1, 支持AI生成文案预览

2, 支持自定义视频文案，关键词
3, 可选择是否启用字幕
4, UI优化
5, 一些其他bug修复和优化

webui/Main.py

@@ -1,11 +1,12 @@
 import asyncio
 import sys
 import os
+import time
 from uuid import uuid4
 import streamlit as st
 from loguru import logger
-from app.models.schema import VideoParams, VideoAspect, VoiceNames
-from app.services import task as tm
+from app.models.schema import VideoParams, VideoAspect, VoiceNames, VideoConcatMode
+from app.services import task as tm, llm
 
 st.set_page_config(page_title="MoneyPrinterTurbo", page_icon="🤖", layout="wide",
                    initial_sidebar_state="auto")
@@ -15,6 +16,15 @@ root_dir = os.path.dirname(os.path.dirname(os.path.realpath(__file__)))
 font_dir = os.path.join(root_dir, "resource", "fonts")
 song_dir = os.path.join(root_dir, "resource", "songs")
 
+# st.session_state
+
+if 'video_subject' not in st.session_state:
+    st.session_state['video_subject'] = ''
+if 'video_script' not in st.session_state:
+    st.session_state['video_script'] = ''
+if 'video_terms' not in st.session_state:
+    st.session_state['video_terms'] = ''
+
 
 def get_all_fonts():
     fonts = []
@@ -65,17 +75,61 @@ def init_log():
 
 init_log()
 
-panel = st.columns(2)
+panel = st.columns(3)
 left_panel = panel[0]
-right_panel = panel[1]
+middle_panel = panel[1]
+right_panel = panel[2]
 
 # define cfg as VideoParams class
 cfg = VideoParams()
 
 with left_panel:
+    with st.container(border=True):
+        st.write("**文案设置**")
+        cfg.video_subject = st.text_input("视频主题（给定一个关键词，:red[AI自动生成]视频文案）",
+                                          value=st.session_state['video_subject']).strip()
+        if st.button("点击使用AI根据**主题**生成 【视频文案】 和 【视频关键词】", key="auto_generate_script"):
+            with st.spinner("AI正在生成视频文案和关键词..."):
+                script = llm.generate_script(cfg.video_subject)
+                terms = llm.generate_terms(cfg.video_subject, script)
+                st.toast('AI生成成功')
+                st.session_state['video_script'] = script
+                st.session_state['video_terms'] = ", ".join(terms)
+
+        cfg.video_script = st.text_area(
+            "视频文案（:blue[①可不填，使用AI生成  ②合理使用标点断句，有助于生成字幕]）",
+            value=st.session_state['video_script'],
+            height=190
+        )
+        if st.button("点击使用AI根据**文案**生成【视频关键词】", key="auto_generate_terms"):
+            if not cfg.video_script:
+                st.error("请先填写视频文案")
+                st.stop()
+
+            with st.spinner("AI正在生成视频关键词..."):
+                terms = llm.generate_terms(cfg.video_subject, cfg.video_script)
+                st.toast('AI生成成功')
+                st.session_state['video_terms'] = ", ".join(terms)
+
+        cfg.video_terms = st.text_area(
+            "视频关键词（:blue[①可不填，使用AI生成 ②用**英文逗号**分隔，只支持英文]）",
+            value=st.session_state['video_terms'],
+            height=40)
+
+with middle_panel:
     with st.container(border=True):
         st.write("**视频设置**")
-        cfg.video_subject = st.text_area("视频主题", help="请输入视频主题")
+        video_concat_modes = [
+            ("顺序拼接", "sequential"),
+            ("随机拼接", "random"),
+        ]
+        selected_index = st.selectbox("视频拼接模式",
+                                      index=1,
+                                      options=range(len(video_concat_modes)),  # 使用索引作为内部选项值
+                                      format_func=lambda x: video_concat_modes[x][0]  # 显示给用户的是标签
+                                      )
+        cfg.video_concat_mode = VideoConcatMode(video_concat_modes[selected_index][1])
+
         video_aspect_ratios = [
             ("竖屏 9:16（抖音视频）", VideoAspect.portrait.value),
             ("横屏 16:9（西瓜视频）", VideoAspect.landscape.value),
@@ -87,8 +141,10 @@ with left_panel:
                                       )
         cfg.video_aspect = VideoAspect(video_aspect_ratios[selected_index][1])
 
+        cfg.video_clip_duration = st.slider("视频片段最大时长(秒)", 2, 5, 3)
+
     with st.container(border=True):
-        st.write("**声音设置**")
+        st.write("**音频设置**")
         # 创建一个映射字典，将原始值映射到友好名称
         friendly_names = {
             voice: voice.
@@ -100,38 +156,53 @@ with left_panel:
             replace("en-US", "英文").
             replace("Neural", "") for
             voice in VoiceNames}
-        selected_friendly_name = st.selectbox("声音", options=list(friendly_names.values()))
+        selected_friendly_name = st.selectbox("朗读声音", options=list(friendly_names.values()))
         voice_name = list(friendly_names.keys())[list(friendly_names.values()).index(selected_friendly_name)]
         cfg.voice_name = voice_name
 
-        song_names = [
+        bgm_options = [
             ("无背景音乐 No BGM", ""),
             ("随机背景音乐 Random BGM", "random"),
-            *[(song, song) for song in get_all_songs()]
+            ("自定义背景音乐 Custom BGM", "custom"),
         ]
         selected_index = st.selectbox("背景音乐",
                                       index=1,
-                                      options=range(len(song_names)),  # 使用索引作为内部选项值
-                                      format_func=lambda x: song_names[x][0]  # 显示给用户的是标签
+                                      options=range(len(bgm_options)),  # 使用索引作为内部选项值
+                                      format_func=lambda x: bgm_options[x][0]  # 显示给用户的是标签
                                       )
-        cfg.bgm_name = song_names[selected_index][1]
+        # 获取选择的背景音乐类型
+        bgm_type = bgm_options[selected_index][1]
+
+        # 根据选择显示或隐藏组件
+        if bgm_type == "custom":
+            custom_bgm_file = st.text_input("请输入自定义背景音乐的文件路径：")
+            if custom_bgm_file and os.path.exists(custom_bgm_file):
+                cfg.bgm_file = custom_bgm_file
+                # st.write(f":red[已选择自定义背景音乐]：**{custom_bgm_file}**")
 
 with right_panel:
     with st.container(border=True):
         st.write("**字幕设置**")
+        cfg.subtitle_enabled = st.checkbox("生成字幕（若取消勾选，下面的设置都将不生效）", value=True)
         font_names = get_all_fonts()
         cfg.font_name = st.selectbox("字体", font_names)
-        cfg.text_fore_color = st.color_picker("字幕颜色", "#FFFFFF")
-        cfg.font_size = st.slider("字幕大小", 30, 100, 60)
-        cfg.stroke_color = st.color_picker("描边颜色", "#000000")
-        cfg.stroke_width = st.slider("描边粗细", 0.0, 10.0, 1.5)
+        font_cols = st.columns([0.3, 0.7])
+        with font_cols[0]:
+            cfg.text_fore_color = st.color_picker("字幕颜色", "#FFFFFF")
+        with font_cols[1]:
+            cfg.font_size = st.slider("字幕大小", 30, 100, 60)
+
+        stroke_cols = st.columns([0.3, 0.7])
+        with stroke_cols[0]:
+            cfg.stroke_color = st.color_picker("描边颜色", "#000000")
+        with stroke_cols[1]:
+            cfg.stroke_width = st.slider("描边粗细", 0.0, 10.0, 1.5)
 
 start_button = st.button("开始生成视频", use_container_width=True, type="primary")
 if start_button:
     task_id = str(uuid4())
-    st.session_state['started'] = True
-    if not cfg.video_subject:
-        st.error("视频主题不能为空")
+    if not cfg.video_subject and not cfg.video_script:
+        st.error("视频主题 或 视频文案，不能同时为空")
         st.stop()
 
     st.write(cfg)