation not enabled/ready. Creating placeholder.")
             return self._create_placeholder_image_content(
-                image_prompt_text, scene_identifier_filename, size=self.video_frame_size
+                image_prompt_text, scene_identifier_filename 
             )
 
     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:
-            # print("ElevenLabs not enabled, client not initialized, or no text. Skipping audio.") # Less verbose
+            # print("INFO: ElevenLabs not enabled, client not initialized, or no text. Skipping audio.")
             return None
         
         audio_filepath = os.path.join(self.output_dir, output_filename)
         try:
-            print(f"Generating ElevenLabs audio (Voice: {self.elevenlabs_voice_id}) for: {text_to_narrate[:70]}...")
+            print(f"INFO: Generating ElevenLabs audio (Voice: {self.elevenlabs_voice_id}) for: {text_to_narrate[:70]}...")
             
-            voice_param = self.elevenlabs_voice_id # Default to string ID
+            voice_param = self.elevenlabs_voice_id 
             if Voice and self.elevenlabs_voice_settings: # Check if Voice & VoiceSettings were imported
                 voice_param = Voice(
                     voice_id=self.elevenlabs_voice_id,
@@ -236,29 +238,28 @@ class VisualEngine:
             audio_data_iterator = self.elevenlabs_client.generate(
                 text=text_to_narrate,
                 voice=voice_param, 
-                model="eleven_multilingual_v2" # Or other models e.g. "eleven_turbo_v2"
+                model="eleven_multilingual_v2" 
             )
             
             with open(audio_filepath, "wb") as f:
                 for chunk in audio_data_iterator: 
                     if chunk: f.write(chunk)
             
-            print(f"ElevenLabs audio saved: {audio_filepath}")
+            print(f"INFO: ElevenLabs audio saved: {audio_filepath}")
             return audio_filepath
         except AttributeError as ae:
-             print(f"AttributeError with ElevenLabs client (method name like 'generate' might differ): {ae}")
+             print(f"ERROR: AttributeError with ElevenLabs client (method 'generate' might be different or client not fully init): {ae}")
         except Exception as e:
-            print(f"Error generating ElevenLabs audio: {e}")
+            print(f"ERROR: Generating ElevenLabs audio: {e}")
         return None
 
     def create_video_from_images(self, image_data_list, overall_narration_path=None, output_filename="final_video.mp4", fps=24, duration_per_image=4.5):
-        if not image_data_list: print("No image data for video."); return None
-        # print(f"Creating video from {len(image_data_list)} image sets.") # Less verbose
+        if not image_data_list: print("WARNING: No image data for video."); return None
         processed_clips = []; narration_audio_clip = None; final_video_clip_obj = None
 
         for i, data in enumerate(image_data_list):
             img_path, scene_num, key_action = data.get('path'), data.get('scene_num', i+1), data.get('key_action', '')
-            if not (img_path and os.path.exists(img_path)): print(f"Img not found: {img_path}"); continue
+            if not (img_path and os.path.exists(img_path)): print(f"WARNING: Img not found: {img_path}"); continue
             try:
                 pil_img = Image.open(img_path); 
                 if pil_img.mode != 'RGB': pil_img = pil_img.convert('RGB')
@@ -281,9 +282,9 @@ class VisualEngine:
                     final_scene_clip = CompositeVideoClip([img_clip, txt_clip], size=self.video_frame_size)
                 else: final_scene_clip = img_clip
                 processed_clips.append(final_scene_clip)
-            except Exception as e: print(f"Error creating video clip for {img_path}: {e}.")
+            except Exception as e: print(f"ERROR: Creating video clip for {img_path}: {e}.")
         
-        if not processed_clips: print("No clips processed for video."); return None
+        if not processed_clips: print("WARNING: No clips processed for video."); return None
         transition = 0.8 
         final_video_clip_obj = concatenate_videoclips(processed_clips, padding=-transition, method="compose")
         if final_video_clip_obj.duration > transition*2: 
@@ -292,24 +293,26 @@ class VisualEngine:
         if overall_narration_path and os.path.exists(overall_narration_path):
             try:
                 narration_audio_clip = AudioFileClip(overall_narration_path)
-                current_video_duration = final_video_clip_obj.duration
-                # If narration is shorter, trim video. If narration is longer, audio will be cut by video duration.
-                if narration_audio_clip.duration < current_video_duration:
+                # Ensure video duration matches audio if audio is shorter, or cap audio at video length.
+                # MoviePy's set_audio will truncate the longer of the two to match the shorter one IF the video has no audio track yet.
+                # If video already has audio, it replaces. If video is shorter than new audio, video extends with last frame.
+                # We want audio to dictate length if it's shorter than visual sequence.
+                if narration_audio_clip.duration < final_video_clip_obj.duration:
                     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)
-                print("Overall narration added to video.")
-            except Exception as e: print(f"Error adding overall narration: {e}.")
+                print("INFO: Overall narration added to video.")
+            except Exception as e: print(f"ERROR: Adding overall narration: {e}.")
         
         output_path = os.path.join(self.output_dir, output_filename)
         try:
-            print(f"Writing final video to: {output_path}")
+            print(f"INFO: Writing final video to: {output_path}")
             final_video_clip_obj.write_videofile(output_path, fps=fps, codec='libx264', preset='medium', 
                                         audio_codec='aac',
                                         temp_audiofile=os.path.join(self.output_dir, f'temp-audio-{os.urandom(4).hex()}.m4a'), 
-                                        remove_temp=True, threads=os.cpu_count() or 2, logger='bar', bitrate="5000k") # Consider 'medium' preset
-            print(f"Video successfully created: {output_path}"); return output_path
-        except Exception as e: print(f"Error writing video file: {e}"); return None
+                                        remove_temp=True, threads=os.cpu_count() or 2, logger='bar', bitrate="5000k")
+            print(f"INFO: Video successfully created: {output_path}"); return output_path
+        except Exception as e: print(f"ERROR: Writing video file: {e}"); return None
         finally:
             for c_item in processed_clips: 
                 if hasattr(c_item, 'close'): c_item.close()