feat: Add smart paragraph merging mode with AI grouping

- Add "smart" split mode that uses LLM to intelligently merge related paragraphs
- Implement two-step approach: analyze text structure, then group by semantic relevance
- Add paragraph_merging.py with analysis and grouping prompts
- Update UI to support smart mode selection with auto-detect hint
- Add i18n translations for smart mode (en_US, zh_CN)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

web/components/content_input.py

@@ -65,6 +65,7 @@ def render_content_input():
                     "paragraph": tr("split.mode_paragraph"),
                     "line": tr("split.mode_line"),
                     "sentence": tr("split.mode_sentence"),
+                    "smart": tr("split.mode_smart"),
                 }
                 split_mode = st.selectbox(
                     tr("split.mode_label"),
@@ -73,8 +74,16 @@ def render_content_input():
                     index=0,  # Default to paragraph mode
                     help=tr("split.mode_help")
                 )
+                
+                # Show info for smart mode (auto-detect segment count)
+                if split_mode == "smart":
+                    st.info(tr("split.smart_auto_hint"))
+                    target_segments = None  # Auto-detect
+                else:
+                    target_segments = None  # Not used for other modes
             else:
                 split_mode = "paragraph"  # Default for generate mode (not used)
+                target_segments = None
             
             # Title input (optional for both modes)
             title = st.text_input(
@@ -105,7 +114,8 @@ def render_content_input():
                 "text": text,
                 "title": title,
                 "n_scenes": n_scenes,
-                "split_mode": split_mode
+                "split_mode": split_mode,
+                "target_segments": target_segments
             }
         
         else:

web/components/output_preview.py

@@ -52,6 +52,7 @@ def render_single_output(pixelle_video, video_params):
     title = video_params.get("title")
     n_scenes = video_params.get("n_scenes", 5)
     split_mode = video_params.get("split_mode", "paragraph")
+    target_segments = video_params.get("target_segments", 8)
     bgm_path = video_params.get("bgm_path")
     bgm_volume = video_params.get("bgm_volume", 0.2)
     
@@ -116,6 +117,7 @@ def render_single_output(pixelle_video, video_params):
                         "title": title if title else None,
                         "n_scenes": n_scenes,
                         "split_mode": split_mode,
+                        "target_segments": target_segments,
                         "media_workflow": workflow_key,
                         "frame_template": frame_template,
                         "prompt_prefix": prompt_prefix,
@@ -222,6 +224,7 @@ def render_single_output(pixelle_video, video_params):
                     "title": title if title else None,
                     "n_scenes": n_scenes,
                     "split_mode": split_mode,
+                    "target_segments": target_segments,
                     "media_workflow": workflow_key,
                     "frame_template": frame_template,
                     "prompt_prefix": prompt_prefix,

web/i18n/locales/en_US.json

@@ -26,6 +26,8 @@
     "split.mode_paragraph": "📄 By Paragraph (\\n\\n)",
     "split.mode_line": "📝 By Line (\\n)",
     "split.mode_sentence": "✂️ By Sentence (。.!?)",
+    "split.mode_smart": "🧠 Smart Merge (AI Grouping)",
+    "split.smart_auto_hint": "🤖 AI will analyze text structure, recommend optimal segment count, and intelligently merge related paragraphs (dialogues, same scene)",
     "input.content": "Content",
     "input.content_placeholder": "Used directly without modification (split by strategy below)\nExample:\nHello everyone, today I'll share three study tips.\n\nThe first tip is focus training, meditate for 10 minutes daily.\n\nThe second tip is active recall, review immediately after learning.",
     "input.content_help": "Provide your own content for video generation",

web/i18n/locales/zh_CN.json

@@ -26,6 +26,8 @@
     "split.mode_paragraph": "📄 按段落（\\n\\n）",
     "split.mode_line": "📝 按行（\\n）",
     "split.mode_sentence": "✂️ 按句号（。.!?）",
+    "split.mode_smart": "🧠 智能合并（AI 分组）",
+    "split.smart_auto_hint": "🤖 AI 将自动分析文本结构，推荐最佳分镜数量，并智能合并相关段落（对话、同一场景）",
     "input.content": "内容",
     "input.content_placeholder": "直接使用，不做改写（根据下方分割方式切分）\n例如：\n大家好，今天跟你分享三个学习技巧。\n\n第一个技巧是专注力训练，每天冥想10分钟。\n\n第二个技巧是主动回忆，学完立即复述。",
     "input.content_help": "提供您自己的内容用于视频生成",