🐛 fix(GenerationContext.py): fix indentation issue in process() method

✨ feat(GenerationContext.py): add support for z-index of moviepy clips to improve video rendering
The indentation issue in the process() method has been fixed. The z-index of moviepy clips has been added to improve the rendering of the video. This allows the clips to be rendered in different layers based on their index, resulting in a more visually appealing video.

src/chore/GenerationContext.py

@@ -63,27 +63,33 @@ class GenerationContext:
     def process(self):
         # ⚠️ IMPORTANT NOTE: All methods called here are expected to be defined as abstract methods in the base classes, if not there is an issue with the engine implementation.
 
-        # we start by loading the settings
+        # Kinda like in css, we have a z-index of moviepy clips (any). Then the engines append some clips to this, and we render it all with index 0 below, and index 9 at the top.
+        self.index_0 = []
+        self.index_1 = []
+        self.index_2 = []
+        self.index_3 = []
+        self.index_4 = []
+        self.index_5 = []
+        self.index_6 = []
+        self.index_7 = []
+        self.index_8 = []
+        self.index_9 = []
 
         self.progress(0.1, "Loading settings...")
+        self.setup_dir()
         self.settingsengine.load()
 
-        self.setup_dir()
-
         self.progress(0.2, "Generating script...")
-        self.script = self.scriptengine.generate()
+        self.scriptengine.generate()
 
         self.progress(0.3, "Generating synthtetizing voice...")
-        self.timed_script = self.ttsengine.synthesize(
+        self.ttsengine.synthesize(self.script, self.get_file_path("tts.wav"))
-            self.script, self.get_file_path("tts.wav")
+        self.duration: float #for type hinting
-        )
-
-        self.assets = []
 
         if not isinstance(self.backgroundengine, engines.NoneEngine):
             self.progress(0.4, "Generating background...")
-            self.background = self.backgroundengine.get_background()
+            self.backgroundengine.get_background()
-            self.assets.append(self.background)
+
         self.assetsengine = [
             engine
             for engine in self.assetsengine
@@ -91,21 +97,32 @@ class GenerationContext:
         ]
         if len(self.assetsengine) > 0:
             self.progress(0.5, "Generating assets...")
-            self.assets.extend(self.assetsengineselector.get_assets())
+            self.assetsengineselector.get_assets()
 
         if not isinstance(self.captioningengine, engines.NoneEngine):
             self.progress(0.6, "Generating captions...")
-            self.captions = self.captioningengine.get_captions()
+            self.captioningengine.get_captions()
         else:
             self.captions = []
 
         # add any other processing steps here
 
         # we render to a file called final.mp4
-        # using moviepy CompositeVideoClip
         self.progress(0.7, "Rendering video...")
-        clips = [*self.assets, *self.captions]
+        clips = [
+            *self.index_0,
+            *self.index_1,
+            *self.index_2,
+            *self.index_3,
+            *self.index_4,
+            *self.index_5,
+            *self.index_6,
+            *self.index_7,
+            *self.index_8,
+            *self.index_9,
+        ]
         clip = mp.CompositeVideoClip(clips, size=(self.width, self.height))
+        clip.set_duration(self.duration)
         audio = mp.AudioFileClip(self.get_file_path("tts.wav"))
         clip = clip.set_audio(audio)
         clip.write_videofile(self.get_file_path("final.mp4"), fps=60)

src/engines/AssetsEngine/AssetsEngineSelector.py

@@ -33,4 +33,4 @@ class AssetsEngineSelector:
             assets_opts = [asset for asset in assets if asset["engine"] == engine.name]
             assets_opts = [asset["args"] for asset in assets_opts]
             clips.extend(engine.get_assets(assets_opts))
-        return clips
+        self.ctx.index_3.extend(clips)

src/engines/BackgroundEngine/BaseBackgroundEngine.py

@@ -6,5 +6,5 @@ from moviepy.editor import VideoClip
 
 class BaseBackgroundEngine(BaseEngine):
     @abstractmethod
-    def get_background(self) -> VideoClip:
+    def get_background(self) -> None:
         ...

src/engines/BackgroundEngine/VideoBackgroundEngine.py

@@ -10,7 +10,7 @@ from moviepy.video.fx.crop import crop
 from . import BaseBackgroundEngine
 
 
-class SimpleBackgroundEngine(BaseBackgroundEngine):
+class VideoBackgroundEngine(BaseBackgroundEngine):
     name = "SImple Background Engine"
     description = "A basic background engine to set the background of the video from a local file."
     num_options = 1
@@ -48,13 +48,15 @@ class SimpleBackgroundEngine(BaseBackgroundEngine):
         start = random.uniform(0, background_max_start)
         clip = background.subclip(start, start + self.ctx.duration)
         w, h = clip.size
-        return crop(
+        self.ctx.index_0.append(
+            crop(
                 clip,
                 width=self.ctx.width,
                 height=self.ctx.height,
                 x_center=w / 2,
                 y_center=h / 2,
             )
+        )
 
     @classmethod
     def get_settings(cls) -> list:

src/engines/BackgroundEngine/__init__.py

@@ -1,2 +1,2 @@
 from .BaseBackgroundEngine import BaseBackgroundEngine
-from .SimpleBackgroundEngine import SimpleBackgroundEngine
+from .VideoBackgroundEngine import VideoBackgroundEngine

src/engines/BaseEngine.py

@@ -19,7 +19,7 @@ class BaseEngine(ABC):
 
     @classmethod
     @abstractmethod
-    def get_options():
+    def get_options(cls):
         ...
 
     def get_video_duration(self, path: str) -> float:

src/engines/CaptioningEngine/BaseCaptioningEngine.py

@@ -6,5 +6,5 @@ from moviepy.editor import TextClip
 
 class BaseCaptioningEngine(BaseEngine):
     @abstractmethod
-    def get_captions(self) -> list[TextClip]:
+    def get_captions(self) -> None:
         ...

src/engines/CaptioningEngine/SimpleCaptioningEngine.py

@@ -39,7 +39,7 @@ class SimpleCaptioningEngine(BaseCaptioningEngine):
         punctuations = (".", "?", "!", ",", ":", ";")
         return text.strip().endswith(tuple(punctuations))
 
-    def get_captions(self) -> list[TextClip]:
+    def get_captions(self):
         # 3 words per 1000 px, we do the math
         max_words = int(self.ctx.width / 1000 * 3)
 
@@ -78,7 +78,7 @@ class SimpleCaptioningEngine(BaseCaptioningEngine):
                 )
             )
 
-        return clips
+        self.ctx.index_7.extend(clips)
 
     @classmethod
     def get_options(cls) -> list:

src/engines/MetadataEngine/ShortsMetadataEngine.py

@@ -4,6 +4,11 @@ from ...utils.prompting import get_prompt
 
 
 class ShortsMetadataEngine(BaseMetadataEngine):
+    name = "ShortsMetadata"
+    description = "Generate metadata for YouTube Shorts / TikTok format videos"
+
+    num_options = 0
+
     def __init__(self, **kwargs) -> None:
         ...
 
@@ -13,9 +18,12 @@ class ShortsMetadataEngine(BaseMetadataEngine):
         )
         chat_prompt = chat_prompt.replace("{script}", self.ctx.script)
 
-        return self.ctx.simplellmengine.generate(
+        result = self.ctx.simplellmengine.generate(
             chat_prompt=chat_prompt, system_prompt=sytsem_prompt, json_mode=True
         )
+        self.ctx.title = result["title"]
+        self.ctx.description = result["description"]
 
-    def get_options(self):
+    @classmethod
+    def get_options(cls):
         return []

src/engines/ScriptEngine/BaseScriptEngine.py

@@ -6,7 +6,7 @@ class BaseScriptEngine(BaseEngine):
     pass
 
     @abstractmethod
-    def generate(self) -> str:
+    def generate(self) -> None:
         pass
 
     def time_script(self):

src/engines/ScriptEngine/CustomScriptEngine.py

@@ -12,7 +12,7 @@ class CustomScriptEngine(BaseScriptEngine):
         super().__init__()
 
     def generate(self, *args, **kwargs) -> str:
-        return self.script
+        self.ctx.script = self.script.strip().copy()
 
     @classmethod
     def get_options(cls) -> list:

src/engines/ScriptEngine/ShowerThoughtsScriptEngine.py

@@ -23,13 +23,17 @@ class ShowerThoughtsScriptEngine(BaseScriptEngine):
         )
         sys_prompt = sys_prompt.format(n_sentences=self.n_sentences)
         chat_prompt = chat_prompt.format(n_sentences=self.n_sentences)
-        return self.ctx.powerfulllmengine.generate(
+        self.ctx.script = (
+            self.ctx.powerfulllmengine.generate(
                 system_prompt=sys_prompt,
                 chat_prompt=chat_prompt,
                 max_tokens=20 * self.n_sentences,
                 temperature=1.3,
                 json_mode=False,
             )
+            .strip()
+            .copy()
+        )
 
     @classmethod
     def get_options(cls) -> list:

src/engines/TTSEngine/BaseTTSEngine.py

@@ -16,7 +16,7 @@ class Word(TypedDict):
 
 class BaseTTSEngine(BaseEngine):
     @abstractmethod
-    def synthesize(self, text: str, path: str) -> list[Word]:
+    def synthesize(self, text: str, path: str) -> None:
         pass
 
     def remove_punctuation(self, text: str) -> str:

src/engines/TTSEngine/CoquiTTSEngine.py

@@ -127,7 +127,7 @@ class CoquiTTSEngine(BaseTTSEngine):
 
         self.ctx.duration = self.get_audio_duration(path)
 
-        return self.time_with_whisper(path)
+        self.ctx.timed_script = self.time_with_whisper(path)
 
     @classmethod
     def get_options(cls) -> list:

src/engines/TTSEngine/ElevenLabsTTSEngine.py

@@ -1,19 +0,0 @@
-from .BaseTTSEngine import BaseTTSEngine
-import gradio as gr
-
-
-class ElevenLabsTTSEngine(BaseTTSEngine):
-    name = "ElevenLabs"
-    description = "ElevenLabs TTS engine."
-    num_options = 0
-
-    def __init__(self, options: list[list | tuple | str | int | float | bool | None]):
-        # self.voice = options[0][0]
-        super().__init__()
-
-    def synthesize(self, text: str, path: str) -> str:
-        pass
-
-    @classmethod
-    def get_options(cls) -> list:
-        return []

src/engines/__init__.py

@@ -50,7 +50,7 @@ ENGINES: dict[str, EngineDict] = {
         "multiple": True,
     },
     "BackgroundEngine": {
-        "classes": [BackgroundEngine.SimpleBackgroundEngine, NoneEngine],
+        "classes": [BackgroundEngine.VideoBackgroundEngine, NoneEngine],
         "multiple": False,
     },
     "MetadataEngine": {

ui/gradio_ui.py

@@ -75,11 +75,8 @@ class GenerateUI:
                             inputs.append(engine_dropdown)
                             engine_rows = []
                             for i, engine in enumerate(engines):
-                                with gr.Group(visible=(i == 0)) as engine_row:
+                                with gr.Row(visible=(i == 0)) as engine_row:
-                                    gr.Markdown(
+                                    gr.Markdown(value=f"## {engine.name}")
-                                        value=f"## {engine.name}",
-                                        render=False,
-                                    )
                                     engine_rows.append(engine_row)
                                     options = engine.get_options()
                                     inputs.extend(options)