auto backup: 2026-06-05 08:10:01

2026-06-05 08:10:01 +08:00 · 2026-06-05 08:10:01 +08:00 · fa5d896a2c
commit fa5d896a2c
parent 01072f4003
36 changed files with 91929 additions and 82 deletions
--- a/business_production/组件配置/interactive-component-json/SKILL.md
+++ b/business_production/组件配置/interactive-component-json/SKILL.md
@ -233,7 +233,7 @@ interactive-component-json/
 | 造句互动 | mid_sentence_makeSentence | WfHMbzBYPau6hZsmFD0cBwWuncc |
 | 发音互动 | mid_pron_pron | CeH5bO3FhaG5cNsPZ4Xc72axnrb |

-### 核心互动组件（已实现12种，其余待定义）
+### 核心互动组件（已实现13种，其余待定义）
 - **合作阅读** `core_reading_order` — 7步LLM调用，输出 taskData/textData/sequenceData/learningData
 - **看图选词** `core_reading_imageDrag` — 5步LLM调用，输出 taskData/questionGroupData/studyData
 - **口语快答** `core_speaking_reply` — 7步LLM调用，输出 taskData/settingData/configData/learningData
@ -246,8 +246,9 @@ interactive-component-json/
 - **看图撰写** `core_writing_imgWrite` — 6步LLM调用，输出 taskInfo/textInfo/studyInfo/evalInfo
 - **邮件组句** `core_writing_questionMakeSentence` — 6步LLM调用，输出 taskInfo/textInfo/studyInfo/evalInfo
 - **邮件撰写** `core_writing_questionWrite` — 6步LLM调用，输出 taskInfo/textInfo/studyInfo/evalInfo
+- **看图拼词** `core_writing_imgMakeWord` — 7步(Step5规则)，输出 taskInfo/preDialog/imageInfo/questionList/studyInfo

-其余7种详见 `scripts/match_component.py` 中的 `CORE_INTERACTION_TYPES`
+其余6种详见 `scripts/match_component.py` 中的 `CORE_INTERACTION_TYPES`

 ## LLM 配置

--- a/business_production/组件配置/interactive-component-json/db/components.db
+++ b/business_production/组件配置/interactive-component-json/db/components.db
--- a/business_production/组件配置/interactive-component-json/outputs/L1-S2-U17-L1
+++ b/business_production/组件配置/interactive-component-json/outputs/L1-S2-U17-L1
--- a/business_production/组件配置/interactive-component-json/outputs/L1-S2-U17-L2
+++ b/business_production/组件配置/interactive-component-json/outputs/L1-S2-U17-L2
--- a/business_production/组件配置/interactive-component-json/outputs/L1-S2-U17-L2
+++ b/business_production/组件配置/interactive-component-json/outputs/L1-S2-U17-L2
--- a/business_production/组件配置/interactive-component-json/outputs/L1-S2-U17-L2
+++ b/business_production/组件配置/interactive-component-json/outputs/L1-S2-U17-L2
--- a/business_production/组件配置/interactive-component-json/outputs/L1-S2-U17-L2
+++ b/business_production/组件配置/interactive-component-json/outputs/L1-S2-U17-L2
--- a/business_production/组件配置/interactive-component-json/outputs/L1-S2-U17-L2
+++ b/business_production/组件配置/interactive-component-json/outputs/L1-S2-U17-L2
--- a/business_production/组件配置/interactive-component-json/outputs/L1-S2-U17-L4
+++ b/business_production/组件配置/interactive-component-json/outputs/L1-S2-U17-L4
--- a/business_production/组件配置/interactive-component-json/outputs/L1-S2-U17-L4
+++ b/business_production/组件配置/interactive-component-json/outputs/L1-S2-U17-L4
--- a/business_production/组件配置/interactive-component-json/outputs/L1-S2-U17-L4
+++ b/business_production/组件配置/interactive-component-json/outputs/L1-S2-U17-L4
--- a/business_production/组件配置/interactive-component-json/outputs/L1-S2-U17-L4
+++ b/business_production/组件配置/interactive-component-json/outputs/L1-S2-U17-L4
--- a/business_production/组件配置/interactive-component-json/outputs/L1-S2-U17-L5
+++ b/business_production/组件配置/interactive-component-json/outputs/L1-S2-U17-L5
--- a/business_production/组件配置/interactive-component-json/outputs/L1-S2-U17-L5
+++ b/business_production/组件配置/interactive-component-json/outputs/L1-S2-U17-L5
--- a/business_production/组件配置/interactive-component-json/outputs/L1-S2-U18-L2
+++ b/business_production/组件配置/interactive-component-json/outputs/L1-S2-U18-L2
--- a/business_production/组件配置/interactive-component-json/outputs/L1-S2-U18-L3
+++ b/business_production/组件配置/interactive-component-json/outputs/L1-S2-U18-L3
--- a/business_production/组件配置/interactive-component-json/outputs/L1-S2-U18-L3
+++ b/business_production/组件配置/interactive-component-json/outputs/L1-S2-U18-L3
--- a/business_production/组件配置/interactive-component-json/outputs/L1-S2-U18-L5
+++ b/business_production/组件配置/interactive-component-json/outputs/L1-S2-U18-L5
--- a/business_production/组件配置/interactive-component-json/outputs/L1-S2-U18-L5
+++ b/business_production/组件配置/interactive-component-json/outputs/L1-S2-U18-L5
--- a/business_production/组件配置/interactive-component-json/outputs/L1-S2-U19-L5
+++ b/business_production/组件配置/interactive-component-json/outputs/L1-S2-U19-L5
--- a/business_production/组件配置/interactive-component-json/project.md
+++ b/business_production/组件配置/interactive-component-json/project.md
@ -110,6 +110,18 @@ bitable数据可用于验证和参考。
 - 组句类inputType=sentence(optionList+answer)；撰写类inputType=write(answerText+hint)
 - prompt模板: `prompts/core_writing_img_prompts.py`, `prompts/core_writing_question_prompts.py`

+### v8.1 — 看图拼词（1种，已完成）
+
+| 类型 | cType | LLM步数 | 输出字段 | bitable wiki |
+|------|-------|---------|---------|-------------|
+| 看图拼词 | core_writing_imgMakeWord | 7 (Step5规则) | taskInfo, preDialog, imageInfo, questionList, studyInfo | QDyhwiY7PiSER8kHGZncDsrinde |
+
+- 输入: 单一教研配置块（【任务标题】【任务背景】【任务描述】【通关知识】【前置对话】【互动问题】）
+- 中间步骤: 学习流程配置(LLM生成) + 互动描述(LLM生成)
+- Step 5 imageInfo 为规则生成: `{"questionImg": "{cId}_question.png"}`
+- 独立7步流水线 `generate_core_writing_imgmakeword()`
+- prompt模板: `prompts/core_writing_imgmakeword_prompts.py`
+
 ### 其他未实现类型
 - 口语独白、口语探讨
 - 写作互动
@ -246,3 +258,26 @@ bitable数据可用于验证和参考。
  }
 }
 ```
+
+## v14 — 重新生成拆分 + 知识点推送失败判定（已完成）
+
+### 重新生成拆分
+将原来的单一"重新生成"按钮拆分为两个独立操作：
+- **重新生成组件**: 只重新生成 jsonData/core 字段，保留现有 kpInfo 不变
+- **重新生成知识点**: 只重新生成 kpInfo，保留现有 jsonData/core 字段不变
+
+实现方式：
+- 前端拆分为 `regenerateComponentJson(idx)` 和 `regenerateComponentKp(idx)` 两个函数
+- 后端 `/api/regenerate` 通过 `mode` 参数区分：`mode='component'` 或 `mode='kp'`
+- `mode='component'`: 调用 `generate_component()` 但跳过 kpInfo
+- `mode='kp'`: 只调用 `generate_kp_info()`
+- `knowledge_text` 嵌入到 HTML 报告的 COMPONENTS 前端数据中，供重新生成知识点使用
+
+### 知识点推送失败判定
+- 单个组件"知识推送"按钮: 无知识点或所有 kpId 为空时，标记为推送**失败**（原来算成功/info）
+- 批量"推送全部知识点": 无知识点或 kpId 为空时，计入 failCount 并显示失败详情（原来静默跳过或算成功）
+
+### 涉及文件
+- `scripts/pipeline.py`: report entry 添加 `knowledge_text` 字段
+- `scripts/html_report.py`: 前端按钮/面板/JS 逻辑拆分 + 推送失败判定
+- `scripts/proxy_server.py`: `_do_regenerate` 按 mode 分流
--- a/business_production/组件配置/interactive-component-json/prompts/pycache/core_writing_imgmakeword_prompts.cpython-312.pyc
+++ b/business_production/组件配置/interactive-component-json/prompts/pycache/core_writing_imgmakeword_prompts.cpython-312.pyc
--- a/business_production/组件配置/interactive-component-json/prompts/core_writing_imgmakeword_prompts.py
+++ b/business_production/组件配置/interactive-component-json/prompts/core_writing_imgmakeword_prompts.py
@ -0,0 +1,340 @@
+#!/usr/bin/env python3
+"""
+看图拼词 (core_writing_imgMakeWord) 多步 LLM prompt 模板
+
+生成流水线（7步，Step 5 为规则生成）：
+  Step 1: 学习流程配置 (intermediate text)
+  Step 2: 互动描述 (intermediate text)
+  Step 3: taskInfo → JSON
+  Step 4: preDialog → JSON
+  Step 5: imageInfo → 规则生成 (不需要 prompt)
+  Step 6: questionList → JSON
+  Step 7: studyInfo → JSON
+"""
+
+# ── Step 1: 学习流程配置 ──
+STEP1_LEARNING_FLOW = r"""你是一名儿童英语学习助手，负责为学习者生成互动式的巩固练习题。请根据以下信息生成练习内容：
+
+**输入信息**
+$任务信息$
+
+**生成要求**
+
+1. 为每个【通关知识】的知识词汇生成一个小题，包含以下结构：
+
+   ```
+   # 小题 X
+   ## 互动题目
+   1. 抄写题
+   Brainy：如图所示，首先/接着/最后要帮 Sarah 整理的物品是"中文词"，该词的英文是"英文词"。先来抄写一下这个词吧！
+
+   2. 听写题
+   Brainy：很好。接下来，记一下"英文词"的拼写，是"字母拆分"。然后看着图片，试试听写出"中文词"这个词吧！
+
+   Brainy：没错！"中文词"对应的英文词汇是"英文词"，你的拼写完全正确！
+   ```
+
+2. 每个知识词汇生成一题，顺序与知识词汇列表一致。
+
+3. 拼写拆解用字母加"-"表示，多个单词间用空格分隔（如 teddy bear → t-e-d-d-y  b-e-a-r）。
+
+4. 最后生成一个"完结反馈"：
+
+   ```
+   Brainy：太棒了！你复习了所有的关键知识点，拼对了每一个词，相信下一次你一定可以做得更好！
+   ```
+
+5. 输出中文引号全部使用英文引号""，不要使用中文引号"「」"。
+
+6. 不输出额外说明或过程，只生成完整互动题内容。
+
+### 三、示例输出（参考格式）
+```
+# 小题 1
+## 互动题目
+1. 抄写题
+Brainy：如图所示，首先要帮 Sally 整理的物品是"照片"，"照片"的英文是"photo"。先来抄写一下这个词吧！
+
+2. 听写题
+Brainy：很好。接下来，记一下"photo"的拼写，是"p-h-o-t-o"。然后看着图片，试试听写出"照片"这个词吧！
+
+Brainy：没错！"照片"对应的英文词汇是"photo"，你的拼写完全正确！
+
+# 小题 2
+## 互动题目
+1. 抄写题
+Brainy：如图所示，接着要帮 Sally 整理的物品是"奖牌"，"奖牌"的英文是"medal"。先来抄写一下这个词吧！
+
+2. 听写题
+Brainy：很好。接下来，记一下"medal"的拼写，是"m-e-d-a-l"。然后看着图片，试试听写出"奖牌"这个词吧！
+
+Brainy：没错！"奖牌"对应的英文词汇是"medal"，你的拼写完全正确！
+
+
+# 完结反馈
+Brainy：太棒了！你复习了所有的关键知识点，拼对了每一个词，相信下一次你一定可以做得更好！
+```
+
+#输入
+任务信息：
+$任务信息$
+
+# 输出
+现在请按照上述信息和要求，直接输出目标内容。"""
+
+
+# ── Step 2: 互动描述 ──
+STEP2_INTERACTION_DESC = r"""# 任务目标
+你是一个儿童英语学习内容生成助手。现在需要根据给定的任务信息、知识词汇和前置对话，生成一句符合场景的英文描述句，并附上对应中文翻译。
+
+# 输入信息
+* 任务信息：
+$任务信息$
+
+* 前置对话：
+$前置对话$
+
+# 生成要求
+只输出：
+一句完整、自然、符合儿童英语学习场景的英文描述句
+对应中文翻译（放在括号中）
+英文句子要求：
+必须结合「任务名称」与「知识」中的词汇
+尽量自然使用多个知识词汇
+句子简洁、口语化、符合少儿英语水平
+与「前置对话」场景保持一致
+中文翻译要求：
+与英文句子语义完全对应
+表达自然通顺
+输出格式固定如下（禁止输出任何解释）：
+English sentence.（中文翻译。）
+
+# 输出示例
+Dear Kate, I'd like to say thank you in this letter. Thank you for taking care of me. I have to go away now. Let's meet in the holidays.（亲爱的凯特，我想在这封信里向你表达感谢。感谢你对我的照顾。我现在必须离开了。咱们假期见。）"""
+
+
+# ── Step 3: taskInfo ──
+STEP3_TASK_INFO = r"""# 字段内容要求
+{
+    "cId": "{ID}", // 字符串类型
+    "cType": "core_writing_imgMakeWord",
+    "title": {任务信息}中的【任务名称】,
+    "sceneDesc": {任务信息}中的【任务背景】,
+    "cDesc": "xxx", // {互动描述}的全部内容
+    "key": [ // {任务信息}中的【通关知识】
+        {
+            "desc": "",
+            "keyList": [
+                {
+                    "type": "default", //固定
+                    "content": 知识点内容（英文部分）,
+                    "desc": 知识点解释（中文部分，如果是单词前面会带有英文缩写的词性标识）
+                },
+                 ... ...
+            ]
+        },
+        ...
+    ],
+    "resourceMapping"(抽取出{角色配置}中的全部 name:id映射 可为空。id必须为int类型):
+    {
+    样例:
+    "Tom":43,
+    "Jerry":10
+    }
+}
+
+# 输出格式
+- 整个输出必须在一行内完成，**不要换行**。
+- **不要使用 markdown 代码块**（如 ```json）包裹输出。
+
+样例:
+{"cId":"0000999","cType":"core_writing_imgMakeWord","title":"帮助sarah整理物品","sceneDesc":"场景描述，场景描述，看图说话场景描述","cDesc":"xxxxxx","key":[{"desc":"","keyList":[{"type":"default","content":"passport"},{"type":"default","content":"suitcase"},{"type":"default","content":"clothes"},{"type":"default","content":"pack"}]}],"resourceMapping":{"brainy":291,"jack":210}}
+
+# 输入
+{ID}:
+$ID$
+
+{任务信息}：
+$任务信息$
+
+{互动描述}：
+$互动描述$
+
+{角色配置}：
+$角色配置$
+
+# 输出
+现在基于输入字段和格式要求，生成***可以直接解析的单行json文本***，***注意前后不要输出任何额外符号！！！***
+如果{任务配置}中没有有效文本，则输出"无"。"""
+
+
+# ── Step 4: preDialog ──
+STEP4_PRE_DIALOG = r"""# 字段内容要求
+[ // 从{互动配置}中的【前置对话】中抽取
+    {
+        "type": "npc",
+        "npcName": "xxx",
+        "content": "xxxxxxxx"
+    },
+    {
+        "type": "user", // 如果是user则没有npcName
+        "content": "xxxxxxxx"
+    },
+    ......
+]
+
+# 输出格式
+- 整个输出必须在一行内完成，**不要换行**。
+- **不要使用 markdown 代码块**（如 ```json）包裹输出。
+
+样例:
+[{"type":"npc","npcName":"Brainy","content":"OK,let's try to the scene the scene together."},{"type":"npc","npcName":"Amy","content":"OK,let's try to the scene the scene together."},{"type":"npc","npcName":"Brainy","content":"OK,let's try to the scene the scene together."}]
+
+# 输入
+{互动配置}：
+$前置对话$
+
+# 输出
+现在基于输入字段和格式要求，生成***可以直接解析的单行json文本***，***注意前后不要输出任何额外符号！！！***
+如果{任务配置}中没有有效文本，则输出"无"。"""
+
+
+# ── Step 6: questionList ──
+STEP6_QUESTION_LIST = r"""# 字段内容要求
+[ // 从{互动配置}中的内容抽取为下面格式，有几个题目则有几个item
+    {
+        "desc": "(x)", // 题干中抽取
+        "imageInfo": {
+            "questionImg": "{ID}_question_{图片编号}.png" // 图片编号从【互动问题】中该小题对应的抽取
+        },
+        "optionList": [ // 选项中抽取
+            {
+                 "option": "a"
+            },
+           {
+                 "option": "b"
+            },
+            ......
+        ],
+        "answerText": "xxxxx" // 答案中抽取
+    },
+    ......
+]
+
+# 输出格式
+- 整个输出必须在一行内完成，**不要换行**。
+- **不要使用 markdown 代码块**（如 ```json）包裹输出。
+
+样例:
+[{"desc":"(5)","imageInfo":{"questionImg":"012310_question_00.png"},"optionList":[{"option":"o"},{"option":"h"},{"option":"x"},{"option":"P"},{"option":"t"},{"option":"o"},{"option":"k"}],"answerText":"Photo"},{"desc":"(5)","imageInfo":{"questionImg":"012310_question_01.png"},"optionList":[{"option":"m"},{"option":"d"},{"option":"l"},{"option":"q"},{"option":"e"},{"option":"d"},{"option":"m"},{"option":"a"}],"answerText":"medal"},{"desc":"(6)","imageInfo":{"questionImg":"012310_question_02.png"},"optionList":[{"option":"a"},{"option":"j"},{"option":"k"},{"option":"t"},{"option":"e"},{"option":"v"},{"option":"c"},{"option":"z"},{"option":"a"}],"answerText":"jacket"},{"desc":"(5) (4)","imageInfo":{"questionImg":"012310_question_03.png"},"optionList":[{"option":"b"},{"option":"e"},{"option":"t"},{"option":"r"},{"option":"y"},{"option":"a"},{"option":"d"},{"option":"e"},{"option":"t"},{"option":"d"},{"option":"y"}],"answerText":"teddy bear"}]
+
+# 输入
+{互动配置}：
+$互动问题$
+
+{ID}:
+$ID$
+
+# 输出
+现在基于输入字段和格式要求，生成***可以直接解析的单行json文本***，***注意前后不要输出任何额外符号！！！***
+如果{任务配置}中没有有效文本，则输出"无"。"""
+
+
+# ── Step 7: studyInfo ──
+STEP7_STUDY_INFO = r"""# 字段内容要求
+{
+    "learningPart": [ // {学习流程配置}中有几个小题则learningPart中有几个元素
+    {
+        "type": "npc",
+        "npcName": "Brainy",
+        "content": "xxxx" // {任务信息}里【通关知识】和该小题对应的单词
+        "image": "{ID}_question_{两位自增索引}.png"  // 索引从00开始
+        "learning": [
+        {
+            "question": {
+                "type": "npc",
+                "npcName": "Brainy",
+                "content": "xxx", // 抄写题题干
+                "desc": "xxx", // {任务信息}里【通关知识】和该小题对应的单词
+                "answerText": "xxx",   // {任务信息}里【通关知识】和该小题对应的单词
+            },
+          "optionList": [ // 单词的所有字母
+            {"option": "a"},
+            {"option": "b"},
+            ......
+          ]
+    },
+    {
+            "question": {
+                "type": "npc",
+                "npcName": "Brainy",
+                "content": "xxx", // 听写题题干
+                "desc": "（x）", // x为{任务信息}里【通关知识】和该小题对应的单词的字母个数
+                "answerText": "xxx",   // {任务信息}里【通关知识】和该小题对应的单词
+            },
+           "optionList": [ // 单词的所有字母
+                {"option": "a"},
+                {"option": "b"},
+                ......
+          ],
+          "feedback": {
+                "type": "npc",
+                "npcName": "Brainy",
+                "content": "xxxxxx"
+           }
+    },
+    {
+     ......
+     "image": "{ID}_question_01.png"
+     ......
+     },
+......
+    "closing": { //{学习流程配置}中的 完结反馈中提取
+        "type": "npc",
+        "npcName": "Brainy",
+        "content": "xxxxx"
+    }
+}
+
+# 输出格式
+- 整个输出必须在一行内完成，**不要换行**。
+- **不要使用 markdown 代码块**（如 ```json）包裹输出。
+
+样例:
+{"learningPart":[{"type":"npc","npcName":"Brainy","content":"Photo","image":"012310_question_00.png","learning":[{"question":{"type":"npc","npcName":"Brainy","content":"如图所示，首先要帮 Sally 整理的物品是「照片」，「照片」的英文是「photo」。先来抄写一下这个词吧！","desc":"Photo","answerText":"Photo"},"optionList":[{"option":"P"},{"option":"h"},{"option":"o"},{"option":"t"},{"option":"o"}]},{"question":{"type":"npc","npcName":"Brainy","content":"很好。接下来，记一下「photo」的拼写，是「p-h-o-t-o」。然后看着图片，试试听写出「照片」这个词吧！","desc":"(5)","answerText":"Photo"},"optionList":[{"option":"P"},{"option":"h"},{"option":"o"},{"option":"t"},{"option":"o"}],"feedback":{"type":"npc","npcName":"Brainy","content":"没错!「照片」对应的英文词汇是「photo」,你的拼写完全正确!"}}]},{"type":"npc","npcName":"Brainy","content":"medal","image":"012310_question_01.png","learning":[{"question":{"type":"npc","npcName":"Brainy","content":" 如图所示，接着要帮 Sally 整理的物品是「奖牌」，「奖牌」的英文是「medal」。先来抄写一下这个词吧！","desc":"medal","answerText":"medal"},"optionList":[{"option":"m"},{"option":"e"},{"option":"d"},{"option":"a"},{"option":"l"}]},{"question":{"type":"npc","npcName":"Brainy","content":"很好。接下来，记一下「medal」的拼写，是「m-e-d-a-l」。然后看着图片，试试听写出「奖牌」这个词吧！","desc":"(5)","answerText":"medal"},"optionList":[{"option":"m"},{"option":"e"},{"option":"d"},{"option":"a"},{"option":"l"}],"feedback":{"type":"npc","npcName":"Brainy","content":"没错！「奖牌」对应的英文词汇是「medal」，你的拼写完全正确！"}}]},{"type":"npc","npcName":"Brainy","content":"jetcak","image":"012310_question_02.png","learning":[{"question":{"type":"npc","npcName":"Brainy","content":" 如图所示，接着要帮 Sally 整理的物品是「奖牌」，「奖牌」的英文是「medal」。先来抄写一下这个词吧！","desc":"jetcak","answerText":"jetcak"},"optionList":[{"option":"j"},{"option":"e"},{"option":"t"},{"option":"c"},{"option":"a"},{"option":"k"}]},{"question":{"type":"npc","npcName":"Brainy","content":"很好。接下来，记一下「medal」的拼写，是「m-e-d-a-l」。然后看着图片，试试听写出「奖牌」这个词吧！","desc":"(6)","answerText":"jetcak"},"optionList":[{"option":"j"},{"option":"e"},{"option":"t"},{"option":"c"},{"option":"a"},{"option":"k"}],"feedback":{"type":"npc","npcName":"Brainy","content":"没错！「夹克」对应的英文词汇是「jetcak」，你的拼写完全正确！"}}]},{"type":"npc","npcName":"Brainy","content":"teddy bear","image":"012310_question_03.png","learning":[{"question":{"type":"npc","npcName":"Brainy","content":"很好。接下来，记一下「medal」的拼写，是「m-e-d-a-l」。然后看着图片，试试听写出「奖牌」这个词吧！","desc":"teddy bear","answerText":"teddy bear"},"optionList":[{"option":"t"},{"option":"e"},{"option":"d"},{"option":"d"},{"option":"y"},{"option":"b"},{"option":"e"},{"option":"a"},{"option":"r"}]},{"question":{"type":"npc","npcName":"Brainy","content":"很好。接下来，记一下「medal」的拼写，是「m-e-d-a-l」。然后看着图片，试试听写出「奖牌」这个词吧！","desc":"(5) (4)","answerText":"teddy bear"},"optionList":[{"option":"t"},{"option":"e"},{"option":"d"},{"option":"d"},{"option":"y"},{"option":"b"},{"option":"e"},{"option":"a"},{"option":"r"}],"feedback":{"type":"npc","npcName":"Brainy","content":"没错！「泰迪熊」对应的英文词汇是「teddy bear」，你的拼写完全正确！"}}]}],"closing":{"type":"npc","npcName":"Brainy","content":"太棒了！你复习了所有的关键知识点，拼对了每一个词，相信下一次你一定可以做得更好！"}}
+
+# 输入
+{学习流程配置}：
+$学习流程配置$
+
+{任务信息}:
+$任务信息$
+
+{ID}:
+$ID$
+
+# 输出
+现在基于输入字段和格式要求，生成***可以直接解析的单行json文本***，***注意前后不要输出任何额外符号！！！***直接以"{"开头，输出一定保持单行！如果{学习流程配置}中没有有效文本，则输出"无"。"""
+
+
+# ══════════════════════════════════════════════════════════════════════════════
+# Steps Registry & Builder
+# ══════════════════════════════════════════════════════════════════════════════
+
+CORE_WRITING_IMGMAKEWORD_STEPS = {
+    "step1_learning_flow": {"template": STEP1_LEARNING_FLOW, "description": "学习流程配置", "output_type": "text"},
+    "step2_interaction_desc": {"template": STEP2_INTERACTION_DESC, "description": "互动描述", "output_type": "text"},
+    "step3_task_info": {"template": STEP3_TASK_INFO, "description": "taskInfo", "output_type": "json"},
+    "step4_pre_dialog": {"template": STEP4_PRE_DIALOG, "description": "preDialog", "output_type": "json"},
+    "step6_question_list": {"template": STEP6_QUESTION_LIST, "description": "questionList", "output_type": "json"},
+    "step7_study_info": {"template": STEP7_STUDY_INFO, "description": "studyInfo", "output_type": "json"},
+}
+
+
+def build_writing_imgmakeword_prompt(step_key, replacements):
+    """Build a prompt for a given step by replacing $placeholder$ tokens."""
+    info = CORE_WRITING_IMGMAKEWORD_STEPS[step_key]
+    prompt = info["template"]
+    for key, val in replacements.items():
+        prompt = prompt.replace(f"${key}$", str(val) if val else "")
+    return prompt
--- a/business_production/组件配置/interactive-component-json/scripts/pycache/generate_json.cpython-312.pyc
+++ b/business_production/组件配置/interactive-component-json/scripts/pycache/generate_json.cpython-312.pyc
--- a/business_production/组件配置/interactive-component-json/scripts/pycache/html_report.cpython-312.pyc
+++ b/business_production/组件配置/interactive-component-json/scripts/pycache/html_report.cpython-312.pyc
--- a/business_production/组件配置/interactive-component-json/scripts/pycache/kp_matcher.cpython-312.pyc
+++ b/business_production/组件配置/interactive-component-json/scripts/pycache/kp_matcher.cpython-312.pyc
--- a/business_production/组件配置/interactive-component-json/scripts/pycache/match_component.cpython-312.pyc
+++ b/business_production/组件配置/interactive-component-json/scripts/pycache/match_component.cpython-312.pyc
--- a/business_production/组件配置/interactive-component-json/scripts/generate_json.py
+++ b/business_production/组件配置/interactive-component-json/scripts/generate_json.py
@ -130,6 +130,11 @@ from prompts.core_writing_question_prompts import (
    build_writing_question_prompt,
 )

+from prompts.core_writing_imgmakeword_prompts import (
+    CORE_WRITING_IMGMAKEWORD_STEPS,
+    build_writing_imgmakeword_prompt,
+)
+
 # Fallback: 旧版 few-shot 方式（对无 prompt 模板的类型）
 FALLBACK_SYSTEM_PROMPT = """你是一个互动组件配置JSON生成专家。根据教研配置文本生成对应的 jsonData JSON。

@ -382,7 +387,7 @@ def generate_kp_info(cType, cId, knowledge_text, llm_client=None, level=None):
    return match_knowledge_points(knowledge_text, cType, cId, level=level, llm_client=llm_client)


-CORE_TYPES = {"core_reading_order", "core_reading_imageDrag", "core_speaking_reply", "core_speaking_inquiry", "core_speaking_image", "core_listening_order", "core_listening_drag", "core_listening_choose", "core_writing_imgMakeSentence", "core_writing_imgWrite", "core_writing_questionMakeSentence", "core_writing_questionWrite"}
+CORE_TYPES = {"core_reading_order", "core_reading_imageDrag", "core_speaking_reply", "core_speaking_inquiry", "core_speaking_image", "core_listening_order", "core_listening_drag", "core_listening_choose", "core_writing_imgMakeSentence", "core_writing_imgWrite", "core_writing_questionMakeSentence", "core_writing_questionWrite", "core_writing_imgMakeWord"}


 def parse_core_reading_fields(teaching_config):
@ -1995,6 +2000,162 @@ def generate_core_writing_question_write(cId, teaching_config, character_map=Non
    return _generate_core_writing(cId, "core_writing_questionWrite", teaching_config, character_map, llm_client, level=level)


+# ============ 看图拼词 (core_writing_imgMakeWord) ============
+
+def parse_core_writing_imgmakeword_fields(teaching_config):
+    """
+    从组件配置文本中提取 core_writing_imgMakeWord 所需的输入字段。
+
+    教研配置为单一文本块，包含 【任务标题】【任务背景】【任务描述】【通关知识】【前置对话】【互动问题】 等 sections。
+
+    Returns:
+        dict: 字段名 → 值
+    """
+    fields = parse_teaching_config_fields(teaching_config)
+
+    任务标题 = fields.get("任务标题", "")
+    任务背景 = fields.get("任务背景", "")
+    任务描述 = fields.get("任务描述", "")
+    通关知识 = fields.get("通关知识", "")
+    前置对话 = fields.get("前置对话", "")
+    互动问题 = fields.get("互动问题", "")
+
+    # 构建任务信息（composite）
+    任务信息_parts = []
+    if 任务标题:
+        任务信息_parts.append(f"【任务标题】\n{任务标题}")
+    if 任务背景:
+        任务信息_parts.append(f"【任务背景】\n{任务背景}")
+    if 任务描述:
+        任务信息_parts.append(f"【任务描述】\n{任务描述}")
+    if 通关知识:
+        任务信息_parts.append(f"【通关知识】\n{通关知识}")
+    任务信息 = "\n\n".join(任务信息_parts)
+
+    return {
+        "任务标题": 任务标题,
+        "任务背景": 任务背景,
+        "任务描述": 任务描述,
+        "通关知识": 通关知识,
+        "前置对话": 前置对话,
+        "互动问题": 互动问题,
+        "任务信息": 任务信息,
+    }
+
+
+def generate_core_writing_imgmakeword(cId, teaching_config, character_map=None, llm_client=None, level=None):
+    """
+    看图拼词（core_writing_imgMakeWord）7步生成流水线。
+
+    Step 1: 学习流程配置  → 中间文本
+    Step 2: 互动描述      → 中间文本
+    Step 3: taskInfo      → JSON
+    Step 4: preDialog     → JSON
+    Step 5: imageInfo     → 规则生成
+    Step 6: questionList  → JSON
+    Step 7: studyInfo     → JSON
+
+    Returns:
+        dict: {
+            "taskInfo": dict, "preDialog": list,
+            "imageInfo": dict, "questionList": list,
+            "studyInfo": dict, "kpInfo": None,
+            "intermediate": {"学习流程配置": str, "互动描述": str}
+        }
+    """
+    if llm_client is None:
+        from llm_client import get_client
+        llm_client = get_client()
+
+    cType = "core_writing_imgMakeWord"
+
+    # 解析输入字段
+    fields = parse_core_writing_imgmakeword_fields(teaching_config)
+
+    # 角色配置：优先教研配置，回退到 character_map
+    角色配置 = ""
+    if character_map:
+        角色配置 = "\n".join(f"{k}：{v}" for k, v in character_map.items())
+
+    logger.info(f"[core_writing_imgMakeWord] 开始7步生成: cId={cId}")
+
+    # ── Step 1: 学习流程配置 ──
+    logger.info(f"[core_writing_imgMakeWord] Step 1/7: 学习流程配置")
+    prompt1 = build_writing_imgmakeword_prompt("step1_learning_flow", {
+        "任务信息": fields["任务信息"],
+    })
+    学习流程配置, usage1 = llm_client.call("", prompt1, max_tokens=8192, temperature=0.3)
+    学习流程配置 = 学习流程配置.strip()
+    logger.info(f"  Step 1 完成: {len(学习流程配置)} 字符, tokens={usage1}")
+
+    # ── Step 2: 互动描述 ──
+    logger.info(f"[core_writing_imgMakeWord] Step 2/7: 互动描述")
+    prompt2 = build_writing_imgmakeword_prompt("step2_interaction_desc", {
+        "任务信息": fields["任务信息"],
+        "前置对话": fields["前置对话"],
+    })
+    互动描述, usage2 = llm_client.call("", prompt2, max_tokens=4096, temperature=0.3)
+    互动描述 = 互动描述.strip()
+    logger.info(f"  Step 2 完成: {len(互动描述)} 字符, tokens={usage2}")
+
+    # ── Step 3: taskInfo ──
+    logger.info(f"[core_writing_imgMakeWord] Step 3/7: taskInfo")
+    prompt3 = build_writing_imgmakeword_prompt("step3_task_info", {
+        "ID": str(cId),
+        "任务信息": fields["任务信息"],
+        "互动描述": 互动描述,
+        "角色配置": 角色配置,
+    })
+    taskInfo, usage3 = llm_client.call_for_json("", prompt3, max_tokens=4096, temperature=0.1)
+    logger.info(f"  Step 3 完成: taskInfo keys={list(taskInfo.keys()) if isinstance(taskInfo, dict) else type(taskInfo)}, tokens={usage3}")
+
+    # ── Step 4: preDialog ──
+    logger.info(f"[core_writing_imgMakeWord] Step 4/7: preDialog")
+    prompt4 = build_writing_imgmakeword_prompt("step4_pre_dialog", {
+        "前置对话": fields["前置对话"],
+    })
+    preDialog, usage4 = llm_client.call_for_json("", prompt4, max_tokens=4096, temperature=0.1)
+    logger.info(f"  Step 4 完成: preDialog len={len(preDialog) if isinstance(preDialog, list) else 'dict'}, tokens={usage4}")
+
+    # ── Step 5: imageInfo (规则生成) ──
+    logger.info(f"[core_writing_imgMakeWord] Step 5/7: imageInfo (规则生成)")
+    imageInfo = {"questionImg": f"{cId}_question.png"}
+    logger.info(f"  Step 5 完成: imageInfo={imageInfo}")
+
+    # ── Step 6: questionList ──
+    logger.info(f"[core_writing_imgMakeWord] Step 6/7: questionList")
+    prompt6 = build_writing_imgmakeword_prompt("step6_question_list", {
+        "互动问题": fields["互动问题"],
+        "ID": str(cId),
+    })
+    questionList, usage6 = llm_client.call_for_json("", prompt6, max_tokens=8192, temperature=0.1)
+    logger.info(f"  Step 6 完成: questionList len={len(questionList) if isinstance(questionList, list) else 'dict'}, tokens={usage6}")
+
+    # ── Step 7: studyInfo ──
+    logger.info(f"[core_writing_imgMakeWord] Step 7/7: studyInfo")
+    prompt7 = build_writing_imgmakeword_prompt("step7_study_info", {
+        "学习流程配置": 学习流程配置,
+        "任务信息": fields["任务信息"],
+        "ID": str(cId),
+    })
+    studyInfo, usage7 = llm_client.call_for_json("", prompt7, max_tokens=16384, temperature=0.1)
+    logger.info(f"  Step 7 完成: studyInfo keys={list(studyInfo.keys()) if isinstance(studyInfo, dict) else type(studyInfo)}, tokens={usage7}")
+
+    logger.info(f"[core_writing_imgMakeWord] 7步生成完成: cId={cId}")
+    return {
+        "taskInfo": taskInfo,
+        "preDialog": preDialog,
+        "imageInfo": imageInfo,
+        "questionList": questionList,
+        "studyInfo": studyInfo,
+        "kpInfo": None,
+        "intermediate": {
+            "学习流程配置": 学习流程配置,
+            "互动描述": 互动描述,
+        },
+    }
+
+
 def generate_component(component, character_map=None, section_char_map=None, llm_client=None, level=None):
    """
    一站式生成单个组件的 jsonData + kpInfo
@ -2171,6 +2332,20 @@ def generate_component(component, character_map=None, section_char_map=None, llm
                "intermediate": result["intermediate"],
            }

+        elif cType == "core_writing_imgMakeWord":
+            result = generate_core_writing_imgmakeword(**_core_call_args)
+            core_out = {
+                **_core_base,
+                "jsonData": result["taskInfo"],
+                "taskInfo": result["taskInfo"],
+                "preDialog": result["preDialog"],
+                "imageInfo": result["imageInfo"],
+                "questionList": result["questionList"],
+                "studyInfo": result["studyInfo"],
+                "kpInfo": result["kpInfo"],
+                "intermediate": result["intermediate"],
+            }
+
        else:
            raise ValueError(f"核心互动类型 {cType} 尚未实现生成器")

--- a/business_production/组件配置/interactive-component-json/scripts/html_report.py
+++ b/business_production/组件配置/interactive-component-json/scripts/html_report.py
@ -53,6 +53,7 @@ def _build_component_data(result):
        "status": status,
        "raw_config": result.get("raw_config", ""),
        "config_info": result.get("config_info", ""),
+        "knowledge_text": result.get("knowledge_text", ""),
    }

    if status not in ("generated", "saved"):
@ -254,9 +255,12 @@ def _render_component(idx, result):
          知识推送<span class="push-icon" id="kp-status-{idx}"></span>
        </button>'''

-    # Regenerate button (for all components)
-    regen_btn = f'''<button class="push-btn regen-btn" data-idx="{idx}" onclick="event.stopPropagation();toggleRegenPanel({idx})">
-          重新生成<span class="push-icon" id="regen-status-{idx}"></span>
+    # Regenerate buttons (split: component + kp)
+    regen_btn = f'''<button class="push-btn regen-btn" data-idx="{idx}" onclick="event.stopPropagation();toggleRegenPanel({idx},'component')">
+          重新生成组件<span class="push-icon" id="regen-status-{idx}"></span>
+        </button>
+        <button class="push-btn regen-kp-btn" data-idx="{idx}" onclick="event.stopPropagation();toggleRegenPanel({idx},'kp')">
+          重新生成知识点<span class="push-icon" id="regen-kp-status-{idx}"></span>
        </button>'''

    # Build body
@ -303,17 +307,25 @@ def _render_component(idx, result):
        reason = html.escape(result.get("reason", ""))
        body_html = f'<div class="skip-msg">跳过: {reason}</div>'

-    # Regenerate panel (editable teaching_config) — for all components
+    # Regenerate panel — split into component regen + kp regen
    raw_config = result.get("raw_config", "")
    escaped_config = html.escape(raw_config)
    config_rows = min(max(raw_config.count('\n') + 1, 5), 25)
+    knowledge_text = result.get("knowledge_text", "")
+    escaped_kp_text = html.escape(knowledge_text)
+    kp_rows = min(max(knowledge_text.count('\n') + 1, 3), 15)
    # For failed/skipped: show config panel expanded; for success: collapsed
    regen_display = "block" if status in ("failed", "skipped") else "none"
    regen_panel = f'''
-    <div class="regen-panel" id="regen-panel-{idx}" style="display:{regen_display};">
-      <div class="regen-label">原始组件配置 <span class="regen-hint">(可编辑后点击"执行重新生成")</span></div>
+    <div class="regen-panel" id="regen-panel-{idx}" style="display:none;">
+      <div class="regen-label">原始组件配置 <span class="regen-hint">(可编辑后点击"执行重新生成组件")</span></div>
      <textarea class="regen-textarea" id="regen-config-{idx}" rows="{config_rows}">{escaped_config}</textarea>
-      <button class="regen-exec-btn" id="regen-exec-{idx}" onclick="regenerateComponent({idx})">执行重新生成</button>
+      <button class="regen-exec-btn" id="regen-exec-{idx}" onclick="regenerateComponentJson({idx})">执行重新生成组件</button>
+    </div>
+    <div class="regen-panel regen-kp-panel" id="regen-kp-panel-{idx}" style="display:none;">
+      <div class="regen-label regen-kp-label">知识点文本 <span class="regen-hint">(来自sheet知识点列，可编辑后点击"执行重新生成知识点")</span></div>
+      <textarea class="regen-textarea regen-kp-textarea" id="regen-kp-text-{idx}" rows="{kp_rows}">{escaped_kp_text}</textarea>
+      <button class="regen-exec-btn regen-kp-exec-btn" id="regen-kp-exec-{idx}" onclick="regenerateComponentKp({idx})">执行重新生成知识点</button>
    </div>'''

    return f'''
@ -465,6 +477,8 @@ body {{ font-family:-apple-system,BlinkMacSystemFont,"Segoe UI",Roboto,"Noto San
 .kp-btn:hover {{ background:#dcfce7; }}
 .regen-btn {{ background:#fef3c7; color:#d97706; border-color:#fde68a; }}
 .regen-btn:hover {{ background:#fde68a; }}
+.regen-kp-btn {{ background:#fdf2f8; color:#db2777; border-color:#fbcfe8; }}
+.regen-kp-btn:hover {{ background:#fce7f3; }}
 .push-btn:disabled {{ opacity:0.5; cursor:not-allowed; }}
 .push-icon {{ display:inline-block; width:8px; height:8px; border-radius:50%; background:#94a3b8; }}
 .push-icon.ok {{ background:#10b981; }}
@ -485,6 +499,12 @@ body {{ font-family:-apple-system,BlinkMacSystemFont,"Segoe UI",Roboto,"Noto San
                   cursor:pointer; transition:background 0.2s; }}
 .regen-exec-btn:hover {{ background:#b45309; }}
 .regen-exec-btn:disabled {{ opacity:0.5; cursor:not-allowed; }}
+.regen-kp-panel {{ background:#fdf2f8; border-color:#fbcfe8; }}
+.regen-kp-label {{ color:#9d174d; }}
+.regen-kp-textarea {{ border-color:#fbcfe8; }}
+.regen-kp-textarea:focus {{ border-color:#db2777; box-shadow:0 0 0 2px rgba(219,39,119,0.15); }}
+.regen-kp-exec-btn {{ background:#db2777; }}
+.regen-kp-exec-btn:hover {{ background:#9d174d; }}

 /* JSON sections */
 .json-section {{ margin-bottom:12px; }}
@ -1009,8 +1029,9 @@ async function pushComponent(idx) {{
 async function pushKp(idx) {{
  const body = buildKpPushBody(idx);
  if (!body) {{
-    showToast('组件 #' + idx + ' 无知识点数据', 'info');
-    return true;
+    setStatus('kp-status-' + idx, 'fail');
+    showToast('组件 #' + idx + ' 知识点推送失败: 无知识点或kpId为空', 'error');
+    return false;
  }}

  setStatus('kp-status-' + idx, 'loading');
@ -1079,7 +1100,11 @@ async function _runPushAll(mode) {{
        const r2 = await doPush(body2);
        if (r2.ok) {{ setStatus('kp-status-' + idx, 'ok'); okCount++; }}
        else {{ setStatus('kp-status-' + idx, 'fail'); failCount++; _batchErrors.push('[' + _batchContext + '] 知识点推送失败: ' + r2.error); }}
-      }} else {{ if (mode === 'both') okCount++; }}
+      }} else {{
+        setStatus('kp-status-' + idx, 'fail');
+        failCount++;
+        _batchErrors.push('[' + _batchContext + '] 知识点推送失败: 无知识点或kpId为空');
+      }}
      done++; updateProgress();
    }}
  }}
@ -1103,18 +1128,34 @@ function pushAllComponents() {{ _runPushAll('config'); }}
 function pushAllKpAll() {{ _runPushAll('kp'); }}

 // ============ Regenerate ============
-function toggleRegenPanel(idx) {{
-  const panel = document.getElementById('regen-panel-' + idx);
-  if (!panel) return;
-  const showing = panel.style.display !== 'block';
-  panel.style.display = showing ? 'block' : 'none';
+function toggleRegenPanel(idx, mode) {{
+  const compPanel = document.getElementById('regen-panel-' + idx);
+  const kpPanel = document.getElementById('regen-kp-panel-' + idx);
+
+  if (mode === 'kp') {{
+    if (compPanel) compPanel.style.display = 'none';
+    if (kpPanel) {{
+      const showing = kpPanel.style.display !== 'block';
+      kpPanel.style.display = showing ? 'block' : 'none';
      if (showing) {{
        const card = document.getElementById('card-' + idx);
        if (card) card.classList.add('open');
      }}
    }}
+  }} else {{
+    if (kpPanel) kpPanel.style.display = 'none';
+    if (compPanel) {{
+      const showing = compPanel.style.display !== 'block';
+      compPanel.style.display = showing ? 'block' : 'none';
+      if (showing) {{
+        const card = document.getElementById('card-' + idx);
+        if (card) card.classList.add('open');
+      }}
+    }}
+  }}
+}}

-async function regenerateComponent(idx) {{
+async function regenerateComponentJson(idx) {{
  const comp = COMPONENTS[idx - 1];
  const ta = document.getElementById('regen-config-' + idx);
  const btn = document.getElementById('regen-exec-' + idx);
@ -1137,6 +1178,7 @@ async function regenerateComponent(idx) {{
      method: 'POST',
      headers: {{ 'Content-Type': 'application/json' }},
      body: JSON.stringify({{
+        mode: 'component',
        teaching_config: teachingConfig,
        type_name: comp.type_name,
        cId: comp.cId,
@ -1153,7 +1195,7 @@ async function regenerateComponent(idx) {{
      throw new Error(data.error || ('HTTP ' + resp.status));
    }}

-    // Update COMPONENTS data
+    // Update COMPONENTS data (only jsonData/core fields, keep kpInfo)
    comp.status = 'generated';
    comp.category = data.category || comp.category;
    comp.cType = data.cType || comp.cType;
@ -1167,13 +1209,15 @@ async function regenerateComponent(idx) {{
    }} else {{
      if (data.jsonData) comp.jsonData = data.jsonData;
    }}
-    if (data.kpInfo) comp.kpInfo = data.kpInfo;
+    // kpInfo not updated in component-only mode

-    // Rebuild body HTML
-    rebuildComponentBody(idx, data);
+    // Rebuild body HTML (pass existing kpInfo from comp, not from data)
+    const displayData = {{ ...data }};
+    if (!displayData.kpInfo && comp.kpInfo) displayData.kpInfo = comp.kpInfo;
+    rebuildComponentBody(idx, displayData);

    setStatus('regen-status-' + idx, 'ok');
-    btn.textContent = '执行重新生成';
+    btn.textContent = '执行重新生成组件';
    btn.disabled = false;

    // Update status label
@ -1183,12 +1227,83 @@ async function regenerateComponent(idx) {{
      statusEl.textContent = '生成成功';
    }}

-    showToast('组件 #' + idx + ' 重新生成成功', 'success');
+    showToast('组件 #' + idx + ' 组件配置重新生成成功', 'success');
  }} catch(e) {{
    setStatus('regen-status-' + idx, 'fail');
-    btn.textContent = '执行重新生成';
+    btn.textContent = '执行重新生成组件';
    btn.disabled = false;
-    showToast('组件 #' + idx + ' 重新生成失败: ' + e.message, 'error');
+    showToast('组件 #' + idx + ' 组件重新生成失败: ' + e.message, 'error');
+  }}
+}}
+
+async function regenerateComponentKp(idx) {{
+  const comp = COMPONENTS[idx - 1];
+  const ta = document.getElementById('regen-kp-text-' + idx);
+  const btn = document.getElementById('regen-kp-exec-' + idx);
+  if (!ta || !btn) return;
+
+  const knowledgeText = ta.value.trim();
+  if (!knowledgeText) {{
+    showToast('知识点文本不能为空', 'error');
+    return;
+  }}
+
+  btn.disabled = true;
+  btn.textContent = '生成中...';
+  setStatus('regen-kp-status-' + idx, 'loading');
+
+  const regenUrl = getRegenUrl();
+
+  try {{
+    const resp = await fetch(regenUrl, {{
+      method: 'POST',
+      headers: {{ 'Content-Type': 'application/json' }},
+      body: JSON.stringify({{
+        mode: 'kp',
+        type_name: comp.type_name,
+        cId: comp.cId,
+        knowledge_text: knowledgeText,
+        level: LEVEL,
+      }}),
+    }});
+
+    const data = await resp.json();
+
+    if (!resp.ok) {{
+      throw new Error(data.error || ('HTTP ' + resp.status));
+    }}
+
+    // Update kpInfo only
+    if (data.kpInfo) {{
+      comp.kpInfo = data.kpInfo;
+    }} else {{
+      comp.kpInfo = null;
+    }}
+    comp.knowledge_text = knowledgeText;
+
+    // Rebuild body HTML (keep existing jsonData/core fields)
+    const displayData = {{}};
+    if (comp.category === 'core') {{
+      const fields = PUSH_FIELD_MAP[comp.cType] || [];
+      for (const key of fields) {{
+        if (comp[key]) displayData[key] = comp[key];
+      }}
+    }} else {{
+      if (comp.jsonData) displayData.jsonData = comp.jsonData;
+    }}
+    displayData.kpInfo = comp.kpInfo;
+    rebuildComponentBody(idx, displayData);
+
+    setStatus('regen-kp-status-' + idx, 'ok');
+    btn.textContent = '执行重新生成知识点';
+    btn.disabled = false;
+
+    showToast('组件 #' + idx + ' 知识点重新生成成功', 'success');
+  }} catch(e) {{
+    setStatus('regen-kp-status-' + idx, 'fail');
+    btn.textContent = '执行重新生成知识点';
+    btn.disabled = false;
+    showToast('组件 #' + idx + ' 知识点重新生成失败: ' + e.message, 'error');
  }}
 }}

@ -1200,8 +1315,9 @@ function rebuildComponentBody(idx, data) {{
  // Clear old content above regen-result (error-msg, json-sections etc)
  const body = document.getElementById('body-' + idx);
  const regenPanel = document.getElementById('regen-panel-' + idx);
-  // Remove everything except regen-result and regen-panel
-  while (body.firstChild && body.firstChild !== resultDiv && body.firstChild !== regenPanel) {{
+  const regenKpPanel = document.getElementById('regen-kp-panel-' + idx);
+  // Remove everything except regen-result and regen panels
+  while (body.firstChild && body.firstChild !== resultDiv && body.firstChild !== regenPanel && body.firstChild !== regenKpPanel) {{
    body.removeChild(body.firstChild);
  }}

@ -1301,7 +1417,7 @@ function ensurePushButtons(idx) {{
 }}

 // ============ LocalStorage Auto-Save ============
-const SAVE_KEY = 'icj_' + document.title;
+const SAVE_KEY = 'icj_' + document.title + '_{timestamp}';

 function saveEdits() {{
  const state = {{}};
--- a/business_production/组件配置/interactive-component-json/scripts/kp_matcher.py
+++ b/business_production/组件配置/interactive-component-json/scripts/kp_matcher.py
@ -267,7 +267,26 @@ def _extract_english_word_only(text):

 # ============ 单条知识点匹配 ============

-def _match_single_kp(kp_text, level=None, context="", llm_client=None):
+def _filter_candidates_by_unit(candidates, cId):
+    """
+    根据组件ID前4位过滤候选知识点，确保知识点属于同一单元。
+    kpid前4位与cId前4位相同才属于同一单元。
+
+    Args:
+        candidates: 粗召回的候选列表
+        cId: 组件ID
+
+    Returns:
+        list: 过滤后的候选列表（仅保留同单元的知识点）
+    """
+    if not cId or len(str(cId)) < 4:
+        return candidates
+    unit_prefix = str(cId)[:4]
+    filtered = [c for c in candidates if c.get("kp_id") and str(c["kp_id"])[:4] == unit_prefix]
+    return filtered
+
+
+def _match_single_kp(kp_text, level=None, context="", llm_client=None, cId=None):
    """
    匹配单个知识点文本到数据库记录。

@ -309,6 +328,14 @@ def _match_single_kp(kp_text, level=None, context="", llm_client=None):
        logger.info(f"未找到匹配: '{kp_text}' (level={level})")
        return default_result

+    # 根据组件ID前4位过滤候选，确保知识点属于同一单元
+    if cId:
+        filtered = _filter_candidates_by_unit(candidates, cId)
+        if not filtered:
+            logger.info(f"单元过滤后无匹配: '{kp_text}' (cId前4位={str(cId)[:4]})")
+            return default_result
+        candidates = filtered
+
    # 保存候选列表供 HTML 下拉展示
    candidates_for_ui = [
        {"kp_id": c["kp_id"], "type": c.get("type", ""), "title": c.get("title", ""),
@ -378,7 +405,7 @@ def match_knowledge_points(knowledge_text, cType, cId, level=None, llm_client=No
            logger.debug(f"跳过无英文字符行: '{stripped}'")
            continue

-        kp_entry = _match_single_kp(stripped, level, context, llm_client)
+        kp_entry = _match_single_kp(stripped, level, context, llm_client, cId=cId)
        kp_list.append(kp_entry)

    if not kp_list:
--- a/business_production/组件配置/interactive-component-json/scripts/match_component.py
+++ b/business_production/组件配置/interactive-component-json/scripts/match_component.py
@ -51,6 +51,7 @@ CORE_INTERACTION_TYPES = {
    "合作听力":   {"cType": "core_listening_order", "bitable_wiki": "FrxtwNRQDizqiikPkATcBzTCnYe", "db_table": "tblBuFU6wVKdBh8B"},
    "听力拖拽":   {"cType": "core_listening_drag", "bitable_wiki": "K3QrwQnWqiPBm1krhnNcWDTqnhe", "db_table": "tblOOAsgYoZWsTGJ"},
    "听力选择":   {"cType": "core_listening_choose", "bitable_wiki": "Kwrcw6A4jip2sxkdLn4czd1knvf", "db_table": "tblH4xNpAmKJ7gEq"},
+    "看图拼词":   {"cType": "core_writing_imgMakeWord", "bitable_wiki": "QDyhwiY7PiSER8kHGZncDsrinde", "db_table": "tblyPBqx9IRYQ3a9"},
    "写作互动":   {"cType": None, "bitable_wiki": "DVkcwVr2giswTckcgD0cpHwNnv2"},
    "看图组句":   {"cType": "core_writing_imgMakeSentence", "bitable_wiki": "BkmtwUBwMiHd5Ak7VS6ccE9SnHd", "db_table": "tblpILmFc9ec17DY"},
    "看图撰写":   {"cType": "core_writing_imgWrite", "bitable_wiki": "KwPHwnaqdiWlvNkDm5fcFreDnQh", "db_table": "tblAMaM3g7500Lmm"},
--- a/business_production/组件配置/interactive-component-json/scripts/pipeline.py
+++ b/business_production/组件配置/interactive-component-json/scripts/pipeline.py
@ -390,6 +390,7 @@ def process_script(wiki_url_or_token, db_path=None, dry_run=False, target_cids=N
                "result": result,
                "raw_config": comp.get("teaching_config", ""),
                "config_info": comp.get("config_info", ""),
+                "knowledge_text": comp.get("knowledge_text", ""),
            }

            if dry_run:
@ -505,6 +506,7 @@ def process_script(wiki_url_or_token, db_path=None, dry_run=False, target_cids=N
                "kpInfo": result_data.get("kpInfo"),
                "raw_config": res.get("raw_config", ""),
                "config_info": res.get("config_info", ""),
+                "knowledge_text": res.get("knowledge_text", ""),
            }
            if status == "saved":
                entry["component_id"] = res.get("component_id")
--- a/business_production/组件配置/interactive-component-json/scripts/proxy_server.py
+++ b/business_production/组件配置/interactive-component-json/scripts/proxy_server.py
@ -159,35 +159,66 @@ rate_limiter = RateLimiter(CONFIG['rate_limit_per_minute'])

 # --- Regenerate Handler ---
 def _do_regenerate(body_data):
-    """调用 generate_component 重新生成单个组件"""
+    """调用 generate_component 或 generate_kp_info 重新生成单个组件/知识点"""
    if str(SCRIPTS_DIR) not in sys.path:
        sys.path.insert(0, str(SCRIPTS_DIR))

+    mode = body_data.get("mode", "component")
+    type_name = body_data.get("type_name", "")
+    cId = body_data.get("cId", "")
+    level = body_data.get("level") or None
+
+    if not type_name:
+        return 400, {"error": "缺少 type_name"}
+
+    if mode == "kp":
+        # 仅重新生成知识点
+        from generate_json import generate_kp_info
+        from match_component import match_component_type
+        from llm_client import LLMClient
+
+        knowledge_text = body_data.get("knowledge_text", "")
+        if not knowledge_text.strip():
+            return 400, {"error": "缺少 knowledge_text"}
+
+        try:
+            type_info = match_component_type(type_name)
+            cType = type_info["cType"]
+        except ValueError as e:
+            return 400, {"error": str(e)}
+
+        try:
+            llm_client = LLMClient()
+            kp_info = generate_kp_info(cType, cId, knowledge_text, llm_client, level=level)
+            return 200, {"kpInfo": kp_info, "cType": cType}
+        except Exception as e:
+            logger.error(f"Regenerate KP failed: {traceback.format_exc()}")
+            return 500, {"error": str(e)}
+    else:
+        # 仅重新生成组件配置（不生成 kpInfo）
        from generate_json import generate_component
        from llm_client import LLMClient

        teaching_config = body_data.get("teaching_config", "")
-    type_name = body_data.get("type_name", "")
-    cId = body_data.get("cId", "")
        character_map = body_data.get("character_map") or {}
        section_char_map = body_data.get("section_char_map") or []
-    level = body_data.get("level") or None

-    if not teaching_config or not type_name:
-        return 400, {"error": "缺少 teaching_config 或 type_name"}
+        if not teaching_config:
+            return 400, {"error": "缺少 teaching_config"}

        component = {
            "type_name": type_name,
            "cId": cId,
            "teaching_config": teaching_config,
            "has_image": body_data.get("has_image", False),
-        "knowledge_text": body_data.get("knowledge_text", ""),
+            "knowledge_text": "",  # 强制为空，跳过 kpInfo 生成
            "config_info": body_data.get("config_info", ""),
        }

        try:
            llm_client = LLMClient()
            result = generate_component(component, character_map=character_map, section_char_map=section_char_map, llm_client=llm_client, level=level)
+            result["kpInfo"] = None  # 确保不返回 kpInfo
            return 200, result
        except Exception as e:
            logger.error(f"Regenerate failed: {traceback.format_exc()}")
@ -316,8 +347,8 @@ class ProxyHandler(BaseHTTPRequestHandler):
            self._send_json(err[0], err[1])
            return

-        if not isinstance(body_data, dict) or not body_data.get('teaching_config'):
-            self._send_json(400, {'error': 'body must contain teaching_config'})
+        if not isinstance(body_data, dict) or not body_data.get('type_name'):
+            self._send_json(400, {'error': 'body must contain type_name'})
            return

        logger.info(f'[{client_ip}] Regenerating: type_name={body_data.get("type_name")}, cId={body_data.get("cId")}')
--- a/memory/.dreams/events.jsonl
+++ b/memory/.dreams/events.jsonl
@ -84,3 +84,11 @@
 {"type":"memory.recall.recorded","timestamp":"2026-06-01T10:44:47.489Z","query":"王璐辰 反馈 组件配置 中互动 核心互动 问题","resultCount":2,"results":[{"path":"memory/2026-04-30.md","startLine":50,"endLine":63,"score":1},{"path":"memory/2026-05-07.md","startLine":354,"endLine":368,"score":1}]}
 {"type":"memory.recall.recorded","timestamp":"2026-06-02T01:49:17.841Z","query":"英文台词生产 L1-S2-U17-L1 摩托怪客","resultCount":6,"results":[{"path":"memory/2026-05-28.md","startLine":46,"endLine":62,"score":1},{"path":"memory/2026-05-28.md","startLine":60,"endLine":66,"score":1},{"path":"memory/2026-05-29.md","startLine":1,"endLine":14,"score":1},{"path":"memory/2026-05-28.md","startLine":35,"endLine":49,"score":1},{"path":"memory/2026-05-18.md","startLine":791,"endLine":810,"score":1},{"path":"memory/2026-05-08.md","startLine":33,"endLine":55,"score":1}]}
 {"type":"memory.recall.recorded","timestamp":"2026-06-02T07:02:08.195Z","query":"interactive-component-json 组件配置-json skill","resultCount":1,"results":[{"path":"memory/2026-05-20.md","startLine":40,"endLine":59,"score":1}]}
+{"type":"memory.recall.recorded","timestamp":"2026-06-04T03:23:22.968Z","query":"主题图片 L1 L2 theme picture","resultCount":2,"results":[{"path":"memory/2026-05-12.md","startLine":141,"endLine":155,"score":1},{"path":"memory/2026-05-12.md","startLine":132,"endLine":143,"score":1}]}
+{"type":"memory.recall.recorded","timestamp":"2026-06-04T03:23:36.292Z","query":"主题 topic theme L1 L2 图片 单元","resultCount":2,"results":[{"path":"memory/2026-05-12.md","startLine":141,"endLine":155,"score":1},{"path":"memory/2026-05-12.md","startLine":132,"endLine":143,"score":1}]}
+{"type":"memory.recall.recorded","timestamp":"2026-06-04T03:24:24.356Z","query":"主题图片 theme image 单元主题 topic picture L1 L2 课程","resultCount":2,"results":[{"path":"memory/2026-05-12.md","startLine":141,"endLine":155,"score":1},{"path":"memory/2026-05-12.md","startLine":132,"endLine":143,"score":1}]}
+{"type":"memory.recall.recorded","timestamp":"2026-06-04T03:43:35.423Z","query":"瓦拉能力体系 Vala 产品能力","resultCount":2,"results":[{"path":"memory/2026-05-18.md","startLine":702,"endLine":726,"score":1},{"path":"memory/2026-05-18.md","startLine":672,"endLine":707,"score":1}]}
+{"type":"memory.recall.recorded","timestamp":"2026-06-04T06:32:47.988Z","query":"组件生产 多维表格 bitable app_token 1112315","resultCount":2,"results":[{"path":"memory/2026-05-07.md","startLine":354,"endLine":368,"score":1},{"path":"memory/2026-04-22.md","startLine":1,"endLine":8,"score":1}]}
+{"type":"memory.recall.recorded","timestamp":"2026-06-04T06:33:43.353Z","query":"组件生产 bitable app_token 多维表格 互动组件","resultCount":2,"results":[{"path":"memory/2026-05-07.md","startLine":354,"endLine":368,"score":1},{"path":"memory/2026-04-22.md","startLine":1,"endLine":8,"score":1}]}
+{"type":"memory.recall.recorded","timestamp":"2026-06-04T06:34:06.417Z","query":"组件生产 多维表格 app_token CMHSbUUjka","resultCount":2,"results":[{"path":"memory/2026-05-07.md","startLine":354,"endLine":368,"score":1},{"path":"memory/2026-04-22.md","startLine":1,"endLine":8,"score":1}]}
+{"type":"memory.recall.recorded","timestamp":"2026-06-04T09:21:44.866Z","query":"儿童英语学习通病 常见错误 不同阶段 学习难点","resultCount":2,"results":[{"path":"memory/2026-05-11.md","startLine":78,"endLine":89,"score":1},{"path":"memory/2026-05-26.md","startLine":161,"endLine":187,"score":1}]}
--- a/memory/.dreams/short-term-recall.json
+++ b/memory/.dreams/short-term-recall.json
@ -1,6 +1,6 @@
 {
  "version": 1,
-  "updatedAt": "2026-06-02T07:02:08.195Z",
+  "updatedAt": "2026-06-04T09:21:44.866Z",
  "entries": {
    "memory:memory/2026-05-07.md:57:74": {
      "key": "memory:memory/2026-05-07.md:57:74",
@ -876,13 +876,13 @@
      "endLine": 368,
      "source": "memory",
      "snippet": "- **需求：** 将 020102（I am...）和 020103（I am ready / Thank you）两套题合并为一个 `{first:..., second:...}` JSON，统一 questionSetID=0000001 - **状态：** ✅ 已完成 - **核心考点分析（用户强调）：** 需分析每个句型的核心考点（孩子最容易犯错的地方），挖空对准核心考点 - I am/from 组：am（系动词第一人称）、from（介词选择）、student（a+名词结构） - Thank you for 组：for（介词选择，非 you）、helping（for+动名词，非 help/to help） - **输出文件：** `output/writing_pic_qa_combined.json` ### 刘彦江 — 组件配置-json 请求（L1-S2-U13-L4 沙漠之花） - **时间：** 16:45 ~ 17:51 - **文档：** `https://makee-interactive.feishu.cn/wiki/K5E1wzwk7it9t7kXvcbc6Xugnhc` - **状态：** ⚠️ 未完成 — pipeline 识别到 0 组件 - **根因：** 剧本文档的13个组件数据存储在 markdown 内联表格中（lark-table，5列×36行），而非内嵌 Sheet。当前 pipeline 的 parse_script 只从内嵌 Sheet 读取组件数据，不支持 markdown 表格组件解析 - **已识别组件（ma",
-      "recallCount": 12,
+      "recallCount": 15,
      "dailyCount": 0,
      "groundedCount": 0,
-      "totalScore": 12,
+      "totalScore": 15,
      "maxScore": 1,
      "firstRecalledAt": "2026-05-15T07:13:08.147Z",
-      "lastRecalledAt": "2026-06-01T10:44:47.489Z",
+      "lastRecalledAt": "2026-06-04T06:34:06.417Z",
      "queryHashes": [
        "08364c8746ab",
        "4f08741ab4fd",
@ -895,7 +895,10 @@
        "8742c0bf4e2b",
        "eb0902db0156",
        "229e5d3943bb",
-        "f53c9769e692"
+        "f53c9769e692",
+        "5d025e291984",
+        "472b243ec77c",
+        "fd4c8aca2cb6"
      ],
      "recallDays": [
        "2026-05-15",
@ -904,7 +907,8 @@
        "2026-05-21",
        "2026-05-25",
        "2026-05-28",
-        "2026-06-01"
+        "2026-06-01",
+        "2026-06-04"
      ],
      "conceptTags": [
        "am/from",
@ -924,13 +928,13 @@
      "endLine": 8,
      "source": "memory",
      "snippet": "[李应瑛 2026-04-22 提出要求] 所有需要包含对话的内容（如剧本、互动组件等）必须要有【后置对话】字段，无后置对话时填写“无”。 [李应瑛 2026-04-22 确认规则] 剧本内嵌表格组件填写位置规则：仅当表格第一列（A列）明确标注为对话类类型（对话朗读/对话挖空/对话选读/对话组句等）时，才在同一行的H列（【组件】列）填写对应的组件内容，其他类型行（TL/场景/角色/图片/非对话类等）无需填写。 [李应瑛 2026-04-22 确认格式规则] 对话类组件字段换行规则：每个结构单独占一行，格式为： 【任务标题】xxx 【情境引入】xxx 【互动内容】xxx 【后置对话】xxx 单元格内使用\\n作为换行符实现，后续所有组件均遵循此格式。",
-      "recallCount": 10,
+      "recallCount": 13,
      "dailyCount": 0,
      "groundedCount": 0,
-      "totalScore": 10,
+      "totalScore": 13,
      "maxScore": 1,
      "firstRecalledAt": "2026-05-15T07:13:08.147Z",
-      "lastRecalledAt": "2026-06-01T10:44:19.887Z",
+      "lastRecalledAt": "2026-06-04T06:34:06.417Z",
      "queryHashes": [
        "08364c8746ab",
        "4f08741ab4fd",
@ -941,7 +945,10 @@
        "d6a04b711fd9",
        "8742c0bf4e2b",
        "eb0902db0156",
-        "229e5d3943bb"
+        "229e5d3943bb",
+        "5d025e291984",
+        "472b243ec77c",
+        "fd4c8aca2cb6"
      ],
      "recallDays": [
        "2026-05-15",
@ -950,7 +957,8 @@
        "2026-05-21",
        "2026-05-25",
        "2026-05-28",
-        "2026-06-01"
+        "2026-06-01",
+        "2026-06-04"
      ],
      "conceptTags": [
        "对话朗读/对话挖空/对话选读/对话组句等",
@ -1396,19 +1404,21 @@
      "endLine": 726,
      "source": "memory",
      "snippet": "- 颜色题: \"What colour is it?\"，物体题: \"What's this?\" - ability: 图文匹配 - explanation: 含回答要点、图片内容、能力、评估标准、回答指导 ### 审校结果 - ✅ 全部27题 jsonData 结构完整、explanation 全中文、题目格式正确 - ✅ 题目集合ID一致、图片编号连续（00-10/00-11/00-09） - ✅ 题目配置（人类可读）格式规范 - ⚠️ 口语-P4表无「审校结果」列，无法标记审校状态 - ℹ️ 110101题组2 \"colour\"出现2次（调色盘+彩虹）— 按规格生成 - ℹ️ 110201 \"mice\"(复数) 使用 \"What are these?\" 区分于 \"mouse\"(单数) 使用 \"What's this?\" - ℹ️ 以下词汇未在L1词库中检索到（词库仅160词/声称980词，疑似文件不完整）：jacket, T-shirt, hair, monster, colour, bread, cake, candy, chocolate, ice cream, mice, mouse — 按用户指定词汇生产，未做替换 ### Bitable信息 - App Token: CMHSbUUjka3TrUsaxxEc297ongf - Table ID: tblsD2dxaRpLmkXD - 操作方式: Python subprocess + curl（避免shell转义） ## [李应瑛] L1-S2-U17-L3 紧急拜访 组件配置（2026-05-18 20:5",
-      "recallCount": 2,
+      "recallCount": 3,
      "dailyCount": 0,
      "groundedCount": 0,
-      "totalScore": 2,
+      "totalScore": 3,
      "maxScore": 1,
      "firstRecalledAt": "2026-05-19T03:08:27.099Z",
-      "lastRecalledAt": "2026-05-19T03:08:50.132Z",
+      "lastRecalledAt": "2026-06-04T03:43:35.423Z",
      "queryHashes": [
        "34bf2e5fbdee",
-        "3f9471df9e50"
+        "3f9471df9e50",
+        "19f5f93f3465"
      ],
      "recallDays": [
-        "2026-05-19"
+        "2026-05-19",
+        "2026-06-04"
      ],
      "conceptTags": [
        "00-10/00-11/00-09",
@ -1428,19 +1438,21 @@
      "endLine": 707,
      "source": "memory",
      "snippet": "- `/tmp/unit_challenge_ability_doc.md` — 对外分享版（已发送文件 + 飞书文档） --- ## L1-S2-U18-L5 双龙会 台词润色审校（[王璐辰]） ### 审校结果 - 语法/拼写硬错误：5处（Rows 41, 55, 126, 153, 178） - 表达优化建议：4处（Rows 77, 84, 63, 207） - 一致性遗漏：4行无润色 + dragon/Loong 混用不统一 - D列知识点：6个知识点，共20次 ### 交互模式收获 - 用户（王璐辰）偏好直接在 bitable 中写入统计表格 → 后续同类需求优先写入原表 ## [刘彦江] 口语-P4-看图识物 内容生产（2026-05-18 19:34） ### 生产范围（3条记录，全部双题组，全部回填验证通过 ✅） | ID | record_id | 题组1 | 题组2 | 状态 | |----|-----------|-------|-------|------| | 100001 | recvjYhcXkYXIM | 6题(颜色) | 5题(衣物) | ✅ | | 110101 | recvjYhdvUxDgs | 6题(身体) | 6题(颜色) | ✅ | | 110201 | recvjYhe4opOGm | 6题(食物) | 4题(动物) | ✅ | ### 题型结构 **口语-P4 (speaking_pic_recognize)**: - textTitle: \"Look and answer.\" - 每词1题，配1图（白底中央展",
-      "recallCount": 2,
+      "recallCount": 3,
      "dailyCount": 0,
      "groundedCount": 0,
-      "totalScore": 2,
+      "totalScore": 3,
      "maxScore": 1,
      "firstRecalledAt": "2026-05-19T03:08:27.099Z",
-      "lastRecalledAt": "2026-05-19T03:08:50.132Z",
+      "lastRecalledAt": "2026-06-04T03:43:35.423Z",
      "queryHashes": [
        "34bf2e5fbdee",
-        "3f9471df9e50"
+        "3f9471df9e50",
+        "19f5f93f3465"
      ],
      "recallDays": [
-        "2026-05-19"
+        "2026-05-19",
+        "2026-06-04"
      ],
      "conceptTags": [
        "l1-s2-u18-l5",
@ -2979,6 +2991,134 @@
        "token",
        "u1zewhzaaie07tkjmoochurgnih"
      ]
+    },
+    "memory:memory/2026-05-12.md:141:155": {
+      "key": "memory:memory/2026-05-12.md:141:155",
+      "path": "memory/2026-05-12.md",
+      "startLine": 141,
+      "endLine": 155,
+      "source": "memory",
+      "snippet": "- **最终结果：10/10 条记录 sentenceMeaningMeaning 均可 json.loads 正确解析，每条含 2 题，无残留垃圾数据** - **受影响记录：** 1213004, 1213006, 1213010, 1214008, 1215005, 1216001, 1216004, 1216007, 1216008, 1216010 - **审校结果同步更新：** 移除所有 10 条的 `❌ JSON 解析失败` 标记 - **注意：** Q1 explanation 在重建时词表述可能微调（内容含义一致），如需精确原始 wording 可从中文列恢复后人工微调 ## 经验教训 ### bitable 写入需严格流程管控（2026-05-12） - 批量更新 JSON 字段时必须：写入前完整提取原始数据 → 写入后逐字段 diff 验证 → 发现异常立即回滚 - 本次 JSON 修复中首次写入不完整导致 Q2 丢失，原因：修复脚本未正确处理多题 JSON 结构 - **教训：** 任何涉及 JSON 重建/修复的批量操作，先单条试运行验证通过后再批量执行，不可并行甩出多条 ### sentenceMeaningMeaning 的 explanation 引号规范（2026-05-12） - explanation 字段中的中文双引号极易被误用为 ASCII `\"` (U+0022) - 建议生产侧统一使用中文引号 `\"` (U+201C) 和 `\"` (U+201D)，或在 explanation 写入前做引号归一化预处理",
+      "recallCount": 3,
+      "dailyCount": 0,
+      "groundedCount": 0,
+      "totalScore": 3,
+      "maxScore": 1,
+      "firstRecalledAt": "2026-06-04T03:23:22.968Z",
+      "lastRecalledAt": "2026-06-04T03:24:24.356Z",
+      "queryHashes": [
+        "defa0daee9d7",
+        "92343053bee2",
+        "302eafc578a7"
+      ],
+      "recallDays": [
+        "2026-06-04"
+      ],
+      "conceptTags": [
+        "10/10",
+        "json.loads",
+        "重建/修复的批量操作",
+        "最终",
+        "结果",
+        "记录",
+        "sentencemeaningmeaning",
+        "均可"
+      ]
+    },
+    "memory:memory/2026-05-12.md:132:143": {
+      "key": "memory:memory/2026-05-12.md:132:143",
+      "path": "memory/2026-05-12.md",
+      "startLine": 132,
+      "endLine": 143,
+      "source": "memory",
+      "snippet": "- 检查题型：句意选择题 ×30、句子补全题、听句作答题、场景选择题、看图选择题 - 验证方法：提取 explanation 中的英文引句 vs question 中的英文句子、正则提取\"空处应填X\" vs options[answer] - 结果：✅ **0 条实质性不对应** - **10 条 sentenceMeaningMeaning JSON 修复：** - **根因：** explanation 文本中的 ASCII `\"` (U+0022) 被当作中文双引号使用，JSON 解析器将其视为字符串终止符 - **修复策略：** 从「句意选择题」中文列提取英文原句和选项 → json.dumps 重建标准 JSON（只修格式，不修内容） - **1216007 特殊处理：** 原始 JSON 损坏严重，需从中文列完全重建 - **二次修复（Q2 丢失问题）：** 首次不完全修复导致 9 条记录的 Q2 被覆盖丢失，且 Q1 被 JSON 垃圾污染 → 从中文列全面重建 Q1+Q2 - **最终结果：10/10 条记录 sentenceMeaningMeaning 均可 json.loads 正确解析，每条含 2 题，无残留垃圾数据** - **受影响记录：** 1213004, 1213006, 1213010, 1214008, 1215005, 1216001, 1216004, 1216007, 1216008, 1216010 - **审校结果同步更新：** 移除所有 10 条的 `❌ JSON 解析失败` 标记",
+      "recallCount": 3,
+      "dailyCount": 0,
+      "groundedCount": 0,
+      "totalScore": 3,
+      "maxScore": 1,
+      "firstRecalledAt": "2026-06-04T03:23:22.968Z",
+      "lastRecalledAt": "2026-06-04T03:24:24.356Z",
+      "queryHashes": [
+        "defa0daee9d7",
+        "92343053bee2",
+        "302eafc578a7"
+      ],
+      "recallDays": [
+        "2026-06-04"
+      ],
+      "conceptTags": [
+        "json.dumps",
+        "10/10",
+        "json.loads",
+        "检查",
+        "选择",
+        "句子",
+        "作答",
+        "场景"
+      ]
+    },
+    "memory:memory/2026-05-11.md:78:89": {
+      "key": "memory:memory/2026-05-11.md:78:89",
+      "path": "memory/2026-05-11.md",
+      "startLine": 78,
+      "endLine": 89,
+      "source": "memory",
+      "snippet": "- **状态：** ✅ 全部完成（两阶段延迟写入策略，避免自动化覆盖） - **脚本：** `scripts/batch_reading_matchInfo.py` | ID | 题组1 | 题组2 | record_id | |----|-------|-------|-----------| | 021301 | actor/beard/different/brave/neighbour（人物） | dinner/cooking/fire/album/diary（生活） | recvjjk7oAT1YN | | 021401 | plan/repair/newspaper/backpack/seat（日程） | garden/melon/water/plant/summer（自然） | recvjjk7TU4ZFs | | 021501 | visitor/welcome/invite/laugh/guest（社交） | mistake/instruction/reason/technology/brush（学习） | recvjjk8mju8ou | | 021601 | record/event/enter/return/sort（管理） | handbag/camera/gate/wheel/entrance（场所） | recvjjk8Q5ZMIu | | 021701 | driver/street/market/guidebook/shopping（出行） | machine/draw/basketball/bookcase/invention（活动） | r",
+      "recallCount": 1,
+      "dailyCount": 0,
+      "groundedCount": 0,
+      "totalScore": 1,
+      "maxScore": 1,
+      "firstRecalledAt": "2026-06-04T09:21:44.866Z",
+      "lastRecalledAt": "2026-06-04T09:21:44.866Z",
+      "queryHashes": [
+        "3d95a503721c"
+      ],
+      "recallDays": [
+        "2026-06-04"
+      ],
+      "conceptTags": [
+        "record-id",
+        "dinner/cooking/fire/album/diary",
+        "garden/melon/water/plant/summer",
+        "record/event/enter/return/sort",
+        "状态",
+        "全部",
+        "完成",
+        "阶段"
+      ]
+    },
+    "memory:memory/2026-05-26.md:161:187": {
+      "key": "memory:memory/2026-05-26.md:161:187",
+      "path": "memory/2026-05-26.md",
+      "startLine": 161,
+      "endLine": 187,
+      "source": "memory",
+      "snippet": "- 文档：`https://makee-interactive.feishu.cn/wiki/XoGZwgrRMiGk9vkYNWmcegs9nyh`（L1-S2-U19-L1 客厅游戏） - Bot当前无该知识空间权限，无法写入 - 已完成 5 个组件配置文本生成，待权限恢复后写入 G 列 ### 修正格式规范（用户指定） - 不含【组件ID】字段 - 【互动内容】顶部加「题目：选择正确的单词补全句子」 - 挖空用 `___`（3个下划线） - 选项后加「辅助信息」行（如 `afternoon 意为下午`） - 【后置对话】缺省写「空」，不用「无」 - 【互动反馈】正确：「正确：无」；错误：「错误 角色名: 台词」 - 任务标题和反馈需贴合剧情 - 有【教研图】的组件：资源配置需写「图片时机」 ### 5个组件配置摘要 | ID | 句子 | 正确答案 | 干扰项 | 图片时机 | |----|------|---------|--------|---------| | 1219102 | This ___ , we go to space! | afternoon | evening | 无 | | 1219105 | This ___ is the captain's seat! | armchair | sofa | 互动内容 | | 1219110 | Push the ___ , Dad! | armchair | table | 互动内容 | | 1219111 | The ___ is a glowing star! | lamp | armchair |",
+      "recallCount": 1,
+      "dailyCount": 0,
+      "groundedCount": 0,
+      "totalScore": 1,
+      "maxScore": 1,
+      "firstRecalledAt": "2026-06-04T09:21:44.866Z",
+      "lastRecalledAt": "2026-06-04T09:21:44.866Z",
+      "queryHashes": [
+        "3d95a503721c"
+      ],
+      "recallDays": [
+        "2026-06-04"
+      ],
+      "conceptTags": [
+        "l1-s2-u19-l1",
+        "https",
+        "makee",
+        "interactive.feishu.cn",
+        "wiki",
+        "xogzwgrrmigk9vkynwmcegs9nyh",
+        "u19",
+        "客厅"
+      ]
    }
  }
 }
--- a/memory/2026-06-04-1218.md
+++ b/memory/2026-06-04-1218.md
@ -0,0 +1,45 @@
+# Session: 2026-06-04 12:18:44 GMT+8
+
+- **Session Key**: agent:xiaoyan:feishu:direct:ou_8b79b7f711e42a4004f3f3c225a1a99c
+- **Session ID**: bed05a1e-fc0d-4ea2-b319-ed5f913c4a60
+- **Source**: feishu
+
+## Conversation Summary
+
+user: https://makee-interactive.feishu.cn/wiki/PgWrwllW2ikQKgkLTYKcmKFonXc组件配置-json
+assistant: 文档是「L1-S2-U18-L1 中国周」的组件配置文档，包含 17 个互动组件。我先读取相关技能和嵌入的 Sheet 数据。
+
+
+assistant: 文档已读取，标题是「L1-S2-U18-L1 中国周」，包含 17 个互动组件。消息含"组件配置-json"，需要调用 `interactive-component-json` 技能。
+
+
+assistant: 匹配到 `interactive-component-json` 技能。文档「L1-S2-U18-L1 中国周」包含 17 个互动组件，启动流水线处理。
+
+
+assistant: 处理完成。17/17 组件全部成功，HTML 报告已发送给你。
+
+**L1-S2-U18-L1 中国周 — 组件配置 JSON 生成结果：**
+
+| # | 组件ID | 类型 | cType |
+|---|--------|------|-------|
+| 1 | 1218101 | 对话朗读 | mid_dialog_repeat |
+| 2 | 1218102 | 图片单选 | mid_image_choose |
+| 3 | 1218103 | 对话朗读 | mid_dialog_repeat |
+| 4 | 1218104 | 图片单选 | mid_image_choose |
+| 5 | 1218105 | 对话挖空 | mid_dialog_fillin |
+| 6 | 1218106 | 对话朗读 | mid_dialog_repeat |
+| 7 | 1218107 | 图片多选 | mid_image_multiple |
+| 8 | 1218108 | 对话挖空 | mid_dialog_fillin |
+| 9 | 1218109 | 对话朗读 | mid_dialog_repeat |
+| 10 | 1218110 | 对话朗读 | mid_dialog_repeat |
+| 11 | 1218111 | 对话选择 | mid_dialog_choose |
+| 12 | 1218112 | 对话挖空 | mid_dialog_fillin |
+| 13 | 1218113 | 对话组句 | mid_dialog_sentence |
+| 14 | 1218114 | 对话朗读 | mid_dialog_repeat |
+| 15 | 1218115 | 对话组句 | mid_dialog_sentence |
+| 16 | 1218116 | 听力拖拽 | core_listening_drag |
+| 17 | 1218117 | 对话选读 | mid_dialog_select |
+
+⚠️ 1 个注意项：组件 1218115（对话组句）的知识点 `proud of` 在 L1 词库中未匹配到 kpId，需人工确认。
+
+HTML 报告内嵌了推送按钮，你可以在浏览器中打开后直接推送配置到远程数据库。